Dave Winkler, CSIRO Molecular & Health Technologies, Monash University, Clayton South VIC 3169, Australia 

Preamble

The Beilstein Institute is a Trust set up to promote chemical research and communication. It supports several endowed Chairs in Europe, and organizes several small workshop on specialist areas of chemistry, the major one being the Beilstein Bozen conference.  This meeting is a small invited group that includes eminent chemists, who present and discuss research on a defined leading edge chemical frontier.  The conference is biannual, and has been operating for 16 years, the 2008 meeting on Systems Chemistry being the 9^th.  It is traditionally held at the Schloss Hotel Korb, a beautiful, small, family establishment near the Tyrolean Italian town of Bolzano/Bozen.  The meeting is structured to allow substantial time for both lectures and discussion.

Highlights of the presentations and discussions

Günter von Kiedrowski – Systems chemistry and origins of life.
This paper largely discussed synthetic biology and autocatalyic chemical systems.  Kaufman suggested that autocatalytic systems of reactions need to be large (many components) before systems can generate emergent, self-replication properties.  He discussed what the minimum size of a cell or genome can be, based on work by Tibor Ganti, Rebek et al (1990), and Wang and Sutherland (1996).  He discussed his minimal replicator theory based on this prior work and related it to how selection can operate to improve or optimize autocatalytic self-replication chemical systems. He also presented three regimes under which self-replicating chemistries can operate: linear growth (no feedback required); parabolic growth (intermediate feedback); exponential growth (regime in which Darwinian evolution operates).  He gave an example of a self-assembling object (virus particle dimensions) obtained from tris oligonucleotides (see also Stahl JACS 2006, 128, 14014)

Hans Westerhoff – Chemistry in 3D: How Systems Biology affects Systems Chemistry
This paper was largely about the impressive work Westerhoff has done on metabolic networks. These provide an area of substantial overlap between chemistry and biology. He discussed how experiments can be carried out that allow metabolic networks to be deduced, and how sensitivity analyses can be run that identify the key enzymes (hub enzymes) in the network (see Westerhoff J. Theor. Biol. 2008, 252, 555). He found that metabolic networks often exhibit fractal behaviour when perturbed and used this to identify a Law of Systems Biology that states that control (of networks) tends to be conserved.  He illustrated the power of his metabolic network theory using some drug targeting examples.  He showed that the network approach using flux analysis shows that drug targets previously chosen and actually poor choices for controlling the metabolic network.  His analysis of the same problem identified different targets that have a much larger influence on the metabolic network and make better drug targets. He showed that control of organisms is a sum of genetic control and metabolic control, and that experiments where certain chemical species were withheld can identify the relative contributions of these control mechanisms. 

Antoine Danchin – designing a synthetic cell
He defined a cell as a computer that makes a computer.  The machine replicates and the program replicates. 'Life' involves a minimum metabolism (at least ~800 molecules are required), compartmentalization; and information flow.  In cells the machine and the program and the same object.  He used hydrophobicity analysis to identify integral membrane proteins, and has found this works reliably across psycophilic, mesophilic and thermophilic microbes. He employs correspondence analysis in place of principal component analysis for feature selection.  He also discussed how quickly proteins age in biological systems (in E coli as fast as < 2 min) and discussed situations where cells divide to produce one 'young' progeny and one 'aged' (see recent PNAS article).   He noted that approximately 400-500 genes persist in bacteria, and this may represent the minimal requirements for cell operation: essential genes, plus those dealing with stressors, and maintenance. Clustering of gene frequency across species allows identification of persistent genes and orphan genes. The persistent genes may be 'left over' from evolution, while the orphan genes recapitulate the origins of life (see Proteomics 2007, 7, 875).

Athel Cornish-Bowden – Catalysis at the origin of life.
This is related to several of the 125 most compelling questions in science posed recently (Science, 2005,309. no. 5731, 78 - 102). He discussed the (M,R) systems of of Robert Rosen which integrate the theories of Kauffman and Ganti et al.  (M, R) systems is a theory of life that incorporates autocatalytic sets (Kauffman); closure and causation (Rosen); compartmentalization (Muturana and Vanela); self-replications (Ganti et al); and catalytic 'closure'. Closure means that catalytic systems that use enzymes need enzymes to synthesize these enzymes, and these in turn require more enzymes to contract them & etc in an 'infinite regress'.  (M,R) systems provide a solution to this infinite regress so that autocatalytic systems can be finite.  It requires some essential promiscuity in enzymes so that they can play multiple roles and allow organisms to be closed with respect to structure, but open with respect to energy and matter fluxes (see Biodiversity 2007, 4, 2396).

Benjamin List – New concepts for catalysis
This was a very elegant lecture showing that relatively simple molecules like proline and substituted versions can catalyze chiral organic reactions with very high ee. He used the aldol reaction as an example but stated that the catalysis could be generalized to N=O, N=C, C=C, N=N etc.  For the aldol example his listed four types of aldolization: intermolecular; intramolecular (endo/endo); intramolecular (endo/exo); transannular.  He found fluoroprolines work best and that asymmetric counterions (e.g. TRIP) can greatly enhance ee in products (see List, JACS 2000)

Steven Ley – New tools for molecule makers: emerging technologies
This paper was on ways of using technology to improve the efficiency of organic synthesis.  He used immobilized reagents to make reactions cleaner, putting them in 'teabags' that could be removed from the reaction without need for filtering, chromatography etc.  He also used immobilized reagents in packed column flow systems, and was working towards doing this in microfluidic systems. Rates of reaction were found to be greatly enhanced in microfluidic systems, and such systems can be coupled directly to analysis and biological screening systems. Methods are very useful for optimizing reaction conditions and yields very quickly. Using these automated systems he stated that chemists could make 30 molecules in one night. Alternatively, a small scale reaction can be repeated many times with high efficiency to produce scale-up to 15kg in a normal research lab. (see Baumann Org. Lett. 2008, 8, 5231). 

Peter Seeberger – Microreactors as tools or organic synthesis
These systems were very versatile and could use or generate many different compounds and fast (4 reactions/day at 10-500mg scale).  The work was driven by the realization that the needs of discovery and process chemistry did not match. This mismatch is addressed by these microreactors because scale is achieved by simply running the reaction for as long as necessary to obtain the required amount of materials.  Problems with converting lab scale syntheses to large scale are avoided. He used Design of Experiments with computer-controlled microreactors to optimize reactions then carry out production runs.  Has assisted in developing commercial microreactor systems such as SYRIS (AFRICA), CPC (VAPOTECH) (see Helv. Chim. Acta 2005, 88, 1).

As with Ley, rate enhancements of up to an order of magnitude have been achieved in microreactors compared to normal bench reactions.  Reaction yields can be enhanced and some reactions can generate 10g/hr of product (see Chem. Eur. J., 2006, 12, 8434). Microreactors can be set up that are cheap and low tech., and the commercial microreactor chips cost E200, less for quantity. In some cases, PhD students have run 120 reactions in 3 afternoons.. Microreactors are especially suitable for exothermic/dangerous reactions, photochemical reactions, radical reactions, and for making excellent nanoparticles (see Seeberger, Chem. Comm. 2008, 9, 1100). 

Eric Meggers – Chemical biology with organometallics
This was a fascinating talk that started with the question "Why are natural products so complicated?". Generally the very complex scaffolds of natural products provide a relatively rigid scaffold onto which are arrayed the binding groups hat interact with biological targets.  His group uses metal complexes to generate scaffolds that may be analogous to those generated by Nature.  Octahedral complexes of metals can generate 30 stereoisomers providing a rich source of scaffold presentation.  The 3D structure of the complexes is encoded by the central metal atom.  Clearly, the choice of metal and of coordinative bonds must provide stability (some Pt and Ir complexes have biological stabilities of decades).  Meggers' group chose Ru because of its cost ($20/g) and low toxicity.  He made a metal complex analogue of staurosporine (a kinase inhibitor) which had an IC₅₀ value of <1 nM.  Optimizing this initial activity, he obtained kinase inhibitors with IC₅₀ values of <5 pM. He has applied this metal complex scaffold idea to a number of targets with equally impressive results (e.g. GSK3 in the Wnt signaling pathway).  The most potent mimetics have four ligands rather than six. (see Meggers, ChemBioChem 2006, 7). The complexes are easily synthesized, have low toxicity, good pharmacokinetics. 

Tom Blundell – Exploring biological and chemical space with high-throughput crystallography
His paper was essentially on the x-ray structure-based fragment method of design that he pioneered.  He discussed the immensity of chemical space, and the impractibility of trying to sample this with single molecules and libraries.  The fragment method allows more chemistry space to be explored because of the combinatorial efficiencies afforded by combinations of fragments (e.g. a binding site accommodating three fragments (assuming a fragment library of 300) could essentially sample 300³ or 27 million molecules. He discussed the application of fragment-based design to 'diseases of poverty' in particular. 

Douglas Kell – Drug and xenobiotic transport via membrane carriers
The main reasons for attrition of drug candidates are efficacy (40%), toxicity (20%), and pharmacokinetics/metabolism (20%). Kell developed a large metabolic network using an eight-compartment cell model with 1500 chemical species. Used SMILS to encode molecules and SBML (systems biology markup language) to model network. He discussed how the model could also be used to model transport and the role of lipophilicity, as accumulation of drugs in tissue can exacerbate toxicity. 

Sara Linse – Protein interaction, association, and fibrillation
This, and Michele Vendruscolo's talk (Life on the edge: Proteins are close to their solubility limits), concerned the concentrations of proteins in biological systems and the implications of these concentrations on protein behaviour such as folding, precipitation, and aggregate formation (esp. prions). Linse's talk was largely about how proteins overcome the Levinthal Paradox and fold in very short timescales. She explained rapid and efficient folding terms of discrimination between kinetics of contacts during the folding process.  She illustrated the process with respect to amyloid fibril formation. Vendruscolo's talk focused more on the astonishingly close connection between protein expression and aggregation propensities. 

Joseph Lehar – System biology from synergistic chemical combinations
This paper was synergistic to the ideas I presented in my talk on sparse representations of biological networks, and the need to understand the robustness and vulnerabilities of these networks in order to engineer the polypharmacy or promiscuity (side effects) of drugs to have the most beneficial outcomes. The latter issue has been the subject of several important papers in Nature, Science, etc in the past year. Lehar's company (CombinatoRx) is using these ideas to find combinations of registered drugs that produce a more beneficial outcome in disease treatments than single drugs. They have found many synergistic combinations, and have studies the impact of how the modulation of multiple targets produces an improved response from the biological network. They have a full pipeline of fifteen drug combinations in clinical trials. The fact that combinations of registered drugs are used substantially reduces the regulatory hurdle the drugs most overcome.

Dave Winkler