RNA World with Inhibitors Authors: Jaroslaw Synak, Agnieszka Rybarczyk, Marta Kasprzak, Jacek Blazewicz Published: 2024 ABSTRACT During the evolution of the RNA World, compartments, which were fragments of space surrounded by a primitive lipid membrane, had to have emerged. These led eventually to the formation of modern cellular membranes. Inside these compartments, another process had to take place—switching from RNA to DNA as a primary storage of genetic information. The latter part needed a handful of enzymes for the DNA to be able to perform its function. A natural question arises, i.e., how the concentration of all vital molecules could have been kept in check without modern cellular mechanisms. The authors propose a theory on how it could have worked during early stages, using only short RNA molecules, which could have emerged spontaneously. The hypothesis was analysed mathematically and tested against different scenarios by using computer simulations. 1. Introduction Originating in the 1960s, the hypothesis of the “RNA World” emphasises the pivotal role of RNA in early life evolution [1,2,3]. It suggests that RNA was essential both to encoding genetic information and to driving enzymatic reactions, marking a significant era in which RNA served uniquely as both a genetic messenger and a catalyst, possessing the capability to replicate itself [4,5,6]. Ongoing research on ribozymes, aptamers, and various cellular RNAs involved in a wide range of biological activities further highlights the versatile role of RNA in the molecular framework of life [7,8,9,10]. This exploration has led to focused efforts to develop and refine RNA polymerase ribozymes, with the aim of assessing the potential for RNA self-replication. Despite the absence of known biological examples of RNA enzymes with RNA-dependent RNA polymerase activity (RNA replicase), substantial advances have been achieved in laboratories through directed evolution methods [10,11,12,13,14]. In particular, the most advanced examples of laboratory-created polymerase ribozymes used in the RNA-catalysed synthesis of complex RNA molecules are evolved descendants of the class I ligase [10,12,14]. Unfortunately, the synthesis fidelity of RNA ribozymes that have been developed experimentally so far, particularly those that involve more intricate RNAs, has been persistently low [15]. Nonetheless, recent efforts suggest that an RNA enzyme with polymerase activity holds potential for replicating and evolving functional RNA strands. While the replication process may still rely on a larger replicase, which limits its applicability, ongoing improvements in RNA polymerase ribozyme fidelity offer promising opportunities to support the evolution of fit RNA-cleaving enzyme variants [16]. Advances like these are encouraging, yet the discovery of an ancient ribozyme with full replication capability remains elusive, and significant challenges lie ahead to demonstrate that RNA alone could sustain an evolving genetic system. Another key challenge in understanding the origins of life is the lack of error correction mechanisms, which led to frequent mutations in prebiotic replicators. These mutations resulted in a population of quasi-species, consisting of RNA replicases and mutants (parasites defined as RNA molecules lacking polymerase function but capable of being replicated by RNA replicase). Due to high mutation rates, replicase length is limited by replication accuracy. If mutations exceed a threshold, a catastrophe occurs, causing a loss of information. Maynard Smith estimated that the polynucleotide chain size, without enzymes, cannot exceed 100 nucleotides [17], a problem known as the Eigen paradox: RNA must be long enough to function, but being such, it will be quickly overtaken by parasites [18,19,20,21]. Eigen proposed hypercycles as a solution to this problem, but Maynard Smith argued they remain vulnerable to parasites in homogeneous environments [22,23,24]. Additionally, recent advances indicate that RNA did not exist in isolation but coexisted with small peptides and lipids, supporting the compartmentalisation of RNA sequences within prebiotic vesicles (protocells) [25]. These vesicles, formed by a series of self-aggregation processes, could have offered numerous benefits in a world dominated by RNA, particularly supporting complex functions like RNA-catalysed replication reactions [26,27,28]. Compartmentalisation within these vesicles could have enabled the positive feedback essential for evolution by keeping beneficial products close to their catalysts and maintaining higher proximity among closely related molecules compared with those that are more distantly related [29,30,31]. However, membrane-based protocells also face significant challenges, particularly as vesicle boundaries limit access to environmental chemicals, such as charged nucleotides, especially in the absence of developed transport systems. Additionally, the nonenzymatic copying of RNA templates often requires high concentrations of divalent cations for optimal activity, yet these ions also accelerate RNA degradation and destabilise fatty acid vesicles. Chemical compatibility, therefore, becomes a critical concern and may necessitate added components, increasing system complexity [27,28,32]. This is why a wide variety of protocell model systems have been proposed, spanning from entirely inorganic structures to macromolecular aggregates and polymolecular assemblies of amphiphiles [27,33,34,35]. These models include vesicular compartments formed from potentially prebiotic components such as lipidic amphiphiles [27,33], peptides [36], inorganic nanoparticles [37], and polysaccharides [38]. The necessity of compartmentalisation is further underscored by its role in protecting against system crashes due to invasion by parasites [39]. In an unbounded solution, parasitic RNAs, if they serve as superior templates, will proliferate at the cost of functional replicase RNAs, leading to population collapse [40,41]. In a population of small compartments, the rise of a parasitic sequence in one compartment will not affect the overall population, even if it signifies disaster for the descendants of that compartment. Empirical evidence suggests that even temporary compartmentalisation is sufficient to prevent parasite-triggered extinction [27,42,43,44]. To date, many modelling methods have been proposed to study prebiotic systems, with theoretical modelling and computer simulation emerging as the most effective approaches to addressing issues related to evolutionary dynamics [45]. Among these, mathematical models, such as the minimal replicator network, which includes one replicase and one parasite species (shortly referred to as the RP model), have attracted significant attention [46]. At the same time, various computational strategies have been used, including probabilistic models [47], discrete models such as cellular automata [46,48], multi-agent-based approaches (MAS) [40,41,49], artificial chemistry [50], and protocell simulation using molecular dynamics models [47]. Prebiotic reactions have also been studied in detail with algorithms based on quantum dynamics [51]. This paper focuses on two particular modelling strategies, multi-agent systems (MASs) and ordinary differential equations (ODEs), to explore how concentrations of vital molecules might have been regulated in early stages without modern cellular mechanisms, using short spontaneously emerged RNA molecules. These short RNAs can bind to larger ones via Watson–Crick pairing, blocking replication when attached, effectively acting as “inhibitors”. On the contrary, “anti-inhibitors” can bind and neutralise these inhibitors, ensuring continued molecular interactions. All the reactions described are fully reversible. The proposed model aligns with the findings of previous studies, which have highlighted the importance of the tag mechanism in enhancing the expansion and distribution of polymerase ribozymes in both naked and cellular environments [31,45], which remains relevant even in more complex systems that include multiple ribozymes [31]. The “tag” in this context refers to a short specific sequence, often located at the 5′ end of RNA. Additionally, the model presented here parallels the concept explored in earlier experiments with a hammerhead ribozyme [52]. The key distinction, however, is that in the previous study, the control mechanisms relied solely on the concentration of specific molecules. In contrast, the current model introduces the possibility of regulation through inhibitor–anti-inhibitor interactions, offering a more sophisticated layer of control. Moreover, we carried out comprehensive simulation tests for both models. In the end, we compared them to discern the universal laws and norms that control their dynamics and evolution. These simulations aligned with those of previous research in which we demonstrated that tackling the description and analysis of the RP system from various points of view (microscopic and macroscopic) produces congruent outcomes [49]. 2. Materials and Methods 2.1. Biological Model The RNA World hypothesis posits that RNA molecules could have gained the ability to catalyse replication, which allowed stable populations of RNA molecules to arise [4]. Eventually, these had to be enclosed in membranes, creating protocells, which, after aeons, of evolution turned into modern cells [27]. According to the RNA World hypothesis, RNA molecules were the first enzymes, so inside a protocell, they could perform different functions, resulting in several groups of specialised molecules. Modern cells require the concentration of all their enzymes to be tightly controlled, so it is not beyond the realm of possibility that millions of years ago, some mechanisms maintaining the proper concentration of each type of RNA were necessary. While unrestricted replication could have been beneficial during the early stages of the RNA World, the moment multiple different enzymes emerged and began to interact, the uncontrolled proliferation of any one of them could have starved the rest and destroyed the metabolism. Thus, protocells with said mechanisms in place could have had a huge advantage, since they were more resilient, as any (random or caused by the external environment) changes to RNA concentrations inside them could have been corrected. The first such mechanisms had to be simple, so that they could arise by chance. The one proposed in this article and further described below fulfils this condition due to its relative simplicity. Furthermore, it does not rely on any new molecule, as it is based purely on RNA. It should be mentioned that a very similar idea to the one described below has been proposed several years ago and tested with the use of a hammerhead ribozyme, yielding promising results. The main difference is that in the aforementioned work the activity of control mechanisms depended solely on the concentration of proper molecules, while in the model presented in this article, it can happen also due to inhibitor–anti-inhibitor pairing [52]. The model proposed in this publication is based on the following assumptions:One of the steps of RNA World emergence was the transition β→γ, when ribonucleotides started to form longer chains [40]. Apart from functional molecules, which, during the next stage (γ→δ), self-organised into more complicated systems, a lot of shorter, non-functional RNAs should have also emerged. Their creation could be explained by primitive, imperfect polymerisation, which can easily stop ahead of time, when anything goes wrong.Short RNA molecules can bind to larger ones based on Watson–Crick pairing.An RNA molecule with a shorter RNA attached cannot be replicated, since there is no access to its sequence. If the molecule is also an enzyme, then it cannot fold into its proper structure and is rendered non-functional. Molecules with the ability to block longer RNAs like this are called “inhibitors” in this model. While oligonucleotides can also act as replication intermediates, only their inhibiting activity is investigated in this article [52].Inhibitors can bind with their complementary molecules, which prevents them from blocking anything; the latter group is called “anti-inhibitors”.The reactions described above are fully reversible. They can go in both directions. One of the steps of RNA World emergence was the transition β→γ, when ribonucleotides started to form longer chains [40]. Apart from functional molecules, which, during the next stage (γ→δ), self-organised into more complicated systems, a lot of shorter, non-functional RNAs should have also emerged. Their creation could be explained by primitive, imperfect polymerisation, which can easily stop ahead of time, when anything goes wrong. One of the steps of RNA World emergence was the transition β→γ, when ribonucleotides started to form longer chains [40]. Apart from functional molecules, which, during the next stage (γ→δ), self-organised into more complicated systems, a lot of shorter, non-functional RNAs should have also emerged. Their creation could be explained by primitive, imperfect polymerisation, which can easily stop ahead of time, when anything goes wrong. Short RNA molecules can bind to larger ones based on Watson–Crick pairing. Short RNA molecules can bind to larger ones based on Watson–Crick pairing. An RNA molecule with a shorter RNA attached cannot be replicated, since there is no access to its sequence. If the molecule is also an enzyme, then it cannot fold into its proper structure and is rendered non-functional. Molecules with the ability to block longer RNAs like this are called “inhibitors” in this model. While oligonucleotides can also act as replication intermediates, only their inhibiting activity is investigated in this article [52]. An RNA molecule with a shorter RNA attached cannot be replicated, since there is no access to its sequence. If the molecule is also an enzyme, then it cannot fold into its proper structure and is rendered non-functional. Molecules with the ability to block longer RNAs like this are called “inhibitors” in this model. While oligonucleotides can also act as replication intermediates, only their inhibiting activity is investigated in this article [52]. Inhibitors can bind with their complementary molecules, which prevents them from blocking anything; the latter group is called “anti-inhibitors”. Inhibitors can bind with their complementary molecules, which prevents them from blocking anything; the latter group is called “anti-inhibitors”. The reactions described above are fully reversible. They can go in both directions. The reactions described above are fully reversible. They can go in both directions. This approach can be corroborated to some extent by the following facts:Advanced mechanisms controlling cell metabolism had to evolve from something much more primitive, a system which could have emerged by chance and function in early lifeforms [52].In the present, there are sequence fragments called riboswitches, which sometimes occur in the mRNA of modern cells. Riboswitches can regulate the function of mRNA by binding to various molecules. Their presence and the way they function may suggest that they are remnants of much older mechanisms existing back in the RNA World [53]. Advanced mechanisms controlling cell metabolism had to evolve from something much more primitive, a system which could have emerged by chance and function in early lifeforms [52]. Advanced mechanisms controlling cell metabolism had to evolve from something much more primitive, a system which could have emerged by chance and function in early lifeforms [52]. In the present, there are sequence fragments called riboswitches, which sometimes occur in the mRNA of modern cells. Riboswitches can regulate the function of mRNA by binding to various molecules. Their presence and the way they function may suggest that they are remnants of much older mechanisms existing back in the RNA World [53]. In the present, there are sequence fragments called riboswitches, which sometimes occur in the mRNA of modern cells. Riboswitches can regulate the function of mRNA by binding to various molecules. Their presence and the way they function may suggest that they are remnants of much older mechanisms existing back in the RNA World [53]. The general schema of the model is presented in Figure 1. 2.2. Mathematical Model To better understand the properties of such a system, a simpler case is considered first with anti-inhibitors excluded. Then, anti-inhibitors are introduced to observe what difference they can make. The main tools used to describe the model are differential equations, which are extensively analysed in the following subsections. 2.2.1. Agents and Inhibitors In this scenario, there are only two types of molecules: agents (active molecules denoted by A) and inhibitors (X). They can form inactive complexes in a reversible reaction: A+X⇌k2k1AX All reactions are assumed to be in accordance with the law of mass action. The (positive) constants k1 and k2 in the reaction above are its rate constants. By modelling the molar concentrations [A], [X], and [AX] as functions of time, one can instantly infer three differential equations describing their evolution:(1)d[A]dt=k2[AX]−k1[A][X] (2)d[X]dt=d[A]dt (3)d[AX]dt=−d[A]dt If the right side of the Equation (1) is positive, the concentrations [A] and [X] will be increasing, while [AX] will decrease, so the value of k2[AX]−k1[A][X] will also drop. If the right side of Equation (1) is negative, k2[AX]−k1[A][X] will be increasing, meaning that this value always tends to 0. In other words, the system tends towards an equilibrium, which can be described by the equation below:(4)k2[AX]−k1[A][X]=0(5)k2k1[AX]=[A][X] By denoting the concentrations in the equilibrium by eA, eX, and eAX and introducing the equilibrium constant K=k2k1, one obtains the following equation:(6)KeAX=eAeX It is convenient to introduce two additional positive constants—the total concentrations of A and X in the solution:(7)cA=[A]+[AX]=eA+eAX (8)cX=[X]+[AX]=eX+eAX For the purpose of this analysis, the equations above can be written as (9)eAX=cA−eA (10)eX=cX−cA+eA By using the relations above and denoting eA by x, we can rewrite Equation (6):(11)K(cA−x)=x(cX−cA+x)(12)x2+(cX−cA+K)x−KcA=0 In order to check for potential solutions, it is convenient to define the left side of the equation above as the value of a real function f(y) for y=x:(13)f(y)=y2+(cX−cA+K)y−KcA Let us calculate two chosen points:(14)f(0)=−KcA<0(15)f(cA)=cXcA>0 The function changes its sign between 0 and cA, meaning there is an odd number of roots in this range. As a second-degree polynomial, it can have two roots at most, so there is always exactly one root in the aforementioned range. The coefficient of y2 is positive, so f(y) is always positive for sufficiently big arguments and there is no root y≥cA. The roots of f(y) can be calculated with the quadratic formula (16)Δ=(cX−cA+K)2+4KcA>0 because the biggest of the roots is the one being searched for:(17)x=−(cX−cA+K)+Δ2 By treating x as a function of constants cA, cX, and K, we can define the capacity of a solution:(18)β=1/∂x∂cA This value can be a measure of how well the system can counter any changes in concentrations in order to keep them within acceptable levels (how sensitive x is to any changes in cA). The higher the capacity, the better it performs. One of the ways to analyse the formula above is to use Equation (17), but the calculations quickly get complicated and make it hard to see the most important properties of β. One can approach the problem from another angle and start by differentiating Equation (12) with respect to cA:(19)2x∂x∂cA+(cX−cA+K)∂x∂cA−x−K=0(20)β=1/∂x∂cA=2x+cX−cA+Kx+K We multiply the numerator and the denominator by x:(21)β=2x2+cXx−cAx+Kxx2+Kx Equation (12) can be written as (22)cXx=−x2+(cA−K)x+KcA By substituting this into Equation (21), we obtain (23)β=x2+KcAx2+Kx For fixed K and cA, all values of x between 0 and cA are possible, so one only has to choose a correct value of cX. This can be proven with Equation (22):(24)cXx=−x2+(cA−K)x+KcA=(cA−x)(x+K)(25)cX=(cA−x)(x+K)x>0 The parameter cX calculated like this is always positive, so it fulfils the conditions. Returning to Equation (23), we know that x0 (which has already been proven), the function is positive everywhere except the equilibrium point (0,0), so the latter is a local minimum. The gradient of V at every point can be easily calculated:(53)∂V∂x=2k3y0x (54)∂V∂y=2k1x0y The total derivative of V with respect to time can be written as (55)dVdt=∂V∂xx˙+∂V∂yy˙ We then substitute the partial derivatives, x˙ and y˙:(56)dVdt=−2k2k3y0x2−2k1k4x0y2−2k1k3x0y0(x+y)2−2k1k3(y0x2+x0y2)[X] All the components are nonpositive and the value of the total derivative is negative everywhere except the equilibrium point. In other words, the value of V always decreases with time until it reaches its local minimum—the point of equilibrium (0,0). There is also another important conclusion—if the point of equilibrium exists, it is unique. Determining the Equilibrium Point In the previous subsection, the existence of an equilibrium point was merely assumed; now, it will be actually proven. In order not to introduce several new symbols unnecessarily, the concentrations [A], [X], [Y], [AX], and [AY] will not be treated as functions of time anymore, but rather as concentrations in the state of equilibrium. As a first step, Equation (39) with the substitution of (41) has the form (57)K(cA−[A])=[A][X] For convenience, one can denote [A]=x:(58)[X]=KcA−xx By substituting Formulas (41) and (42) in Equation (37), (59)[Y]=[X]+cA−x−cX+cY By using Formulas (42) and (59), we can write Equation (40) as (60)k(cX+x−[X]−cA)=[X]([X]+cA−x−cX+cY) After multiplying both sides by x2, (61)kx(cXx+x2−[X]x−cAx)=[X]x([X]x+cAx−x2−cXx+cYx) By substituting Formula (58) and rearranging, we obtain (62)(k−K)x3+[(k−K)(cX+K−cA)+K(cA+cY)]x2+KcA(2K−k−cA+cX−cY)x−K2cA2=0 Because the polynomial above is quite long, it is convenient to denote its coefficients with letters:(63)a=k−K (64)b=(k−K)(cX+K−cA)+K(cA+cY) (65)c=KcA(2K−k−cA+cX−cY) (66)d=−K2cA2 We denote the polynomial by L(x); naturally, the equilibrium state corresponds to a root within the range (0,cA), so it is necessary to check for its existence. (67)L(0)=−K2cA2<0 (68)L(cA)=kcXcA2>0 Since L(x) is continuous, according to the intermediate value theorem, there should be at least one root within the range (0,cA). In the previous subsection, it has been proven that there can be at most one equilibrium state, so the function L(x) always has exactly one root within this range. One thing has been assumed without a proof, that is, every root within the range (0,cA) corresponds to a correct equilibrium state. It can be proven as follows: The question of whether a root corresponds to a proper equilibrium is the question of whether it does not violate the constraint discussed in the previous subsections (no concentration can be negative). The polynomial L(x) has been derived from Equations (36)–(40); in fact, it is just Equation (40) after some substitutions, and it allows us to calculate [A]. The remaining formulas can be rewritten to obtain [AX], [X], [Y], and [XY] (Equations (41), (42), (58), and (59)). Now, the signs of all concentrations can be analysed. [A] is positive and lower than cA, since it has to lie within the range (0,cA), which means that [X] and [AX] also have to be positive. The concentrations [Y] and [XY] can be both negative or both positive in order to fulfil Equation (40). However, Equation (38) allows only for the second option. System Capacity Just like in the simpler version of the model, the system capacity can be defined. The parameter x=[A] (the concentration of A in the state of equilibrium) can be treated as a function of five constants: K, k, cA, cX, and cY. This approach is justified because their values uniquely determine the value of x. The capacity can be defined with the same formula as previously:(69)β=1/∂x∂cA The capacity represents how a small change of cA influences x, in other words, how the system reacts if some portion of A is added or removed. Since writing the exact formula for x would be very tedious, it is easier to calculate β by differentiating Equation (62) with respect to cA:(70)(3ax2+2bx+c)∂x∂cA−ex2−fx−g=0 where (71)e=−∂b∂cA=k−2K (72)f=−∂c∂cA=−K(2K−k−2cA+cX−cY) (73)g=−∂d∂cA=2K2cA So, β can be written as (74)β=3ax2+2bx+cex2+fx+g By multiplying the numerator and the denominator by cAx, (75)β=cAx3ax3+2bx2+cxcAex2+cAfx+cAg The variable x is a solution of the equation L(x)=0, so L(x) can be safely added to the denominator and subtracted from the numerator:(76)β=cAx2ax3+bx2−dax3+(b+cAe)x2+(c+cAf)x+(d+cAg) It is easy to show that (77)b+cAe=(k−K)(cX+K)+KcY (78)c+cAf=KcA2 (79)d+cAg=K2cA2 So, β can be rewritten as (80)β=cAx2ax3+bx2+K2cA2ax3+(b+cAe)x2+(c+cAf)x+K2cA2 Now, it can be proven that the second factor is always positive and less than 1. First, one can define a polynomial being the denominator subtracted from the numerator:(81)G(x)=ax3−cAex2−KcA2x=(k−K)x3+cA(2K−k)x2−KcA2x It can be factored as (82)G(x)=−x(cA−x)[kx+K(cA−x)] The value of x lies within the range (0,cA), so G(x)<0. It means that in Equation (80), the numerator is always smaller than the denominator. Moreover, the numerator in Equation (74) is the derivative of L(x) with respect to x, and since, at x, the sign of L(x) changes from minus to plus, the derivative has to be nonnegative at this point. It should be mentioned that L(x) was subtracted from the numerator along the way, but x is its root, so L(x)=0. One can easily conclude that (83)0<β