Molecular foundations of Precambrian uniformitarianism

Abstract

Uniformitarian assumptions underlie the oldest evidence for living organisms on Earth, the distinct isotope fractionation between inorganic and organic carbon. Aside from a handful of compelling deviations, the ¹³C/¹²C isotopic mean of preserved organic carbon (δ¹³C_org) has remained remarkably unchanged through time. RuBisCO is the principal carboxylase/oxygenase biomolecular component that is thought to primarily account for the generation of these distinct carbon isotopic signals. However, it is difficult to reconcile a mostly unchanging mean δ¹³C_org with several known factors that can affect the isotope fractionation of RuBisCO, such as atmospheric composition and the amino acid composition of the enzyme itself, which have each changed markedly over Earth history. Here we report the resurrection and genetic incorporation of a Precambrian-age, Form IB RuBisCO in a modern cyanobacterial host. The isotopic composition of biomass relative to CO₂ (ε_p) in ancestral and control strains were much greater when grown under Precambrian CO₂ concentrations compared to modern ambient levels, but displaying values within a nominal envelope of modern-day RuBisCO IB enzyme variants. We infer that these isotopic differences derive indirectly from the decreased fitness of the AncIB strain, which includes diminished growth capacity and total cell RuBisCO activity. We argue that to answer the greatest questions of deep-time paleobiology, ancient biogeochemical signals should be reproduced in the laboratory through the synthesis of the geologic record with experimentally-derived constraints on underlying ancient molecular biology.

Significance Statement The earliest geochemical indicators of microbes, and the enzymes that powered them, extend back almost 3.8 billion years on our planet. Paleobiologists often attempt to understand these indicators by assuming that the behaviors of modern microbes and enzymes are consistent (uniform) with those of their predecessors. This assumption seems uncomfortably at odds with the great variability of Earth’s environment and its highly adaptive microbes. Here we examine whether a uniformitarian assumption for an enzyme thought to generate these indicators, RuBisCO, can be corroborated by independently studying the history of changes recorded within RuBisCO’s genetic sequences. We outline a new approach to paleobiology that informatively links molecule-level evolutionary changes with planet-level geochemical conditions in Earth’s deep past.

Introduction

The history of life on Earth may be broadly subdivided into two, mutually exclusive macroevolutionary phases, the Phanerozoic and the Precambrian (1). The Phanerozoic (~542 million years ago to present) may be characterized by physiological and anatomical innovations and their resultant effects on ecosystem expansion, trophic tiering, and sociality (2–4). Hard- and soft-anatomical preservation provides a rich template for reconstructing Phanerozoic adaptive trends, correlating them with geographical and climatological changes (5, 6), and for testing observed diversity trends against possible systematic effects of preservation bias (7–9). By contrast, the Precambrian (the ~4 billion years preceding the Phanerozoic) is primarily characterized by genetic and metabolic biomolecular innovations, traded amongst microscopic organisms of uncertain phylogenetic assignment (10, 11). The Precambrian record of evolutionary change appears to be cryptic and may have been comparatively static. This may be attributable to macroevolutionary dynamics that were distinctly non-Phanerozoic, or it may merely indicate a lack of direct paleontological and geological evidence of the specific timing and extent of intermediate biomolecular adaptive steps (1).

Comparative analyses of extant organisms have traditionally been the most informative means of interpreting the scant direct evidence of Precambrian life, but such analyses inevitably face pitfalls. A reasonable null hypothesis is that evolution is largely a uniformitarian process, such that rates or tempos may change but the underlying processes or modes of evolutionary change (as established through observations of the more comprehensive Phanerozoic record) are likely invariant in deep time (12). Uniformitarian assumptions of ancient biology inferred from extant or Phanerozoic phenotypes are often employed to make sense of the Precambrian record, including body plan function (13, 14), cladistic assignment (15–18), and isotope biosignature traces (19–23). However, more recent studies complicate the use of uniformitarian assumptions, namely indicating that modes of biological variation can actually vary through time (12). A major crux of the problem is that even the simplest modern organisms, as well as the macromolecules that compose them, differ from their ancient predecessors in having been shaped by the cumulative effects of billions of years of Earth-life co-evolution and ecosystem upheaval. For this reason, Precambrian functional or sequence comparisons may be of limited utility at organismal or biomolecular levels of adaptation, undercutting interpretations made possible by uniformitarian assumptions. Novel experimental approaches may help to distinguish inferred paleobiological phenotypes from characteristically modern adaptive overprints.

The interpretation of the Precambrian carbon isotope record, comprising the oldest signatures of life on Earth, may be aided by novel experimental constraints on ancient phenotypes. Interpretations of this record is conventionally subject to uniformitarian assumptions regarding ancient biogeochemistry (19, 24). The ¹³C/¹²C isotopic mean of preserved organic carbon (δ¹³C_org ≈ −25‰) has remained notably static over geologic time (19, 21, 24), and is leveraged as a general signature of ancient biological activity (25–27). Given its role in the Calvin-Benson-Bassham cycle—likely the predominant mode of carbon fixation for much of Earth history (19)—the majority of contextual information used to assess how carbon isotope biosignatures might have been generated over Earth’s deep history comes from studies of the modern enzyme RuBisCO (Ribulose 1,5-Bisphosphate (RuBP) Carboxylase/Oxygenase, EC 4.1.1.39). RuBisCO catalyzes the uptake of inorganic CO₂ from the environment and facilitates CO₂ reduction and incorporation into organic biomass. The ¹³C/¹²C isotopic fractionation of modern Form I RuBisCO variants in photosynthetic organisms consistently measures ~-25‰ (28–31), approximately the same isotopic difference observed between inorganic and organic carbon in the Precambrian rock record. The carbon isotope discrimination of RuBisCO has therefore been presumed to have remained constant over the history of life. Recent data, however, demonstrate that RuBisCO can produce significantly different carbon isotope signatures within organic matter in response to external factors, such as levels of atmospheric CO₂ and/or O₂ (32–35) or cellular carbon concentrating mechanisms, which affect the catalytic efficiency of RuBisCO (36–38). Internal factors may also affect fractionation, such as single-point mutations that can alter the interaction between RuBisCO and CO₂ (39). Given the sensitivity of RuBisCO to both external and internal variables, it seems unlikely that ancestral forms under ancient environmental conditions generated the same isotope fractionation signal as descendent homologs in modern organisms. An experimental assessment that combines the phylogenetic history of RuBisCO with the study of intra- and extracellular conditions may provide a more insightful basis for comparing extant and Precambrian carbon isotope fractionation patterns.

The Form I clade of the RuBisCO phylogeny (and its macroevolutionary tractability afforded through green plant, algal and cyanobacterial fossil lineages) makes it an exemplary paleomolecular system for assessing uniformitarian assumptions applied to Precambrian biosignatures. Here, we establish an experimental system for the reconstruction of ancestral biomolecules with which to interpret evidence of Precambrian biological activity. Specifically, we report the resurrection and genetic incorporation of a phylogenetically reconstructed, ancient Form IB RuBisCO variant in a modern strain of cyanobacteria, Synechococcus elongatus PCC 7942 (thereafter S. elongatus) (40, 41). We compared expression and activity levels of RuBisCO variants and the resulting changes in the growth of S. elongatus and isotope fractionation under ambient air as well as a CO₂ concentration that reflects Precambrian conditions.

Results

Construction of a S. elongatus strain harboring ancestral RuBisCO

To experimentally investigate the generation of carbon isotope biosignatures in deep time, we designed a paleomolecular system to engineer computationally inferred, ancestral RuBisCO enzymes in extant cyanobacteria. We previously reconstructed a comprehensive phylogenetic history of RuBisCO and inferred maximum-likelihood ancestral RuBisCO large-subunit (RbcL) protein sequences (42) (Fig. 1A). For this study, we selected the ancestor of the Form IB RuBisCO clade (cyanobacteria, green algae, and land plants) for laboratory resurrection, designated “AncIB.” Chlorophyte and land plant RuBisCO homologs are nested among cyanobacterial sequences within the Form IB clade. The Form IB topology therefore recapitulates a primary plastid endosymbiotic history from cyanobacterial to Chlorophyte ancestors (43–45) and constrains the minimum age of ancestral Form IB to older than the Archaeplastida. As a conservative estimate, AncIB is thus likely older than ~1 Ga (the age of the oldest well-characterized, crown-group red and green algal fossils (16, 46)) and younger than maximum age estimates of cyanobacteria (~3 Ga, as constrained by oxidized sediments potentially indicating the early presence of oxygenic photosynthesis (47, 48)).

Figure 1.

Reconstruction of ancestral RuBisCO biogeochemical signatures. (A) Maximum likelihood Form I RuBisCO RbcL phylogeny (derived from full RuBisCO phylogeny described in Kacar et al. (42)). Ancestral AncIB node and descendent Form IB clade highlighted in blue. Approximate likelihood ratio (aLR) branch support indicated by asterisks (**: >10, ****: >1000). Carbon isotope record figure adapted from Garcia et al. (21), with data from Schidlowski et al. (25) (grey) and Krissansen-Totton et al. (24) (dark grey). Approximate age range of AncIB indicated by blue field (see text for discussion). (B) Genetic engineering of S. elongatus strains. Strain Syn01 was constructed by inserting a second copy of the rbc operon in the chromosomal neutral site 2 (NS2). Strain AncIB was constructed by inserting the genetic sequence encoding for the ancestral AncIB rbcL within the NS2 rbc operon. The native rbc operon was removed in both strains Syn01 and AncIB. (C) Photosynthetic carbon isotope fractionation (ε_p) of S. elongatus strains in this study, cultured in ambient air or 2% CO₂. n = 3 for each data point and error bars indicate 1σ (error bars smaller than some datapoints).

The ancestral AncIB and S. elongatus native RbcL proteins differ at 37 sites and share 92% amino acid identity (Fig. 2). This site variation is evenly spread across the length of the protein, except for a highly conserved region between approximately site 170 to 285 (site numbering here and hereafter based on aligned WT S. elongatus RbcL; Fig. 2A) that constitutes a portion of the catalytic C-terminal domain and is proximal to the L-L interface and active site. Critical residues for carboxylase activity, including the Lys-198 site that binds CO₂, are conserved in AncIB. Homology modeling of AncIB using the S. elongatus RbcL template (PDB: 1RBL (49)) indicates high structural conservation without predicted disruption to secondary structure (Fig. 2B). The nucleotide sequence for the reconstructed AncIB RbcL protein was codon-optimized for S. elongatus and cloned within a copy of the rbc operon into pSyn02 (50) for insertion into the S. elongatus chromosomal neutral site 2 (NS2). The native rbc operon was subsequently deleted to create the AncIB strain. In addition, we generated the control strain Syn01 harboring WT rbcL at NS2 (see Materials and Methods for full genome strategy). Thus, the AncIB and Syn01 strains carry a single ectopic copy of the rbc operon at NS2 and only differ in the coding sequence of the large RuBisCO subunit (Table 1).

Figure 2.

Structure and sequence features of ancestral RuBisCO. (A) Amino acid sequence alignment between ancestral AncIB and extant S. elongatus RbcL. Ancestral site variation relative to the S. elongatus template is highlighted in red. (B) Modeled structure of the ancestral AncIB L2 dimer (blue), aligned to the active conformation of the extant S. elongatus L₈S₈ hexadecamer (grey; PDB: 1RBL (49)). Highlighted residues in (A) are also highlighted in (B). Site numbering from extant S. elongatus. Conserved residues are indicated by dots and secondary structure is indicated above the sequences (blue rectangle: α-helix; blue arrow: β-sheet).

Table 1.

Strains and plasmids used in this study.

Ancestral RuBisCO complements photoautotrophic growth of extant S. elongatus

We cultured wild-type (WT) and engineered (AncIB and Syn01) strains of S. elongatus in both ambient air and 2% CO₂ to evaluate the physiological impact of ancestral RuBisCO under estimated Precambrian CO₂ concentrations (51) (Fig. 3A). The AncIB strain was capable of photoautotrophic growth in both ambient air and 2% CO₂. Maximum growth rates for all strains were relatively comparable under each atmospheric condition (averaging doubling times of ~15 to 20 hours), and generally increased under 2% CO₂ relative to air (p < 0.001; Table 2). The AncIB strain exhibited a significantly diminished maximum cell density (OD₇₅₀ ≈ 5) relative to both WT and Syn02 strains (OD₇₅₀ ≈ 8; p < 0.001). No significant difference between WT and Syn01 growth rate or maximum cell density was observed under ambient air or 2% CO₂. In addition, a moderate increase in the midpoint time for the AncIB strain was observed under air, indicating lag in growth (p < 0.01). These growth parameters taken together suggest decreased fitness of the AncIB strain relative to WT and Syn01.

Figure 3.

(A) Growth curves and (B) photosynthetic oxygen evolution of S. elongatus strains cultured in ambient air or 2% CO₂. (A, B) n = 3 for each data point or bar and error bars indicate 1σ.

Table 2.

Growth parameters of S. elongatus strains. Values represent the mean of three replicates ± 1σ. Asterisks indicate significance relative to WT for the same atmospheric condition, determined by one-way ANOVA and post-hoc Tukey HSD tests (**: p < 0.01; ***: p < 0.001).

AncIB RuBisCO protein produced a more modest impact on the oxygen evolution of S. elongatus compared to that on growth parameters. Cell suspensions were briefly incubated in the dark and subsequently exposed to saturated light in an electrode chamber to detect evolution of molecular oxygen (normalized to chlorophyll a concentration (52)). For cells sampled from cultures grown in air, there was no statistical difference detected between any of the WT, Syn01, or AncIB strains (Fig. 3B). We did find a modest but significant decrease in photosynthetic activity for AncIB relative to WT for cells cultured in 2% CO₂, generating ~200 and ~320 nmol O₂·h^-1·μg^-1 chlorophyll a, respectively. However, a slower O₂ evolution rate was also observed for Syn01 at 2% CO₂.

Ancestral RuBisCO is overexpressed and less catalytically active relative to WT

We assessed the impact of ancestral RuBisCO on gene expression at both the transcript and protein levels. S. elongatus rbcL transcript was measured by quantitative reverse-transcription PCR (RT-qPCR) and normalized to that of the secA reference gene (53). For strains cultured in ambient air, we found that the AncIB strain produced a ~29-fold increase in rbcL transcript relative to WT or the control strain Syn01 (p < 0.001; Fig. 4A). The magnitude of AncIB rbcL overexpression was lower in 2% CO₂, with a ~4 and ~2-fold increase observed relative to WT and the control strain, respectively (p < 0.001).

Figure 4.

Expression and total cell activity of RuBisCO in S. elongatus strains. (A) Expression of rbcL in Syn01 and AncIB detected by RT-qPCR (secA reference gene), relative to WT. (B) Immunodetection of RbcL protein. Top, RbcL signal intensities normalized to those for total soluble protein load. Bottom, Western blot showing RbcL protein detected by anti-RuBisCO antibody and total protein stain from crude cell lysates (C) Total cell RuBisCO activity, measured with 2.5 mM and 5 mM HCO₃^- concentrations. (A-C) n = 3 for each bar (except 2% CO₂ RbcL signal intensity data, n = 6) and error bars indicate 1σ.

RbcL protein was quantified for all S. elongatus strains by immunodetection using rabbit anti-RbcL antibody. We found that the amount of RbcL protein was also increased in the AncIB strain by ~3-fold (p < 0.05) and ~5-fold (p < 0.001) relative to WT or Syn01 in air and 2% CO₂, respectively (Fig. 4B). No difference in RbcL quantity was detected between WT and Syn01 strains at either atmospheric condition. Finally, we confirmed assembly of the hexadecameric L₈S₈ RuBisCO complex in the AncIB strain by native PAGE and detection by anti-RbcL antibody (Fig. S1).

The total carboxylase activity of S. elongatus harboring ancestral RuBisCO was measured from crude cell lysates. Activity was assessed by an in vitro spectrophotometric coupled-enzyme assay that measures NADH oxidation and is reported as the RuBP consumption rate normalized to total soluble protein content (54). For two sets of assays using either 2.5 mM or 5 mM HCO₃⁻, AncIB lysate generated less than half the carboxylase activity of WT lysate (p < 0.05) (Fig. 4C).

S. elongatus harboring ancestral RuBisCO produces greater carbon isotopic fractionation than wild-type

We measured the carbon isotope discrimination of S. elongatus strains cultured in ambient air and 2% CO₂ to evaluate how the fractionation behavior of ancestral RuBisCO might influence the interpretation of ancient isotopic biosignatures preserved in the geologic record. The ¹³C/¹²C carbon isotope composition of S. elongatus was measured for biomass (δ¹³C_biomass) as well as dissolved inorganic carbon (DIC; δ¹³C_DIC) in the growth medium (Table S1). We calculated the carbon isotope fractionation associated with photosynthetic CO₂ fixation (ε_p) following Freeman and Hayes (33) after estimating δ¹³C_CO2 from measured δ¹³C_DIC (55, 56) (see Materials and Methods).

Overall, ε_p values were greater for S. elongatus strains cultured in 2% CO₂ compared to ambient air. At 2% CO₂, ε_p ranged between 21‰ and 26‰ compared to only 8‰ and 14‰ in air. We found that S. elongatus engineered with AncIB RbcL had an ε_p ~5‰ greater than both WT and Syn01 when cultured in air (p < 0.001) and ~2‰ to 4‰ greater than WT and Syn01 when cultured at 2% CO₂ (p < 0.01) (Fig. 1C), though these differences in ε_p appear driven by the inorganic carbon pool composition rather than biomass (Table S1). A substantially smaller increase in ε_p (~1‰; p < 0.001) was observed for Syn01 relative to WT under ambient air. Conversely, under 2% CO₂, Syn01 ε_p was decreased by ~2‰ (p < 0.001) relative to WT.

Discussion

Form IB ancestral RuBisCO, when engineered into an extant strain of S. elongatus, decreased both the organismal growth capacity and the total cell RuBisCO activity. The genetic engineering strategy for insertion of the AncIB rbcL sequence in the cyanobacterial genome cannot solely account for these physiological differences since differences between the WT and Syn01 control strains were insignificant for most measured properties, or not comparable in magnitude to differences between the WT and AncIB strains. Rather, the observed differences for the AncIB strain indicate that the resultant phenotype is likely attributable to the functionality of the ancestral enzyme itself. The unique phenotype of the AncIB strain could be a direct result of the ancestral RbcL subunit or due to impediments to the assembly and activation of a hexadecamer RuBisCO complex containing both the ancestral RbcL and modern RbcS subunits. Another possibility is hampered integration of ancestral RbcL given a modern suite of associated proteins required for RuBisCO folding and assembly (57). Further, while overexpression of the ancestral RbcL occurred at the level of transcription and translation, the ancestral strain appears to have comparable levels of assembled hexadecamer RuBisCO, suggesting lower rates of RuBisCO assembly (or faster degradation). Even lower rates of measured total carboxylase activity suggest that the AncIB has decreased efficacy, which could be directly representing ancestral RuBisCO kinetics as well as the challenges associated with hybrid enzyme activation and activity.

Overexpression of the amount of ancestral RuBisCO shown by RT-qPCR and immunodetection assays is a common physiological response to decreased enzymatic efficacy throughout the cell (e.g., (58, 59)). However, expression compensation is insufficient to fully restore the extant WT phenotype, as indicated by the reduced fitness (i.e., decreased maximum cell density, oxygen evolution, and total carboxylase activity) of the ancestral strain harboring the ancestral RuBisCO compared to WT and Syn01.

There are few in vitro measurements of the kinetic isotope effect of Form IB RuBisCO in modern-day organisms, but those available range from ~22‰ to 28‰ for cyanobacteria and C₃ plants, respectively (28–31). The ~26‰ ε_p of AncIB strain biomass grown under 2% CO₂ suggests that the ancestral RuBisCO also fractionates within this range. It has been theorized that RuBisCO kinetics have adapted in response to CO₂ availability, either due to increased environmental CO₂ (36) or the emergence of CCMs (e.g., C4 photosynthesis in plants (60)). Considering the positive relationship between enzymatic fractionation and RuBisCO’s specificity to CO₂ (36), reconstructing ancient RuBisCO kinetic isotope effect could provide insights into the co-evolution of atmospheric concentrations of CO₂ and O₂ and carbon fixation strategies during the Precambrian, in particular the emergence of carbon concentrating mechanisms (CCMs) (34, 61–63). This is relevant as precise estimates of the magnitude of atmospheric CO₂ elevation during the Precambrian relative to the present, as well as the emergence and effectiveness of Precambrian CCMs, are unknown.

We did observe statistically significant differences in ε_p of the ancestral strain compared to WT S. elongatus under both ambient air and 2% CO₂ atmospheric conditions. However, upon inspection it appears that the diminished activity of the AncIB strain is influencing the composition of the DIC pool (both δ¹³C and concentration) in our cultures, and it is in fact the differences in DIC composition driving the calculated differences in ε_p. Though strains were harvested at similar cell densities, small differences in cell concentrations at high densities can strongly influence carbonate chemistry of the media (64). The lower CO₂ availability in the air treatment is more sensitive to cellular influence, resulting in a larger difference in ε_p compared to the 2% CO₂ treatment. These differences in DIC are unlikely to be due to experimental setup (e.g., variations in CO₂ bubbling), as biological replicates showed similar values. Therefore, the differences in fractionation reported here are likely implicated indirectly with the less efficient AncIB ancestral enzyme. Further comparative biomolecular characterization of AncIB and WT S. elongatus RbcL forms is needed to determine the degree to which enzymatic inefficiencies are contributing directly to the AncIB strain phenotype.

The observed carbon isotopic fractionation values corroborate a uniformitarian assumption for applying the maximal range of extant organism-level isotope fractionation values to interpret deep time isotopic biosignatures. There are, however, several potentially important contextual caveats. The most obvious is that the isotopic fractionation values of all strains (WT, Syn01, and AncIB) are increased under simulated Precambrian conditions with elevated CO₂. The Form IB ancestor represents predecessors that are at least 1 billion years old, but it is also genetically and functionally still likely to be very different from the putative ‘root’ or common ancestor of all RuBisCO variants that emerged much earlier. Reconstruction of older ancestors may further expand this maximal envelope of RuBisCO-generated carbon fractionation, or it may indicate that the extant maximal envelope is pervasive (and perhaps characteristic) across all functional variants of RuBisCO.

Another important caveat lies in the observation that, whereas all strains produce increased isotopic fractionation under elevated CO₂, the comparative difference between ancestral AncIB and WT RbcL fractionation is relatively muted under 2% CO₂ relative to ambient air. One possibility is that elevated CO₂ brings the intrinsic fractionation properties of RuBisCO into relief (35, 65–67), at least compared to fractionation effects deriving from the overlying organismal physiology. By contrast, in present-day conditions, RuBisCO-mediated fractionation processes may be more significantly overprinted by physical factors that can affect RuBisCO catalytic efficiency, including cellular diffusion of O₂/CO₂ or other factors such as the presence of carbon concentration mechanisms.

There are many fundamental attributes of extant and ancestral metabolism for which the systemic effects on biosignature production have yet to be characterized. Disentangling these effects is critical for interpretation of the oldest biogeochemical record. A host cyanobacterium S. elongatus engineered with a Form IB RbcL ancestor confirms that organism- and enzyme-level effects on biosignature production are not always synonymous but differ in nuanced ways. These differences are contingent upon changes to internal (cellular, physiological) and external (environmental) conditions that have demonstrably varied over Earth’s long history. Cyanobacteria, with well-characterized genetic and morphological features (40, 41, 61–63) and a tractable paleobiological history (18, 68), are ideal hosts for investigating a range of early Precambrian metabolic processes (68–70).

Discernible trends (or steadfast consistencies) in metabolic outputs over macroevolutionary timescales can lead to foundational uniformitarian approaches to deep time molecular paleobiology. The available rock record becomes vanishingly sparse with greater age, but it is arguably well-sampled across key global-scale and biotically relevant isotopic systems at least through the early Archaean. Greater geologic sampling will therefore likely generate diminishing returns for shedding new light on deep time paleobiological trends. Innovative approaches that can chart a comprehensive envelope of biomolecular variability over time are a promising new means of reconciling coarse geochemical data with the nuance and complexity of ancient biological activity.

The engineering of ancient-modern hybrid organisms and their characterization can be used to complement the existing array of fossil remains, biogeochemical signatures, and modern organismal and molecular proxies to assess and contextualize plausible ranges of Precambrian carbon isotope biosignature production. Hybrid organisms may be particularly useful to disentangle the regulatory, physiological, and inter- and intramolecular factors that have impacted isotope fractionation, none of which are individually expressed in the geologic record. These factors must be systematically accounted for when interpreting bulk fractionation signals, even if only to elucidate the evolutionary molecular underpinnings of uniformitarian phenomena over geologic time.

Conclusions

After engineering a cyanobacterium with an ancient RuBisCO large protein subunit and cultivating it under conditions that mimic those prevailing through much of the Precambrian, we found the resultant carbon isotope fractionation to be within the range of organisms utilizing modern Form IB RuBisCO. The underlying biomolecular and organismal adjustments made by the cell to accommodate the ancestral gene were tracked, and we conclude that the small fractionation differences observed are likely attributable indirectly to decreased fitness of the AncIB strain, which influenced the inorganic carbonate chemistry of the media. The consistency of isotopic signatures generated by this strain indicates that uniformitarian assumptions based on the range of phenotypes of modern RuBisCO variants may apply for Precambrian environmental conditions, but that further study is warranted to discern organism- and enzyme-level trends in carbon isotope fractionation that may extend deeper into the early Precambrian.

Materials and Methods

Inference of ancestral AncIB RbcL protein sequence

A RuBisCO RbcL phylogeny was reconstructed as previously described (42). Briefly, RbcL orthologs were identified from the NCBI protein database by BLAST (sequence dataset and the tree can be found at https://github.com/kacarlab/rubisco). Phylogenetic analysis was performed by Phylobot (71), a web portal that integrates alignment, phylogenetic reconstruction by RAxML (72), and ancestral sequence inference by PAML (73). A maximum likelihood phylogeny was built using a MSAProbs alignment (74) and the best-fit PROTCATWAG model (75, 76), determined by the Akaike information criterion (77). Ancestral states were reconstructed at each amino acid site for all phylogenetic nodes, and gap characters were inferred according to Fitch’s parsimony (78).

Cyanobacterial growth and maintenance

S. elongatus PCC 7942 strains were cultured in BG-11 medium (79) as liquid cultures or on agar plates (1.5% (w/v) agar and 1 mM Na₂S₂O₃·5H₂O). Liquid cultures were grown at 30°C, continuous shaking at 120 rpm, sparged with ambient air or 2% CO₂, and with 115 μmol photon·m^-2·s^-1 (with the exception of cultures used to prepare samples for O₂ evolution, which were grown with 80 μmol photon·m^-2·s^-1). The 2% CO₂ gas mix was controlled by an environment chamber (Percival, Cat. No. I36LLVLC8) with a CO₂ tank input. For recombinant strains, liquid and solid media were supplemented with appropriate antibiotics: 2 μg·ml^-1 Spectinomycin (Sp) plus 2 μg·ml^-1 Streptomycin (Sm), 5 μg·ml^-1 Kanamycin (Km). Cyanobacterial growth was measured at an optical density of 750 nm (OD₇₅₀) and growth parameters were estimated using the Growthcurver package for R (80) (Growthcurver analysis script can be found at https://github.com/kacarlab/rubisco). Cultures were sampled at the middle exponential growth phase, i.e., at an OD₇₅₀ of ~2.5 (AncIB) or ~4.5 (WT and Syn01) for all subsequent experiments.

Genetic engineering of cyanobacteria

Recombinant strains of S. elongatus were constructed by natural transformation using standard protocols (81) with minor modifications (50). The plasmids and strains used in this study are listed in Table 1. The construction of S. elongatus Syn01, carrying a single ectopic copy of the rbc operon at NS2, as well as the plasmids pSyn01 and pSyn02 used to construct strain Syn01, were described previously (50). The construction of strain AncIB was performed similarly to Syn01. Briefly, pSyn03, which carries the AncIB nucleotide sequence within the entire rbc operon (including flanking sequences and homologous regions for recombination at neutral site 2 (NS2) of the S. elongatus chromosome), was transformed in S. elongatus. Transformation of WT S. elongatus with pSyn03 generated strain Syn03 carrying a second copy of the rbc operon at NS2. Strain Syn03 was subsequently transformed with pSyn01 to replace the native rbc operon with a spectinomycin/streptomycin resistance gene as described previously (50), producing strain AncIB. Transformants of Syn03 and AncIB were screened for complete segregation by colony PCR using primers F06, R06, F07, and R07 (Fig. S2; Table S2) and the strain sequences at the deletion and insertion sites were further verified by Sanger sequencing using the primers R07, F08, R08, F15, and F16 (Table S2). To construct plasmid pSyn03, pSyn02 excluding the rbcL coding sequence was PCR-amplified and linearized using primers F13/R13 and assembled with the AncIB RbcL coding sequence codon-optimized for S. elongatus and synthesized by Twist Bioscience. Both DNA fragments were assembled using the GeneArt™ Seamless Cloning and Assembly Kit (Invitrogen, Cat. No. A13288).

Analysis of rbcL expression by RT-qPCR

Cells were pelleted by centrifugation and resuspended in TE buffer (10 mM Tris, pH 8.0, 1 mM EDTA). Total RNA was extracted using the RNeasy® Protect Bacteria Mini Kit (QIAGEN, Cat. No. 74524). DNase I-treated RNA was then used in reverse transcription (RT) performed with the SuperScript™ IV First-Strand Synthesis System (Invitrogen, Cat. No. 18091050). F09/R09, F11/R11, F14/R14 pairs of qPCR primers (Table S2) were designed with Primer3Plus (http://www.bioinformatics.nl/cgi-bin/primer3plus/primer3plus.cgi). The quality of cDNA and primer specificity was assessed by PCR using cDNA templates (RT positive reactions) and RT negative controls. qPCR was performed by the real-time thermal cycler qTOWER³ G (Analytik Jena AG) using qPCRsoft software. The relative expression of native and AncIB rbcL was calculated as the average fold change normalized to the secA reference gene (53) using the delta-delta Ct method. The experiment was carried out using three biological replicates and three technical replicates.

Immunodetection of RbcL protein

Cells were pelleted by centrifugation and resuspended in 95°C TE buffer supplemented with 1% (w/v) SDS and incubated at 95°C for 10 min. The mixture was sonicated and centrifuged to remove cell debris. Total protein concentration in the crude cell lysates was measured using the Pierce™ BCA Protein Assay Kit (Thermo Scientific, Cat. No. 23225). Lysates containing 5 μg of total protein in Laemmli sample buffer were loaded onto a 6% (v/v) polyacrylamide stacking gel. Proteins were electrophoresed in a 12% polyacrylamide resolving gel and blotted onto a nitrocellulose membrane. Detection of RbcL and total protein load was performed as previously described (50). The densitometric analysis of RbcL signal intensity, normalized to total protein load, was performed with Quantity One® software (Bio-Rad) for three to six biological replicates.

Confirmation of RuBisCO assembly

Assembly of the RuBisCO large and small subunits into a hexadecameric complex in each strain was evaluated by native gel electrophoresis and immunodetection, as previously described (50). Immunodetection of the RuBisCO complex was performed for three biological replicates with the same primary and secondary antibodies that were used to detect RbcL, as described above.

Catalytic activity of RuBisCO

The activity of RuBisCO in cyanobacterial lysates was measured using a spectrophotometric coupled-enzyme assay that links this activity with the rate of NADH oxidation (82). Cell lysis and the activity assay were carried out as previously described (50) with either 2.5 mM or 5 mM NaHCO₃. After 20 min at 25 °C for activation of Rubisco, the reaction was initialized with the addition of ribulose 1,5-bisphosphate (RuBP) (0.5 mM) and the absorbance at 340 nm was recorded using a Synergy H1 plate reader (BioTek). RuBisCO activity was reported as the RuBP consumption rate normalized to total soluble protein content. The assay was performed for three biological replicates.

Photosynthetic oxygen evolution rate

S. elongatus strain photosynthetic activity was assayed using a Clark-type oxygen electrode chamber to measure the level of molecular oxygen produced in cyanobacterial cultures. Cells were pelleted and resuspended in fresh BG-11 to an OD₇₅₀ of ~1 following De Porcelinis (83). Concentration of chlorophyll a (for normalization) was measured following the protocol by Zavrel et al. (84). The remaining suspension was incubated in the dark for 20 min with gentle agitation. Samples from each suspension were analyzed in an oxygen electrode chamber under saturated light, using the Oxygraph+ System (Hansatech Instruments) equipped with the OxyTrace+ software. Oxygen evolution rate was monitored for 10 min and expressed as nanomoles of molecular oxygen evolved per hour per microgram of chlorophyll a. The assay was performed for three biological replicates.

Carbon isotope fractionation in bulk cyanobacterial biomass

Cells were pelleted by centrifugation and washed in 10 mL of 10 mM NaCl (OD₇₅₀ for Syn-1 ~ 4.5, OD₇₅₀ for AncIB ~2.5). Pellets were then dried at 50°C. In parallel, the supernatant from centrifuged culture samples was sterilized through 0.2 μm filtration for DIC isotopic analysis of growth media. Sterilized media was transferred to Exetainer vials leaving no headspace and stored at 4°C until analysis. Isotopic analysis was performed for three biological replicates. The carbon isotope composition of bulk biomass (δ¹³C_biomass) and DIC (δ¹³C_DIC) was determined at the UC Davis Stable Isotope Facility. δ¹³C_biomass was analyzed using a PDZ Europa ANCA-GSL elemental analyzer interfaced to a PDZ Europa 20-20 isotope ratio mass spectrometer (Sercon Ltd.). DIC samples were analyzed by gas evolution and composition was measured by a GasBench II system interfaced to a Delta V Plus IRMS (Thermo Scientific). The carbon isotopic composition values were reported relative to the Vienna PeeDee Belemnite standard (V-PDB):

The isotopic composition of dissolved molecular CO₂ (δ¹³C_CO2) was estimated from δ¹³C_DIC following Rau et al. (55) and Mook et al. (56):

The carbon isotope fractionation associated with photosynthetic CO₂ fixation (ε_p) was calculated relative to δ¹³C_CO2 in the post-culture medium according to Freeman and Hayes (1992):

Statistical analyses

Results for experimental analyses were presented as the mean and the sample standard deviation (1σ) values of at least three biological replicates. For comparisons of two groups, statistical significance was analyzed by an unpaired, two-tailed t-test assuming equal variance. For comparisons of three or more groups, significance was analyzed by one-way ANOVA and a post-hoc Tukey HSD test.

Supplemental Information

Figure S1.

Immunodetection of assembled RuBisCO. Western blot showing the assembly of RbcL (WT and AncIB) and RbcS into the L₈S₈ hexadecameric complex (520 kDa), detected by anti-RbcL antibody.

Figure S2.

Genotyping of S. elongatus strains. Primers F06/R06 (Table S2) were used to confirm the rbc operon insertion at the neutral site 2 (NS2), either with WT rbcL (for Syn01) or AncIB rbcL (rbc insert: 8,617 bp; no insert: 1,970 bp). Primers F07/R07 were used to confirm the presence or absence of the rbc operon at the native site (rbc present at its native site: 5,140 bp; rbc knocked out and replaced with the aadA gene: 3,342 bp).

Table S1.

Isotopic composition of biomass and DIC in growth medium of S. elongatus cultures.

Table S2.

Primers used in this study.

Acknowledgements

We sincerely thank Emily Peñaherrera, Jenan Kharbush, Ryan Ward, Sky Dominguez, and the University of California-Davis Stable Isotope Facility for assistance. This work was supported by the National Aeronautics and Space Administration Early Career Faculty (ECF) Award No. 80NSSC19K1617 (BK), the National Science Foundation Emerging Frontiers Program Award No. 1724090 (BK), National Aeronautics and Space Administration Postdoctoral Fellowship (AKG), Simons Foundation Early Career Award No. 561645 (JNY and ML), and National Institute of General Medical Sciences of the National Institutes of Health award number R01GM118815 (AT, to James W. Golden at the University of California-San Diego).