Enzymic Approach to Eurythermalism of Alvinella pompejana and Its Episymbionts

Abstract

The equilibrium model, which describes the influence of temperature on enzyme activity, has been established as a valid and useful tool for characterizing enzyme eurythermalism and thermophily. By introducing K_eq, a temperature-dependent equilibrium constant for the interconversion between E_act, the active form of enzyme, and E_inact, a reversibly inactive form of enzyme, the equilibrium model currently provides the most complete description of the enzyme-temperature relationship; its derived parameters are intrinsic and apparently universal and, being derived under reaction conditions, potentially have physiological significance. One of these parameters, T_eq, correlates with host growth temperature better than enzyme stability does. The vent-dwelling annelid Alvinella pompejana has been reported as an extremely eurythermal organism, and the symbiotic complex microbial community associated with its dorsal surface is likely to experience similar environmental thermal conditions. The A. pompejana episymbiont community, predominantly composed of epsilonproteobacteria, has been analyzed metagenomically, enabling direct retrieval of genes coding for enzymes suitable for equilibrium model applications. Two such genes, coding for isopropylmalate dehydrogenase and glutamate dehydrogenase, have been isolated from the A. pompejana episymbionts, heterologously expressed, and shown by reverse transcription-quantitative PCR to be actively expressed. The equilibrium model parameters of characterized expression products suggested that enzyme eurythermalism constitutes part of the thermal adaptation strategy employed by the episymbionts. Moreover, the enzymes' thermal characteristics correspond to their predicted physiological roles and the abundance and expression of the corresponding genes. This paper demonstrates the use of the equilibrium model as part of a top-down metagenomic approach to studying temperature adaptation of uncultured organisms.

Temperature variation is an intrinsic property of almost all ecosystems, and many environments feature large temporal and/or spatial temperature gradients. Organisms adapted to such wide ranges of temperatures are termed eurythermal. While eurythermal poikilotherms can achieve adaptation through behavioral means, strict ectotherms and microbes have to be metabolically and structurally adapted. Therefore, enzymes of a eurythermal ectotherm or microbe must be adapted accordingly to facilitate cellular functions. The question of how this is achieved has challenged scientists (4), and important discoveries have been made during the last few decades on how cellular mechanisms and enzymes of eurythermal organisms collectively contribute to such adaptations (34).

However, one aspect of enzyme temperature adaptation had not been satisfactorily addressed, i.e., how to assess and compare enzymes for their eurythermal qualities. Traditionally, biochemists have used enzyme catalytic efficiency (k_cat/K_m) as a measure of how well an enzyme operates (18). Catalytic efficiency has been shown to be an inadequate and even misleading parameter for comparing enzymes and is unsatisfactory as a measure of enzyme efficiency (17). Furthermore, an enzyme's in vivo activity depends on its stability, as well as catalytic qualities (33), but stability is usually assessed in the absence of substrates, making the interpretation of these data difficult and potentially less physiologically significant.

While amino acid usage has been suggested to associate with enzyme thermophily, the associations tend to lack predictive value for individual proteins (22, 29). A clear demonstration of this comes from a study of 26 glyceraldehyde phosphate dehydrogenases from organisms with optimal growth temperatures between 20°C and 102°C (3), in which findings from individual thermophilic enzymes conflicted with the “predictions” from the overall data in several aspects.

The recently proposed and experimentally validated equilibrium model (13, 27, 28) has provided a solution to the above problem. By incorporating a reversibly inactive form of enzyme (E_inact) in addition to the active form (E_act) and the irreversibly denatured form (X), the equilibrium model proposes that an equilibrium exists between the active and inactive forms of enzyme, as expressed in the following equation:

where K_eq is the equilibrium constant describing the E_inact/E_act ratio in the reversible E_act-E_inact interconversion and k_inact is the rate constant for irreversible thermal denaturation. K_eq gives rise to two new, intrinsic, temperature-dependent parameters of enzymes, ΔH_eq and T_eq, which are the change in enthalpy and temperature midpoint, respectively, associated with the E_act-E_inact equilibrium. These two parameters, along with two previously established parameters, ΔG^‡_cat (the free energy of activation of the catalytic reaction) and ΔG^‡_inact (the free energy of activation of the thermal denaturation process), provide the most complete quantitative description of how temperature influences enzyme activity and stability to date. Enzyme thermodynamic parameters derived from the equilibrium model, being intrinsic and universal, can be used to quantitatively compare enzymes of diverse origins and nature, and since ΔH_eq and T_eq are obtained under assay conditions, they potentially reflect in vivo conditions.

The equilibrium model's usefulness in studying enzyme temperature adaptation has been demonstrated through a survey of 21 enzymes from a wide variety of backgrounds (24). The survey uncovered a strong link between T_eq and optimal growth temperature and showed that T_eq correlated with optimal growth temperature better than enzyme stability (ΔG^‡_inact) did. The study also established ΔH_eq as the first intrinsic and quantitative measure of enzyme eurythermalism, with findings from the survey supporting the theory that ΔH_eq and T_eq are independent of ΔG^‡_cat and ΔG^‡_inact (28). Overall, the findings established the equilibrium model as a valid tool for assessing and comparing enzyme eurythermalism and thermophily (24).

Although eurythermalism is a common theme in many ecosystems, examples of highly eurythermal organisms are predominantly metazoans adapted to relatively gradual temperature shifts (35). The vent-dwelling annelid Alvinella pompejana and its episymbionts, meanwhile, have been reported to be adapted to an extremely steep temperature gradient, with a measured temperature difference of up to 60°C or more along the worm's body length (9). A. pompejana is a polychaetous annelid colonizing active chimneys at deep-sea hydrothermal vents along the East Pacific Rise (10); it resides in the diffuse-flow regions around the vents, one of the most extreme and thermally dynamic environment known to science. Apart from being possibly the most eurythermal eukaryote discovered to date, A. pompejana is also likely to be highly thermotolerant, since frequent temperature spikes of 80°C or more have been observed inside tubes in which it resides (9).

The dorsal surface of a typical A. pompejana specimen is covered by an attached fleece of filamentous epsilonproteobacteria (5, 8, 19). The bacterial consortium appears to maintain a unilateral nonobligatory symbiotic relationship with the host since A. pompejana has never been found without the episymbionts, and members of the consortium can be found on surfaces around A. pompejana colonies (8, 19). The bacterial episymbionts associated with A. pompejana are collectively referred to as the A. pompejana Proteobacteria episymbionts (APS) and are dominated by a few phylotypes (8, 19). The A. pompejana episymbionts have been suggested to be involved in the nutrition (1, 6) and detoxification (1, 21) of the host, and the bacteria and their associated structures (i.e., biofilm) may play an important role in shielding the host from rapid high temperature fluctuations (9, 14, 16).

There have been some controversies surrounding the thermal adaptedness of A. pompejana. While several studies have reported temperature spikes of 80°C or more in the immediate vicinities of live A. pompejana worms (9, 16), others have argued that A. pompejana itself lacks the adaptations necessary for coping with such temperatures (12, 14, 23). Answers to whether A. pompejana experiences such high temperatures may be provided by examining the thermophilic and eurythermal features of A. pompejana episymbiont enzymes, because although A. pompejana may employ physiological, behavioral, and multiple levels of cellular adaptation to achieve this unique eurythermal adaptation, its bacterial episymbionts are much more limited in terms of available adaptation mechanisms while facing conditions just as severe as, if not more severe than, those experienced by the host.

An international collaborative effort, funded by the National Science Foundation, is under way to study the A. pompejana episymbionts by a metagenomic approach (7). More than 300,000 sequences have been obtained through community-wide shotgun sequencing and put through a custom-designed annotation pipeline (Project Metagenome, http://ocean.dbi.udel.edu/ [log in with “guest”]). Not only can the wealth of information available in the metagenome reveal valuable information about the consortium and its relationship with A. pompejana, it also provides access to sequence information about the episymbiont enzymes which would otherwise be very difficult to obtain.

By utilizing the equilibrium model, this study aimed to characterize and assess the eurythermal qualities of enzymes derived from genes identified in the A. pompejana episymbiont metagenome database in the hope of uncovering new information related to the potential thermal protective roles of the episymbionts and, consequently, the relationship between A. pompejana and its episymbionts. In this study, suitable genes were identified by using bioinformatic tools in the metagenome and cloned into an expression vector. The thermal properties of the heterologously expressed and functional products were then characterized by using the equilibrium model. To further relate the results to in vivo data, the expression levels of target genes were also examined by reverse transcription-quantitative PCR (RT-qPCR) of RNA extracted from A. pompejana episymbiont samples.

Through the development and validation of the equilibrium model, it is now possible to quantitatively and objectively assess an enzyme's eurythermal qualities (13, 24, 28). A general relationship between enzyme intrinsic property and host optimal growth temperature has also been observed based on parameters derived from the equilibrium model, potentially allowing prediction of host habitat properties from those of its enzymes (24). By applying the equilibrium model to enzymes from A. pompejana episymbionts, one may gather new information on the microbial consortium and its potential contribution to the thermal adaptation of A. pompejana. Such information may be instrumental in resolving long-standing controversies surrounding the eurythermalism and thermal tolerance of A. pompejana. To our knowledge, this is the first attempt at elucidating the eurythermalism of a complex microbial system by quantitatively characterizing enzymes isolated from its metagenome.

MATERIALS AND METHODS

Genomic DNA and RNA.

A. pompejana episymbiont genomic DNA sample AP201 was extracted from the episymbiont biomass of a single A. pompejana specimen collected at Bio9 vent (9°N East Pacific Rise) during dive 3836 of the Extreme 2002 cruise, and episymbiont genomic DNA sample AP243 originated from an A. pompejana specimen collected at Q vent (9°N East Pacific Rise) during dive 3837 of the same cruise.

To extract episymbiont metagenomic DNA, the worms were slightly thawed on an iced aluminum block and the episymbiont biomass was removed with sterile forceps and placed in Tris-sodium dodecyl sulfate (SDS)-proteinase K lysis buffer for 1 h. The nucleic acids were recovered with high-salt-cetyltrimethylammonium bromide and phenol-chloroform extractions and subsequently precipitated with isopropanol (2). RNA was removed with RNase Cocktail (Ambion Inc., Austin, TX), and DNA concentration and integrity were measured by UV spectrophotometry and agarose gel electrophoresis.

For RNA extraction, A. pompejana specimens were immersed in RNAlater (Ambion Inc., Austin, TX) immediately after collection on the seafloor to preserve their RNA. Episymbiont total RNA sample 4064 was extracted from the episymbiont biomass of an A. pompejana specimen collected during the Extreme 2004 cruise with a Total RNA Isolation kit (Ambion Inc., Austin, TX). The RNA sample was quantified (∼200 ng/μl) with a NanoDrop (NanoDrop Technologies, Wilmington, DE), and its integrity was verified by agarose gel electrophoresis.

Metagenomics.

Episymbiont genomic DNA samples were sheared and then cloned into high-copy clone vectors (pPCR4-TOPO from Invitrogen Corp., Carlsbad, CA, or pSmart from Lucigen Corp., Middleton, WI). More than 180,000 shotgun clones with 1.5- to 3-kb inserts were generated, with clones derived from genomic DNA sample AP201 accounting for more than 90% of them. The clones were sequenced in the forward and reverse directions, totaling around 134 Mb of high-quality sequence data.

The metagenomic sequences were annotated with a custom annotation pipeline based on the FGenesB package (SoftBerry Inc., Mount Kisco, NY), blastX, and PRIAM analysis (11). The sequences were assembled with stringent criteria and resulted in ∼26,000 contigs and ∼100,000 singletons; this data set was designated the complete assembled Vent Epibionts Environmental Gene (VEEG) data set and was also annotated through the same pipeline. The sequence and annotation data were compiled into a web-accessible database.

Bioinformatics.

NCBI clusters of orthologous groups of proteins (COG) database (37) entries related to genes of interest (GOIs) were identified and used to retrieve chromatograms of metagenomic sequences related to the GOIs. The sequence chromatograms were assembled at high stringency (91% or 96%) with SeqMan (version 5.51; DNASTAR Inc., Madison, WI); the assemblies were edited to identify consensus sequences and SNPs. Additional sequence chromatograms were retrieved based on sequence homology to assist assembly editing, as were consensus sequences from the Complete Assembled Vent Epibionts Environmental Gene data set and gene homologs from other epsilonproteobacteria.

The identified GOI consensus sequences were checked for the presence of NcoI and SmaI cut sites and used for designing PCR cloning primers. The sequences of constructed clones were compared to the original sequence assemblies, and their codon usage patterns were analyzed with Graphical Codon Usage Analyser (http://gcua.schoedl.de/index.html).

Cloning primer design.

PCR cloning primers were designed for APS leuB and APS gdhA on the basis of their consensus sequences. The PCR cloning primers are as follows: aps_leuB forward, 5′-ATGGTTAAGAGATATAAGATTGGTATT; aps_leuB reverse, 5′-TTAGCTTCTGCTAACCAAACTAG; aps_gdhA forward, 5′-ATGAGTGCAAGAGAGTATGTTAATAGTA; aps_gdhA reverse, 5′-TTAGTTAGATTAAAGAGCCAAGAGG. Restriction endonuclease cut sites for NcoI and SmaI were added to the forward and reverse primers, respectively, along with extra overhangs (ATCGAT) at the 5′ ends of primers to improve the efficiency of digestion. The primers were confirmed to be in frame with the expression vector and have similar melting temperatures (T_ms).

Cloning of GOIs.

The GOIs were PCR amplified with the designed PCR cloning primers by using AP201 genomic DNA as the template. PCRs were carried out with a final volume of 25 μl containing 5 ng of AP201 genomic DNA template, 100 nM each respective primer, 100 μM deoxynucleoside triphosphates, 3 mM MgCl₂, 2.5 μl of 10× PCR buffer, and 1 U of Taq DNA polymerase (Roche Molecular Diagnostics, Pleasanton, CA). PCRs were performed under the following conditions: initial denaturation at 94°C for 2 min, followed by 45 cycles of denaturation at 94°C for 30 s, annealing at 58°C for 30 s, and extension at 72°C for 30 s, with a final extension of 5 min at 72°C. The PCR amplicons were size verified by agarose gel electrophoresis, restriction digested with NcoI and SmaI, and ligated into expression vector pIVEX2.4d (catalog no. 3269019; Roche Molecular Diagnostics, Pleasanton, CA), which places the recombinant gene under the control of a T7 promoter and includes an N-terminal His tag for easy purification. The ligation mixtures were transformed into Invitrogen One Shot TOP10 competent Escherichia coli (Invitrogen Corp., Carlsbad, CA). The transformation cultures were spread on LB agar plates containing 100 μg/ml ampicillin and incubated overnight.

The transformants were screened by PCR with T7 primers, and cells picked from transformant colonies were added to PCR mixtures as templates. Colonies with PCR amplicons of the correct size (determined by agarose gel electrophoresis) were chosen for further inoculation and used for plasmid preparation with a GenScript QuickClean 5 M Miniprep Kit (GenScript Corp., Piscataway, NJ). Multiple colonies were selected for both genes, and the resulting plasmids were sequenced in both directions.

Homology analysis of cloned sequences.

Nucleic acid sequences of GOI clones were queried against the GenBank nr and env_nt databases, and sequences of the top three matches were retrieved along with select homologous genes from related organisms. Sequence analyses were performed with CLC Free Workbench (version 4.0.2; CLC bio, Cambridge, MA), including multiple-sequence alignment with ClustalW, and phylogenetic trees were generated by the neighbor-joining algorithm with 100 iterations of bootstrapping.

Overexpression of GOIs.

Clones of GOIs were transformed into Invitrogen One Shot BL21 Star (DE3)pLysS chemically competent E. coli (Invitrogen Corp., Carlsbad, CA) and screened by PCR as described above to verify successful transformation. The transformants were added to LB medium containing 100 μg/ml ampicillin and 34 μg/ml chloramphenicol and incubated at 37°C. Isopropyl-β-d-thiogalactopyranoside (IPTG) was added at a 1 mM concentration to the cultures at an optical density of 0.6 to induce overexpression for 4 to 6 h. Cells were harvested by centrifugation, and cell lysates were purified by following the instructions from the Qiagen Ni-NTA Spin Kit (Qiagen, Hilden, Germany). Purified expression products, APS IPMDH and APS GDH, were analyzed by SDS-polyacrylamide gel electrophoresis and native protein gel electrophoresis.

Enzyme assays.

In general terms, enzyme assays were carried out as described previously (27). All of the enzyme assays performed in this work were continuous spectrophotometric assays measured with a Thermo Spectronic Helios Gamma spectrophotometer equipped with a Thermo Spectronic single-cell Peltier effect cuvette holder. Assays were performed across a temperature range suitable for the enzyme, and absorbance data were collected at 1-s intervals for 3 min with Vision (version 1.25; Thermo Spectronic Inc.) on a Windows personal computer connected to the Helios. The formation of NADH and the oxidation of NADPH were both monitored at 365 nm [_ɛNAD(P)H = 1,700 M⁻¹ cm⁻¹]. A quartz cuvette with a 5-mm light path length was used for all assays.

All reactions were started by the rapid addition and mixing of microliter quantities of enzyme solution held at 0°C into 400 or 500 μl of temperature-equilibrated reaction mixture. This allowed data collection to begin within the first few seconds, and the temperature of the reaction mixture remained within ±1°C of the desired temperature during data collection. Substrate concentrations were set to at least 10 times K_m at all temperatures where possible to keep the enzyme operating at close to V_max; otherwise, K_m values were determined and reaction curves were adjusted accordingly at each temperature point to compensate for the effect of temperature-related K_m increases. No signs of substrate or product inhibition were observed under the experimental conditions described. Blank rates were measured at all temperatures and used to correct reaction rates, if significant. All of the criteria set here and in reference 27 for the accurate determination of equilibrium model parameters were met.

Because of its low ΔpK_a/Δt value, potassium phosphate buffer was used and was pH optimized for both enzymes. APS IPMDH was assayed with the following reaction mixture: 7.5 mM NAD⁺ and 2 to 6 mM isopropylmalate in 100 mM K⁺ phosphate buffer, pH 8.0, with 1 mM MgCl₂. An aliquot of enzyme (1.06 μg in 10 μl of working stock) was added to the assay mixture to give a total volume of 400 μl, and the final enzyme concentration was 35.4 nM. APS GDH was assayed with the following reaction mixture: 1 mM NADPH, 35 mM α-ketoglutarate, and 600 mM NH₄Cl in 100 mM K⁺ phosphate buffer (pH 8.0). An aliquot of APS GDH (77 ng in 1 μl of working stock) was added to the assay mixture to give a total volume of 500 μl, and the final enzyme concentration was 430 pM.

Enzyme data analysis.

The method for processing experimental data is as described in detail in reference 27, except that Scientist (version 2.01; MicroMath Inc.) was used for data processing. Briefly, absorbance data from Vision were first converted to progress curves of product concentration (M) versus time (seconds) in Excel (version 11 for Windows; Microsoft Corp.). The data were subsequently imported into Scientist, where a set of initial estimates of thermodynamic parameters was optimized by Simplex searches through the complete data set (product concentration versus time versus temperature). The optimized initial estimates of parameters were then used to perform least-squares fits of the complete data set to the equilibrium model to generate the final values of the parameters ΔG^‡_cat, ΔG^‡_inact, ΔH_eq, and T_eq. Three-dimensional (3D) plots derived from the resulting parameters were compared to 3D plots generated from smoothed raw activity data to verify the fittings. The parameters generated were then used in the zero-time model (27) to draw a zero-time activity-versus-temperature plot. Based on the variation between the individual triplicate rates from which the parameters are derived, for all of the enzymes we have assayed thus far, we find that the experimental errors in the determination of ΔG^‡_cat, ΔG^‡_inact, and T_eq are less than 0.5% and that they are less than 6% in the determination of ΔH_eq (27).

An alternative package for processing the equilibrium model data, presented as a stand-alone application based on MATLAB (The MathWorks, Inc.) is available on compact disc. The application enables facile derivation of the equilibrium model parameters from a Microsoft Office Excel file of experimental progress curves (product concentration versus time) and is suitable for computers running Microsoft Windows XP or Vista and is for noncommercial research purposes only.

Enzyme half-lives were calculated from enzyme assay data with Prism (GraphPad Software, Inc.).

cDNA generation.

Episymbiont RNA sample 4064 was treated with 7 μg/ml ethidium monoazide bromide (EMA; catalog no. 40015; Biotium Inc., Hayward, CA) to deactivate any residual genomic DNA as described in reference 32. cDNA preparations were generated from the RNA sample with a RETROscript kit (Ambion Inc., Austin, TX) with both gene-specific primers (GDH_qR and IPMDH_qR, cDNA preparation 4064GS) and random decamers provided with the RETROscript kit (cDNA preparation 4064RD). RT was performed according to the manufacturer's instruction (except that 2 μl of Moloney murine leukemia virus reverse transcriptase was used), and 1 μg of EMA-treated total episymbiont RNA was used as the template.

qPCR.

To examine in vivo expression levels of APS leuB and APS gdhA in the A. pompejana episymbiont consortium, RT-qPCR was used to amplify internal fragments of said genes from the episymbiont cDNA pool generated as described above. RT-qPCR primers based on verified clone sequences were designed for APS leuB and APS gdhA with Primer3 (31). The primers used are as follows: GDH_qL, 5′-AAAGCCATCCAGAAGCTACC; GDH_qR, 5′-TTTGCACCCTCATTAACCAT; IPMDH_qL, 5′-TGCGACAATCTCTAGCCTCT; and IPMDH_qR, 5′-CCCATCTCAGTAGAGGAGCA. The product sizes are 150 bp for APS gdhA and 165 bp for APS leuB, and the predicted T_m for all primers is 58 ± 0.3°C.

qPCRs were carried out with a final volume of 25 μl containing either 1 μl of genomic DNA, cDNA, or plasmid template, 100 nM each respective primer, 100 μM deoxynucleoside triphosphates, 5 mM MgCl₂, 2.5 μl of 10× PCR buffer, 1 U of Taq DNA polymerase (Roche Molecular Diagnostics, Pleasanton, CA), and SYBR green (Invitrogen Corp., Carlsbad, CA). For APS leuB, 1 μg/ml PCR grade bovine serum albumin was also added to the reaction mixture to improve reaction performance. qPCR for APS leuB was performed under the following conditions: an initial denaturation at 94°C for 2 min, followed by 45 cycles of denaturation at 94°C for 7 s, annealing at 58°C for 10 s, and extension at 68°C for 10 s with no final extension. qPCR for APS gdhA was performed under the following conditions: an initial denaturation at 94°C for 2 min, followed by 45 cycles of denaturation at 94°C for 7 s and annealing/extension at 60°C for 20 s with no final extension. All amplifications were performed with a Rotor-Gene 6000 Real-Time Rotary Analyzer (Corbett Life Science, Sydney, Australia), which automatically determines the C_T (cycle threshold) values (the cycle number where the target signal level reached a preset threshold during qPCR) for the reactions. Fluorescence readings were acquired immediately after the extension step on the green channel with maximal gain.

For unknown samples, 1 μl of undiluted cDNA per genomic DNA preparation was used. Standard curves were created separately for APS leuB and APS gdhA with a 10-fold dilution series (100 ng to 1 fg) of plasmids containing the respective genes. All unknown samples were analyzed in triplicate.

RESULTS

Candidate genes.

GOIs were selected on the basis of assay availability, potential physiological significance, and sequence representation in the metagenome data set. Two genes, leuB and gdhA, which code for 3-isopropylmalate dehydrogenase (IPMDH; EC 1.1.1.85) and glutamate dehydrogenase (GDH; EC 1.4.1.4), respectively, were chosen, and sequences related to these genes were identified in the metagenome database by using annotation data (COG database number) and sequence homology.

DNA sequence chromatograms associated with the genes were retrieved and assembled locally, and the sequence assemblies were edited to reveal the consensus sequences of each gene and to examine the level of single-nucleotide polymorphism (SNP) present in the genes. Only one consensus sequence was identified for each GOI, although both sequence assemblies contained large numbers of singletons that cannot be incorporated into the prominent contigs even when assembled at a lower stringency (80% similarity). The candidate genes are hereafter referred to as APS leuB and APS gdhA, and PCR cloning primers were designed on the basis of their consensus sequences. Although SNPs were identified in both APS leuB and APS gdhA, none of the SNPs in APS gdhA were noisy (resulting in an amino acid change), whereas APS leuB contained one noisy SNP.

Cloning and expression of GOIs.

Both APS leuB and APS gdhA were successfully amplified from genomic DNA sample AP201. The use of an AP201 genomic DNA sample ensured a definitive correspondence between the sequences in the metagenome database and clones constructed for this study, and all of the genomic DNA samples used in this study were confirmed to contain insignificant amounts of eukaryotic DNA. PCR amplicons of the GOIs were cloned into the pIVEX2.4d vector, and constructed clones were sequenced in both forward and reverse directions. The sequence of the APS gdhA clone corresponded perfectly to the predicted consensus sequence, and those of the APS leuB clones fell into two types distinguishable by one noisy SNP at residue 314 (Lys or Arg). Codon usage analysis of clone sequences showed that the APS leuB and APS gdhA clones contained fewer than 10 amino acids below the 10% relative adaptiveness threshold for E. coli, none of which was in the first 50 amino acids encoded by the genes.

APS IPMDH and APS GDH, the respective overexpression products of APS leuB and APS gdhA clones, were purified with a Qiagen Ni-NTA Spin Kit column and analyzed by SDS-polyacrylamide gel electrophoresis and native protein gel electrophoresis, which confirmed the enzyme preparations to be essentially pure. However, the overexpression product of the APS leuB clone with Arg at residue 314 was nonfunctional; thus, APS IPMDH refers to only the overexpression product of the APS leuB clone with Lys at residue 314.

Identities of cloned genes.

To further verify the identities of the isolated genes, amino acid sequences translated from clone DNA sequences were analyzed by aligning them with homologous genes retrieved from GenBank databases (see Fig. S1 and S2 in the supplemental material), and phylogenetic trees based on these alignments were constructed (see Fig. S3 and S4 in the supplemental material). The closest match for both enzymes came from Sulfurovum sp. strain NBC37-1, a member of the epsilonproteobacteria isolated from a deep-sea hydrothermal field (25). Phylogenetic analysis confirmed that the A. pompejana episymbiont genes were most closely related to those from Sulfurovum sp. (see Fig. S3 and S4 in the supplemental material). APS IPMDH is less than 80% similar to its counterpart found in Sulfurovum sp., while APS GDH and its homolog from Sulfurovum sp. share less than 70% of their amino acid sequences. The results indicate that the two isolated genes were novel and likely of marine epsilonproteobacterial origin.

cDNA generation.

Typically, genomic DNA contamination in RNA extractions is removed by DNase I treatment. For the episymbiont RNA sample, however, a standard DNase I treatment could not completely remove genomic DNA contamination, and PCR amplification of DNase I-treated RNA samples with RT-qPCR primers produced significant amounts of amplicon. EMA is a photoreactive analogue of ethidium bromide that selectively eliminates double-stranded DNA from PCR amplifications by covalently cross-linking with nucleic acid after light exposure (32). By treating RNA samples with EMA, all of the double-stranded genomic DNA was eliminated from further reactions while the single-stranded RNA remained unaffected. PCR amplifications of the EMA-treated episymbiont RNA sample yielded no amplicon detectable by agarose gel electrophoresis, suggesting that the EMA treatment had effectively rendered all of the genomic DNA in the sample unavailable for amplification. The genomic-DNA-free RNA samples were then reverse transcribed with both random decamers (4064RD) and equal amounts of gene-specific reverse primers for both APS leuB and APS gdhA (4064GS). The resulting samples both tested positive for amplifiable cDNA by PCR as described above.

qPCR.

To verify that the GOIs were actively expressed in vivo, qPCR studies of the GOIs were carried out with cDNA samples reverse transcribed from episymbiont total RNA (4064RD and 4064GS), as well as episymbiont genomic DNA samples (AP201 and AP243), as templates; quantified clone plasmid DNA was used as standards. Table 1 lists the qPCR results, including the C_T values and calculated copy numbers of each gene in all samples, and the reaction statistics presented in Table 2 demonstrate the robustness of both assays. Compared to the qPCR assay for APS gdhA, the qPCR assay for APS leuB was more prone to primer dimer formation and consequently loss of linearity at low template concentrations. Therefore, the APS leuB qPCR assay had a relatively high background signal and a C_T value for the negative control substantially lower than that of the APS gdhA qPCR assay (Fig. 1). However, both standard curves were extensive enough to cover all unknown samples, and the signal levels of all unknown samples were significantly higher than those of the negative controls (Fig. 1).

TABLE 1.

C_T values and copy numbers of APS gdhA and APS leuB in A. pompejana episymbiont genomic DNA and cDNA samples

Sample	CT (mean ± SE)a		No. of copies/μl (103) (mean ± SE)		APS gdhA/APS leuB ratiob
	APS gdhA	APS leuB	APS gdhA	APS leuB
AP201 genomic DNA	19 ± 0.1	18.8 ± 0.2	70 ± 7	1,860 ± 430	26.7
AP243 genomic DNA	19.6 ± 0.2	19.4 ± 0.2	47 ± 12	1,234 ± 212	26.0
4064GS cDNA	12.4 ± 0.1	15.8 ± 0.2	6,564 ± 760	15,007 ± 2,929	2.3
4064RD cDNA	23 ± 0.1	23.9 ± 0.1	4 ± 1	53 ± 8	12.3

^aC_T value is the cycle number where the target signal level reached a preset threshold during qPCR.

^bRatios of APS gdhA to APS leuB were calculated from average copy numbers.

TABLE 2.

Correlation coefficients, slopes, and efficiencies of standard curves of the qPCR assays

Assay	r2	Slope	Efficiency
APS gdhA	0.998	−3.33	0.997
APS leuB	0.996	−3.31	1.003

FIG. 1.

qPCR plot. Symbols: ○, APS gdhA plasmid standards (1 ng to 10 fg); •, APS leuB plasmid standards (1 ng to 100 fg); □, APS gdhA 4064GS cDNA; ▪, APS leuB 4064GS cDNA; +, APS gdhA 4064RD cDNA; white + in black square, APS leuB 4064RD cDNA; ▵, APS gdhA AP201 genomic DNA; ▴, APS leuB AP201 genomic DNA; ⋄, APS gdhA AP243 genomic DNA; ⧫, APS leuB AP243 genomic DNA; ×, APS gdhA negative control; white × in black square, APS leuB negative control.

The ratios of APS leuB to APS gdhA in AP201 and AP243 were nearly identical, suggesting that APS leuB is more abundant than APS gdhA in the metagenome. Meanwhile, the ratios of APS leuB to APS gdhA in 4064RD and 4064GS cDNA metatranscriptomic samples showed some disparities, which may have arisen from fundamental differences between the ways the two cDNA samples were generated. In both cases, the ratios indicated that the relative levels of APS gdhA mRNA were higher than that of genomic DNA. Overall, results from both cDNA preparations confirmed that both genes are expressed in vivo, with APS leuB potentially at a collectively higher level than APS gdhA in the epsilonproteobacterial consortium.

Enzyme properties.

The purified APS GDH and APS IPMDH enzymes were characterized to obtain their kinetic parameters. APS GDH is a relatively stable (high ΔG^‡_inact) enzyme. More than 90% of its activity remained after 24 h of incubation at room temperature (data not shown), and its apparent loss of activity after 10 min at 70°C is less than 50% of the initial activity at the same temperature (see Fig. 3). Meanwhile, APS IPMDH is significantly less stable since it denatured significantly at temperatures above 60°C and completely and irreversibly lost activity at 70°C within 1 min.

FIG. 3.

Cross sections of 3D activity plots at zero time and 10 min. Solid line, APS GDH activity plot at zero time. Dashed line, APS GDH activity plot at 10 min. Dotted line, APS IPMDH activity plot at zero time. Dash-dot-alternating line, APS IPMDH activity plot at 10 min. The 10-min plots were extrapolated on the basis of data presented in Table 3 and Fig. 2.

Equilibrium model parameters and plots.

Temperature profiles of both APS GDH and APS IPMDH were obtained and fitted to the equilibrium model, and the resulting equilibrium model parameters, which were derived directly from raw data, are listed in Table 3. The free energy of activation of the catalytic reaction (ΔG^‡_cat) for APS IPMDH is 70 kJ·mol⁻¹, and that for APS GDH is 58 kJ·mol⁻¹. Simulated 3D plots of both enzymes generated by using the parameters listed in Table 3 are shown in Fig. 2. The contrast in the widths of the peaks between the two plots highlights the different ranges of temperatures over which these enzymes work optimally, and the positions of the activity peaks suggest that both enzymes are adapted for working at relatively high temperatures. Zero-time activity plots of the enzymes are shown in Fig. 3; they are the zero-time cross sections of the plot in Fig. 2 but are shown here to further demonstrate the difference in temperature adaptation of the two enzymes. Also shown in Fig. 3 are cross sections of 3D enzyme activity plots at 10 min, representing a more physiologically significant comparison. It is clear from Fig. 3 that APS GDH works over a significantly wider and higher range of temperatures compared to APS IPMDH.

TABLE 3.

Thermodynamic parameters^a

Enzyme or parameter	ΔG‡inact (kJ·mol−1)	ΔHeq (kJ·mol−1)	Teq (°C)	ΔG‡cat (kJ·mol−1)
APS GDH	101	93	81	58
APS IPMDH	92	793	67	70
25th percentileb	91	112	53	NAc
75th percentileb	97	394	62	NA
Mean ± SDb	94 ± 4	277 ± 218	56 ± 9	NA

^aΔG^‡_inact is the free energy of activation of the thermal denaturation process. ΔH_eq is the enthalpic difference between E_act and E_inact. T_eq is the temperature at which the concentration of E_act equals that of E_inact.

^bCalculated from mesophilic enzyme data from reference 24.

^cNA, not applicable.

FIG. 2.

APS GDH (A) and APS IPMDH (B) 3D activity plots. The plots show the effects of time and temperature on enzyme activity, determined by fitting the experimental data to the equilibrium model.

DISCUSSION

Two genes have been identified and isolated from the A. pompejana episymbiont metagenome, and heterologous overexpression of the isolated genes has resulted in functional enzymes. The product of APS leuB, APS IPMDH, catalyzes the oxidative decarboxylation of 3-isopropylmalate to 2-oxoisocaproate, the penultimate step in leucine biosynthesis. Although the role of IPMDH is confined to a peripheral pathway, it is widely found in all types of organisms (26, 38). APS GDH is the expressed product of APS gdhA and plays central roles in nitrogen, amino acid, and glutamine metabolism, forming a key link between catabolic and metabolic pathways, and is therefore ubiquitous in most organisms (20).

APS IPMDH and APS GDH were characterized by using the equilibrium model, and their thermodynamic parameters (ΔG^‡_inact, ΔH_eq, T_eq, and ΔG^‡_cat) were compared with reference data compiled from parameters of 14 mesophilic enzymes (from organisms with optimal growth temperatures between 20°C and 40°C) surveyed in an earlier study (24) (Table 3). APS GDH and APS IPMDH are very different enzymes in terms of both temperature-activity response and stability, and the most significant difference is perhaps that between their ΔH_eq values (Table 3). ΔH_eq is the enthalpic change associated with the reversible, temperature driven interconversion between E_act and E_inact and has been established as a quantitative measure of eurythermalism at the enzymatic level (24); a large ΔH_eq implies stenothermal behavior (i.e., a sharp decline in activity with increasing temperature), and vice versa. The small ΔH_eq of APS GDH predicts eurythermal behavior, and the large ΔH_eq of APS IPMDH implies a very narrow operating temperature range for the enzyme. However, the physiological significance of ΔH_eq cannot be elucidated without considering ΔG^‡_inact, since the usefulness of an enzyme in situ depends on its stability as much as on its activity.

The exceptionally high ΔG^‡_inact of APS GDH clearly indicates that it is a thermally stable enzyme, as suggested by its half-life at high temperatures (half-life at 70°C, >10 min) and resistance to activity loss at room temperature. This combination of low ΔH_eq and high ΔG^‡_inact suggests that APS GDH is a highly stable enzyme capable of working across a relatively wide range of temperatures (i.e., eurythermal), as shown in Fig. 2A and 3. This corresponds ostensibly to the eurythermal and thermotolerant nature of the A. pompejana episymbionts. APS IPMDH appears to work optimally within a relatively narrow range of temperatures, as inferred from its high ΔH_eq. Its ΔG^‡_inact is also significantly below the average ΔG^‡_inact of enzymes isolated from mesophiles in reference 24 (94 kJ·mol⁻¹), suggesting that other mechanisms of adaptation (e.g., high turnover, temperature-linked gene expression) may be employed by the episymbionts to adapt functions associated with APS IPMDH to the high temperature fluctuations it regularly experiences or that the episymbionts can tolerate brief periods of inaction.

T_eq is the temperature at which [E_act] is equal to [E_inact], and it may be helpful to conceptualize it as the thermal equivalent of K_m for E_act-E_inact equilibrium. Although a difference of 14°C exists between the T_eq values of the two enzymes, such discrepancies (i.e., variation in T_eq among enzymes from the same organism) have been previously observed (24) and seem common. Predictions based on T_eq are tentative but better than those based on ΔG^‡_inact (24). Since the average T_eq of the reference data is 56 ± 9°C, the T_eqs of both APS GDH and APS IPMDH are above the normal T_eq range of mesophilic enzymes. This implies that, overall, the A. pompejana episymbionts are adapted to a relatively thermophilic lifestyle. As described previously (24), the mean difference between T_eq and optimal growth temperature is 21 ± 15°C, and this gap closes with increasing optimal growth temperature. Considering that APS IPMDH is stable at temperatures of up to 60°C (Fig. 2B) and that its apparent activity at 50°C after 10 min is less than half of the maximum (Fig. 3), APS IPMDH is likely to work mostly in the temperature range of 50 to 60°C and the operational temperature range of APS GDH is potentially higher and wider (Fig. 3).

Stability is usually, but not necessarily, linked to temperature adaptation, and it has been suggested that ΔG^‡_cat has a weak positive correlation with stability/ΔG^‡_inact (24). The facts that APS GDH has a ΔG^‡_cat higher than that of Candida utilis GDH and that the average ΔG^‡_cat of IPMDHs in the earlier survey (24) is higher than that of APS IPMDH appear to support previous observations. The free energy of activation of the catalytic reaction (ΔG^‡_cat) for APS IPMDH is 70 kJ·mol⁻¹, close to the average value of three IPMDHs (74 kJ·mol⁻¹) described in the earlier survey (24), and APS GDH has a ΔG^‡_cat of 59 kJ·mol⁻¹, close to that of C. utilis GDH (57 kJ·mol⁻¹) described previously (24).

The qPCR results in Table 1 suggest that the total copy number of APS leuB is much higher than that of APS gdhA in the metagenome, and the ratios of APS leuB to APS gdhA in two individual genomic DNA samples agree remarkably well. Although RT-qPCR results from 4064GS and 4064RD cDNA preparations both suggested that APS leuB mRNA is more abundant in the collective community, results from the two data sets do not agree on the ratio of APS leuB to APS gdhA; this difference may have come from the different ways in which the cDNA samples were prepared. The 4064GS cDNA sample was generated with gene-specific reverse (antisense) qPCR primers for both APS leuB and APS gdhA during the RT step. cDNA generated by this approach may be more adequate for metagenomic applications since the resulting target cDNA concentrations are much higher (Table 1) compared to the 4064RD cDNA sample, which was generated by using random decamer nucleotides (supplied with the Ambion RETROscript kit) as RT primers. In 4064GS cDNA, the final ratio of APS gdhA cDNA versus APS leuB cDNA should, in theory, closely reflect the ratio of two mRNA populations in the total RNA sample since RT is a linear reaction performed at temperatures far below the T_ms of both primers and thus relatively insensitive to different primer efficiencies.

It may be relevant to interpret RT-qPCR results in the context of the enzymes' properties, and although mRNA levels do not 100% reflect in vivo protein levels (15, 30), they do provide a glimpse into intracellular protein expressions. The presence of APS gdhA transcript in the episymbiont mRNA pool is expected, since it is a housekeeping gene. The eurythermal and thermophilic nature of APS GDH also agrees with the high environmental temperature variability and occasional temperature spikes likely experienced by the episymbionts. Meanwhile, the observation that APS leuB exists in 25 times higher abundance than APS gdhA in the metagenomic DNA implies that the episymbiont consortium, as a whole, may be capable of expressing APS leuB at a high level rapidly under suitable circumstances, such as when the host, A. pompejana, resides in a lower-temperature area for an extended period of time.

The controversy between the high-temperature characteristics of A. pompejana's natural habitat and the absence of evidence for thermophilic and eurythermal adaptation in the worm itself so far may lie in the thermal adaptations of its episymbionts. During a temperature spike in the vent flow, the internal temperature of the host worm equilibrates less rapidly than the milieu surrounding its episymbionts; therefore, the episymbionts may be exposed to more direct and severe temperature changes. Although there is currently little information on the physical protection that the episymbionts and the associated biofilm may provide to the host, much can be learned about the thermal tolerance of the A. pompejana episymbionts from the temperature adaptation of episymbiont enzymes. The very stable and eurythermal APS GDH enzyme may indicate that the episymbiont consortium is capable of withstanding a steep temperature gradient and the frequent high temperature spikes characteristic of the habitat. However, it must also be pointed out that host behavior and other acclimatization mechanisms are also likely to be important since APS IPMDH is unlikely to function sustainably in the same manner as APS GDH. The high pressure of the episymbionts' native environment may also contribute to the adaptation of its proteins, although this hypothesis was not tested here. Additionally, the facts that only one dominant consensus sequence could be identified for leuB and gdhA and that the consensus sequences themselves lack heterogeneity at the amino acid level potentially reflect the structure of the episymbiont community, which is dominated by a few closely related phylotypes.

The identification, isolation, overexpression, and characterization of novel enzymes from the A. pompejana epsilonproteobacterial episymbionts have revealed relevant and useful information about the episymbionts. The use of the equilibrium model has provided insights on enzyme properties that were previously unavailable, and metatranscriptomic analysis of the GOIs by RT-qPCR has not only provided information on the in vivo expression of the GOIs but has also led to a hypothesis on intricate relationships between A. pompejana and its episymbionts. Overall, the findings are in harmony with the general consensus that a fine balance between stability and flexibility (i.e., thermolability) is a consistent feature of the evolution of virtually all proteins (36) and that temperature adaptations occur at multiple levels of cellular functions (34). Lastly, this study has shown how insights into the temperature adaptation of a complex microbial community can be gained by retrieving and expressing GOIs from its metagenomic sequences through bioinformatic investigations and characterizing the resulting enzymes by using the equilibrium model.