Charge-density reduction promotes ribozyme activity in RNA–peptide coacervates via RNA fluidization and magnesium partitioning

It has long been proposed that phase-separated compartments can provide a basis for the formation of cellular precursors in prebiotic environments. However, we know very little about the properties of coacervates formed from simple peptides, their compatibility with ribozymes or their functional significance. Here we assess the conditions under which functional ribozymes form coacervates with simple peptides. We find coacervation to be most robust when transitioning from long homopeptides to shorter, more pre-biologically plausible heteropeptides. We mechanistically show that these RNA–peptide coacervates display peptide-dependent material properties and cofactor concentrations. We find that the interspacing of cationic and neutral amino acids increases RNA mobility, and we use isothermal calorimetry to reveal sequence-dependent Mg2+ partitioning, two critical factors that together enable ribozyme activity. Our results establish how peptides of limited length, homogeneity and charge density facilitate the compartmentalization of active ribozymes into non-gelating, magnesium-rich coacervates, a scenario that could be applicable to cellular precursors with peptide-dependent functional phenotypes.


Nucleic acids
RNA sequences for the R3C ligase system were obtained from Lincoln and Joyce (2009) 1 . S1 and S2 RNAs correspond to A' and B'. Fluorescently labelled RNAs were purchased from Biomers or Integrated DNA technologies (IDT); these include FAM-E, FAM-S1, FAM-S2, Cy5-(AC)9, Cy5-S1, Cy5-S2. Lyophilized RNA was resuspended in RNA-free water (Ambion) to a stock concentration of 100 µM. Torula yeast RNA type VI was purchased from Sigma-Aldrich. DNAs used for in vitro transcription were purchased from Biomers.

RNA sequences
Name Sequence

DNA sequences
Name Sequence a, 12.5% polyacrylamide gels of MgCl2 titration to 5 µM E, 6 µM S2 and 0.1 µM Cy5-S1. Three independent repeats were carried out and reactions were incubated for 1 hour. Control lanes show the starting substrate. b, Quantification of gels on (a); data are shown as the mean ± 95% CI. Data points were fitted to a Hill equation.

Supplementary Fig. 2 | Rn coacervates are stable under peptide excess.
Transmission electron micrographs of R9, R10, R18 and poly-R show that coacervates are still present at high peptide excess. Scale bar: 1 µm.

Supplementary Fig. 3 | Differential dissolution of Kn coacervates under peptide excess.
Transmission electron micrographs show that K9, K10, K18 and poly-K. K9 coacervates are stable and do not dissolve under peptide excess. Peptides longer than K9 result in coacervates that dissolve at high peptide ratios. Dissolution is accompanied by the formation of small filamentous structures, indicating the formation of soluble RNA-peptide complexes. Scale bar: 1 µm.

Supplementary Fig. 4 | (RGG)n coacervates are stable under peptide excess.
Transmission electron micrographs of (a) (RGG)5 and (b) (RGG)4. For (RGG)4 and (RGG)5, greater than 1:1 charge ratios are needed to trigger phase separation. Numbers in the top left corner of the micrographs represent the amino acid-to-nucleotide ratio used for sample preparation. Supplementary Fig. 7 | (RGG)4 shows reduced activity in the coacervate phase.
a, Three experimental repeats of R3C ligation at different amino acid-to-nucleotide ratios of (RGG)4. Control lane shows the starting substrate. b, Enzyme present in the supernatant after centrifugation of the samples prepared at different amino acid-to-nucleotide ratios of (RGG)4.

Supplementary Fig. 8 | Cy5-P RNA shows reduced or no mobility inside Rn coacervates.
FRAP curves for Rn coacervates show the mean ± 95% CI of n experiments with the 95% confidence interval as a shaded region. In R4 and R5 coacervate RNA diffusion was fitted to a single exponential curve assuming total RNA recovery. All other coacervates did not show RNA recovery after photobleaching. Apparent diffusion coefficients were derived from individual traces.

Supplementary Fig. 11 | RNA-(RGG)5 coacervates show liquid-like properties.
a, FRAP experiments of FITC-(RGG)5 and Cy5-P show differential mobility of the RNA and peptide components of the coacervate phase and increased RNA mobility compared to Kn peptides. Apparent diffusion coefficients were derived from individual traces. Data are shown as the mean of n individual determinations ± 95% CI. b, Droplet fusion experiments show liquid-like behaviour of the coacervate protocells. Inverse capillary velocity is reported as the mean ± 95% CI. c, Droplets prepared with Cy5-P or FITC-P show complete RNA rearrangement after fusion with droplets containing differently labelled RNA. Rearrangement is completed after 1 hour of incubation. Pearson correlation coefficients fitted to guide with a single exponential equation.  a, Relative turbidity plots of (KGG)9 prepared at 5 mM (light purple) and 20 mM (dark purple) mM MgCl2. A right shift in the turbidity curve is observed when increasing MgCl2 concentration, indicative of RNA-peptide interaction shielding. n = 3, error bars show the 95% confidence interval. b, R3C ligation activity in the presence of different aa/nt ratios of (KGG)9 at 5 (top panel) and 20 (bottom panel) mM MgCl2. (KGG)9 coacervates can maintain enzymatic activity even at high peptide ratios. c, enzyme present in the supernatant at different aa/nt of (KGG)9 in the presence of 20 mM MgCl2. d, quantification of relative product formation (red lines) and relative enzyme concentration in the coacervate phase for (KGG)9 coacervates prepared at 5 mM (left panel) and 20 mM (right panel) MgCl2. n = 3 , error bars show the 68% confidence interval. e-f, Droplet fusion experiments further indicate a 120-fold difference in droplet fusion times between coacervates prepared with K9 and those prepared with (KGG)9. The plots show the characteristic relaxation time of fusion for two droplets plotted against their geometric length. Data were fitted with a linear fit while constraining the y intercept to 0. The inverse capillary velocity is inferred from the slope of the fit. The shaded region represents the 95% confidence interval of the fit. Scale bar: 5 µm

In-out-diffusion of RNA from coacervates
To further substantiate the formation of production in coacervates, partition behaviour, remaining concentrations in solution as well as rate of in-out diffusion was considered in quantitative assays shown below.
7. FRAP results on total droplets formed at 40 times (RGG)5 and 20 mM MgCl2 show that the exchange is not high enough to completely replenish the supernatant pool between 4 to 7 times. Supplementary  Fig. 19).
8. If we follow this argument for the rest of peptide concentrations we observe that even at the highest aa/nt ratio the amount of product formed highly exceeds what would be expected if only the supernatant fraction reacted (a). In particular at high peptide excess (e.g. 60 fold) the product formation is about 50 times as high as can be expected by in/out diffusion and reaction in the buffer (b).
RNA ligation occurs inside (RGG)5 coacervates at 20 mM MgCl2. a, Product formation at different aa/nt ratios of (RGG)5 exceeds the expected concentration if product was being formed in solution. The red line shows the concentration of product formed if the reaction proceeded at different RNA dilutions. The reaction dilutions were obtained from the amount of RNA-E remaining in the supernatant after coacervate centrifugation. A 1X dilution corresponds to 5 µM E, 0.1 µM S1 and 6 µM S2. The aa/nt ratios are shown as numbers next to the datapoints. n = 3. b. Ratio of observed over expected product if the R3C ligase reaction would take place in the supernatant.