Recent evidence suggests that a polyaneuploid cancer cell (PACC) state may play a key role in the adaptation of cancer cells to stressful environments and in promoting therapeutic resistance. The PACC state allows cancer cells to pause cell division and to avoid DNA damage and programmed cell death. Transition to the PACC state may also lead to an increase in the cancer cell’s ability to generate heritable variation (evolvability). One way this can occur is through evolutionary triage. Under this framework, cells gradually gain resistance by scaling hills on a fitness landscape through a process of mutation and selection. Another way this can happen is through self-genetic modification whereby cells in the PACC state find a viable solution to the stressor and then undergo depolyploidization, passing it on to their heritably resistant progeny. Here, we develop a stochastic model to simulate both of these evolutionary frameworks. We examine the impact of treatment dosage and extent of self-genetic modification on eco-evolutionary dynamics of cancer cells with aneuploid and PACC states. We find that under low doses of therapy, evolutionary triage performs better whereas under high doses of therapy, self-genetic modification is favored. This study generates predictions for teasing apart these biological hypotheses, examines the implications of each in the context of cancer, and provides a modeling framework to compare Mendelian and non-traditional forms of inheritance.
Despite major advances in cancer research, advanced cancers largely remain incurable due to the evolution of resistance, killing more than 10 million people per year1,2. Traditionally, it is assumed that cancers randomly mutate in the presence of therapy3. Some of these stochastic mutations confer a fitness advantage to cells by making them more resistant to the therapy. These cells then go on to proliferate at higher rates in the population. In this way, the cancer becomes increasingly resistant to therapy, eventually rendering the therapy ineffective4. However, recent experimental evidence indicates another key factor in the emergence of resistance: the polyaneuploid cancer cell (PACC) state1,5,6,7. Cancer cells in the aneuploid 2N+ state can enter the PACC state by bypassing mitosis, cyotkinesis, or both and resulting in a whole genome doubling5,7,8. The non-proliferative PACC cell state is characterized by two distinct features: 1) an exit of the cell cycle after S phase to pause division and allow cancer cells to withstand stressful conditions9 and 2) an increase in genomic material due to polyploidization which is assumed to contribute to a greater level of heritable variation (evolvability)10,11,12,13,14 that can be passed on to their progeny when they undergo depolyploidization into the 2N+ state15,16,17,18. In this way, the PACC state provides a mechanism for facultative evolvability in the cancer cell population in which cells can increase their evolvability (and thus, their rate of adaptation) during times of stress and decrease it during times of stability.
The ecological and evolutionary dynamics underlying these processes remain an enigma. Under classical notions of Mendelian inheritance and evolutionary triage (ET)19, organisms scale adaptive landscapes in a process of creative destruction. Due to the struggle for existence, this inherently requires the proliferation of more fit 2N+ cells and the death of less fit 2N+ cells in the population (in the context of therapy, this leads to the death of more sensitive cells and the proliferation of more resistant ones)20. However, observations of cancer cell population dynamics with 2N+ and PACC states under therapy cannot be reconciled solely with this theory. Under extreme stress, we notice a drastic transition of cells from the 2N+ state to the non-proliferative PACC state5,21,22,23,24,25,26. After a delay, cells in the PACC state begin depolyploidizing into 2N+ progeny. Critically, these progeny immediately display high levels of resistance to the stressor. Unlike as would be expected under ET, we do not observe the death of mutant clones. In other words, this evolutionary process appears to “solve the therapeutic problem” without mortality of less successful variants relative to more successful ones27,28.
One way this can occur is through the process of self (epi-)genetic modification, hereafter referred to as self-genetic modification (SGM)29,30,31. SGM, also described as genetic assimilation, was originally described by Waddington as an interplay between traits acquired during an organism’s lifetime and the genetic encoding of these traits32,33. In Lamarckian fashion, reverse transcription allows changes that a cell undergoes during its own lifetime to be written back into its DNA and passed onto its offspring34,35,36. This process was first identified in the context of the immune system and was proposed as an explanation for why homologous DNA sequences from VDJ gene regions of parent mice are found in their germ cells and were passed on to offspring for subsequent generations36,37,38. The concept of reverse transcriptional inheritance is closely related to the idea of adaptive or directed mutations, a process in which mutations are produced to relieve the selective pressure39 in that death of less fit variants is not required for evolution to proceed.
In the context of cancer, under stress, cells in the PACC state may be able to assess their lack of viability, devise a viable solution, and code this back into their DNA30,31,40. These adaptations are heritable and are thus directly passed on to their 2N+ progeny upon depolyploidization. Evolution in this sense does not require death of 2N+ cells: rather than climbing hills on their adaptive landscape due to mutation and selection, cells under SGM “jump” up the landscape, displaying short bursts of rapid evolutionary change. Such processes are not unique to PACCs; similar phenomena have also been observed in yeast41, Drosophila32,42,43,44, invasive herbs45,46, butterflies47, tiger snakes48, Anolis lizards49,50, grasshoppers51,52, yarrow53,54,55,56, and bacteria39,57,58, in which a variety of stressors such as nutrient depletion, change in temperature, DNA damage, and exposure to antibiotics can induce mutagenesis via global response systems such as activation of DNA polymerases, downregulation of error-correcting enzymes, and movement of mobile genetic elements57. Though these mechanisms increase heritable mutation, evidence also supports the selective capture of mutations directed towards the stressors at hand59,60,61,62,63,64,65,66,67,68.
Despite several theoretical studies on mechanisms of inheritance similar to reverse transcription69,70,71,72,73,74,75,76,77,78, robust methods do not exist to broadly model coupled ecological and evolutionary dynamics under such inheritance and to compare these results to those under classical Mendelian inheritance. Furthermore, none of these studies investigate reverse transcription in particular and of those that consider Lamarckian inheritance more generally, none do so in the context of cancer. Although we specifically focus on creating stochastic mathematical models of Mendelian and reverse transcriptional inheritance in the context of cancer cell populations with 2N+ and PACC states under therapy, our approach is broadly applicable and can serve as a general modeling framework within which mechanisms of reverse transcriptional and Mendelian inheritance can be explored in a variety of contexts. To construct this model, we build on a model we developed in prior work and outline a birth–death switching process algorithm to simulate the ET and SGM hypotheses stochastically. We examine how the eco-evolutionary dynamics of the cancer cell populations are affected by drug dosage and extent of SGM. Finally, we run in silico competition experiments to help us understand whether ET or SGM performs better under soft (low drug dosage) and hard (high drug dosage) selection regimes.
ODE model of 2N+ and PACC ecology
In order to stochastically simulate the eco-evolutionary dynamics of cancer cell populations with 2N+ and PACC states, we first construct an underlying ODE model, from which we can derive birth, death, and switching rates. To do this, we first create a model of the population dynamics and state transitions for the 2N+ and PACC states in Eq. (1).
Here, \(x_1\) and \(x_2\) represent the population of cells in the 2N+ and PACC states respectively and v represents a resistance strategy such as the number of drug antiporters, receptor levels, gene expression levels, allocation to different metabolic pathways, or upregulation of DNA repair processes. We make several assumptions: cells in the PACC do not die from therapy, competition, or natural causes (i.e., no background death); our resistance strategy is a continuous quantitative trait; cells within the 2N+ and PACC states are identical and posess the same resistance strategy (though it is not necessary while a cell is in the PACC state); cells in the PACC state are all 4N+ in ploidy (actual measures of PACCs find this and even higher levels of ploidy); and the anueploid cancer cell population is pre-existing.
The 2N+ dynamics emerge from the following. We assume logistic growth, in which cells in the 2N+ state are inhibited equally by cells in the 2N+ and PACC states, until they reach their carrying capacity. We allow for an obligate (constant) transition of cells to the PACC state, an assumption supported by experimental literature that observes a baseline level of PACCs in most cancer cell populations5. We let drug induced mortality take on a Michaelis–Menten form, in which mortality is a decreasing function of resistance. We also include a facultative or condition-dependent transition to the PACC state. The higher the death in 2N+ cells due to the drug, the greater their transition rate to the PACC state. These facultative transitions are observed experimentally, with an increase in PACC frequency and number upon administration of therapy22. Finally, we imagine a constant transition rate of cells from the PACC state into the 2N+ state. Note that since we assume cells in the PACC state are 4N+ in ploidy, a depolyplodization event produces two 2N+ cells.
The PACC cell dynamics emerge from the following. Since cells in the PACC state cannot divide without depolyploidizing into the 2N+ state, cells in the PACC state have no inherent growth rate. We also assume that cells in the PACC state are fully resistant to therapy. PACC dynamics are solely influenced by obligate and facultative transitions from the 2N+ state and a constant transition back to the 2N+ state. For the former, we also include a successful polyaneuploid transition probability, \(\eta \), to account for the possibility of failed polyploidization that leads to cell death, as is sometimes observed experimentally.
Table 1 provides parameter values and their interpretations. All values were chosen to be biologically realistic, numerically convenient, and to clearly show differences between simulation results. Many of these parameter values can qualitatively alter our results, however. Anytime the intrinsic growth rate of the 2N+ cells, r, is higher than the maximum potential kill rate of the drug, \(m/\lambda \), the cancer will survive therapy even in the absence of evolving resistance. As expected, a higher carrying capacities lead to higher equilibrium population sizes. We present the results of sensitivity analyses of the key parameters in the supplemental material. With this model, we can now outline a method for stochastically stimulating the ET and SGM hypotheses.
Stochastic simulation: birth–death-switching process
To simulate the eco-evolutionary dynamics of this state-structured cancer cell population, we use a birth–death-switching process. We provide an intuitive explanation of this process that parallels Hammerstein’s streetcar theory of evolution with several temporary stops determined by the underlying genetic architecture and a “final stop” determined by phenotypic selection that promotes ecological stability79. The implementation of this process is closely related to Dieckmann’s directed random walks in adaptive dynamics80,81,82 in which evolution proceeds as a sequence of trait substitutions, as mutant traits with positive invasion fitness replace the resident trait in the population83. The simulation implements the stochastic element at the population level, tracking changes in trait values and population sizes at different cell state compartments.
Under ET, when the cancer cell population is under stress, cells transition from the 2N+ to the PACC state. This state provides a refuge for cells under therapy and, like Goldschmidt’s hopeful monster84, produces profound mutant progeny “in hopes” that they will remain viable in the face of therapy. This high mutational breadth is the result of the increased genomic content that accompanies the polyploidization program of cells as they transition into the PACC state. If the mutant progeny has a higher fitness than the residents in the population, it will eventually sweep through the population and replace the resident trait. If the mutant has a lower fitness than the resident, it will be driven to extinction. In this process of creative destruction, the cancer cells will climb an adaptive landscape via the mortality of less successful variants relative to more successful ones, gradually gaining resistance in a step-wise fashion. Whether or not the cells can gain resistance fast enough before succumbing to the therapy determines whether they will remain extant or go extinct.
In SGM, as in the case of ET, an environmental stressor promotes the transition of cells from the 2N+ state into the PACC state. However, what happens next depends on how extreme this stressor is or, more accurately, what the fitness and internal condition of the cell is. If cells are near or within their fundamental niche (either above or slightly below “sea level”)1,85, evolutionary triage will occur, just as described above. However, if cells are “drowning”, far from their fundamental niche, the SGM program will commence. Namely, cells in the PACC state will assess lack of viability, remain in this state until they find a sufficient solution, and then resume proliferation and evolution via ET. There are two main components to this: generating and assessing a solution. Cells may generate solutions via a “single-shot” approach: a mutation is produced and, if it’s sufficient, the mutation is kept, a mutant progeny with this trait is produced, and proliferation resumes. If it’s not sufficient, the mutant trait is discarded and the process repeats. However, we assume that cells generate solutions via a gradualistic approach that parallels ET. In this process, a mutant trait is produced. If the fitness of cells with this trait would be higher than that with the current trait, the mutant trait will be adopted and evolution will build upon this. If it is not better than the resident trait, it will be discarded. It is critical to note that in this process, 2N+ progeny will not be produced by cells in the PACC state until a sufficient solution is reached. Although the mechanism by which the assessment of solutions is unknown, we propose a conceptual framework to help us crudely understand how this happens. When cells in the PACC state produce a mutant trait under SGM, a mutant progeny is not produced. Rather, via an internal process of trial-and-error, a “hypothetical progeny” is produced and tested to examine how viable it would be in the harsh external environment. If it would be more viable than a progeny with the current trait, the PACC will adopt this mutant trait as its new baseline and continue this process. Theoretically, an SGM threshold controls just how viable this progeny needs to actually be produced by the PACC. When this threshold is met, cells in the PACC state can undergo depolyploidization into the 2N+ state with this trait and can continue proliferating and evolving via ET. An outline of how the ET and SGM hypotheses were simulated can be seen in Fig. 1. To more clearly grasp when the phases of evolutionary triage and self-genetic modification occur in the SGM hypothesis, consider Fig. 2.
In addition to considering how the fitness of a cancer cell population influences the mode of evolution, it is critical to understand how this fitness is achieved. Namely, what conditions can cause an organism to be pushed outside its fundamental niche? We separate these factors into two groups: density-dependent and density-independent factors. Density-dependent effects refer to any process in which the fitness of an organism is regulated by the density of the population, e.g., via competition for resources. Density-independent factors on the other hand, impact an individual’s fitness regardless of population density, e.g., therapy (viewed simplistically).
To distinguish between these two forms of fitness regulation, we introduce the concept of soft and hard selection regimes. Under soft selection, ecological disturbances may greatly reduce environmental quality, thereby reducing the carrying capacity and supporting fewer cells. However, there is still a positive ecological equilibrium under the disturbance. As the population size decreases, the fitness of extant cells increases dramatically and some cells are able to persist in the population. Although evolution may help increase population size in this case, density dependent factors play a much stronger role at regulating the fitness of the cells. Under hard selection, ecological catastrophes do not permit a positive ecological equilibrium. Under this scenario, evolution is critical since the current strategy cannot maintain a population. Although a reduction in population size may increase the fitness of the remaining cells, this is not sufficient to prevent extinction. In this paper, we assume that SGM comes into play when cells are under hard selection regimes and when quickly evolving a viable strategy is essential for cells to persist. As we will soon see, we enforce this by using the effect of therapy as a marker for when SGM is implemented.
Using this framework, we examined how drug dosage and extent of self-genetic modification affect the population and resistance strategy dynamics of cancer cell populations under ET and SGM. To supplement these investigations, we ran in silico competition experiments by competing cancer populations under ET with populations under SGM under soft (low drug dosage) and hard (high drug dosage) selection regimes. In all our simulations, we ran 100 trials for robustness. We hypothesized that higher doses of drug will favor SGM as it allows cancer cells to persist in a stable PACC state until they find a viable solution to the stressor. Conversely, we expected ET to perform better under low doses of therapy due to density-dependent competition factors that become more critical, as the importance shifts away from simply surviving the stressor to also competing with other cancer cells. We anticipated that the higher the degree of self-genetic modification, the longer the cancer cell population will remain in the stable PACC state.
Cancer eco-evolutionary dynamics
First, we simulated basic dynamics of a cancer cell population with 2N+ and PACC states using the technique outlined above under evolutionary triage and self-genetic modification. Specifically, we analyzed population and strategy dynamics under the cases of no therapy, continuous therapy, and intermittent therapy. Each line in all the following stochastic simulations corresponds to the results of one simulation. For continuous therapy, the drug was administered from time 200 to 800. Under intermittent therapy, the drug was administered for 100 time steps, followed by a drug holiday for 100 time steps, repeated over the duration of the simulation. In both cases, we used a drug dosage of m = 0.8 and a self-genetic modification threshold of \(\zeta =0.4\). For all simulations in this subsection, the cancer cell population was initialized with 10 cells in the 2N+ state, 0 cells in the PACC state, and with a resistance strategy of 0. The results from these simulations can be seen in Fig. 3.
In each of the following simulations, the top panel captures the population dynamics of cancer cells in the 2N+ (red curves) and PACC (blue curves) states. The bottom panel captures the evolutionary dynamics of resistance (black lines). The regions shaded in gray depict the intervals during which therapy is administered; unshaded regions are those during which no therapy is given.
First, consider the control case of no therapy (Fig. 3a). As we expect, there is no evolution in the drug resistance trait (as no selection pressure has been induced) and the population quickly equilibrates to predominantly cells in the 2N+ state. No population goes extinct and the eco-evolutionary dynamics under ET and SGM are virtually identical. Next, let’s consider the case of continuous therapy (Fig. 3b). Under ET, 27% of populations went extinct, whereas none went extinct under SGM. In both cases, there is a drastic shift of cells from the 2N+ state to the PACC state upon induction of therapy. As cells gain resistance, the need for great heritable variation diminishes. This leads to a gradual decrease of cells in the PACC state, accompanied by a corresponding increase of cells in the 2N+ state. Under SGM, as soon as therapy is administered, cells enter a stable PACC state that lasts about 100 time steps. This gives the cancer cells time to achieve adequate levels of resistance. After this, cells in the PACC state begin to undergo depolyploidization and evolutionary triage commences. This trend is paralleled in the evolutionary dynamics of resistance, where we see a period of evolutionary stasis until around time 300, at which point we see rapid evolutionary change followed by more gradualistic evolution until therapy is removed. Similar trends can be seen in the intermittent therapy simulations (Fig. 3c), in which 13% of ET populations and 0% of SGM populations went extinct. During periods of treatment, the frequency of cells in the PACC state increases and the frequency of cells in the 2N+ state decreases. However, as resistance emerges, a lower frequency of cells in the PACC state is noticed during times of treatment, in accordance with the reduced need for great heritable variation. In the SGM case, we note that the stable PACC state lasts for the first treatment cycle and ET begins soon after the second cycle begins.
Next, we examined how different drug doses impact the eco-evolutionary dynamics of cancer cell populations under ET and SGM. We expected that higher doses of drugs will result in a higher extinction rate and a longer stable PACC state. To test our intuitions, we simulated population and resistance strategy dynamics under low dose (m = 0.6) and high dose (m = 1) therapy that is continuously administered from time step 200 to 800. The results of these simulations can be seen in Fig. 4.
First, let’s consider the low dose simulation results (Fig. 4a). In this case, none of the ET or SGM populations go extinct. The similarity of the eco-evolutionary dynamics under the ET and SGM is due to the shorter stable PACC state. Since a low dose of therapy was given, death due to drug is lower and it takes less time for the cancer cell population to evolve an adequate resistance strategy. Now, let’s shift our focus to the high dose simulation results (Fig. 4b). In this case, 85% of ET populations went extinct, whereas all SGM populations avoided extinction. Note that it is the immediate transition to the stable PACC state that allows cancer cells under SGM to remain extant in the population while evolving resistance. When exposed to extreme environmental conditions such as a high dose of therapy, this stable PACC state exists for longer, giving the cancer cell population enough time to evolve resistance and effect evolutionary rescue86. On the other hand, ET populations cannot access this mechanism. As a result, cells in the 2N+ state often go extinct, followed shortly after by the eradication of cells in the PACC state.
In the SGM simulations thus far, we assumed that cells in the PACC state undergo depolyploidization into the 2N+ state only when drug kill is less than 0.4. To explore how the extent of self-genetic modification impacts eco-evolutionary dynamics of cancer cell populations under SGM, we compared extreme (\(\zeta =0.2\)), moderate (\(\zeta =0.4\)), and minor (\(\zeta =0.6\)) SGM. Note that when \(\zeta =1\), the SGM hypothesis is analogous to ET and when \(\zeta =0\), the SGM hypothesis functionally serves as a “hibernation state” that cancer cells enter every time they are exposed to therapy. To clearly show the differences among these levels of self-genetic modification, we used a high dose therapy (m = 1) that is administered continuously from time step 200 to 800. The results of these simulations can be seen in Fig. 5.
As expected, none of the populations go extinct in any of the simulations, regardless of the degree of SGM. We note that the more extreme the degree of SGM, the longer the cells remain in their stable PACC state. This is clearly paralleled in the evolutionary resistance strategy dynamics, in which we see a longer period of evolutionary stasis and a greater burst of evolutionary change for more extreme SGM mechanisms.
In silico competition experiments
From the simulations thus far, it seems that SGM is favored over ET in nearly all cases. However, since we simulated the ET and SGM populations independently, the density-dependent factors between these populations were inherently ignored. This competition becomes particularly important under soft selection regimes in which external selection pressures are not significant. ET allows cells to continue proliferating at all times, unlike SGM that halts growth of the population during the stable PACC phase. Thus, we hypothesized that when cells are under low stress environments, ET will be favored due to its density-dependent competitive advantage. Conversely, when cells are under high stress environments and must undergo evolutionary rescue to avoid extinction, SGM will be favored since cells under ET will not be able to evolve fast enough to persist in harsh conditions without access to the PACC state. When cells are in an environment with no or very little stress, the SGM and ET hypotheses are identical and thus, both will do equally well. Here, we competed cancer cell populations under ET and SGM against each other under low (m = 0.6) and high (m = 1) dose continuous therapy. To clearly note the difference between ET and SGM, we used a self-genetic modification threshold of \(\zeta =0.2\). To avoid initial stochastic effects that can cause extinction of either population before therapy is administered, we gave therapy from time 0 to 600 and ran the simulation for 800 time steps. We initialized the cancer cell populations with 100 cells in the 2N+ state, 0 cells in the PACC state, and resistance strategies of 0 for both the ET and SGM populations. The results of these simulations can be seen in Fig. 6.
In these plots, the red and blue population curves represent cells in the 2N+ and PACC states under SGM. The yellow and purple population curves represent cells in the 2N+ and PACC states under ET. The black and magenta curves in the bottom panel capture the evolutionary resistance dynamics of SGM and ET populations, respectively. As we hypothesized, under the low dose treatment regimen (Fig. 6a), 88% of the SGM populations went extinct as compared to only 8% of the ET populations. Conversely, under high dose therapy (Fig. 6b), 98% of the ET populations went extinct, whereas none of the SGM populations went extinct. To more finely explore the interplay between drug dosage and ET/SGM evolutionary mechanisms, we run simulations competing these populations against each other for drug doses between 0 and 1 (at intervals of 0.1). A summary of the results of these simulations can be seen in Fig. 7.
In this plot, we depicted the number of extinction events (out of 100) for ET and SGM populations competed against each other for a range of drug doses. The magenta curve captures ET extinctions and the black curve captures SGM extinctions. The region shaded in gray is the “no selection” regime, in which the ET and SGM mechanisms are identical. The aqua shaded portion is the “low selection” regime, where the ET mechanism is favored. The yellow shaded part is the “high selection” regime, in which SGM performs better. As per our hypothesis, evolutionary triage seems to achieve an advantage through density-dependent competition under the low selection regime. However, in the high selection regime, the importance of density-dependent factors is overshadowed by the necessity to simply survive in the face of extreme stressors. The stable PACC state afforded by self-genetic modification gives it this advantage.
Therapeutic resistance is a major cause of treatment failure in patients with advanced cancers. We created a stochastic model to examine two potential mechanisms by which a key contributor to therapeutic resistance, the PACC state, allows cancer cells to persist and adapt to extreme environmental stressors: evolutionary triage and self-genetic modification. In doing so, we provided a general framework that allows us to model mechanisms of reverse transcriptional inheritance and compare them to traditional Mendelian inheritance. We used this framework to simulate the eco-evolutionary dynamics of cancer cell populations under therapy. We investigated how these dynamics change due to drug dosage and extent of self-genetic modification. Finally, we supplemented these studies with in silico competition experiments where we competed cancer cell populations under evolutionary triage and under self-genetic modification against each other under low and high dose therapeutic regimens. Most importantly, we found that self-genetic modification is favored under extreme stressors, in which simply surviving is more important than density-dependent competition. Conversely, evolutionary triage performs better under more modest stressors, as the importance of density-dependent competition comes to the fore.
These findings have broad implications for both cancer biology and evolutionary biology. Our results provide qualitative predictions for how the PACC state contributes to therapeutic resistance through evolutionary triage and self-genetic modification. These results can help guide future experimental work to empirically examine how SGM and ET work together to contribute to resistance. Our work also provides a generalizable framework within which hypotheses and mechanisms of Lamarckian and Mendelian inheritance can be simulated and compared. Lamarckian inheritance is often erroneously conceptualized as antithetical to classical evolution by natural selection. However, in our work, we show how it can be naturally placed in a Darwinian framework. Though we focus on cancer here, we expect that these findings are broadly applicable to any system capable of evolving.
The datasets generated during and/or analysed during the current study are available in the Self-Genetic Modification repository in Anuraag Bukkuri’s GitHub, at https://github.com/abukkuri/Self-Genetic-Modification/.
Pienta, K. J., Hammarlund, E. U., Austin, R. H., Axelrod, R., Brown, J. S. & Amend, S. R. Cancer cells employ an evolutionarily conserved polyploidization program to resist therapy. In Seminars in Cancer Biology, 1–15 (2020).
Siegel, R. L., Miller, K. D. & Jemal, A. Cancer statistics, 2020. CA A Cancer J. Clin. 70(1), 7–30 (2020).
Duesberg, P. & Rasnick, D. Aneuploidy, the somatic mutation that makes cancer a species of its own. Cell Motil. Cytoskelet. 47(2), 81–107 (2000).
Hanahan, D. & Weinberg, R. A. Leading edge review hallmarks of cancer: The next generation. Cell 144, 646–674 (2011).
Amend, S. R. et al. Polyploid giant cancer cells: Unrecognized actuators of tumorigenesis, metastasis, and resistance. Prostate 79(13), 1489–1497 (2019).
Pienta, K. J. et al. Convergent evolution, evolving evolvability, and the origins of lethal cancer. Mol. Cancer Res. 18(6), 801–810 (2020).
Pienta, K. J., Hammarlund, E. U., Axelrod, R., Brown, J. S. & Amend, S. R. Poly-aneuploid cancer cells promote evolvability, generating lethal cancer. Evol. Appl. 13(7), 1626–1634 (2020).
Roychowdhury, S. et al. Personalized oncology through integrative high-throughput sequencing: A pilot study. Sci. Transl. Med. 3(111), 1–12 (2011).
Kuczler, M. D., Olseen, A. M., Pienta, K. J. & Amend, S. R. ROS-induced cell cycle arrest as a mechanism of resistance in polyaneuploid cancer cells (PACCs). Prog. Biophys. Mol. Biol. 165, 3–7 (2021).
Brown, R. L. What evolvability really is. Br. J. Philos. Sci.65(3), 549–572 (2014).
Crother, B. I. & Murray, C. M. Early usage and meaning of evolvability. Ecol. Evol. 9(7), 3784–3793 (2019).
Payne, J. L. & Wagner, A. The causes of evolvability and their evolution. Nat. Rev. Genet. 20, 24–38 (2019).
Pigliucci, M. Is evolvability evolvable?. Genetics 9, 75–82 (2008).
Sniegowski, P. D. & Murphy, H. A. Evolvability. Curr. Biol. 16, R831–R834 (2006).
Kostecka, L. G., Pienta, K. J. & Amend, S. R. Polyaneuploid cancer cell dormancy: Lessons from evolutionary phyla. Front. Ecol. Evol. 9, 439 (2021).
Rajaraman, R., Rajaraman, M. M., Rajaraman, S. R. & Guernsey, D. L. Neosis–a paradigm of self-renewal in cancer. Cell Biol. Int. 29(12), 1084–1097 (2005).
Rajaraman, R., Guernsey, D. L., Rajaraman, M. M. & Rajaraman, S. R. Neosis–A parasexual somatic reduction division in cancer. Int. J. Hum. Genet. 7(1), 29–48 (2007).
Sundaram, M., Guernsey, D. L., Rajaraman, M. M. & Rajaraman, R. Neosis: A novel type of cell division in cancer. Cancer Biol. Ther. 3(2), 207–218 (2004).
Gatenby, R. A., Cunningham, J. J. & Brown, J. S. Evolutionary triage governs fitness in driver and passenger mutations and suggests targeting never mutations. Nat. Commun. 5(1), 1–9 (2014).
Bukkuri, A. & Brown, J. S. Evolutionary game theory: Darwinian dynamics and the G function approach. MDPI Games 12(4), 1–19 (2021).
Lopez-Sánchez, L. M. et al. CoCl2, a mimic of hypoxia, induces formation of polyploid giant cells with stem characteristics in colon cancer. PLoS ONE 9(6), e99143 (2014).
Mittal, K. et al. Multinucleated polyploidy drives resistance to Docetaxel chemotherapy in prostate cancer. Br. J. Cancer 116(9), 1186–1194 (2017).
Niu, N., Mercado-Uribe, I. & Liu, J. Dedifferentiation into blastomere-like cancer stem cells via formation of polyploid giant cancer cells. Oncogene 36(34), 4887–4900 (2017).
Ogden, A., Rida, P. C. G., Knudsen, B. S., Kucuk, O. & Aneja, R. Docetaxel-induced polyploidization may underlie chemoresistance and disease relapse. Cancer Lett. 367, 89–92 (2015).
Puig, P. E. et al. Tumor cells can escape DNA-damaging cisplatin through DNA endoreduplication and reversible polyploidy. Cell Biol. Int. 32(9), 1031–1043 (2008).
Zhang, S. et al. Generation of cancer stem-like cells through the formation of polyploid giant cancer cells. Oncogene 33(1), 116–128 (2014).
Lin, K. C. et al. The role of heterogeneous environment and docetaxel gradient in the emergence of polyploid, mesenchymal and resistant prostate cancer cells. Clin. Exp. Metastasis 36(2), 97–108 (2019).
Lin, K.-C. et al. Epithelial and mesenchymal prostate cancer cell population dynamics on a complex drug landscape. Converg. Sci. Phys. Oncol. 3(4), 045001 (2017).
Boe, L. Mechanism for induction of adaptive mutations in Escherichia coli. Mol. Microbiol. 4(4), 597–601 (1990).
Cairns, J. Mutation and cancer: The antecedents to our studies of adaptive mutation. Genetics 148(4), 1433–1440 (1998).
Hall, B. G. Adaptive mutagenesis: A process that generates almost exclusively beneficial mutations. Genetica 102, 109 (1998).
Waddington, C. H. Genetic assimilation of an acquired character. Evolution 7(2), 118–126 (1953).
Waddington, C. H. Genetic assimilation. Adv. Genet. 10, 257–293 (1961).
Jablonka, E. V. A. & Raz, G. A. L. Transgenerational epigenetic inheritance: Prevalence, mechanisms, and implications for the study of heredity and evolution. Q. Rev. Biol. 84(2), 131–176 (2009).
Steele, E. J. & Pollard, J. W. Hypothesis: Somatic hypermutation by gene conversion via the error prone DNA\(\longrightarrow \)RNA\(\longrightarrow \)DNA information loop. Mol. Immunol. 24(6), 667–673 (1987).
Steele, E. J. Somatic hypermutation in immunity and cancer: Critical analysis of strand-biased and codon-context mutation signatures. DNA Repair 45, 1–24 (2016).
Steele, E. J. Somatic Selection and Adaptive Evolution (Springer, US, 1979).
Steele, E. J., Lindley, R. A. & Blanden, R. V. Lamarck’s Signature (Perseus Books, 1998).
Foster, P. L. Adaptive mutation: Implications for evolution. Bioessays 22, 1067–1074 (2000).
McCutcheon, J. P. & Moran, N. A. Extreme genome reduction in symbiotic bacteria. Nat. Rev. Microbiol. 10(1), 13–26 (2012).
Badyaev, A. V. Stress-induced variation in evolution: From behavioural plasticity to genetic assimilation. Proc. R. Soc. B Biol. Sci. 272, 877–886 (2005).
Bateman, K. G. The genetic assimilation of four venation phenocopies. J. Genet. 56(3), 443–474 (1959).
Milkman, R. D. The genetic basis of natural variation. VI. Selection of a crossveinless strain of Drosophila by phenocopying at high temperature. Genetics 51(1), 87 (1965).
Waddington, C. H. Genetic assimilation of the bithorax phenotype. Evolution 10(1), 1–13 (1956).
Godoy, O., Saldaña, A., Fuentes, N., Valladares, F. & Gianoli, E. Forests are not immune to plant invasions: Phenotypic plasticity and local adaptation allow Prunella vulgaris to colonize a temperate evergreen rainforest. Biol. Invasions 13(7), 1615–1625 (2011).
Schlichting, C. D. & Wund, M. A. Phenotypic plasticity and epigenetic marking: An assessment of evidence for genetic accommodation. Evolution 68(3), 656–672 (2014).
Otaki, J. M., Hiyama, A., Iwata, M. & Kudo, T. Phenotypic plasticity in the range-margin population of the lycaenid butterfly Zizeeria maha. BMC Evol. Biol. 10(1), 1–13 (2010).
Aubret, F. & Shine, R. Genetic assimilation and the postcolonization erosion of phenotypic plasticity in island tiger snakes. Curr. Biol. 19(22), 1932–1936 (2009).
Losos, J. B., Irschick, D. J. & Schoener, T. W. Adaptation and constraint in the evolution of specialization of Bahamian Anolis lizards. Evolution 48(6), 1786–1798 (1994).
Losos, J. B. et al. Evolutionary implications of phenotypic plasticity in the hindlimb of the lizard Anolis sagrei. Evolution 54(1), 301–305 (2000).
Sword, G. A. Density-dependent warning coloration. Nature 397(6716), 217 (1999).
Sword, G. A. A role for phenotypic plasticity in the evolution of aposematism. Proc. R. Soc. B Biol. Sci. 269(1501), 1639–1644 (2002).
Clausen, J. & Hiesey, W. M. The balance between coherence and variation in evolution. Proc. Natl. Acad. Sci. 46(4), 494–506 (1960).
Gurevitch, J. Variation in leaf dissection and leaf energy budgets among populations of Achillea from an altitudinal gradient. Am. J. Bot. 75(9), 1298–1306 (1988).
Gurevitch, J. & Schuepp, P. H. Boundary layer properties of highly dissected leaves: An investigation using an electrochemical fluid tunnel. Plant Cell Environ. 13(8), 783–792 (1990).
Gurevitch, J. Sources of variation in leaf shape among two populations of Achillea lanulosa. Genetics 130(2), 385–394 (1992).
Foster, P. L. Stress-induced mutagenesis in bacteria. Crit. Rev. Biochem. Mol. Biol. 42(5), 373–397 (2007).
Soppa, J. Polyploidy in archaea and bacteria: About desiccation resistance, giant cell size, long-term survival, enforcement by a eukaryotic host and additional aspects. Microb. Physiol. 24, 409–419 (2014).
Bastide, A. & David, A. The ribosome, (slow) beating heart of cancer (stem) cell. Oncogenesis 7(4), 1–13 (2018).
Cairns, J., Overbaugh, J. & Miller, S. The origin of mutants. Nature 335, 142–145 (1988).
Foster, P. L. Adaptive mutation: The uses of adversity. Annu. Rev. Microbiol. 47, 467–504. https://doi.org/10.1146/annurev.mi.47.100193.002343 (2003).
Lenski, R. E. & Mittler, J. E. The directed mutation controversy and neo-Darwinism. Science 259(5092), 188–194 (1993).
Lenski, R. E. & Sniegowski, P. D. “Adaptive mutation’’: The debate goes on. Science 269, 285–288 (1995).
Noller, H. F., Hoffarth, V. & Zimniak, L. Unusual resistance of peptidyl transferase to protein extraction procedures. Science 256(5062), 1416–1419 (1992).
Pribis, J. P. et al. Gamblers: An antibiotic-induced evolvable cell subpopulation differentiated by reactive-oxygen-induced general stress response. Mol. Cell 74(4), 785–800 (2019).
Silvera, D., Formenti, S. C. & Schneider, R. J. Translational control in cancer. Nat. Rev. Cancer 10(4), 254–266 (2010).
Shcherbakov, D. et al. Ribosomal mistranslation leads to silencing of the unfolded protein response and increased mitochondrial biogenesis. Commun. Biol. 2(1), 1–16 (2019).
Truitt, M. L. & Ruggero, D. New frontiers in translational control of the cancer genome. Nat. Rev. Cancer 16(5), 288–304 (2016).
Alphey, L. S., Crisanti, A., Randazzo, F. & Akbari, O. S. Opinion: Standardizing the definition of gene drive. Proc. Natl. Acad. Sci. USA 117(49), 30864 (2020).
Champer, J., Buchman, A. & Akbari, O. S. Cheating evolution: Engineering gene drives to manipulate the fate of wild populations. Nat. Rev. Genet. 17, 146–159 (2016).
Champer, S. E. et al. Modeling CRISPR gene drives for suppression of invasive rodents using a supervised machine learning framework. PLOS Comput. Biol. 17(12), e1009660 (2021).
Deredec, A., Burt, A. & Godfray, H. C. J. The population genetics of using homing endonuclease genes in vector and pest management. Genetics 179(4), 2013–2026 (2008).
Heffel, M. G. & Finnigan, G. C. Mathematical modeling of self-contained CRISPR gene drive reversal systems. Sci. Rep. 9(1), 1–10 (2019).
Leftwich, P. T. et al. Recent advances in threshold-dependent gene drives for mosquitoes. Biochem. Soc. Trans. 46, 1203–1212 (2018).
Nijhout, H. F., Kudla, A. M. & Hazelwood, C. C. Genetic assimilation and accommodation: Models and mechanisms. Curr. Top. Dev. Biol. 141, 337–369 (2021).
Noble, C., Adlam, B., Church, G. M., Esvelt, K. M. & Nowak, M. A. Current CRISPR gene drive systems are likely to be highly invasive in wild populations. eLife 7, e33423 (2018).
Novozhilov, A. S., Karev, G. P. & Koonin, E. V. Mathematical modeling of evolution of horizontally transferred genes. Mol. Biol. Evol. 22(8), 1721–1732 (2005).
Pigliucci, M. & Murren, C. J. Perspective: Genetic assimilation and a possible evolutionary paradox: Can macroevolution sometimes be so fast as to pass us by?. Evolution 57, 1455–1464 (2003).
Hammerstein, P. Darwinian adaptation, population genetics and the streetcar theory of evolution. J. Math. Biol. 34(5–6), 511–532 (1996).
Dieckmann, U. Coevolutionary Dynamics of Stochastic Replicator Systems (Central Library of the Research Center Jülich, 1994).
Dieckmann, U., Marrow, P. & Law, R. Evolutionary cycling in predator-prey interactions: population dynamics and the red queen. J. Theor. Biol. 176(1), 91–102 (1995).
Dieckmann, U. & Law, R. The dynamical theory of coevolution: a derivation from stochastic ecological processes. J. Math. Biol. 34, 579–612 (1996).
Metz, J. A. J., Nisbet, R. M. & Geritz, S. A. H. How should we define ‘fitness’ for general ecological scenarios?. Trends Ecol. Evol. 7(6), 198–202 (1992).
Goldschmidt, R. Some aspects of evolution. Science 78(2033), 539–547 (1933).
Vincent, T. L., Cohen, Y. & Brown, J. S. Evolution via strategy dynamics. Theor. Popul. Biol. 44(2), 149–176 (1993).
Bell, G. Evolutionary rescue. Annu. Rev. Ecol. Evol. Syst. 48, 605–627 (2017).
AB acknowledges support by the Stiftelsen Längmanska kulturfonden (BA22-0753), the Royal Swedish Academy of Sciences Stiftelsen GS Magnusons fond (MG2022-0019), the Crafoord foundation (20220633), and the National Science Foundation Graduate Research Fellowship Program under Grant No. 1746051. RAH was supported by the US National Science Foundation PHY-1659940, the National Science Foundation through the Center for the Physics of Biological Function (PHY-1734030), and the Princeton Catalysis Initiative. JSB acknowledges funding from NIH/NCI 1U54CA193489-01A1, Cancer as a Complex Adaptive System, NIH/NCI U54 Supplement, The tumor-host evolutionary arms race, and NIH/NCI 1R01CA258089, Eco-evolutionary drivers of clonal dynamics during UV-induced skin carcinogenesis. EUH received funding from the ParadOX-ERC Starting Grant (No. 96948), the Crafoord foundation (20220633), and from the Swedish Research Council (Grant 2019-05254). SRA was funded by the US Department of Defense CDMRP/PCRP (W81XWH-20-10353), the Patrick C. Walsh Prostate Cancer Research Fund and the Prostate Cancer Foundation. KJP acknowledges funding from NCI grants U54CA143803, CA163124, CA093900, and CA143055, and the Prostate Cancer Foundation. This work was also supported by the William and Carolyn Stutt Research Fund, Ronald Rose, MC Dean, Inc., William and Marjorie Springer, Mary and Dave Stevens, Louis Dorfman, the Jones Family Foundation, Timothy Hanson, and the David and June Trone Family Foundation.
KJP is a consultant for CUE Biopharma, Inc., and holds equity interest in CUE Biopharma, Inc., Keystone Biopharma, Inc. and PEEL Therapeutics, Inc. SRA holds equity interest in Keystone Biopharma, Inc. AB, RHA, EUH, and JSB declare no potential conflict of interest.
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
About this article
Cite this article
Bukkuri, A., Pienta, K.J., Austin, R.H. et al. Stochastic models of Mendelian and reverse transcriptional inheritance in state-structured cancer populations. Sci Rep 12, 13079 (2022). https://doi.org/10.1038/s41598-022-17456-w