Highly active rubiscos discovered by systematic interrogation of natural sequence diversity

Abstract

CO₂ is converted into biomass almost solely by the enzyme rubisco. The poor carboxylation properties of plant rubiscos have led to efforts that made it the most kinetically characterized enzyme, yet these studies focused on < 5% of its natural diversity. Here, we searched for fast-carboxylating variants by systematically mining genomic and metagenomic data. Approximately 33,000 unique rubisco sequences were identified and clustered into ≈ 1,000 similarity groups. We then synthesized, purified, and biochemically tested the carboxylation rates of 143 representatives, spanning all clusters of form-II and form-II/III rubiscos. Most variants (> 100) were active in vitro, with the fastest having a turnover number of 22 ± 1 s^-1 -sixfold faster than the median plant rubisco and nearly twofold faster than the fastest measured rubisco to date. Unlike rubiscos from plants and cyanobacteria, the fastest variants discovered here are homodimers and exhibit a much simpler folding and activation kinetics. Our pipeline can be utilized to explore the kinetic space of other enzymes of interest, allowing us to get a better view of the biosynthetic potential of the biosphere.

Keywords: carbon fixation; carboxylation rate; enhanced photosynthesis; metagenomic survey; ribulose-1,5-bisphosphate carboxylase/oxygenase.

Figure 1. Schematic of the workflow developed and employed in this study

1. A

   A computational pipeline to extract rubisco sequences from genomic and metagenomic databases, cluster them based on sequence identity, and select representatives that cover the entire diversity of rubisco variants from nature.

2. B

   An experimental pipeline to screen the representative variants for carboxylation activity.

3. C

   Catalytic outliers are evaluated using radiometric assays for the determination of accurate k _cat and K _M carboxylation values as well as S_C/O.

Figure 2. Systematic exploration of uncharted rubisco representatives outside of the heavily sampled form‐I group

Of 223 total k _cat values reported in the literature, 217 measurements are for form‐I, two for form‐II, three for form‐III, and one for form‐II/III (Flamholz et al, 2019). This previously explored diversity coverage is shown on a phylogenetic tree of rubisco homologs at 90% identity, compared to the diversity explored in this study. Each leaf corresponds to a cluster of sequences such that all leaves share < 90% identity. All subtypes of form‐I, form‐II, and form‐II/III rubiscos are annotated. Form‐III rubiscos were omitted from the tree for clarity, as they were not biochemically tested here and only three variants from this group have reported kinetics. Form‐IV were removed as they apparently lack the carboxylating activity (Tabita et al, 2007) (see Appendix Fig S1 for the full phylogenetic tree).

Figure 3. High‐throughput pipeline for measuring the carboxylation rate of rubisco

1. A

   Coupling rubisco activity to NADH oxidation is done by two enzymatic steps catalyzed by phosphoglucokinase (pgk) and glyceraldehyde 3‐phosphate dehydrogenase (gapdh) (Kubien et al, 2011). A gradient of CABP is used to gradually inhibit rubisco activity.

2. B

   NADH oxidation is monitored at 340 nm in a gas‐controlled plate reader under 4% CO₂ and 0.2% O₂ in order to favor carboxylation over oxygenation. The slope of the curves gives the rate of NADH oxidation, which is equal to twice the carboxylation rate; the rate with no CABP was measured in duplicates.

3. C

   Rate of carboxylation (y‐axis; slopes from panel B) as a function of CABP concentration (x‐axis). The x‐intercept gives the concentration of rubisco active sites ([E]) while the y‐intercept gives the reaction rate without CABP inhibition (V _max); thus, the specific activity per active site is given by dividing V _max by [E]; dashed line is a least‐square linear regression (r ² > 0.99). For this example, rubisco from R. rubrum catalyzes ≈ 7 reactions per second.

Figure 4. Measured carboxylation rates for all the variants that were successfully expressed (N = 105)

Rates were determined using a spectrophotometric coupled assay at 30°C (see Materials and Methods). Rubiscos are ordered by phylogeny, as indicated by the dendrogram, and display a weak relationship between sequence similarity and rate of carboxylation, except for one cluster that corresponds to microaerobic bacteria. In green is the interquartile range of k _cat values for all reported plant rubiscos from Flamholz et al (2019) when corrected for 30°C assuming Q ₁₀ = 2.2 (Cen & Sage, 2005). Hatched bars represent variants that were analyzed by radiometric analysis, as described below. Bars and error bars correspond to the mean ± standard errors for each variant. The number of replicates for each variant is reported in Table EV1. Values below 0.5 s⁻¹ were considered inactive (dashed line). For a full description of the results, see Table EV1.

Figure 5. Comparing the carboxylation kinetics of Gallionella sp. rubisco to reported values from all characterized rubiscos

1. A

   Michaelis–Menten kinetic plots of rubiscos from Gallionella sp. (red; k _cat = 22 ± 1.1 s⁻¹; K _M = 276 ± 6 μM) and S. elongatus (gray; k _cat = 11.7 ± 0.6 s⁻¹; K _M = 200 ± 4 μM), measured by ¹⁴C labeling in the absence of oxygen at 25°C (see Materials and Methods); variants were measured in two biological repeats (circles and triangles), each in duplicate; dark and light green lines represent the median C₃ (k _cat = 3.1 s⁻¹; K _M = 14 μM) and C₄ (k _cat = 4.2 s⁻¹; K _M = 20 μM) plant rubiscos, respectively.

2. B

   k _cat and K _M values for all previously measured rubiscos (Flamholz et al, 2019) (gray) and the seven promising variants tested here (red for Gallionella sp. and blue for the other six); dashed line indicates the least‐square linear regression fit (slope in log scale is 0.4; r ²  = 0.44); histograms for k _cat and K _M are plotted on parallel axes and clearly show that the variants discovered in our screen are outliers (95^th percentile) in both k _cat and K _M for CO₂.