Module 2: The RNA World

Modern biology hands genetic memory to DNA and catalysis to proteins, with RNA as a messenger. But this neat division of labour cannot have been the starting point: information and function must have been unified in a single molecule. The RNA World hypothesis(Woese 1967, Crick 1968, Orgel 1968) proposes that early life used RNA for both. The discovery of catalytic RNA (ribozymes) by Cech and Altman, and the fact that peptide bond formation in the ribosome is ribozyme-catalysed, provide strong empirical support. This module examines the evidence, the in-vitro evolution of ribozymes, Eigen's quasispecies theory, and the pre-RNA-world alternatives.

2.1 Evidence for the RNA World

Five major lines of evidence point toward a historical RNA-first stage:

Catalytic RNA (ribozymes). Tom Cech (1981) showed that the Group I intron of Tetrahymena self-splices, and Sidney Altman (1983) proved that the catalytic component of bacterial RNase P is its RNA (M1), not protein. Both shared the 1989 Nobel in Chemistry.
The ribosome is a ribozyme. X-ray crystallography (Yonath, Ramakrishnan, Steitz; 2009 Nobel) revealed that the peptidyl transferase centre (PTC) contains no protein — peptide bond formation is catalysed entirely by 23S rRNA.
RNA cofactors. Central metabolites (ATP, NAD⁺, FAD, coenzyme A, SAM) are all RNA-derived, fossilised relics of an age when RNA ran metabolism.
Deoxyribose from ribose. In modern cells, ribonucleotide reductase makes DNA from RNA — the synthetic precedence is unambiguous.
Riboswitches. Natural RNAs in bacteria still sense small molecules and regulate gene expression — direct descendants of an RNA-run regulatory logic.

2.2 Ribozymes & Catalytic Chemistry

Natural ribozymes span five structural classes: hammerhead, hairpin, HDV, Varkud satellite, and glmS. All catalyse RNA cleavage or ligation with \(k_\text{cat}/k_\text{uncat}\) rate enhancements from \(10^5\) to \(10^9\) — good but modest compared to protein enzymes ( \(10^{14}\)+). The PTC rate enhancement is only \(\sim 10^7\) but involves positioning rather than chemistry; the RNA backbone templates the attacking α-amino group.

Metal ion catalysis

Most natural ribozymes depend on divalent cations, typically Mg²⁺, which:

Neutralise phosphate backbone charge, allowing compact folding
Activate the 2′-OH nucleophile by deprotonation
Stabilise the pentacoordinate phosphorane transition state
Shield negative charge on the leaving group

The two-metal-ion mechanism proposed by Steitz & Steitz (1993) operates in HDV, group I/II introns, and the RNase P active site. In Hadean oceans (warm, Mg-rich from serpentinite weathering), Mg²⁺ was likely abundant.

In-vitro evolution & SELEX

Systematic Evolution of Ligands by EXponential enrichment (SELEX; Tuerk & Gold 1990; Ellington & Szostak 1990) takes a random pool of \(\sim 10^{15}\) RNAs (40-mer: \(4^{40} \approx 10^{24}\)possibilities, so sampled randomly) and iterates selection + amplification. After 10-15 rounds the best binders or catalysts are enriched by many orders of magnitude. SELEX has produced:

Aptamers binding any small molecule with nM affinity
Ribozymes catalysing Diels-Alder cycloaddition, aldol condensation, amino-acid charging
An RNA polymerase ribozyme (Bartel & Szostak 1993; Wochner et al. 2011) capable of copying short templates

2.3 Self-Replication: Lincoln & Joyce 2009

In 2009, Tracey Lincoln and Gerald Joyce at Scripps published an RNA system that replicates itself indefinitely without protein enzymes. Two “R3C” ribozymes (E and F) each catalyse a ligation that produces the other from two oligonucleotide substrates (A+B and A′+B′). The cross-catalytic kinetics:

\[ \frac{d[E]}{dt} = k_F [F][A][B] - d_E [E], \quad \frac{d[F]}{dt} = k_E [E][A'][B'] - d_F [F] \]

This system grows exponentially and supports Darwinian evolution under laboratory conditions: variants appear through ligation errors, compete for limiting substrates, and the fittest survive. In serial-transfer experiments, new recombinant ribozymes 3-4 fold faster than the parents emerged after 150 generations.

Fitness and selection equation

For a pool of variants with individual rate constants \(k_i\), the concentration of variant \(i\) follows:

\[ \frac{dx_i}{dt} = k_i x_i - \langle k \rangle x_i = (k_i - \langle k \rangle)x_i \]

This is Fisher's selection equation in disguise: variants grow faster than the pool average increase, slower ones decrease. Combined with mutation (imperfect replication) the system explores the fitness landscape.

2.4 Eigen's Quasispecies & the Error Catastrophe

Manfred Eigen (1971) derived a fundamental constraint on the length of any self-replicating molecule given its per-base fidelity. Let \(q\) be the probability that each base is copied correctly. For a sequence of length \(L\):

\[ Q = q^L = \text{probability of error-free copy} \]

The master sequence dominates the population only if the selective advantage \(s\) (fitness of master vs mutants) satisfies:

\[ s > \frac{1}{Q} = q^{-L} \quad \Leftrightarrow \quad L < \frac{\ln s}{-\ln q} \approx \frac{\ln s}{1-q} \]

Above the error threshold, the master sequence dissolves into a cloud of mutants and genetic information is lost: the error catastrophe. For typical ribozyme copying fidelities \((q \approx 0.99)\) and mild selection \((s \approx 10)\), the maximum length is \(L_\text{max} \approx 230\) bases — limits consistent with the shortest natural ribozymes.

The information crisis

This creates a bootstrap problem: a high-fidelity polymerase ribozyme needs \(L \gtrsim 200\) bases, but copying 200 bases reliably requires \(q \gtrsim 0.995\) — already better than the best in-vitro RNA polymerase. Proposed escapes:

Shorter proto-ribozymes functioning in modular cooperative networks
Compartmentalisation in lipid vesicles reducing parasite spread
Surface-bound replication with sequence-specific catalysis
Hypercycles: mutually reinforcing replicator pairs (Eigen & Schuster 1979)

2.5 The Chirality & Instability Problems

The RNA world faces serious chemical challenges:

Cross-inhibition by opposite chirality

Joyce et al. (1984) demonstrated that template-directed polymerisation of D-ribonucleotides is blocked by the presence of L-enantiomers — a chain-terminating effect. Homochirality of the monomer pool is thus a precondition, not a consequence, of the first RNA polymerisation. The Soai autocatalytic amplification (Module 1) or a chiral mineral surface (calcite, quartz) must solve this upstream.

Backbone instability

The 2′-OH of ribose attacks the adjacent phosphodiester, cleaving RNA in neutral aqueous solution with a half-life of \(\sim 4\) years at 25 C and pH 7. In hot hydrothermal environments ( \(T = 85\) C) the half-life collapses to roughly \(\sim 8\) days. This is part of the reason life eventually replaced RNA with the chemically far more stable DNA.

The chicken-and-egg problem

The first RNA molecule capable of self-replication requires a prior pool of activated nucleotides, a folded catalytic structure, and a favourable template — none of which are obvious outcomes of unguided chemistry. Progress depends on finding plausible minimalist systems below the error threshold.

2.6 Pre-RNA Worlds: TNA, PNA, GNA

Given the difficulty of prebiotic ribose synthesis and the chemical frailty of the RNA backbone, several authors have suggested that the first genetic polymer may not have been RNA but a chemically simpler precursor, later replaced by RNA in a “genetic takeover” (Cairns-Smith). Three leading candidates:

TNA (threose nucleic acid): Eschenmoser's group (2000) showed that TNA with threose replacing ribose forms stable Watson-Crick duplexes with itself, DNA, and RNA. Threose is four-carbon, arises directly from the formose reaction without requiring a specific stereoselection.
PNA (peptide nucleic acid): Nielsen (1991) designed PNA with a peptide-like N-aminoethylglycine backbone. PNA hybridises to DNA/RNA with higher affinity than the natural backbones and is chemically much more stable. PNA templates direct their own ligation from activated dimers.
GNA (glycol nucleic acid): Meggers (2005) showed propylene glycol-based nucleic acids, lacking stereogenic centres in the backbone altogether.

These alternative backbones suggest the first living chemistry could have used simpler, more prebiotically plausible components that were eventually overwritten by RNA after the biosynthetic machinery for ribose became available.

Python: Eigen Quasispecies Dynamics

Panel 1 shows a population of 12 sequence types evolving under high copying fidelity (\(q = 0.97\)): the master sequence (0) dominates, its near-mutants form a cloud. Panel 2 drops the fidelity to \(q = 0.58\), past the error threshold: the master sequence loses its grip and the population delocalises — the error catastrophe.

Python

script.py85 lines

import numpy as np
import matplotlib
matplotlib.use('Agg')
import matplotlib.pyplot as plt

# Eigen quasispecies model
# dx_i/dt = sum_j Q_{ij} A_j x_j - E(t) x_i
# with E(t) = sum_j A_j x_j  (fitness normalization)
# Q_{ij} = mutation probability from j to i

np.random.seed(5)

fig, axes = plt.subplots(1, 2, figsize=(16, 7), facecolor='#03120a')
fig.subplots_adjust(wspace=0.35)

# Number of sequences (types)
N = 12
# Fitness landscape: single peak "master sequence"
A = np.ones(N) * 1.0
A[0] = 2.0   # master sequence

def simulate(q, T=500):
    # Mutation matrix: Q[i,j] = prob of j->i
    # Simple metric: d(i,j) in 1..N-1
    Q = np.zeros((N,N))
    for i in range(N):
        for j in range(N):
            d = abs(i - j)
            # Hamming-like distance approximation
            Q[i,j] = q**(N-d) * ((1-q)**d)
        Q[:,j] /= Q[:,j].sum()
    x = np.ones(N) / N
    hist = []
    for step in range(T):
        flux = Q @ (A * x)
        E = np.sum(A * x)
        dx = flux - E * x
        x = x + 0.02 * dx
        x = np.clip(x, 0, None)
        x = x / x.sum()
        hist.append(x.copy())
    return np.array(hist)

# ---- Panel 1: Below error threshold (q high, faithful copying) ----
ax1 = axes[0]
ax1.set_facecolor('#061a0d')
hist_low = simulate(q=0.97, T=400)
for i in range(N):
    alpha = 1.0 if i == 0 else 0.45
    ax1.plot(hist_low[:, i],
             color=plt.cm.viridis(i/N), lw=2.0 if i == 0 else 1.0, alpha=alpha,
             label=f'Seq {i}' if i in [0, 1, 5, 10] else None)
ax1.set_xlabel('Time step', color='#6ee7b7', fontsize=11)
ax1.set_ylabel('Frequency', color='#6ee7b7', fontsize=11)
ax1.set_title('Quasispecies: q = 0.97 (below error threshold)\nMaster sequence dominates',
              color='#a7f3d0', fontsize=12, fontweight='bold')
ax1.legend(fontsize=8, labelcolor='#a7f3d0', facecolor='#061a0d', edgecolor='#10b981')
ax1.tick_params(colors='#6ee7b7')
ax1.grid(True, alpha=0.2, color='#10b981')
for sp in ax1.spines.values(): sp.set_edgecolor('#10b981')

# ---- Panel 2: Above error threshold ----
ax2 = axes[1]
ax2.set_facecolor('#061a0d')
hist_high = simulate(q=0.58, T=400)
for i in range(N):
    alpha = 1.0 if i == 0 else 0.45
    ax2.plot(hist_high[:, i],
             color=plt.cm.plasma(i/N), lw=2.0 if i == 0 else 1.0, alpha=alpha,
             label=f'Seq {i}' if i in [0, 1, 5, 10] else None)
ax2.set_xlabel('Time step', color='#6ee7b7', fontsize=11)
ax2.set_ylabel('Frequency', color='#6ee7b7', fontsize=11)
ax2.set_title('Quasispecies: q = 0.58 (above error threshold)\nError catastrophe - information lost',
              color='#a7f3d0', fontsize=12, fontweight='bold')
ax2.legend(fontsize=8, labelcolor='#a7f3d0', facecolor='#061a0d', edgecolor='#10b981')
ax2.tick_params(colors='#6ee7b7')
ax2.grid(True, alpha=0.2, color='#10b981')
for sp in ax2.spines.values(): sp.set_edgecolor('#10b981')

fig.suptitle('Eigen Quasispecies Model & the Error Catastrophe',
             color='#a7f3d0', fontsize=14, fontweight='bold')
plt.savefig('output.png', bbox_inches='tight', facecolor=fig.get_facecolor(), dpi=120)
plt.close()
print('Quasispecies simulation complete.')

Click Run to execute the Python code

Code will be executed with Python 3 on the server

Python: SELEX & Cross-Replicating Ribozymes

Panel 1 runs a simple SELEX-style in-vitro evolution: a pool of 200 ribozyme variants with log-normal activity undergoes selection and amplification for 15 rounds, converging on the best catalyst while diversity collapses. Panel 2 integrates the Lincoln-Joyce cross-replicator ODEs for three parameter sets, showing symmetric exponential growth, failure when one partner is too rare, and slower asymmetric growth under unbalanced kinetics.

Python

script.py97 lines

import numpy as np
import matplotlib
matplotlib.use('Agg')
import matplotlib.pyplot as plt

# In-vitro selection (SELEX) style simulation
# Pool of ribozymes with variable catalytic efficiency k_cat
# Amplification + selection pressure -> best ones dominate

np.random.seed(11)
fig, axes = plt.subplots(1, 2, figsize=(16, 7), facecolor='#03120a')
fig.subplots_adjust(wspace=0.35)

# ---- Panel 1: SELEX convergence ----
ax1 = axes[0]
ax1.set_facecolor('#061a0d')

# Initial pool: 200 sequences with log-normal activity
n_pool = 200
activity = np.exp(np.random.randn(n_pool) * 1.2)
activity = activity / activity.max()

n_rounds = 15
mean_hist = []
max_hist = []
div_hist = []

pool = activity.copy()
for rd in range(n_rounds):
    # Selection probability prop to activity
    probs = pool / pool.sum()
    # Amplification: resample with replacement
    idx = np.random.choice(len(pool), size=len(pool), p=probs)
    pool = pool[idx]
    # Mutation: small Gaussian perturbation
    pool = np.clip(pool + 0.03 * np.random.randn(len(pool)), 0.001, 1.5)
    mean_hist.append(pool.mean())
    max_hist.append(pool.max())
    div_hist.append(np.std(pool))

rounds = np.arange(1, n_rounds+1)
ax1.plot(rounds, mean_hist, color='#34d399', lw=2.5, marker='o', label='Mean activity')
ax1.plot(rounds, max_hist, color='#fbbf24', lw=2.5, marker='s', label='Best-in-pool')
ax1.fill_between(rounds,
                 np.array(mean_hist) - np.array(div_hist),
                 np.array(mean_hist) + np.array(div_hist),
                 alpha=0.2, color='#34d399')

ax1.set_xlabel('SELEX round', color='#6ee7b7', fontsize=11)
ax1.set_ylabel('Relative catalytic activity', color='#6ee7b7', fontsize=11)
ax1.set_title('In-Vitro Ribozyme Selection\nConvergence toward optimum',
              color='#a7f3d0', fontsize=12, fontweight='bold')
ax1.legend(fontsize=9, labelcolor='#a7f3d0', facecolor='#061a0d', edgecolor='#10b981')
ax1.tick_params(colors='#6ee7b7')
ax1.grid(True, alpha=0.2, color='#10b981')
for sp in ax1.spines.values(): sp.set_edgecolor('#10b981')

# ---- Panel 2: Self-replication dynamics ----
ax2 = axes[1]
ax2.set_facecolor('#061a0d')

# Lincoln-Joyce cross replicator: two RNA enzymes E, F
# E catalyses production of F, F catalyses E; both require substrates A, B
# Simplified ODE: dE/dt = k_F * F * S - d*E, dF/dt = k_E * E * S - d*F
from scipy.integrate import odeint

def model(y, t, kE, kF, d, S):
    E, F = y
    dEdt = kF * F * S - d * E
    dFdt = kE * E * S - d * F
    return [dEdt, dFdt]

t = np.linspace(0, 25, 500)
sol1 = odeint(model, [0.01, 0.01], t, args=(0.25, 0.25, 0.05, 1.0))
sol2 = odeint(model, [0.01, 0.00001], t, args=(0.25, 0.25, 0.05, 1.0))
sol3 = odeint(model, [0.01, 0.01], t, args=(0.15, 0.25, 0.1, 1.0))

ax2.plot(t, sol1[:, 0], color='#34d399', lw=2.3, label='E (symmetric)')
ax2.plot(t, sol1[:, 1], color='#60a5fa', lw=2.3, label='F (symmetric)')
ax2.plot(t, sol2[:, 0], color='#f97316', lw=2.0, ls='--', label='E (F seed depleted)')
ax2.plot(t, sol3[:, 0], color='#f472b6', lw=2.0, ls=':', label='E (asymmetric kE)')

ax2.set_xlabel('Time (arbitrary)', color='#6ee7b7', fontsize=11)
ax2.set_ylabel('Ribozyme concentration', color='#6ee7b7', fontsize=11)
ax2.set_title('Cross-Replicating Ribozymes\n(Lincoln & Joyce 2009 style)',
              color='#a7f3d0', fontsize=12, fontweight='bold')
ax2.legend(fontsize=8, labelcolor='#a7f3d0', facecolor='#061a0d', edgecolor='#10b981')
ax2.tick_params(colors='#6ee7b7')
ax2.grid(True, alpha=0.2, color='#10b981')
for sp in ax2.spines.values(): sp.set_edgecolor('#10b981')

fig.suptitle('Ribozyme Selection & Cross-Replication Dynamics',
             color='#a7f3d0', fontsize=14, fontweight='bold')
plt.savefig('output.png', bbox_inches='tight', facecolor=fig.get_facecolor(), dpi=120)
plt.close()
print('Ribozyme SELEX simulation complete.')

Click Run to execute the Python code

Code will be executed with Python 3 on the server

References

Kruger, K. et al. (1982). Self-splicing RNA: autoexcision and autocyclization of the ribosomal RNA intervening sequence of Tetrahymena. Cell, 31, 147-157.
Guerrier-Takada, C., Gardiner, K., Marsh, T., Pace, N. & Altman, S. (1983). The RNA moiety of ribonuclease P is the catalytic subunit. Cell, 35, 849-857.
Nissen, P., Hansen, J., Ban, N., Moore, P.B. & Steitz, T.A. (2000). The structural basis of ribosome activity in peptide bond synthesis. Science, 289, 920-930.
Lincoln, T.A. & Joyce, G.F. (2009). Self-sustained replication of an RNA enzyme. Science, 323, 1229-1232.
Wochner, A. et al. (2011). Ribozyme-catalyzed transcription of an active ribozyme. Science, 332, 209-212.
Eigen, M. (1971). Selforganization of matter and the evolution of biological macromolecules. Naturwissenschaften, 58, 465-523.
Eigen, M. & Schuster, P. (1979). The Hypercycle. Springer.
Bartel, D.P. & Szostak, J.W. (1993). Isolation of new ribozymes from a large pool of random sequences. Science, 261, 1411-1418.
Schöning, K.-U. et al. (2000). Chemical etiology of nucleic acid structure: the α-threofuranosyl-(3′->2′) oligonucleotide system. Science, 290, 1347-1351.
Nielsen, P.E., Egholm, M., Berg, R.H. & Buchardt, O. (1991). Sequence-selective recognition of DNA by strand displacement with a thymine-substituted polyamide. Science, 254, 1497-1500.
Joyce, G.F. et al. (1984). Chiral selection in poly(C)-directed synthesis of oligo(G). Nature, 310, 602-604.
Gilbert, W. (1986). The RNA world. Nature, 319, 618.

←Previous: Prebiotic Chemistry Next: Hydrothermal Origins→