Niche partitioning in the coevolution of 2 distinct RNA enzymes

Abstract

Organisms that compete for limited resources within a common environment may evolve traits that allow them to exploit distinct ecological niches, thus enabling multiple species to coexist within the same habitat. The process of niche partitioning now has been captured at the molecular level, employing the method of continuous in vitro evolution. Mixed populations of 2 different "species" of RNA enzymes were made to compete for limited amounts of one or more substrates, with utilization of the substrate being necessary for amplification of the RNA. Evolution in the presence of a single substrate led to the extinction of one or the other enzyme, whereas evolution in the presence of 5 alternative substrates led to the accumulation of mutations that allowed each enzyme to exploit a different preferred resource. The evolved enzymes were capable of sustained coevolution within a common environment, exemplifying the emergence of stable ecological niche behavior in a model system. Biochemical characterization of the 2 evolved enzymes revealed marked differences in their kinetic properties and adaptive strategies. One enzyme reacted with its preferred substrate approximately 100-fold faster than the other, but the slower-reacting species produced 2- to 3-fold more progeny per reacted parent molecule. The in vitro coevolution of 2 or more species of RNA enzymes will make possible further studies in molecular ecology, including the exploration of more complex behaviors, such as predation or cooperation, under controlled laboratory conditions.

Fig. 1.

Sequence and secondary structure of the evolved CL1 (A) and DSL (B) enzymes, shown with substrates S4 and S5, respectively. Open rectangles indicate primer binding sites at the 5′ end of the substrate and 3′ end of the enzyme. Curved arrow indicates the site of ligation. Filled circles highlight mutations present in typical clones isolated after 50 transfers of coevolution (with 5 substrates) relative to the starting enzymes.

Fig. 2.

Time course of 50 transfers of continuous coevolution in the presence of 5 different substrates. The concentration of the CL1 (orange) and DSL (blue) enzymes was determined (based on their respective cDNAs) before and after each transfer. Paired bar graph at the Top indicates the substrate preference of each enzyme (CL1 at Left, DSL at Right) after various transfers when provided 1 μM S1 (red), S2 (green), S3 (purple), S4 (orange), or S5 (blue). Stepped graph at the Bottom indicates the concentration of S5 present during the coevolution experiment; all other substrates were present at 1 μM concentration throughout.

Fig. 3.

Amplification profiles of the CL1 (circles) and DSL (squares) enzymes operating in the presence of 1 μM of either S4 or S5. Filled symbols indicate behavior in the presence of the cognate substrate (CL1 with S4, DSL with S5); open symbols indicate behavior in the presence of the noncognate substrate. The concentration of RNA enzyme was determined at various times, and the data were fit to the logistic equation: [enzyme] = a/(1 + be^−ct), where a is the maximum extent of amplification and c is the exponential growth rate. Curvilinear regression coefficients were 0.995 for CL1 with S4 and 0.998 for DSL with S5. Inset shows behavior during the linear phase of growth.

Fig. 4.

Sustained coevolution of the CL1 (circles) and DSL (squares) enzymes. Five successive rounds of amplification and 100-fold dilution were performed over a period of 2.5 h. The concentration of each enzyme was determined based on incorporation of [α-³²P]ATP into newly synthesized RNAs. The concentrations of S4 and S5 were 1 and 0.02 μM, respectively.

Fig. 5.

Catalytic activity and transcription rate of the evolved CL1 (circles) and DSL (squares) enzymes. (A) Reaction of CL1 with S4. (B) Reaction of DSL with S5. Data were fit to the Michaelis–Menten equation; curvilinear regression coefficients were 0.984 and 0.981 for CL1 and DSL, respectively. (C) Transcription rates were measured by using various concentrations of RNA/cDNA heteroduplex templates. Linear regression coefficients were 0.994 and 0.983 for CL1 and DSL, respectively.