Adaptive DNA amplification of synthetic gene circuit opens a way to overcome cancer chemoresistance

Significance Cancers still resist treatment too often, creating a need for new insights into cancer drug resistance and new treatment options. Experimental models that are broad, yet simple could aid with the identification of mechanisms and development of countermeasures of drug resistance. Here, by studying engineered mammalian cells that previously evolved resistance to a drug, we found DNA amplification as the most common cause of drug resistance. A nucleotide treatment combined with the original drug efficiently suppressed the growth of hamster and cancer cells with DNA amplification, suggesting broadly applicable therapies that might combat chemoresistance in cancer.


Figures S1 to S7
First, we altered each parameter in the model one at a time, ranging from 1/20 th to 20x the original amount, to see the effect that this would have on the dose response.From this, we could find which parameters had the most drastic effects on the dose response to focus on for more finetuned adjustments.
From these initial findings, we further investigated these changes using a custom MATLAB GUI.
Using this, we could see in real time how parameter changes impacted the dose response, and how close it was to the experimental dose response.We found that changing the value of f (the rate of dox entering the cell) to 2.5x its original value matched the dose response closely.We used this because it only involved changing one parameter for an accurate correspondence.All other parameters were kept the same as in the original model.
We also adjusted some of the reactions to make it more accurate to the current CHO system.Namely, we removed the fusion of eGFP and hTetR, and made them separate proteins, representing the use of a P2A sequence as opposed to a fusion protein.It was assumed that the translation and degradation rates of eGFP would be the same as hTetR.Additionally, the leakage reaction was changed to come from the bound promoters instead of a separate entity.All other reactions were kept the same as in the original model.

Detailed Description of the Current Model
The schematic of the current model is shown in Fig S3 .Protein expression is driven by the promoter, where two conditions need to be met for robust transcription.First, the polymerase machinery Aup needs to be bound to the promoter.It has a binding rate α and an unbinding rate a.Second, the promoter needs to be unoccupied (A) by dimerized hTetR (D).If both of these conditions are met, there is robust transcription at a rate m (thick arrow).hTetR can bind to the promoter once (R01) or twice (R02), at a rate r, and can unbind at a rate ρ.These repressed promoters can undergo transcriptional leakage at a rate λ.
Dox can enter the cell as a zeroth order reaction at rate fc, and exits at rate f, where c is an effective inducer concentration constant multiplier.Once in the cell, Dox can bind to hTetR once (H) or twice (B) to hTetR at a rate b irreversibly.In this dox-bound state, hTetR cannot bind to the promoters, effectively leading to more time in the A state, and consequently more robust transcription.
Once mRNA is produced from transcription, it can be translated into hTetR and eGFP (G) protein at a rate p, or degraded at a rate μ.Consequently, these proteins can be degraded at a rate δ.
To simulate Evo3 through Evo6 (DNA Amplification), the initial values of A and Aup were changed, ranging from 1 to 15.To simulate Evo2 (TetR mutation), the rate of hTetR binding r was set to zero.

Mathematical Derivations
To complement the simulations, we also considered the differential equations to justify transcriptional leakage as the major contributor of protein expression at steady state.Let Y denote the DNA copy number.
Therefore, the probability of the promoter being in the A state is  .These are equal when  = √5−1 2 , or when the hTetR dimer level approaches []~62.Thus, when the hTetR concentration exceeds this value, more promoters are in the bound state, and thus undergo transcriptional leakage.This effect is intensified at higher DNA copy numbers, due to the increase of hTetR protein.Similarly, the steady state of eGFP is  µ , which is also linear, and reflected in Fig. 5D in the main text.S2 Lists of primer sequences and TaqMan probes used for copy number detection and gene expression analysis.

Source
Sequence / Identifier
[A] tends towards zero, [R01] + [R02] tends towards Y.Thus, the production rate is approximately .Consequently, the steady state value becomes  µ .Therefore, the steady state of mRNA is linearly proportional to the DNA copy number at high hTetR levels, which is reflected in Fig 5D in the main text.
Lists of Triplet-forming Oligo sequences used for cell treatment.Mismatched thymine (T) is labeled in red and non-interacting thymine (T) is labeled in blue.