Latent evolutionary potentials under the neutral mutational drift of an enzyme

Abstract

Biological systems exhibit mutational robustness, or neutrality, whereby the impact of mutations is minimized. Does neutrality hamper their ability to adapt in the face of changing environments? We monitored changes in genotype and phenotype that occur within a neutral mutational network of an enzyme, experimentally and computationally (see accompanying article). Using the enzyme PON1 as a model, we performed random mutagenesis and purifying selection to purge deleterious mutations. We characterized approximately 300 variants that are apparently neutral, or close to neutral, with respect to PON1's levels of expression and native lactonase activity. Their activities with promiscuous substrates and ligands indicated significant changes in adaptive potentials. Almost half of the variants exhibited changes in promiscuous activities, specificities, or inhibition, and several of these were found to be one or two mutations, closer to potentially new phenotypes: aryl esterase, thiolactonase, phosphotriesterase, or drug resistance. This empirical measure of phenotypic changes under neutrality supports the notion that sequence changes that are neutral, i.e., non-adaptive, in a current context can facilitate adaptation under changing circumstances, by both expanding the activity range of existing enzymes and thus providing an immediate advantage, and by reducing the number of mutations required for divergence of new functions.

Figure 1. Defining the neutrality categories according to the native lactoanse activity and expression level.

The lactonase activity (with TBBL as substrate), and expression level (GFP emission), were determined for each variant by averaging 2–8 independent measurements, and normalized to wild-type values. Variants were divided into three groups according to these categories: (A). Neutral—both lactonase activity and expression levels within 2SD of the wild type’s values. (B). Nearly neutral—only lactonase activity within 2SD of the wild type’s value. [Note that, the neutral variants [group (A)] comprise a subset of the nearly neutral variants group (B)]. (C) Possibly neutral—lactonase activity and expression levels that deviate from wild-type PON1, yet are well above background (⩾3SD). (D) The lactonase activity and expression levels for multiple independent repeats of wild-type PON1 (N=147). Noted (black box) are the activity and expression levels that define the neutral set [category (A)].

Figure 2. Changes in total activities within apparently neutral variants.

Variants were categorized as detailed in Fig. 1. Presented here are the total activities (rate of product formation relative to wild type) for each variant with all substrates. (D) The total activities measured for a control set of 147 repeats of wild-type PON1.

Figure 3. Changes in selectivity within apparently neutral variants.

Selectivity values were determined by calculating for each variant the ratios of activities between different pairs of substrates and then normalizing to wild type. (D) The selectivity measured for a control set of 147 repeats of the wild type.

Figure 4. Changes in susceptibility to inhibition within apparently neutral variants.

Presented is the relative level of lipolactonase activity in the presence of 100 μM of the inhibitor 2ODQ, for the entire set of 311 apparently neutral variants, and 147 repeats of the wild type (inset). The wild-type lactonase activity in the absence of inhibitor (100%), and in its presence (11%) are marked with red lines.

Figure 5. The apparently neutral sequence changes with respect to PON1’s tertiary and primary structures.

(A) A ribbon representation of the 3D structure of PON1. The active site pocket comprises 32 residues and is presented in mesh. The highly conserved active site core (catalytic histidine-dyad H115 and H134) is shown in cyan spheres. Active site residues found to be mutated in neutral and nearly neutral variants are colored in red and possibly neutral variants in yellow. (B) Mutations found in “neutral” PON1 variants correlates with the natural variability of the PON family. The multiple sequence alignment of the PON gene family indicated three conserved blocks (marked A, B, and C). The conservation of each position and residue is illustrated as its height in bits (LOGO presentation). Residues found to be mutated in neutral PON1 variants are indicated by a red arrow, nearly neutral in green, and possibly neutral in blue.

Figure 6. Schematic representation of PON1 putative neutral network.

The scheme is based on the neutral variants described here (supplementary Table S1), and on the mutations and transition phenotypes identified in a previous directed evolution experiment and in natural paralogs of PON1 (Aharoni et al., 2005a; Aharoni et al., 2005b; Harel et al., 2004). The length of the edges corresponds to their neutrality state. The neutral variants are scattered in a distance (edge length) of 1 from the WT. Nearly neutral variants are scattered around the WT in a distance of 2, and possibly neutral in a distance of 3. The direction is the tendency towards a specific latent activity. The other neutral regions (e.g., aryl esterase, thio-lactonase, etc.) were arbitrarily placed in some point in the sequence space scheme, although some phenotypes might be closer to others (e.g., thio-lactonase and lipo-lactonase). The large circle denotes the boundaries of the neutral network of PON1 native phenotype (lipo-lactonase). The dashed borders and edges are hypothetical. Filled edges are based on common mutations. Each node corresponds to a specific sequence with one or more mutations that do not appear at the point of origin (WT sequence).