Reasonably Random Synthetic Biology at Amyris Tim Gardner - PowerPoint PPT Presentation

Reasonably Random Synthetic Biology at Amyris Tim Gardner Director, Research Programs & Operations October 27, 2010

Overview ► Amyris is an integrated renewable products company producing advanced renewable fuels and chemicals ► Founded in 2003 on principle of social responsibility: use our know-how to address biggest health and environmental challenges ► Public company (IPO September 2010) with R&D, Manufacturing and Distribution facilities in the Emeryville, CA, Campinas, Brazil & Chicago, IL 2

Amyris’ fouding product: Artemsinin Artemisinin is 95% effective against malaria The Challenge: Supplying Artemisinin Anti-Malarials Malaria causes: 1 to 3 million deaths per year Treating malaria would require: 300 to 500 million treatments per year Artemisinin treatments needed: 225 to 400 tons of artemisinin per year This would require: 6,000,000 tons of plant material 50X increase in production Total Chemical Synthesis 10X decrease in price too expensive 3

Non-profit effort to manufacture Artemsinin Glucose G6P FDP Artemisa G3P annua PEP idi DMAP PYR IPP Mevalonate ispA AcCoA Pathway FPP OAA CIT TCA Cycle MAL steroids quinones membranes Amorphadiene (arteminin precurser) 4

Strain performance targets reached Improvement 1 50 Improvement 2 Improvement 3 40 Improvement 4 Amorphadiene [g/L] 30 25g/L target 20 10 0 0.00 20.00 40.00 60.00 80.00 100.00 120.00 140.00 Time (hrs) • Sanofi-aventis now ramping production, formulation and product stability testing • Aim for world-wide distribution in 2012. 5

From drugs to fuels • Hydrocarbons, not alcohols or Isoprenoid technology esters platform capable of • Can be used in existing engines making more than with no performance trade-offs 50,000 molecules • Superior environmental profile – substantially lower greenhouse gas emissions than petroleum – No sulfur – Lower particulates and NOx • Can be delivered using existing distribution infrastructure Phase-contrast micrograph of Amyris engineered microbes producing precursor to Amyris Renewable Diesel 6

Diesel production Glucose G6P FDP G3P PEP idi DMAP PYR IPP Mevalonate ispA AcCoA Pathway FPP OAA CIT TCA Cycle MAL steroids quinones membranes Farenesene 7

Amyris Renewable Diesel: a better fuel Cloud Point (°C) Cetane Number Energy Density (cold temp operation) 1000 BTU/gal 40-55 115-142 -9 to-30 # 2-D # 2-D # 2-D +1 FAME FAME 47 FAME 118 < – 50 Amyris Amyris 58.1 Amyris 121 Diesel Diesel Diesel -75 -50 -25 0 0 20 40 60 0 50 100 150 Additional benefits of Amyris renewable diesel compared to #2-Diesel • 90%+ lower greenhouse gas emissions • No sulfur & produces lower NOx and particulate emissions • Registered with the EPA for 20% blends Note: Amyris diesel will be used in blends with conventional fuels; values shown for Amyris diesel is for our biomass derived blending component; 8 SME = Soy Methyl Esters

>$1 Trillion dollar market accessible • Renewable diesel $809B Farnesene – “plug-in” fuels – meets or exceeds stds – substantially lower emissions $48B • Lubricants fermentation – family of base oils – designed to be high performance biology • Consumer products >$50B – detergents chemistry – Cosmetics – fragrances • Polymers $337B By combining biology – adhesives – oxygen scavenger and chemistry, Biofene – toughening agent becomes a building block of renewable products for a diverse • Other applications set of applications – crop protection – many others 9

Lower cost of production enables access to larger markets 10

Low-cost production drives everything in strain R&D 2M ton/yr plant Capital cost-saving opportunities 8 Uncertain 19 15 14 9 1 18 13 10 Level of risk Unfamiliar Engineering decisions drive strain 16 3 6 performance criteria Multi-parameter strain 7 optimization problem 12 5 17 • yield Familiar 11 • productivity 4 2 • reduced media supplements • temperature • biocatalyst stability Short term Medium term Long term • GMM certification Time to value Amount of savings For fuel synthesis we aim to direct >90% of cell resources to the synthesis of byproducts under stringent productivity, temperature, and media conditions 11

12 like getting a toddler to eat salad

13 But we’ve made rapid progress Fene production Dec Nov Oct Sep Aug Jul 2009 Jun May Apr Fuels strain improvement since program start Mar Feb Jan Dec Nov Oct Sep Aug Jul 2008 Jun May Apr Mar Feb Jan (modified for fuel synthesis) Dec Nov Artemisinin base strain Oct Sep 2007 Aug Jul Jun May Apr

14 development mutagenesis engineering + process + rational + breeding Fene production Dec Nov Oct Sep How we got there Aug Jul 2009 Jun May Apr Mar Feb Jan Dec Nov Oct Sep Aug Jul 2008 Jun May Apr Mar Feb Jan (modified for fuel synthesis) Dec Nov Artemisinin base strain Oct Sep 2007 Aug Jul Jun May Apr

What can we learn from the mutants? Illumina paired-end sequencing performed by Prognosys, Sequence assembly & analysis by Amyris 15

Mutant family tree and performance gains improvement 0 E high X 18 D medium Mutant Strain X C 11 low H I 13 No change 7 G Causal mutations found in J •Post-translational regulation 8 •Cofactor synthesis Post child 29 18 16 Most genes we’d never A B F K L considered. None “on -2 pathway” 3.1 One we had tried rationally but not with the right M mutation O N 20 P 16 20

What about rational engineering? Synthetic Biology: the dream of plug and play biology Are electronics and machines the right paradigm? 17

The neutral chassis hypothesis Add a little synthetic biology 18

19 It works, but it’s not always pretty Biology is designed by natural selection

20 Too many parts kinda complexity

Promoter strength varies depending on its insertion site Gene Expression 3 4 4 1 2 1 3 1 2 3 3 4 3 4 2 1 2 1 2 Promoter E A D B C Genome locus 21

How much does diversity influence pathway production? Farnesene Pathway Sporulate (haploidize) 4 diverse hybrid haploids. Select for production 22

Fold increase in mevalonate titer over reference Impact of diversity on Mevalonate production Reference strain 23

24 Frequencies for top and bottom Mevalonate pools

A practical approach Pathway PoC Pathway Optimization Rational Semi-rational Random 1 2 3 Random Stoichiometry & mRNA Enzyme kinetics Context-effects & post- expression transcriptional regulation Routine Doable but hard Shooting in the dark We always start here We have targeted activities This is where most of the here when bottlenecks strain improvement become clear “action” is. 25

1. Rational: Modeling / Isotopomers Yeast metabolic database Output Input -Balanced models -Public models -Simulatable models -Yeastcyc -Matlab, Excel, GAMS format -Amyris knowledge -Experimental data 26

2. & 3. Industrialize strain improvement Capacity: Screen >70,000 strains / week Random Test >40 2L fermentations / week Mutagenesis Strain Engineering Downstream Rational Strain Screening Fermentation Processing Scale-Up Design (DSP) Knowledge Analytics Management 27

Continuous process improvement is critical Assuming constant absolute yield gain per improved mutant strain. • S/N will drop as yield increases. • So too must CV. 8.0% 7.0% Relative improvement 6.0% for a constant absolute yield gain 5.0% 4.0% CV required to detect winning mutant 3.0% 2.0% 1.0% 0.0% 5 10 15 20 25 30 Yield of parent strain 28

The value of process control Yield in 2L tanks Original strain The reward: screening assay Overall screening process CV <4% Relative production by shake plate assay Enables detection of 4% improvements w/ 5% FN and 5% FPs through 2 Yield in 2L tanks tiered screen New assay Relative production by shake plate assay 29

HT screening pipeline Process improvement is easier said than done 30

31 Diagnosing sources of variation

Better decisions via informatics integration LIMS systems is identifying and eliminating sources of error Y4921 Strain score Count Systematic drops in median plate titer traced to worn posts in one plate shaker 32

Multivariate optimization – picking winners Informatics integration is critical to good decisions (get data out of silos) winner Stress resistance 0.94 0.26 0.00 33

Informatics integration enables assessment of stress resistance Calculate Tank Testing Yield PV DB (PV Proj.) (PV Proj.) HTS DB Strain hit from Plate, pick, assay Screening (HTS proj.) (HTS Proj.) Calculator MAD App DB Filter, calculate, store, visualize (K2Y proj.) Plate re-testing (MAD Proj.) Data Warehouse Stress resistance 34 metric

Let the data guide Use empirical data mining to guide library construction, screening conditions, process dev. 35