1. Biological Replication Today

2. The Chemoton

3. Autocatalysis

4. Replication of Informational Molecules

5. Models for cell division and growth

Chemical replicating systems

Motivation

Self-replication as the central biological phenomenon

Emergence of life

Evolution = Replication + Variation + Selection

Technological aspects: Biosensors, Detection of molecules (e.g., PCR, RCA, etc.)

Self-reproducing technical sytems

Biological replication

cell division in E. Coli

semi-conservative natureof DNA replication

Biological replication

Biological replication

the replication fork

Molecular evolution

The major transitions according to Maynard Smith & Szathmary

self-replicating molecules populations of molecules in compartments

independent replicators chromosomes

RNA as gene and enzyme (RNA world) DNA and proteins

procaryotes eucaryotes

asexual clones sexual populations

protists animals, plants, fungi (cell differentiation)

solitary individuals colonies

primates societies (language)

Molecular evolution

RNA world scenario

Molecular evolution

Evolution of compartments

The Chemoton

Autocatalysis

A + X 2A

X A

A

• autocatalytic systems catalyze their own production (feedback)

• autocatalysis: important step towards self-replication, but no evolution

Autocatalysis

• compensates for losses due to side-reactions

• exponential growth (in principle) competition of replicators

d[A]dt

= k[A]

[A](t) = [A](0)! exp (kt)

exponential growth

A minimal self-replication model (self-complementary molecule)

Autocatalysis

14

Classification of replicators

Simple replicators

Hereditary replicators with limited variation

Hereditary replicators with unlimited variation

15

Prions as simple replicators

©!!""#!Nature Publishing Group!

!

Conversion affected by kinetic or thermodynamic barriers

Native conformers

Prion conformers

De novo prion induction by nucleation

Conformation conversion at fibre ends Dissemination of prion fibresby fragmentation

‘Seed’

Amyloid fibre

+

Start

Stop

Stop

High-fidelitytermination

AAAA

Sup45

Low-fidelitytermination

AAAA

Sup45

Protein

Ribosome

Ribosome

Stop

Stop

Sup35

Sup35 prion aggregates

Genotype: ade1–14

Genotype: ade1–14Phenotype: ADE+

Phenotype: ade–

b

a

c

Seeded Unseeded

Fibr

e as

sem

bley

(%)

Time (h)

100

75

50

25

050403020100

Start

[ psi –]

[PSI+]

CODICAL DOMAIN The domain of natural selection dealing solely with self-replicating information as opposed to material entities.

cervids and its potential transmission to livestock and humans, looms as a serious public health threat5.

The causative agent of these diseases is unlikely to be nucleic-acid based because it is extraordinarily resistant to nucleases and UV radiation2,5,9. These observations invoked several diverse hypotheses concerning the composition of the infectious agent, which included the ‘protein-only’ hypothesis2,5. Infectivity is sensitive to agents that inactivate proteins, strongly indicat-ing a protein-based infectious agent (abbreviated to prion)2,5. Speculations10 were found to be prescient once biochemical purification of the infectious agent revealed protein aggregates that were virtually devoid of nucleic acid and were composed principally of one protein2,5,11. The conundrum of how a protein could confer infectivity was solved when it was shown to be an endogenous host protein, termed prion protein2,5,11

(PrP). Infectivity, therefore, only required PrP to encode self-templating conformational information.

PrP is ubiquitously expressed in its normal, !-heli-cal cellular form, termed PrPC (see TABLE 1 for prion nomenclature), as a glycosylphosphatidylinositol-anchored plasma membrane glycoprotein, the func-tion of which is still obscure5. Although spontaneous prion diseases afflict about one in a million people per year, several independent missense mutations in the PrP ORF almost guarantee the development of disease12. Conversely, the deletion of the gene that encodes PrP renders animals resistant to prion dis-eases13,14. When linked to infectivity, PrP generally adopts an altered, protease-resistant, "-sheet-rich, polymeric conformation known as PrPSc !REFS 5,15" (see TABLE 1 for prion nomenclature). PrPSc propagates by converting PrPC molecules to the PrPSc state (FIG. 1a),

Figure 1 | The prion hypothesis and the yeast prion [PSI+]. a | A nucleation event stabilizes protein conformers in an altered self-replicating prion conformation. The probability of nucleation is dictated by kinetic and thermodynamic considerations and is exceptionally unlikely for most proteins in physiological settings. Once formed, the nucleus, or ‘seed’, recruits other conformers (that are probably in a transiently unfolded state) and converts them to the self-replicating conformation. The nucleus then increases in size to become an AMYLOID fibre and continues to convert other conformers to the self-replicating conformation at the fibre ends. Amplification of conformational replication is achieved by the fragmentation of fibres to liberate new ends. Fragmentation also allows the dissemination of infectious material. b | In Saccharomyces cerevisiae [psi–] cells, the translation-termination factor Sup35 functions with Sup45 to recognize stop codons and terminate translation. Cells that carry a premature stop codon in their ADE1 gene do not make functional Ade1 and accumulate a red metabolite. By contrast, in [PSI+] cells most Sup35 protein is sequestered in self-replicating prion fibres, and is unable to participate in translation termination. Consequently, some ribosomes read through the stop codon and functional Ade1 is produced. [PSI+] cells therefore produce white colonies and can grow on adenine-deficient medium. The red/white colony assay is convenient and frequently used, but [PSI+] can also suppress a wide range of other stop codon mutations. [PSI+] increases the efficiency of readthrough by ~0.2–35%, depending on the specific [PSI+] variant, yeast strain and stop codon in question67–69. Modified with permission from REF. 156 © (2000) Elsevier Science. c | Prion fibres form only after a characteristic lag phase in reactions that are not seeded (blue). By contrast, prion fibres form rapidly without a lag phase in reactions seeded with small quantities (2% w/w) of prion fibres (red).

436 | JUNE 2005 | VOLUME 6 www.nature.com/reviews/genetics

R E V I EWS

A simple self-complementary replicator

template-directedformation of aSchiff base

(Terfort, von Kiedrowski,Angew. Chemie Int. Ed. Engl., 31, 654 (1992))

Replication of informational molecules

17

Replication of informational molecules

L. Orgel, Nature, vol. 358, 203 (1992)

© 1992 Nature Publishing Group

18

Replication of informational molecules

cross-catalytic self-replication

© 1994 Nature Publishing Group

© 1994 Nature Publishing Group

cross-catalysis

autocatalysis

Sievers & v. Kiedrowksi, Nature, vol. 369, 221 (1994)

19

Kinetics & Product inhibition

I cT3 lo4

I m o ~ L “ I

6 71 d

b C

5

4

3

2

1 _ _ - - - a

0 0 0.5 1.0 1.5 2.0

t Ihl - Fig. 1. Time course for the formation of the hexameric 3’-5’-phosphoamidate

T3 in the absence (curve b) and in the presence of template T’ (curves c-e) as measured by HPLC. The data points shown are averages from the experimental

time courses of A’, B3, T3. The set of theoretical curves b e represents the solution of differential equation (a) for the parameters given in Table 1 and the experimental set of initial concentrations shown below. The yield ofT3 is 50%

after 2 h (curve b). Curve a represents the theoretical time course if no auto- catalysis occurs. Initial concentrations and reaction conditions: cA = c. =

O.O8(c), 0.16(d), 0.32 mM (e), HEPES = 2-[4-(2-hydroxyethyl)-l-piperazino]- ethanesulfonic acid/sodium salt.

1 mM, [EDC] = 0.2 M, [HEPES] = 0.1 M, pH = 7.5, T = 3 0 ° C CTo = O(b),

The exponential factor in equation (a) has been omitted for the system presented here, since the hydrolysis of EDC is negligible within the first two hours under the reaction con- ditions. As a consequence a three-parameter fit would pro- duce a less reliable k, . To show the general applicability of our method and for the sake of comparison, we have also evaluated by equation (a) the set of experimental time cours- es for the systems reported by Orgel et aI.[’] and Rebek et

(exp ( - k,t) omitted).[g1

Table 1. Apparent rate constants for the autocatalytic and the nonautocatalytic tem-

plate production (k, and kb) and for the hydrolysis of the activated species ( k J . The experimental time courses of the systems reported by Orgelet al. and Rebek et al. have

been taken from the literature.

A’, B3, T3, T’ (7.43 0.12) [a] (1.77 ? 0.15) x lo-* [a] Orgel [5] (5.71 0 . 2 3 ) ~ lo-’ (1.66 f 0.19) x (1.92 k 0.46) x

Rebek [6] (9.87 +_ 0 . 1 7 ) ~ lo-’ (1.24 f 0 . 0 4 ) ~ lo-’

[a] The computation is conservative in the sense that k , is slightly overestimated at the cost of k, . This is because our method of calibration reflects a concentration depen- dant-recovery of single-component HPLC peaks, assuming that the recovery of the

respective component is not influenced by other components. A multicomponent

calibration method is in the development stages. However, the present method of calibration is more reliable than methods which assume direct proportionality be- tween HPLC peak areas and concentrations.

An efficient self-replicating system should exhibit a high apparent rate constant k, and a low constant k,. The first criterion is especially important if the self-replicating tem- plate is formed via a condensation reaction in water, since template formation has to compete with hydrolysis of the activated species, i.e., the EDC adducts. If all three systems were measured under optimal conditions (especially with re- spect to the temperature), the above system is roughly 130

Angew. Chem. Int. Ed. Engl. 30 (1991) No. 4 0 VCH Verlagsgesellschaft

and 75 times more efficient than Orgel’s and Rebek’s system, respectively, in terms of k,. A low k, is necessary to achieve sigmoidal growth. Thus, the quotient k,/k,, which defines the factor by which the autocatalytic synthesis exceeds the noninstructed process at a template concentration of 1 M,

must be large enough to observe sigmoidicity. With respect

to the autocatalytic excess factor E, as this quotient is termed hereafter, the above system ( E = 420) is comparable to Orgel’s system ( E = 340) but exceeds Rebek’s system (E = 80). The type of growth observed in curve b is, however, an experimental novelty. We term it parabolic since the growth of an autocatalyst with a reaction order of 112 in the rate equation of its synthesis can be expressed as a second order polynomial of time for early reaction times.

The reasoning behind the mechanism proposed for the autocatalytic template production is the same as for the orig- inal system.12] It is generalized in Figure 2 to account for

?? B

z- A

& T

( 2 4

u. m,_._i a

D

Fig. 2. General mechanism for a minimal self-replicating system. Constituent A represents the activated form of trimer A* here. Large arrowheads at the

reaction arrows for the reversible reactions indicate the favored site of the equilibrium.

similarities between the systems known so far. The following should be noted:

1 ) The rate determining step is the irreversible formation of a 3’-5’-phosphoamidate linkage leading from the ternary complex M to the self-complementary duplex D which is

expressed by: d[D]/dt = k [ M ] . 2) The system is in dynamic equilibrium with respect to all

single- and double-stranded oligonucleotide constituents. With K , = [MI [A]-’ [B]-’ [TJ-’ and K2 = [D] [TI-’ it follows for the formation of duplex molecules that d[D]/ dt = k K , K z - 1 / 2 [A] [B] [D]”’.

K , $ I), and for the special case of complete activation, the apparent rate constant k, can be rationalized in terms of the ele- mentary rate and equilibrium constants: k, = 21/2 kK1K2-’ l2 .

According to Eigen and Schuster[”] parabolic growth should lead to Darwinian selection under conditions which allow self-replicating templates to compete for their con- stituents when the total concentration of templates is kept constant (i.e. in a chemostat). This statement has been refut- ed recently by Szathmary who proposes, with reference to experimental results,[2’ that parabolic growth invariably leads to coexistence of self-replicating templates,” ‘](i.e., the most efficient template is not able to supersede its less effi- cient competitors completely). We agree with Szathmary on the point that this type of growth must have preceeded expo- nential growth in an RNA world, since it allows autocatalyt-

3) Under parabolic growth conditions ( K ,

mbH, W-6940 Weinheim, 1991 0S70-0833/91/0404-04ZS $3.50+ .2SjO 425

• problem of product release

•no exponential growth observed

• approximate square-root law

I cT3 lo4

I m o ~ L “ I

6 71 d

b C

5

4

3

2

1 _ _ - - - a

0 0 0.5 1.0 1.5 2.0

t Ihl - Fig. 1. Time course for the formation of the hexameric 3’-5’-phosphoamidate

T3 in the absence (curve b) and in the presence of template T’ (curves c-e) as measured by HPLC. The data points shown are averages from the experimental

time courses of A’, B3, T3. The set of theoretical curves b e represents the solution of differential equation (a) for the parameters given in Table 1 and the experimental set of initial concentrations shown below. The yield ofT3 is 50%

after 2 h (curve b). Curve a represents the theoretical time course if no auto- catalysis occurs. Initial concentrations and reaction conditions: cA = c. =

O.O8(c), 0.16(d), 0.32 mM (e), HEPES = 2-[4-(2-hydroxyethyl)-l-piperazino]- ethanesulfonic acid/sodium salt.

1 mM, [EDC] = 0.2 M, [HEPES] = 0.1 M, pH = 7.5, T = 3 0 ° C CTo = O(b),

The exponential factor in equation (a) has been omitted for the system presented here, since the hydrolysis of EDC is negligible within the first two hours under the reaction con- ditions. As a consequence a three-parameter fit would pro- duce a less reliable k, . To show the general applicability of our method and for the sake of comparison, we have also evaluated by equation (a) the set of experimental time cours- es for the systems reported by Orgel et aI.[’] and Rebek et

(exp ( - k,t) omitted).[g1

Table 1. Apparent rate constants for the autocatalytic and the nonautocatalytic tem-

plate production (k, and kb) and for the hydrolysis of the activated species ( k J . The experimental time courses of the systems reported by Orgelet al. and Rebek et al. have

been taken from the literature.

A’, B3, T3, T’ (7.43 0.12) [a] (1.77 ? 0.15) x lo-* [a] Orgel [5] (5.71 0 . 2 3 ) ~ lo-’ (1.66 f 0.19) x (1.92 k 0.46) x

Rebek [6] (9.87 +_ 0 . 1 7 ) ~ lo-’ (1.24 f 0 . 0 4 ) ~ lo-’

[a] The computation is conservative in the sense that k , is slightly overestimated at the cost of k, . This is because our method of calibration reflects a concentration depen- dant-recovery of single-component HPLC peaks, assuming that the recovery of the

respective component is not influenced by other components. A multicomponent

calibration method is in the development stages. However, the present method of calibration is more reliable than methods which assume direct proportionality be- tween HPLC peak areas and concentrations.

An efficient self-replicating system should exhibit a high apparent rate constant k, and a low constant k,. The first criterion is especially important if the self-replicating tem- plate is formed via a condensation reaction in water, since template formation has to compete with hydrolysis of the activated species, i.e., the EDC adducts. If all three systems were measured under optimal conditions (especially with re- spect to the temperature), the above system is roughly 130

Angew. Chem. Int. Ed. Engl. 30 (1991) No. 4 0 VCH Verlagsgesellschaft

and 75 times more efficient than Orgel’s and Rebek’s system, respectively, in terms of k,. A low k, is necessary to achieve sigmoidal growth. Thus, the quotient k,/k,, which defines the factor by which the autocatalytic synthesis exceeds the noninstructed process at a template concentration of 1 M,

must be large enough to observe sigmoidicity. With respect

to the autocatalytic excess factor E, as this quotient is termed hereafter, the above system ( E = 420) is comparable to Orgel’s system ( E = 340) but exceeds Rebek’s system (E = 80). The type of growth observed in curve b is, however, an experimental novelty. We term it parabolic since the growth of an autocatalyst with a reaction order of 112 in the rate equation of its synthesis can be expressed as a second order polynomial of time for early reaction times.

The reasoning behind the mechanism proposed for the autocatalytic template production is the same as for the orig- inal system.12] It is generalized in Figure 2 to account for

?? B

z- A

& T

( 2 4

u. m,_._i a

D

Fig. 2. General mechanism for a minimal self-replicating system. Constituent A represents the activated form of trimer A* here. Large arrowheads at the

reaction arrows for the reversible reactions indicate the favored site of the equilibrium.

similarities between the systems known so far. The following should be noted:

1 ) The rate determining step is the irreversible formation of a 3’-5’-phosphoamidate linkage leading from the ternary complex M to the self-complementary duplex D which is

expressed by: d[D]/dt = k [ M ] . 2) The system is in dynamic equilibrium with respect to all

single- and double-stranded oligonucleotide constituents. With K , = [MI [A]-’ [B]-’ [TJ-’ and K2 = [D] [TI-’ it follows for the formation of duplex molecules that d[D]/ dt = k K , K z - 1 / 2 [A] [B] [D]”’.

K , $ I), and for the special case of complete activation, the apparent rate constant k, can be rationalized in terms of the ele- mentary rate and equilibrium constants: k, = 21/2 kK1K2-’ l2 .

According to Eigen and Schuster[”] parabolic growth should lead to Darwinian selection under conditions which allow self-replicating templates to compete for their con- stituents when the total concentration of templates is kept constant (i.e. in a chemostat). This statement has been refut- ed recently by Szathmary who proposes, with reference to experimental results,[2’ that parabolic growth invariably leads to coexistence of self-replicating templates,” ‘](i.e., the most efficient template is not able to supersede its less effi- cient competitors completely). We agree with Szathmary on the point that this type of growth must have preceeded expo- nential growth in an RNA world, since it allows autocatalyt-

3) Under parabolic growth conditions ( K ,

mbH, W-6940 Weinheim, 1991 0S70-0833/91/0404-04ZS $3.50+ .2SjO 425

v. Kiedrowski, Angew. Chem. Int. Ed. 30, 423 (1991)

no autocatalysis

d[A]dt

= ! + "[A]p

p = 1/2reaction order

growth rate

20

Kinetics & "Ecology"

• exponential growth, several specieswith different growth rates, limited resources

one species will prevail

• subexponential growth, several specieswith different growth rates, limited resources

coexistence of species possible

growth limited by own copy number(product inhibition !)

21

A self-replicating ribozyme

Paul & Joyce, PNAS 99, 12733 (2002)

5 discontinuous double-helical sections improve product separation

22

A self-replicating ribozyme

Paul & Joyce, PNAS 99, 12733 (2002)

reaction order p = 1

23

Self-replicating peptides

has attained a reaction order higher than the product inhibited order.

One potential cause of the reduction in product inhibition would

be a reduction in the stability of the tetrameric product of RI-26 as

compared to the dimeric product of E1E2 self-replication. To

address this hypothesis the thermal denaturation of both peptides

was examined by circular dichroism (Figure 4).5,6 The melting

temperature of RI-26 displayed an approximately 20 °C shift to

lower temperature as compared to E1E2, confirming the significant

reduction in stability of the coiled-coil of RI-26.

In conclusion, we disclose a successful strategy whereby

modulation of coiled-coil stability results in remarkable catalytic

efficiency for self-replication. By shortening the peptide to the

minimum length necessary for coiled-coil formation a highly

efficient self-replicating system was obtained due to very low

background reaction rates, bringing the efficiency close to naturally

occurring enzymes. An added benefit of the reduction in peptide

size was a decrease in the stability of the coiled-coil leading to

effective suppression of product inhibition. The design of auto-

catalytic peptides with high efficiencies and low product inhibition

enhances their potential for future technological applications and

their consideration as pre-biotic precursor molecules.10

Acknowledgment. The authors are grateful to NSF (78923) and

NASA for support of this work. We are also grateful to Prof. von

Kiedrowski for the generous gift of the software program SimFit.

Supporting Information Available: Characterization (mass spectral

data, amino acid analyses, circular dichroism data, SimFit files and

analytical ultracentrifugation) data (PDF). This material is available

free of charge via the Internet at http://pubs.acs.org.

References

(1) Issac, R.; Ham, Y.-W.; Chmielewski, J. A. Curr. Opin. Struct. Biol. 2001,11, 458.

(2) von Kiedrowski, G. Bioorg. Chem. Front 1993, 3, 113.(3) Luther, A.; Brandsch R.; von Kiedrowski, G. Nature 1998, 396, 245.(4) Yao, S.; Ghosh, I. G.; Zutshi, R.; Chmielewski, J. J. Am. Chem. Soc.

1997, 119, 10599.(5) Fairman, R.; Chao, H.-G.; Mueller, L.; Lavoie T. B.; Shen, L.; Novotny

J.; Matsueda G. R. Protein Sci. 1995, 4, 1457.(6) Su, J. Y.; Hodges, R. S.; Kay, C. M. Biochemistry 1994, 33, 15501.(7) Dawson, P. E.; Muir, T. W.; Clark-Lewis, L.; Kent, S. B. H. Science

1994, 266, 776.(8) (a) von Kiedrowski, G. Angew. Chem., Int. Ed. Engl. 1986, 28, 1235. (b)

von Kiedrowski, G.; Wlotzka, B.; Helbing J.; Matzen, M.; Jordan, S.Angew. Chem., Int. Ed. Engl. 1991, 30, 423. (c) Huc, I.; Pieters, R. J.;Rebek, J., Jr. J. Am. Chem. Soc. 1994, 116, 10296. (d) Lee, D. H.; Granja,J. R.; Martinez, J. A.; Severin, K.; Ghadiri, M. R. Nature 1996, 382, 525.(e) Severin, K.; Lee, D. H.; Martinez, J. A.; Ghadiri, M. R. Chem. Eur.J. 1997, 3, 1017. (f) Yao, S.; Ghosh, I.; Zutshi, R.; Chmielewski, J. Angew.Chem., Int. Ed. Engl. 1998, 37, 478.

(9) Hanson, L. O.; Windersten, M.; Mannervik, B. Biochemistry 1997, 36,11252-11260.

(10) Rode, B. M. Peptides 1999, 20, 773-786.

JA026024I

Figure 2. RI-26 production from two fragments, RI-26a and RI-26b (500µM each), at 23 °C in 100 mMMOPS buffer (with 1% 3-mercaptopropionicacid) at pH 4.0 as a function of time with varying initial concentrations oftemplate: (]) no template, (0) 10 µM RI-26, (4) 20 µM RI-26, and (O)40 µM RI-26. Error bars reflect standard deviations of two independentexperiments. Curves were generated with SimFit2 by simulations based onthe reaction model: RI-26a + RI-26b f RI-26 (kb); RI-26a + RI-26b +0.91 RI-26 f 1.91 RI-26 (ka).

Figure 3. Initial rate of RI-26 formation as a function of the concentrationof added template to the power of 0.91.

Figure 4. Thermal denaturation of 20 µM E1E2 (4) and RI-26 ([) in thepresence of 6 M GdnHCl in a pH 4.0, 100 mM MOPS buffer.

C O M M U N I C A T I O N S

J. AM. CHEM. SOC. 9 VOL. 124, NO. 24, 2002 6809

Approaching Exponential Growth with a Self-Replicating Peptide

Roy Issac and Jean Chmielewski*

Department of Chemistry, Purdue UniVersity, West Lafayette, Indiana 47907

Received February 25, 2002

Self-replicating peptide systems hold great promise for a wide

range of technological applications, as well as to address funda-

mental questions pertaining to the molecular origins of life.1 The

development of peptide self-replicators, however, requires high

catalytic efficiency that is highly dependent on the stability of the

product-template complex.2 Under optimum conditions and in theabsence of product inhibition, self-replicating systems should exhibit

exponential growth while product inhibition causes growth to be

parabolic for dimeric systems. The design of self-replicating

compounds capable of high efficiency has remained elusive,

although recently Kiedrowski and co-workers reported the develop-

ment of a self-replicating DNA analogue in which solid-phase

cycling circumvented product inhibition.3 Here we describe the

development of a highly efficient peptide self-replicator with a

catalytic enhancement that is close to that of known enzymes.

In an effort to improve the catalytic efficiency of the self-

replicating peptide E1E2,4 we sought to destabilize its coiled coil

structure. Fairman and co-workers achieved dramatic decreases in

stability for tetrameric coiled-coil peptides by shortening the chain

lengths,5 and Hodges and co-workers found similar effects with

dimeric coiled coil peptides.6 With these precedents as a basis, a

peptide was designed for self-replication, RI-26, that contains 3

full heptad repeats within the coiled coil, one shorter than the

original E1E2 sequence. RI-26 maintained the design principle of

E1E2 in that glutamate residues were positioned at the e and g

positions of the helical heptad repeats to achieve pH-based control

over helicity (Figure 1). RI-26a and RI-26b, therefore, correspond

to the two fragments of RI-26 that may undergo thioester mediated

chemical ligation to produce RI-26.7

The full length template, RI-26, was found to adopt a helical

conformation in a pH-dependent fashion; at pH 7.0 the helical

content was only 28% as determined by circular dichroism, and

increased to 87% when the pH was lowered to 4.0 as was observed

with E1E2.5 The helicity of RI-26a and RI-26b increased by 45%

and 16%, respectively, in the presence of RI-26 at pH 4.0, indicating

the ability of RI-26 to act as a template for its fragments. Analytical

ultracentrifugation was used to determine the aggregation state of

RI-26 at pH 4.0, and interestingly this peptide was found to exist

as a tetramer. E1E2 by contrast exists as a dimer under similar

conditions.

To determine if RI-26 had self-replicating properties, the

fragments RI-26a and RI-26b were incubated at pH 4.0 with and

without added RI-26. As is indicative of an autocatalytic system,2

adding increasing amounts of the template led to a dramatic

acceleration in product formation (Figure 2). By contrast, the

reaction at pH 7 was insensitive to added template and, therefore,

not autocatalytic. The experimental data were analyzed with the

program SimFit based on the empirical equations developed by

Kiedrowski (Figure 2).2 The SimFit analysis provided an apparent

catalytic rate constant, ka, of 50.6 ( 0.5 M-1.91 s-1 and a

noncatalytic rate constant, kb, of 5.04 ( 0.03 ! 10-4 M-1 s-1 with

a catalytic efficiency (! ) ka/kb) of 1.0 ! 105. This is a remarkably

efficient system when compared to other self-replicating molecules;

self-replicating peptides and oligonucleotides have displayed cata-

lytic efficiencies in the range of 24 to 3700.4,8 The efficiency

observed with RI-26 is comparable to that observed for some

enzymatic systems, such as glutathione transferases.9 The unin-

structed noncatalytic or background reaction, presumably a result

of the association between the two fragments, is also much slower

in this peptide system than any of the other reported peptide self-

replication systems.4,8e-f This is most likely due to the presence of

fewer leucine residues in the shorter fragments, thereby reducing

the hydrophobic interactions between them.

The order of the self-replicating reaction was determined by

finding the best fit for the catalytic and noncatalytic reaction rates,

using SimFit, and was found to be 0.91( 0.04. A linear relationshipwas also found between the initial rate for each reaction as a

function of the concentration of the template to the power of 0.91

(Figure 3). Since RI-26 forms a tetramer in solution, one possible

ligation complex would be composed of RI26a‚RI26b‚(RI-26)3. Fora self-replicating tetramer the reaction order (p) would be expected

to be 0.75 if the system exhibited product inhibition.2,8a For RI-26

a significantly higher reaction order was observed, thereby

classifying this replicating system as weakly exponential

(0.75 < p <1). This is the first self-replicating system to date that* Corresponding author. E-mail: chml@purdue.edu.

Figure 1. Helical wheel diagram (a) and sequence (b) of RI-26 and itsfragments. An arrow indicates the residues where chemical ligation occurs.

Published on Web 05/22/2002

6808 9 J. AM. CHEM. SOC. 2002, 124, 6808-6809 10.1021/ja026024i CCC: $22.00 © 2002 American Chemical Society

chemical ligation of α-helix forming peptides

Issac & Chmielewski, JACS (2002)

p = 0.91

24

Vesicle growth and division

replication of a non-informational structure

Szostak, Bartel & Luisi, Nature 409, 387 (2001)

25

Models for cellular compartments

Hanczyc, Fujikawa, Szostak, Science vol. 302, 618 (2003)

hydrated sodium calcium aluminium magnesium silicate hydroxide (Na,Ca)0.33(Al,Mg)2(Si4O10)(OH)2·nH2O

montmorillonite

• clay enhances vesicle growth

• RNA adsorbed to clay is incorporated

26

Towards a protocell

Szostak, Bartel & Luisi, Nature 409, 387 (2001)

• RNA replicase (self-replicating ribozyme)

• lipid vesicle compartment

• division of vesicles (feeding with micelles)

• coupled replication cycles

27

Evolution

Szostak, Bartel & Luisi, Nature 409, 387 (2001)

28

Eigen's error threshold

xi = k0[AiQi !Di]xi +!

j !=i

wijxj ! w0ixi

phenomenological rate equation

Eigen, Naturwissenschaften, vol. 58, 465 (1971)

leads to (...)

maximum number of bases Nmax which can be stably maintained with base copy fidelity q and minimum fidelity Qmin for "master sequence"

Nmax =lnQmin

1! q

⇨ Eigen's paradoxon

29

The Hypercycle

• Eigen's paradoxon: impossible to store all information necessary to ensure faithful replication

• Hypercycle: Coexistence of several cooperating replicators - members of hypercycle better than mutants; no competition between members; cycle better than other cycles ...

autocatalytic replicators

30

A hypercyclic network of peptide replicators

Nature © Macmillan Publishers Ltd 1997

letters to nature

592 NATURE | VOL 390 | 11 DECEMBER 1997

chemical system that constitutes a clear example of a minimalhypercyclic network, in which two otherwise competitiveself-replicating peptides symbiotically catalyse each others’production.

The present design of a minimal hypercycle is based on two self-replicating coiled coil peptides R1 and R2 (Fig. 1). The replicator R1

was recently reported11,12 and is produced as the ligation product ofthe electrophilic peptide fragment E and the nucleophilic fragmentN1. The replicator R2 is made from the same electrophilic fragmentbut a different nucleophilic peptide fragment N2. The nucleophilicfragments N1 and N2 differ in their sequence at the hydrophobicrecognition surface—N1 is composed of valine and leucine whereasN2 is made up of isoleucine and leucine residues. This difference insequence at the hydrophobic core is known to affect profoundly theaggregation state of coiled coils13,14. Furthermore it is known thatconservative mutations in this region of the structure can drasticallyalter the kinetic behaviour of the replicator11,12,15.

The ability of R2 to self-replicate was determined by observationof characteristics previously established as signatures of self-replication (Fig. 2)11,12. Similar to that of R1, the new replicatorR2 also displays a parabolic growth profile. Numerical fitting ofthe kinetic data obtained for R2 to the empirical rate equationsof von Kiedrowski16 gave a background rate constantkb ! 0:072 ! 0:005 M " 1 s " 1 and an apparent autocatalytic rateconstant ka ! 52 ! 1 M " 3=2 s " 1, making R2 more efficientthan its relative R1 (kb ! 0:063 M " 1 s " 1 and constantka ! 29:4 M " 3=2 s " 1).

A solution containing all three fragments E, N1 and N2 gave acombinatorial synthesis of both replicators. A priori, one would

expect a survival-of-the-fittest situation where the more efficientreplicator R2 would overwhelm R1 by consuming the commonfragment E more quickly. At first glance, this expectation seemed tobe borne out as R2 was produced in greater abundance than R1 (asexpected, when molecular interactions are disrupted in the presenceof guanidinium hydrochloride, no kinetic preference for R2 over R1

was observed). However, the situation is more interesting andcomplex. When we sought to give R1 an advantage in this competi-tion by adding 40% R1 (with respect to the nucleophile concentra-tion) at the start of the reaction, to our surprise the rate of R1 self-production increased by only 1.7 times over the unseeded reactionbut the rate of R2 formation was enhanced to a greater extent, by 5.4times (Table 1, Fig. 3). Thus the two replicators are not mutuallyexclusive in their growth; R1 catalyses the formation of R2 as well asitself. Likewise, perturbation of the reaction by seeding it with 45%R2 not only increased the rate of R2 production 2.9 times but R1 aswell, by 3.5 times. Thus a cross-catalytic cycle is cooperativelycoupled with two self-replicating reactions, making this systemone which is hypercyclic in nature. There are four characteristicoutcomes expected for such a hypercyclic network, depending onthe relative efficiencies of the coupled catalytic and autocatalyticreactions2. The observed greater efficiencies of the catalytic reactionsover the autocatalytic components of the system are the mostdesirable outcomes which assure the stability of the hypercycle:production of one species promotes the production of the other toan even greater degree. This particular mode of catalytic couplingprevents one replicator from overwhelming the other and enablesthe two to reproduce as a single coherent unit.

To verify that R1 and R2 catalyse each other’s production, the

E

N1

N2R2

R1

E

I II

IV III

Figure 1 Schematic diagram of a minimal hypercycle based on two self-replicating peptides. Cycles I and III show the self-producing cycles of replicatorsR1 (dark grey/light grey) and R2 (dark grey/striped) respectively, which pre-organize their constituent fragments thereby promoting peptide ligation. CycleII, where R1 promotes R2 formation, and cycle IV, where R2 promotes R1 formation,comprise the catalytic components of the hypercycle and allow the replicators topositively regulate each others’ production. The mechanistic details of thepresent hypercyclic network may be more complex than the minimal systemdepicted here. Detailed kinetic analyses of the replicator sequences have shownthat the autocatalytically productive intermediates involve, at least in part, qua-ternary complexes in which two template strands pre-organize the reactivepeptide fragments (ref. 12 and K. Kumar, D.H.L., M.R.G., unpublished results).The following peptide sequences were employed in this study: replicator 1 (R1),ArCONH-RMKQLEEKVYELLSKVA-CLEXEVARLKKLVGE-CONH2; replicator 2(R2), ArCONH-RMKQLEEKVYELLSKVA-CLEXEIARLKKLIGE-CONH2; electrophilicfragment (E), ArCONH-RMKQLEEKVYELLSKVA-COSBn; nucleophilic fragment1 (N1), H2N-CLEXEVARLKKLVGE-CONH2; nucleophilic fragment 2 (N2), H2N-CLEXEIARLKKLIGE-CONH2. Bn, benzyl; Ar, 4-acetamidophenyl; and X, lysine-e-NHCO-Ar.

Figure 2 Production of R2 as a function of time in the presence of various initialconcentrations of R2. Open circles, in the absence of any added R2; filleddiamonds, in the presence of 4.0 #M; open triangles, in the presence of21.4 #M; and filled circles, 42.6 #M of initially added R2. Curves were generatedby nonlinear least-squares fit of the data to the empirical rate equation of vonKiedrowski using the program SimFit16. Data are an average of two experiments.

Table 1 Initial rates of product formation

Product No replicators added +40% R1 +45% R2.............................................................................................................................................................................R1 4.8 8.2 17.0R2 5.8 31.1 16.9.............................................................................................................................................................................The data in this table (in units of 10!8 Mmin!1) are for reactions containing the three peptidefragments in the absence and presence of added replicators.

Lee, Severin, Yokobashi, Ghadiri, Nature 390, 591 (1997)

autocatalysis

autocatalysis

cross-catalysis

cross-catalysis

catalysis more efficient in the presence of partner replicator !

31

491

Autocatalytic networks: the transition from molecular

self-replication to molecular ecosystems

David H Lee, Kay Severin and M Reza Ghadiri*

The transition from inanimate to animate chemistry is

thought to involve self-organised networks of molecular

species whose collective emergent property gives rise to

the overall characteristics of living systems. In the past,

simple autocatalytic networks have been constructed that

display basic forms of cooperative behaviour. These include

reciprocal catalysis, autocratic, and hypercyclic networks.

The design and emergent properties of these novel molecular

networks are reviewed here.

Addresses

Departments of Chemistry and Molecular Biology and the Skaggs

Institute for Chemical Biology, The Scripps Research Institute, La

Jolla, CA 92037, USA

*e-mail: ghadiri@scripps.edu

Current Opinion in Chemical Biology 1997, 1:491-496

http://biomednet.com/elecref/1367593100100491

0 Current Biology Ltd ISSN 1367-5931

Introduction

Living systems are autonomous self-reproducing ‘molec-

ular ecosystems’ defined as a collective of self-organized

communities of dynamic, interdependent, interacting,

and computing molecular species. One hallmark of

living systems is their ability to translate molecular

interactions and chemical reactions into complex animate

characteristics that far exceed the simple sum of the

individual properties of their molecular constituents.

These animate characteristics are macroscopic properties

that arise as a result of nonlinear dynamic interactions

and the passage, growth, and change of information

within the molecular ecosystem. Understanding how such

self-organized systems may have established themselves

in the ‘beginning’, how they might have evolved and

grown in complexity, and how they result in the emergent

properties that distinguish living systems from inanimate

matter, remains a major experimental and theoretical

challenge.

It is widely believed that inanimate chemistry em-

barked on its path towards ‘living chemistry’ via the

formation of self-replicating molecules. Now that a

few enzyme-free self-replicating molecular systems have

been experimentally realized, the focus needs to be

increasingly shifted to the remaining paramount issues

of how, and by what mechanism(s), the unimolecular

self-reproducing systems can self-organize into networks

to produce life-like properties. Over the past three

decades, theories of molecular evolution have emerged

which combine principles of biology and chemistry with

concepts of non-equilibrium physics [l-3]. In general,

they provide plausible descriptions of the early stages

of biogenesis wherein biopolymers self-organize to form

autocatalytic networks. Although the main conclusions of

these theories are generally accepted, there have been

very few experimental results that support their tenets.

In this review we summarize recent attempts to fill

this gap. We will focus on multicomponent enzyme-free

systems, the constituents of which are interconnected

via autocatalytic and/or catalytic cycles. Thus far only

three very simple types of autocatalytic networks have

been realized experimentally: cross-catalytic, autocratic,

and hypercyclic. These systems are all based on molecular

species that can directly (or indirectly) catalyze their own

formation.

Molecular Replication

In the most basic form of molecular self-replication

a reaction product serves as a specific catalyst for its

own synthesis. The requirement of a specific recognition

event during the catalytic transformation distinguishes

self-replication from plain autocatalytic reactions, and

allows information to be transferred to the molecular

offspring. The first enzyme-free replication system-a

palindromic hexanucleotide sequence- was reported in

1986 by von Kiedrowski [4*]. Since then a series of other

self-replicating molecules have been designed and charac-

terized (for reviews see [.5”,6,7,8]), including abiological

organic molecules [5”,9] and most recently peptides

[lO”,ll*]. The autocatalytic process for these systems

is based on template-directed condensation reactions

induced by assembly of self-complementary structures

(Figure 1). For replicators containing nucleobases, molec-

ular recognition is mediated by a well defined pattern

of hydrogen bond donor and acceptor groups. In the

case of peptide self-replication however, hydrophobic

interactions-assisted by electrostatic interactions -are

the main recognition driving force. Detailed kinetic

analyses of several such systems, complemented by a

‘minimal replicator theory’ [12], have revealed a linear

relationship between the initial rate of product formation

and the square root of the initial template concentration.

The corresponding parabolic growth profile reflects an

intrinsic shortcoming of current replication systems, that

is the self-inhibitory tendencies of the catalysts (products)

employed. Explicit modeling of a self-replicating system

has also provided great insight into the many complex

processes that occur in solution [13*]. Armed with the

knowledge gained by such studies, more advanced design

strategies are expected to lead to more efficient self-

replicating systems in the near future. Recently, systems

based on peptide nucleic acids (PNA) [14] and pyranosyl

Autocatalytic networks: the transition from molecular self-replication to molecular ecosystems Lee, Severin and Ghadiri 495

Figure 4 Figure 5

Schematic representation of an autocratic network. This system is

composed of a peptide replicator (T, long dark grey cylinder), its

constituent fragments E and N, (short dark grey cylinders), and two

single alanine replicator mutants (raA stipled/darkgrey and TlsA dark

grey/spotted cylinders). These peptides spontaneously self-organize

into an autocatalytic network that corrects the errors by subjugating

the mutants for the exclusive production of the wild type, (adapted

from [20”1).

network of self-replicating molecules. Hypercycles spread

the information content and couple genotype and pheno-

type over all members of the system. As a consequence

of this feature, hypercycles not only enforce genotype

stabilization but also serve as a mechanism for coherent

growth of the system and allow for the coexistence

of otherwise competitive entities. The relationship is

more than mere coexistence, rather it is one of ultimate

cooperation. Working together within this symbiotic

relationship, members of a hypercycle may compete better

for resources. This ‘principle of self-organization’ was

originally conceived by Eigen [l], who later went on

to experimentally demonstrate hypercyclic organization

between coliphage DNA and its replicase apparatus [Zl].

Most recently, a hypercyclic network was constructed out

of two self-replicating peptides that share one common

fragment ([ZP]; Figure 5). At first glance one might

expect a survival-of-the-fittest situation where the more

efficient replicator would commandeer all the resources.

It was discovered, however, that not only could each

peptide self-replicate but that they collaborate in each

others’ synthesis as well! Control experiments verified the

catalytic and autocatalytic abilities of these two peptides.

Interestingly, the catalytic coupling that contributed to

the production of a given peptide was stronger than the

corresponding autocatalytic component of the hypercycle

that is, catalysis was more efficient than autocatalysis. This

result is significant in that this pattern of coupling is

thought to be necessary for a hypercycle to remain stable.

Although not explicitly stated, other multicomponent

systems may have formed hypercycles too. For instance,

one of Rebek and co-workers [23] triacid-based replicators

NZ

Schematic diagram of a hypercyclic network. The hypercycle is

composed of two self-replicating peptides (Rl 1 dark/light grey

cylinder, and R2, dark grey/spotted cylinder). The two replicators

are formed by a reaction of one of two nucleophilic peptides, Nl and

N2 (light grey and spotted respectively) with a common electrophilic

peptide (dark grey). The self-replication cycles I and III are connected

by two catalytic cycles, II and IV, which allows for the distribution of

information content and phenotype over the entire network. This type

of organization unifies two otherwise competitive species into a single

cooperative reproducing entity. (Adapted from [22”).

may form a hypercycle with some of its ‘mutants’,

analogues which were partially protected to prevent

Watson-Crick base-pairing. These mutants have remain-

ing to them the Hoogsteen mode of binding and so can

still self-replicate or catalyze the formation of the original

replicator. In separate reactions, the original replicator

could promote production of the mutants as well as

itself. Similarly, Achilles and von Kiedrowski [24] observed

reciprocal catalysis between self-replicating pentameric

and hexameric oligonucleotides which may have included

a hypercyclic component.

Conclusions

In this review we have discussed how self-organization of

catalytic molecules into simple networks with nonlinear

growth kinetics results in the rapid emergence of a

robustness [ZS] and functional innovation that would be

difficult, if not impossible, to achieve through the gradual

accumulation of mutations in any single molecule. Even a

cursory examination of the internal organization of living

things makes it clear that this transition from individual

self-replicating molecules to the next level in the hierarchy

of self-organized systems should be further explored.

What other forms of self-organization are possible? How

can several networks be productively associated? What

emergent properties might these systems display? The

current challenge is to design and characterize new and

more complex networks, and, in time, even large molecular

ecosystems, in order to answer these questions. Ultimately

496 Model systems

we hope that in designing such self-instructed chemical 12. Von Kiedrowski G: Minimal replicator theory I: parabolic versus

processes we will glean some insights into how life came exponential growth. Bioorg Chem front 1993, 3:113-l 46.

to be. ion Kiedrowski discusses various aspects of empirically modelling self-repli-

cating systems.

Acknowledgements

‘I‘he authors would hkc to thank Krishna Kumar, Alan Kennan, Yohei

Yokobayashi and Jose Antonio hlartinez for many productive brainstorming

sessions and their enthusiasm. The authors are also grateful to Juan R Granja

for invaluable discussions and help with some of the figures.

References and recommended readings

Papers of particular interest, published within the annual period of review,

have been highlighted as:

. of special interest

. . of outstanding interest

1.

2.

Eigen M, Schuster P: The Hypercycfe. A Principle of Natural Seff- Organization. Berlin: Springer; 1979.

Ktippers BO: The Origin of Biological Information. Cambridge,

Massachusetts: MIT Press; 1990.

3. Kauffman SA: The Origins of Order. New York: Oxford University

Press; 1993.

16. Sievers D, von Kiedrowski G: Self-replication of complementary

nucleotide-based oligomers. Nature 1994, 369:221-224.

Currently this paper constitutes the only example of a cross-catalytic system

where the whole system self-replicates. Such cross-catalytic systems are

particularly appealing targets since reproduction is based on the reciprocal

catalysis of complementary sequences as opposed to self-complementary

sequences.

4. Von Kiedrowski G: A self-replicating hexadeoxynucleotide.

Angew Chem fnf Ed Engl 1986, 25:932-934.

;his paper describes the first non-enzymatic template-directed self-replicat-

ing system. The groundwork for molecular self-replicating systems is outlined

here.

1 7. Pieters RJ, Hut I, Rebek J Jr: Reciprocal template effects in a

replication cycle. Angew Chem fnt Ed fngl 1994, 33:1579-l 581.

18. Pieters RJ, Hut I, Rebek J Jr: Reciprocal template effects in

bisubstrate systems: a replication cycle. Tetrahedron 1995,

51:485-498.

5. Wintner EA, Rebek J Jr: Autocatalysis and the generation of

self-replicating systems. Acta Chim Stand 1996, 50:469-485.

Kis review takes the reader through the evolution of Rebek’s abiotic replica-

tors. The design features and subsequent improvements are discussed. The

comprehensive sets of controls that were carried out are also summarized

and placed into perspective.

19. lnoue T, Orgel LE: A nonenzymatic RNA polymerase model.

Science 1983, 219:859-862.

20. Severin K, Lee DH, Antonio JA, Vieth M, Ghadiri MR: Dynamic . . error correction in autocatalytic peptide networks. Angew

Chem fnf Ed Eng/, in press.

6. Bag BG, von Kiedrowski G: Templates, autocatalysis, and

molecular replication. Pure Appl Chem 1996, 68:2145-2152.

z

8.

Orgel LE: Molecular replication. Nature 1992, 356:203-209.

Li T, Nicolaou KC: Chemical self-replication of palindromic

duplex DNA. Nature 1992, 369:218-221.

This study is the first example of an autocratic network. Particularly striking

is how emergent properties can arise out of a multicomponent autocatalytic

system. This unprecedented error-correction mechanism may be significant

for genotype stabilization of self-replicating molecules.

9. Terfort A, von Kiedrowski G: Self-replication by condensation of

I-aminobenzamidines and 2-formylphenoxyacetic acids. Angew Chem Int Ed Engl 1992, 31:654-656.

21. Eigen M, Biebricher CK, Gebinoga M, Gardiner WC: The

hypercycle. Coupling of RNA and protein synthesis in the

infection cycle of an RNA bacteriophage. Biochemistry 1991,

30:11005-l 1018.

10. Lee DH, Granja JR, Martinez JA, Severin K, Ghadiri MR: A self- . . replicating peptide. Nature 1996, 382:525-528.

This manuscript reports the design and characterization of the first self-repli-

cating peptide. Unlike other chemical self-replicating systems, self-comple-

mentarity arises from a recognition surface defined by hydrophobic and elec-

trostatic features, rather that hydrogen-bonding. This peptide is one example

of only a very few functional de nova designed proteins.

22. Lee DH, Severin K, Yokobayashi Y, Ghadiri MR: Emergence of . . symbiosis in peptide self-replication through a hypercycle

network Nature 1997:in press.

This paper reports the construction of the first hypercycle based on self-

replicating peptides.

23. Hong JI, Feng Q, Rotello V, Rebek J Jr: Competition, cooperation,

and mutation: improving a synthetic replicator by light

irradiation. Science 1992, 255:848-850.

11. Severin K, Lee DH, Martinez JA, Ghadiri MR: Peptide self- . replication via template-directed ligation. Chem Eur J 1997,

3:1017-l 024.

24. Achilles T, von Kiedrowski G: A self-replicating system from

three starting materials. Angew Chem Int Ed Engl 1993,

32:1198-l 201.

Expanding on the initial discovery of the self-replicating peptide, a detailed 25. Barkai N, Leibler S: Robustness in simple biochemical

kinetic analysis is described here. networks. Nature 1997, 387:913-917.

13. Reinhoudt DN, Rudkevich DM, de Jong F: Kinetic analysis of the . Rebek self-replicating system: is there a controversy? J Am

Chem Sot 1996,i 18:6880-6889.

Reinhoudt and coworkers demonstrate the utility of explicit modelling in

studying the self-replicating system of Rebek.

14. Bohler C, Nielsen PE, Orgel LE: Template switching between

PNA and RNA oligonucleotides. Nature 1995, 376:578-581.

This manuscript describes the non-enzymatic transition from one type of

informational biopolymer to another, providing some credence for the plau-

sibility of a pre-RNA genetic material.

15. Pitsch S, Krishnamurthy R, Bolli M, Wendeborn S, Holzner A,

Minton M, Lesueur C, Schkinvogt I, Jaun B, Eschenmoser A:

Pyranosyl-RNA (‘p-RNA’): base-pairing selectivity and potential

to replicate. He/v Chim Acta 1995, 78:1621-l 635.

Current Opinion in Chemical Biology 1, 491 (1997)

32

Self-reproducing machines