Exemplary Discussion Draft 4

Hypothesis

The results demonstrated that a bistable switch was present for varying values of signals A⁺ and B^- in the concentrations of the transcription factors OCT4, SOX2, NANOG, and OCT4-SOX2, and in the concentrations of the differentiation and self-renewal target gene products in both the coherent and incoherent models. Further, this bistability was demonstrated to be persistent over varying parameter values for binding strengths of OCT4-SOX2 and OCT4-SOX2-NANOG, kinetic constants of OCT4-SOX2, and degradation rates. Additionally, it was shown that by increasing the binding and repression strengths of OCT4-SOX2 on other transcription factors or increasing the basal transcription rate of NANOG would cause the bistable switch to become irreversible, making it easier for one to control if the differentiation or self-renewal genes are expressed. These directly support the hypothesis.

Limitations

The model simplifies the various outside signals the system could receive into the positive A⁺ signal and the negative B^- signal, which may be an oversimplification of the signals, and signals represented by A⁺ and B^-, such as BMP4 and WNT, have been shown to effect each other, which was not addressed in this model. The model assumes that the signals and transcription factors do not directly interact, which could change results. Additionally, the model fails to determine if the system is a coherent or incoherent network and also relies on nondimentionalized variables, so it is difficult to quantitatively apply this to experiments. Further, this model ignores stochasticity in transcription, which is known to occur.

While this model included the three main transcription factors and the complexes they form, many recent papers have explored the effects that other transcription factors have on this system. Two of the most significant include Klf4, which activates OCT4 and SOX2, PRDM14, which also activates OCT4 and SOX2, and FoxD3, which activates NANOG. In addition, at least thirteen other transcription factors have shown to be part of this network, as well as miRNA.

Regardless, this model still demonstrates biologically significant results that match experimental data.

Discrepancies

In Figure 6B and S5B, the maximum transcription of the differentiation target genes is too high when there is either no autoregulation in Figure 6B or the binding strength between Oct4-SOX2 and the target genes is high. However, the differences are not too large and results are qualitatively the same. In Figure S1, bistability disappears for a comparatively low value of a₂, but the figures are qualitatively the same and it is not feasible to experimentally change the value of a₂. Additionally, bistability is present for a large range of values of e₁ in the created figure while the figure from the paper does not appear to have bistability for e₁.. Through the paper and recreated figures, it has been shown that bistability is present for e₁ equal to 0.0001 through 0.01, and additional figures created in Mathematica that have not been shown demonstrate that the bifurcation diagram created in Mathematica is correct.

Outside Literature

Other work in this field has also shown that the transcription factors create a network that creates a bistable switch between self-renewal and differentiation. The overexpression of OCT4, SOX2, and NANOG, as well as KLf4, a transcription factor not included in this model, was shown to induce pluripotency in cells that had differentiated. The network modeled in this paper was also confirmed to include a large portion of the transcription factor network, though recent papers have discovered the importance of additional proteins. One of these papers have built off of the model presented here, including additional transcription factors, which still demonstrates the bistability found in this model.

Additionally, while the part of the hypothesis suggesting ways to control the pluripotency of stem cells using outside methods was explored through the model, no further research was done on how the findings could be applied to actual cells. This is likely due to the difficulties implementing strategies such as changing binding efficiencies or basal transcription rates. However, one extension of the model showed that an irreversible switch was present for high levels of the decay rate of the additional transcription factor REST, similar to findings from this model. Further, experiments have shown that by forcing the expression of NANOG, expression of differentiation target genes is reduced, which is parallel to increasing the basal transcription rate of NANOG.

Future Work

The immediately following goals are to create time plots of the concentrations of the transcription factors and target gene concentrations for varying combinations of levels of signals A⁺ and B^-, which are valuable when used with the bifurcation diagrams to illustrate how those diagrams relate to the actual progression. Following the time plots based on the original equations, additional time plots involving stochasticity for the transcription of NANOG will be created to demonstrate that the system still follows the bifurcation diagrams recreated in this paper. Additionally, steady-state plots for the concentration of target gene products due to varying concentration of signal A⁺ for several values of e₁ will be created to confirm the discrepancy between Figure S1 in the paper and in Mathematica, as well as a figure similar to Figure S4 for values of η₁ and η₅ between 0 and 0.01, which is expected to show the change between a bistable and irreversible switch.

Other future goals may include determing if the system is coherent or incoherent and extending the model to involve more of the associated transcription factors discovered by other researchers.