The source and brand of each of the esters (purity ≥99%) used in this study was Merck Life Science S.L.U., Madrid, Spain. The oligonucleotides used for DNA amplification were synthesized by Sigma Genosys, Ltd. (Pampisford, Cambridgeshire, United Kingdom). The Escherichia coli EPI-300-T1R strain used for pCCFOS1 fosmid library construction and screening was from Epicentre Biotechnologies (Madison, WI, United States). The E. coli strain GigaSingles used for gene cloning and E. coli strain BL21 (DE3) used for gene expression were from Novagen (Darmstadt, Germany).

The Los Rueldos gallery is located along the northwestern slope of the Morgao Valley (2 km northeast of the town of Mieres and 20 km southeast of Oviedo, which is the capital city of Asturias in northwestern Spain; 43°15′47″N, 5°46′9″W). It is a 70 m-long gallery with 10–30 cm depths in the shallower areas and 40–70 cm depths in the deeper sections (Méndez-García et al., 2014). Microorganisms are developed along the AMD system (pH ∼2), forming a bedded acidic biofilm with uppermost oxic (B1A) and lowermost anoxic (B1B) strata. The DNA samples from B1A and B1B (see below) samples collected and used in this study were the same as those in the previous work (Méndez-García et al., 2014). Briefly, samples were collected in sterile 50 ml tubes at two sampling sites determined by the presence of each different macroscopic microbial growth morphology (B1A: up to 2 cm deep; B1B: from 2 to 15 cm deep) and kept on ice until nucleic acid extraction was performed (within the following 2 h).

The DNA samples from B1A and B1B were the same as those used in a previous work (Méndez-García et al., 2014), which were obtained using the Power Soil DNA extraction kit (Cambio, Cambridge, United Kingdom) according to the manufacturer’s guidelines. Prior to clone library construction, the metagenomics DNA was concentrated by first adding 50 μl of 3 M sodium acetate solution to 50 μl DNA extract. Precipitation was conducted by the addition of 1.25 ml of ethanol and incubation at room temperature for 10 s. Precipitated DNA was pelleted by centrifugation at 20,000 g for 10 min. The resulting pellets were washed with 500 μl of 70% (v/v) ethanol twice, and the traces of ethanol were evaporated by incubation under a fume hood at room temperature for 10 min. The resulting pellets were then dissolved in 20 μl of sterile nuclease-free water. Before cloning in the large-insert pCCFOS1 fosmid libraries using the CopyControl Fosmid Library Kit (Epicentre Biotechnologies, Madison, WI, United States) and the E. coli EPI300-T1^R strain, the DNA (10 μg) that was unsheared by gel electrophoresis was subjected to shearing by pipetting through a 200 μl pipette tip 100 times, following the recommendations of the supplier (Epicentre Biotechnologies, Madison, WI, United States) to reach an approximately size of 30,000 bp. Cells of each pCCFOS1 fosmid library were suspended in glycerol to a final concentration of 20% (v/v) and stored at −80°C until further use. We generated subsets of 94,000 and 81,000 clones for the B1A and B1B samples, respectively. Restriction analysis of 10 randomly selected clones from each library revealed average insert sizes of 34,000 bp (for the B1A samples) and 39,500 bp (for the B1B samples), which included nearly 3.2 Gbp of community genomes per sample. This size is within the range of the average size range of DNA inserts in positive clones found in this study (see below).

Fosmid clones were plated onto large (22.5 × 22.5 cm) Petri plates with Luria Bertani (LB) agar containing chloramphenicol (12.5 μg/ml) and induction solution (Epicentre Biotechnologies; WI, United States) at a quantity recommended by the supplier to induce a high fosmid copy number. Clones were scored by the ability to hydrolyze α-naphthyl acetate (α-NA) and tributyrin, as previously described (Reyes-Duarte et al., 2012). Positive clones presumptively containing carboxylesterases and lipases with the α/β hydrolase fold were selected, and their DNA inserts were sequenced using a MiSeq Sequencing System (Illumina, San Diego, CA, United States) with a 2 × 150-bp sequencing v2 kit at Lifesequencing S.L. (Valencia, Spain). Before sequencing, fosmid DNA was extracted from the fosmid clones containing the metagenomic segments using the QUIAGEN Large-Construct Kit (QUIAGEN, Hilden, Germany), according to the manufacturer’s protocol. Upon the completion of sequencing, the reads were quality-filtered and assembled to generate non-redundant meta-sequences, and genes were predicted and annotated as described previously (Placido et al., 2015).

The predicted protein-coding genes obtained in a previous study (Méndez-García et al., 2014) after the sequencing of DNA material from resident microbial communities in each of the samples (B1A and B1B) with a Roche 454 GS FLX Ti sequencer (Roche Applied Science, Penzberg, Germany) were used in this study. The meta-sequences are available from the National Center for Biotechnology Information (NCBI) non-redundant public database with the IDs PRJNA193663 (for B1A) and PRJNA193664 (for B1B). Protein-coding genes identified from metagenomes (sequence-based screening) and from the DNA inserts of positive clones (naïve screen) were screened (score >45; e-value <10e^–3) using BLASTP and PSI-BLAST searching (Altschul et al., 1997) for enzymes of interest against the ESTerases and alpha/beta-Hydrolase Enzymes and Relatives (ESTHER) and LED databases (Fischer and Pleiss, 2003; Barth et al., 2004; Bauer et al., 2020).

The experimental procedures used for the cloning, expression, and purification of selected proteins (either from naïve or homology sequence screening) in the Ek/LIC 46 vector and E. coli strain BL21 (DE3) were performed as described previously (Alcaide et al., 2015). The primers used for amplification are listed in Supplementary Material. All proteins studied here were N-terminally His₆-tagged, and the soluble His-tagged proteins were produced and purified at room temperature after binding to a nickel–nitrilotriacetic acid (Ni–NTA) His-Bind resin (from Merck Life Science S.L.U., Madrid, Spain) as described previously (Giunta et al., 2020), with slight modifications (the expression culture was scaled up to 1 L using 50 ml pre-inoculum). The purity was assessed as >98% using sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS–PAGE; Supplementary Figure 1) in a Bio-Rad Mini Protein system (Laemmli, 1970). Protein concentrations were determined according to the Bradford method with bovine serum albumin as the standard (Bradford, 1976). A total of approximately 0.8–37 mg of purified recombinant proteins was obtained from each 1 L culture on average, as follows: Est_A1 (6.4 mg/L), Est_A2 (25 mg/L), Est_A3 (13 mg/L), Est_A4 (37 mg/L), Est_A5 (41 mg/L), Est_A6 (7 mg/L), Est_A7 (0.8 mg/L), Est_A8 (19 mg/L), Est_B1 (1.0 mg/L), and Est_B2 (32 mg/L).

The hydrolysis of 2-naphthyl acrylate (ref. 577189), tri(propylene glycol) diacrylate (ref. 246832), dibenzyl terephthalate (ref. PH000126), and bis(2-hydroxyethyl)-terephthalate (BHET; ref. 465151) (all from Merck Life Science S.L.U., Madrid, Spain) was assessed using a pH indicator assay in 384-well plates (ref. 781162, Greiner Bio-One GmbH, Kremsmünster, Austria) at 30°C and pH 8.0 in a Synergy HT Multi-Mode Microplate Reader in continuous mode at 550 nm over 24 h [extinction coefficient (ε) of phenol red, 8,450 M^–1 cm^–1]. The acid produced after ester bond cleavage by the hydrolytic enzyme induced a color change in the pH indicator that was measured spectrophotometrically at 550 nm. The experimental conditions were as detailed previously (Giunta et al., 2020), with the absence of activity defined as at least a twofold background signal. For V_max determination, (protein): 270 μg/ml; (ester): 20 mM; reaction volume: 44 μl; T: 30°C; and pH: 8.0. Activity was calculated by determining the absorbance per minute from the generated slopes and applying the following equation:

The activity toward the model esters p-nitrophenyl acetate (pNPC₂), propionate (pNPC₃), butyrate (pNPC₄), octanoate (pNPC₈), decanoate (pNPC₁₀), and decanoate (pNPC₁₂) was assessed in 50 mM Britton and Robinson (BR) buffer at pH 8.0 and 30°C by monitoring the production of 4-nitrophenol at 348 nm (pH-independent isosbestic point, ε = 4147 M^–1 cm^–1) for over 5 min and determining the absorbance per minute from the generated slopes (Santiago et al., 2018). The reactions were performed at 30°C in 96-well plates (ref. 655801, Greiner Bio-One GmbH, Kremsmünster, Austria) and contained 0.09 to 3 μg proteins and 0.8 mM esters in a total volume of 200 μl. The effect of pH on the activity was determined in 50 mM BR buffer at pH 4.0–12.0, as described previously. Similar assay conditions were used to assay the effects of temperature on the ester hydrolysis of pNPC₃, but in this case, the reactions were performed in 50 mM BR buffer pH 7.0. Note that the BR buffer consists of a mixture of 0.04 M H₃BO₃, 0.04 M H₃PO₄, and 0.04 M CH₃COOH that has been titrated to the desired pH with 0.2 M NaOH. All values were determined in triplicate and were corrected for non-enzymatic transformation. In all cases, the activity was calculated by determining the absorbance per minute from the generated slopes and applying the following equation:

Poly(propylene glycol) diacrylate (ref. 455024, Merck Life Science S.L.U., Madrid, Spain) and poly(DL-lactide) with an average molecular weight 2,000 (ref. AP224, PolySciTech, Akina, IN, United States) were assayed as described previously (Hajighasemi et al., 2018). The hydrolysis of polyethylene terephthalate (PET) films (prepared as reported by Bollinger et al., 2020) and particles, which were prepared using PET from a bottle (from a local shop – Granini brand), as described previously (Pütz, 2006), was evaluated at 30°C and pH 8.0 with 270 μg protein/ml and 2 mg/ml plastic material, as previously reported (Bollinger et al., 2020).

The effect of the inhibitors mercaptoethanol (ref. M7154) and iodoacetamide (ref. I1149), which were both from Merck Life Science S.L.U., Madrid, Spain, was tested as follows. A mixture containing the purified enzymes (final concentration of 1 mg/ml) in 190 μl of 40 mM 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES) at pH 7.0 and the inhibitors (final concentration, 1–10 mM) was incubated for 5 min to 24 h at 30–45°C. The reaction was initiated by adding pNPC₃ (0.8 mM, final concentration), and the activity was measured for over 5 min as described above and compared to control samples without inhibitors.

Circular dichroism (CD) spectra were acquired between 190 and 270 nm with a Jasco J-720 spectropolarimeter equipped with a Peltier temperature controller in a 0.1-mm cell at 25°C. The spectra were analyzed, and denaturation temperature (T_d) values were determined at 220 nm between 10 and 85°C at a rate of 30°C per hour in 40 mM HEPES buffer at pH 7.0. CD measurements were performed at pH 7.0 and not at the optimal pH (8.5–9.0) to ensure protein stability. A protein concentration of 0.5 mg/ml was used. T_d (and the standard deviation of the linear fit) was calculated by fitting the ellipticity (mdeg) at 220 nm at each of the different temperatures using a 5-parameter sigmoid fit with SigmaPlot 13.0.

The sequences were named based on the code “Est,” which refers to Esterase, followed by a letter indicating the origin of the sample, as follows: Est_A, esterase from the uppermost oxic B1A strata; and Est_B, esterase from the lowermost anoxic B1B sediment attached strata. The final number (subscript) is an arbitrary number representing the number of enzymes per site. Sequences encoding enzymes were deposited under the BioProject IDs PRJNA193663 (for B1A) and PRJNA193664 (for B1B) in the NCBI public database, with the accession numbers detailed in Table 1.

The models of the protein structures were predicted with AlphaFold 2.1.0 (Jumper et al., 2021).

Microbial communities inhabiting two distinct compartments within Los Rueldos AMD formation were screened for sequences encoding carboxylesterases or lipases. For that, we used two complementary metagenomics approaches, namely, naïve and homology sequence screens. First, a total of approximately 81,000 pCCFOS1 clones from each clone library (equivalent to 2.8 Gbp for B1A and 3.20 Gbp for B1B) were screened for esterase/lipase activity using plate-based screen with αNA and tributyrin as model substrates. We identified a total of 10 positive clones being active against both substrates in the B1A clone library, whereas no positives were found in the B1B library. The fosmids with insert lengths ranging from 16,545 to 41,280 bp were fully sequenced, from which 10 genes (one per positive clone), encoding presumptive esterases/lipases, were identified. In addition, we searched the predicted protein-coding genes obtained through next-generation sequencing for sequences encoding esterases and lipases by BLAST search against the ESTHER and LED databases. A total of 6 full-length sequences (B1A: 2; B1B: 4), with accession numbers EQD63018.1, EQD66234.1, EQD71191.1, EQD52136.1, EQD55146.1, and EQD26916.1, encoding potential enzymes were identified. Taken together, a total of 16 genes encoding hydrolases from the α/β-hydrolase fold superfamily, specifically, 12 from B1A (EstA₁ to EstA₁₂) and 4 from B1B (EstB₁ to EstB₄), were identified (Table 1). As determined by Matcher (EMBOSS package), the pairwise amino acid sequence identity for 14 of the 16 α/β hydrolases ranged from 4.0 to 44.9%. This, together with the fact that only 2 out of 16 polypeptides were highly similar (EstA₅ and EstA₆ differ in only 2 amino acids: arginine 152 and alanine 179 in Est_A5 are cysteine 152 and threonine 179 in Est_A6), suggests a large divergence at the sequence level within the enzymes examined, and that the diversity of polypeptides was not dominated by a particular type of protein or highly similar clusters of proteins, but rather by diverse non-redundant sequences. Note that only 1 of 10 sequences selected after naïve screens was found in the metagenomic data generated after direct DNA sequencing (EstA₁₁, which is 99% identical to GenBank accession no. EQD37671.1 from the Los Rueldos metagenome) (Méndez-García et al., 2014). This demonstrates that both types of screens (naïve and in silico) are complementary tools for enzyme discovery. However, deeper metagenomic sequencing could potentially detect all enzymes isolated by naïve screens.

The deduced molecular mass and estimated isoelectric point (pI) values ranged from 23.19 to 101.53 kDa and from 4.62 to 10.04, respectively. Putative proteins exhibited a maximum amino acid sequence identity ranging from 39 to 100% to putative esterases/lipases in public databases (Table 1). It is worth mentioning that EstA₃, EstA₈, and EstB₃ are related to presumptive esterase/lipase-like subfamily proteins of the Serine-Glycine-Asparagine-Histidine (SGNH) hydrolases, EstA₁₁ to presumptive glycoside-hydrolase family GH114 (N-terminal domain) and CE4_PelA_like hydrolases (C-terminal domain), and EstA₁₂ to presumptive sialate O-acetylesterases. A further TBLASTX search against metagenomics proteins deposited in databases revealed no similarity of 10 proteins with homologous AMD metagenome proteins. In contrast, five proteins (EstA₃ to EstA₆, and EstA₁₂) do share from 27 to 54% homology to three proteins from the Carnoules arsenic-contaminated mine drainage (GenBank: CBI07622.1, CBH97521.1, and CBI00527.1). This finding suggests that esterases/lipases from microbial communities from the Los Rueldos site are distantly related to proteins from other known homologous proteins from AMD formations with metagenome sequences available. It also reflects the large undiscovered pool of enzymes from bacterial species populating the Los Rueldos site.

Based on the comparison of the primary structures, 14 families of sequence-related esterases and lipases have been reported (Arpigny and Jaeger, 1999; Rao et al., 2013). Sequence analysis categorized 13 enzymes from Los Rueldos into some of these known subfamilies (Figure 1) with most structurally similar homologs as follows: EstA₉ [27%; best hit in Protein Data Bank (PDB) 3DOH_A] and EstB₄ (41%; 3OM8_A) to Family I; EstA₁ (41%; 3V9A_A), EstA₄ (41%; 4YPV_A), EstA₅ (49%; 4YPV_A), and EstA₆ (49%; 4YPV_A) to Family IV; EstA₇ (53%; 4YPV_A) and EstB₂ (41%; 1AUO_A) to Family VI; EstA₂ (37%; 2OGT_A) to Family VII (EstA₂); and EstA₁₀ (41%; 4IVK_A) to beta-lactamase like Family VIII. EstA₃ (43%; PDB code 3P94_A), EstA₈ (44%; PDB code 3P94_A), and EstB₃ (26%; PDB code 3KVN_X) belong to Family II GDSL, but the structural alignment also confirms that they contain a domain that displays the characteristic α–β–α globular fold of the SGNH hydrolase family. In addition, EstB₃ also contains a passenger domain providing the driving force for passenger translocation (Van den Berg, 2010). Three of the sequences could not be assigned to these subfamilies. First, EstA₁₁ contains a 300 amino acid long N-terminal domain most similar to glycoside-hydrolase family 114 and a 616 amino acid long C-terminal domain most similar to the carbohydrate esterase 4 (CE4) superfamily that includes chitin deacetylases (EC 3.5.1.41), N-acetylglucosamine deacetylases (EC 3.5.1.-), and acetylxylan esterases (EC 3.1.1.72), which catalyze the N- or O-deacetylation of substrates such as acetylated chitin, peptidoglycan, and acetylated xylan. Its N-terminal domain is most structurally similar (26%) to that of the glycosidase 2AAM_A and its C-terminal domain is structurally similar to the polysaccharide deacetylase from Bacillus cereus (4HD5). Second, EstA₁₂ is associated with acetylxylan esterases (EC 3.1.1.72), with most similar (21%) structural homolog in PDB being 1ZMB_A. Third, EstB₁ shows homology to small serine alpha/beta-hydrolase/acyl-peptidase (58%; 2FUK_A). The tentative amino acids participating in the typical catalytic triad of esterases and lipases are summarized in Table 1. Together, the analysis of the primary sequence suggests that the diversity of esterases was not dominated by a particular type of protein or a highly similar cluster of proteins, but rather by diverse non-redundant sequences belonging to different microbial groups and distinct esterase/lipase subfamilies.

A search of oligonucleotide patterns against the GOHTAM database (Ménigaud et al., 2012) and TBLASTX analysis revealed compositional similarities between the DNA fragment containing the genes for EstA₁, EstA₂, EstA₁₀, and EstB₄, with genomic sequences of bacteria from the phylum Actinomycetota. Among them, only unambiguous affiliations at lower levels could be achieved for fragments containing Est_A1, Est_A2, and EstA₁₀ that may most likely belong to bacteria from the genera Acidithrix (EstA₁ and EstA₁₀) and Acidimicrobium/Ferrimicrobium (EstA₂), both from the family Acidimicrobiaceae within the order Acidimicrobiales. Note that EstA₁ and EstA₁₀ showed 99–100% sequence identity with uncharacterized esterases and lipases (WP_052605564.1 and WP_052605292.1) from Acidithrix ferrooxidans, and EstA₂ showed 94% sequence identity with an uncharacterized esterase-lipase (NNN14078.1) from Acidimicrobiaceae. EstA₃ was most likely derived from an uncultured bacterium assigned to the phylum Acidobacteria with ambiguous affiliation below the phylum level. The genes for EstA₄ to EstA₉, EstA₁₁, and EstB₁ to EstB₃ were associated with uncultured bacteria of the Proteobacteria phylum, with ambiguous affiliations at a lower taxonomic level, except for EstA₄, which was most likely derived from a bacterium of the genus Acidiphilium from the family Acetobacteraceae within the order Rhodospirillales (best hit OYV70855.1 from Acidiphilium sp., 79% homology). All these bacterial groups have been detected in biofilms thriving in the Los Rueldos mine (Méndez-García et al., 2014). No clear affiliation, other than Bacteria, could be found for EstA₁₂. Note that in some cases no clear affiliation to a taxon of source organism below the level of the phylum could be established, either because of the short fragment length or the low compositional similarity between the metagenomic fragments and the sequences of related bacterial chromosomes and plasmids do not allow proper assignations.

From the 16 sequences selected, 8 from B1A and 2 from B1B were successfully cloned, expressed, and purified as soluble active proteins when expressed in pET Ek/LIC 46 vector and E. coli BL21 as the host. These proteins were herein referred to as EstA₁ to EstA₈, EstB₁, and EstB₂. The remaining six (EstA₉-EstA₁₂ and EstB₃-EstB₄) could not be produced in soluble active form (they formed inclusion bodies) in the expression system applied herein, which consists of the use of the Isopropyl β-D-1-thiogalactopyranoside (IPTG)-inducible Ek/LIC 46 vector and E. coli strain BL21 (DE3) as a host, and their properties are not described herein. Refining the expression conditions, which included variations in the expression conditions (16, 30, and 37°C) and IPTG concentration (from 0.1 to 2 mM), resulted in unsuccessful production of sufficient active protein material for characterization. Further efforts may be needed with different expression vectors, which is beyond the scope of the present study.

The substrate profile of all α/β hydrolases was first examined using a set of esters commonly used to characterize esterases and lipases, namely, pNP esters such as pNPC₂, pNPC₃, pNPC₄, pNPC₈, pNPC₁₀, and pNPC₁₂. All ester hydrolases preferred short-chain-length pNP-esters, particularly pNPC₂ (EstA₃), pNPC₃ (EstA₂, EstA₄, EstA₅, EstA₆, EstB₁, and EstB₂), and pNPC₄ (EstA₁, EstA₇, and EstA₈) (Table 2). Within all six pNP ester tested, all but one (EstA₃) enzyme was able to hydrolyze up to pNPC₁₂, albeit at a much lower level (from 62- to 3,900-fold) compared to shorter derivatives. Considering the preferred pNP esters, the maximum specific activity ranged from 3.06 ± 0.03 to 679.8 ± 9.8 U/mg. We further test the possibility that the enzymes hydrolyze substrates other than pNP esters, particularly, plastic substrates and esters relevant to plastics. Using previously described conditions (Hajighasemi et al., 2018; Bollinger et al., 2020), we did not find that any of the enzymes hydrolyzed large plastic materials such as poly(propylene glycol) diacrylate, poly(DL-lactide), amorphous PET film, and PET nanoparticles. However, by using a pH-indicator assay, we found that the enzymes were able to hydrolyze other terephthalate esters and acrylate esters. Thus, as shown in Table 3, six of the enzymes hydrolyzed esters relevant to acrylic acid plastics, e.g., 2-naphthyl acrylate and tripropylene glycol diacrylate, a commonly used material principally exploited to prepare thermally stable polymers (He et al., 2017). These substrates, herein found to be converted at a maximum rate of 3,915 ± 48 U/g, are rarely hydrolyzed by esterases and lipases. There are only two examples reported, namely, human salivary pseudocholinesterase and cholesterol esterase (Finer and Santerre, 2004; Cai et al., 2014). In addition, one of the enzymes (Est_A8) was capable of hydrolyzing dibenzyl terephthalate (432.2 ± 27.5 U/g), an intermediate produced during chemical PET recycling with benzyl alcohol in the presence of a catalyst (Donahue et al., 2003). No esterase or lipase has been reported to date that degrades this substrate. In addition, six of the esterases (Est_A1, Est_A2, Est_A5, Est_A6, Est_A8, and Est_B2) efficiently hydrolyzed BHET (from 5.0 ± 1.0 to 336.9 ± 3.6 U/g), an intermediate in the degradation of PET (Yoshida et al., 2016). High performance liquid chromatography (HPLC) analysis, performed as described (Bollinger et al., 2020), confirmed the hydrolysis of BHET to mono-(2-hydroxyethyl)-terephthalic acid (MHET) and not to terephthalic acid. To conclude, the enzymes reported herein from the Los Rueldos AMD formation showed high activity for converting and recycling components of synthetic plastics, namely, acrylic- and terephthalate-based plastics, and could be of potential use in developing plastic degradation strategies.

Using pNPC₃ as a substrate, the purified proteins were most active at temperatures ranging from 30 to 65°C (Figure 2). The average annual temperature in Los Rueldos is 13.8 ± 0.6°C, which varied from 10 ± 0.6°C to 17.1 ± 0.6 (Méndez-García et al., 2014), with a temperature of 17°C when samples were taken (July). At this value, the enzymes retained from 20 to 61% of the activity shown at the optimal temperature (Figure 2).

Using pNPC₃ as a substrate, all enzymes showed an optimum pH for activity from neutral to slightly basic, which varied from 7.0 to 9.0 (Figure 2). This finding suggests that these proteins are most likely intracellularly produced, consistent with the absence of signal peptides in their sequences. Even though the enzymes showed a slightly basic optimum pH, all retained 33–68% of their activity at pH 5.5. Interestingly, Est_A6 shows two activity peaks, one at pH 5.5 and one at pH 9.0, while Est_A5, which only differs in two amino acids, has an optimum pH of 9.0 (Figure 2).

Sequence analysis revealed that EstA₅ and EstA₆ which have their origins in a bacterium of the phylum Proteobacteria, differ in only 2 amino acids (99.4% identity). Positions 152 and 172 are occupied by Arg and Ala in EstA₅ and by Cys and Thr in EstA₆, respectively. Notably, EstA₅ was most active at 30°C, retaining more than 80% of the activity at temperatures from 20 to 45°C (Figure 2). The optimum temperature for activity increased up to 45°C for EstA₆, which maintained more than 80% of its activity in the range from 30 to 60°C. Analysis by circular dichroism revealed that Est_A5 showed a sigmoidal curve from which a temperature of denaturation of 60.4 ± 0.2°C could be obtained (Figure 3). However, the curve for Est_A6 shows two transitions, one with a denaturation temperature of 48.1 ± 0.8°C and a second at 75.7 ± 0.2°C. The presence of these two phases may therefore indicate that the presence of these two amino acids may contribute to protein stability and its denaturation under distinct conditions. This result may explain the higher optimum temperature for the activity of this enzyme compared to Est_A5, and the stabilization effect of Cys152 and Thr172. This difference in thermostability between Est_A5 and Est_A6 can probably be explained by the difference in amino acid 152, since Est_A6 has a Cys that would allow it to make a possible disulfide bridge with Cys181, giving it greater thermostability than Est_A5 (since Est_A5 has an Arg at position 152 instead of a Cys), as shown by examination of the 3D models (Supplementary Figure 2 and Figure 4). It is plausible that this difference may also be responsible for the different pH profiles of both enzymes (Figure 2). If the disulfide bridge was present in Est_A6, it could be removed by reduction or chemical modification. Activity tests revealed that both enzymes are resistant to the reducing agent beta-mercaptoethanol, with no activity lost even after 24 h of incubation at 30–45°C in the presence of a 1–10 mM inhibitor. By contrast, in the presence of the cysteine alkylating agent iodoacetamide, both enzymes were rapidly inactivated after 5 min. Thus, we could not verify the presumptive formation of the disulfide bridge in Est_A6, and further studies are needed to test this assumption.