De novo metabolic pathways for isoprene biosynthesis

PRIORITY This Application claims priority to U.S. Provisional Application No. 61/719,198, filed October 26, 2012, which is hereby incorporated by reference in its entirety.

BACKGROUND OF THE INVENTION

The present invention relates generally to biosynthetic processes, and more specifically to the design of non-naturally occurring pathways for producing isoprene and the creation of organisms having such biosynthetic capability. Isoprene is naturally produced by bacteria, animals, humans, and plants; these organisms collectively release an estimated 600 million tons of isoprene into the atmosphere each year. Isoprene (2-methyl- 1,3 -butadiene) is an important commodity chemical used in a wide range of industrial products, such as synthetic rubber for tires and coatings, adhesives, and specialty elastomers. Isoprene is also a versatile building block for the production of hydrocarbon fuels, including diesel, gasoline, and aviation fuels. Approximately 1 million tons of isoprene are made from petrochemical feedstocks every year. ¹ Increasing global demand for isoprene and environmental concerns about greenhouse gas emissions have spurred interest in the

development of a fermentative route for producing this chemical from renewable sugars.

Isoprene consumers are also interested in isoprene bioproduction as a strategy to mitigate their vulnerability to uncertain supply and volatile prices.

Isoprene synthases catalyze the elimination of pyrophosphate from dimethylallyl pyrophosphate (DMAPP) to yield isoprene. Despite isoprene 's pervasiveness in the environment, efforts to identify prokaryotic isoprene synthases have met with limited success. ² Thus far, only plant derived isoprene synthases have been well characterized. Sequence data for these enzymes exist for two plant families: kudzu (the Asian vine, Pueraria montana) and poplar (Populus). The catalytic efficiencies of the characterized isoprene synthases are sub-optimal for industrial applications (K _M ~ 1-10 mM and k _cat ~ 1 s ^"1 ). ¹ ' ³ Accordingly, protein engineering for superior kinetic parameters has been pursued by industrial groups. ^3"4 As a proof of concept, recombinant production of isoprene in microbial hosts has been achieved by expressing plant isoprene synthases in E. coli, S. cerevisiae, and photosynthetic bacteria. ⁵

There are two evolutionarily distinct pathways to biosynthesize the isoprenyl precursors isopentenyl pyrophosphate (IPP) and DMAPP (Fig 1). Archaea and non-plant eukaryotes use the mevalonate (MVA)-dependent pathway exclusively to convert the ubiquitous intermediate acetyl-coenzyme A (A-CoA) to IPP. Most prokaryotes use the l-deoxy-D-xylulose-5- phosphate/2-C-methyl-D-erythritol-4-phosphate (DXP/MEP) pathway to produce IPP and DMAPP from pyruvate and glyceraldehyde-3-phosphate derived from glycolysis. Plants use both the MVA and the DXP/MEP pathways for achieving terpenoid biosynthesis. As shown in Figure 2, the MVA and DXP/MEP pathways have different yields and co-factor requirements. (See

Figure 3 for glucose to isoprene calculations.) The MVA pathway offers a lower theoretical yield of IPP from glucose due to its co-factor imbalance. However, the MVA pathway has proven amenable to genetic and metabolic engineering that has resulted in higher isoprenoid titers in both bacteria and yeast. Although the DXP/MEP pathway provides a higher theoretical yield of isoprene from glucose (Fig 2), it is linked to essential cell functions and contains iron-sulfur cluster enzymes that are poorly characterized on a biochemical level. ⁶ The choice of which pathway to engineer for maximizing flux towards the universal precursors IPP and DMAPP is also dependent on the regulatory elements present in the desired fermentation host. It is important to note that accumulation of prenyl pyrophosphates is toxic to the host, ⁷ such that metabolic engineering experiments have to include a terpene synthase that will produce the volatile hydrocarbon product.

Genencor-DuPont and The Goodyear Tire & Rubber Co. are developing a high efficiency fermentative route for polymer grade isoprene. ¹ Their metabolic engineering strategy imported the heterologous MVA pathway into E. coli to increase the flux towards IPP and DMAPP as well as plant-derived isoprene synthases. The mevalonate pathway was chosen in preference to the DXP/MEP pathway because it is better characterized and has been exploited industrially for isoprenoid production in yeast and bacteria. ^7"8 The MVA pathway was cloned into E. coli as two synthetic operons - a top pathway converting A-CoA to MVA, and a bottom pathway producing DMAPP from MVA (this strategy is similar to that pursued by Martin et al. for the

overproduction of amorphadiene in yeast ⁷ ). The bottom operon was integrated into the chromosome under the control of a constitutive promoter. The top operon, the isoprene synthase and an additional copy of an archaeal mevalonate kinase were expressed on two different plasmids driven from the inducible Vtrc promoter. The reported parameters for this process consist of an isoprene yield of 0.1 1 g g ^"1 , volumetric productivity of 2.0 g L ^"1 h ^"1 and a titer of 60 g L -1.1

The volatile nature of isoprene (b.p. 34 °C) allows gas phase recovery of the product, which simplifies purification and eliminates potential feedback inhibition by virtue of product accumulation. All of these factors drive equilibrium in favor of isoprene synthesis, and enable in situ product removal.

SUMMARY OF INVENTION

Calculations of the maximal theoretical yield for fermentative isoprene production from glucose (Fig 3) reveal that both naturally occurring pathways offer sub-optimal yields [0.324 g g ^"1 (maximum), 0.298 g g ^"1 (DXP/MEP), 0.252 g g ^"1 (MVA)]. Furthermore, as shown in Figure 2, both MVA and DXP/MEP pathways operate optimally under aerobic conditions, thus increasing process costs associated with expensive aeration of large fermentors. Because substrate cost is a significant fraction of the total cost of the desired fermentation product, pathways with superior yields have a better chance of reaching commercialization. Moreover, an anaerobic pathway for isoprene, whilst unprecedented, would enable lower fermentation costs compared to the MVA and DXP/MEP pathways that have thus far been pursued in industry. Since an anaerobic pathway for isoprene at 0.324 g g ^" yield has not been reported in nature, we sought to formulate such a pathway based on existing reaction classes rather than limiting our search to enzymes working on their native substrates. We recognized that the common metabolite 2,3-dihydroxyisovalerate (DHIV) already contains the correct carbon skeleton for isoprene and is only two steps away from pyruvate derived from glycolysis (Fig 4). Herein, we propose 4 major isoprene biosynthetic pathways (Fig 5-1 1), comprising a series of reduction and dehydration steps to produce isoprene from DHIV. The proposed pathways are evaluated with respect to yield, redox balance, ATP balance, number of steps from glucose and number of unknown enzymes. This evaluation is summarized in Table 1.

Table 1. Evaluation of isoprene biosynthetic routes.

As shown in Table 1, all of the proposed pathways are redox balanced. Unlike the MVA and DXP/MEP routes, each of these proposed pathways achieves the maximum yield (0.324 g g ^"1 ). In addition, all 4 pathways described herein produce isoprene in the same number of steps as the native routes. The proposed pathways differ from one another most profoundly in the number of unknown enzymes required and the ATP yield. Pathway 1 (Fig 5) requires 5 new enzymes and is ATP neutral. Pathway 2 (Fig 7) requires 6 new enzymes and produces 1 net ATP. Pathways 3 and 4 (Fig 9-10) each require 7 new enzymes and yield 1 net ATP.

Pathway 1 requires the fewest new enzymes to be engineered, and is the most

thermodynamically favorable conversion of DHIV to isoprene. However, Pathways 2 to 4 all yield 1 equivalent of ATP per isoprene formed. This allows pathway optimization by metabolic evolution, in which selection in serial cultivation for growth improvement is correlated with increased pathway productivity. ⁹ Of the 3 pathways with a net positive ATP yield, Pathway 2 requires the fewest new enzymes. Therefore, Pathway 2 is likely to be the optimal route for isoprene synthesis. However, all pathways, and derivatives thereof, will be developed. BRIEF DESCRIPTION OF THE DRAWINGS

Figure 1: Mevalonate (MVA) and l-deoxy-D-xylulose-5-phosphate/2-C-methyl-D-erythritol-4- phosphate (DXP/MEP) pathways for isoprenoid biosynthesis. Figure 2: Theoretical yield calculations for isoprene biosynthesis via the MVA and the

DXP/MEP pathways.

Figure 3: Calculations of the maximum theoretical yield for isoprene fermentation from glucose. (1) Maximal yield if electrons are supplied exogenously (i.e. electrochemically or

photochemically). (1) + (2) Maximal yield for isoprene fermentation from glucose. Figure 4: Biosynthesis of 2,3-dihydroxyisovalerate from glucose. ALS, acetolactate synthase; KARI, ketol-acid reductoisomerase; PYR, pyruvate; DHIV, 2,3-dihydroxyisovalerate Figure 5: Biosynthesis of isoprene from DHIV via 3,3-dimethylmalate (pathway 1).

Figure 6: Transformations of linaool dehydratase-isomerase.

Figure 7: Biosynthesis of isoprene from DHIV via (3-methyl-2-oxobutanoyl)-CoA (pathway 2).

Figure 8: Alternative syntheses of isoprene from KIV (pathway 2).

Figure 9: Biosynthesis of isoprene from DHIV via (2,3-dihydroxy-3-methylbutanoyl)-CoA. Figure 10: Biosynthesis of isoprene from DHIV via 2-hydroxy-3-methyl-3-butenoic acid. Figure 11: Alternative syntheses of isoprene from DHIV. DETAILED DESCRIPTION OF THE INVENTION

The present invention is directed to the design and production of cells and organisms that have the ability to produce isoprene. The invention, in particular, relates to the design of a microbial organism capable of producing isoprene by introducing one or more exogenous nucleic acids encoding an isoprene pathway enzyme.

Definitions

As used herein, the terms "microbial," "microbial organism" or "microorganism" are intended to mean any organism that exists as a microscopic cell that is included within the domains of archaea, bacteria or eukarya. Therefore, the term is intended to encompass prokaryotic or eukaryotic cells or organisms having a microscopic size and includes bacteria, archaea and eubacteria of all species as well as eukaryotic microorganisms such as yeast and fungi. The term also includes cell cultures of any species that can be cultured for the production of a biochemical. As used herein, the term "non-naturally occurring" when used in reference to a microbial organism or microorganism of the invention is intended to mean that the microbial organism has at least one genetic alteration not normally found in a naturally occurring strain of the referenced species, including wild-type strains of the referenced species. Genetic alterations include, for example, modifications introducing expressible nucleic acids encoding metabolic polypeptides, other nucleic acid additions, nucleic acid deletions and/or other functional disruption of the microbial organism's genetic material. Such modifications include, for example, coding regions and functional fragments thereof, for heterologous, homologous or both heterologous and homologous polypeptides for the referenced species. Additional modifications include, for example, non-coding regulatory regions in which the modifications alter expression of a gene or operon. Exemplary metabolic polypeptides include enzymes or proteins within an isoprene biosynthetic pathway. A metabolic modification refers to a biochemical reaction that is altered from its naturally occurring state. Therefore, non-naturally occurring microorganisms can have genetic

modifications to nucleic acids encoding metabolic polypeptides, or functional fragments.

Exemplary metabolic modifications are disclosed herein.

As used herein, the term "CoA" or "coenzyme A" is intended to mean an organic cofactor or prosthetic group (nonprotein portion of an enzyme) whose presence is required for the activity of many enzymes (the apoenzyme) to form an active enzyme system.

As used herein, the term "substantially anaerobic" when used in reference to a culture or growth condition is intended to mean that the amount of oxygen is less than about 10% of saturation for dissolved oxygen in liquid media. The term also is intended to include sealed chambers of liquid or solid medium maintained with an atmosphere of less than about 1% oxygen.

"Exogenous" as it is used herein is intended to mean that the referenced molecule or the referenced activity is introduced into the host microbial organism. The term as it is used in reference to expression of an encoding nucleic acid refers to introduction of the encoding nucleic acid in an expressible form into the microbial organism. When used in reference to a biosynthetic activity, the term refers to an activity that is introduced into the host reference organism.

Therefore, the term "endogenous" refers to a referenced molecule or activity that is present in the host.

Similarly, the term when used in reference to expression of an encoding nucleic acid refers to expression of an encoding nucleic acid contained within the microbial organism. The term

"heterologous" refers to a molecule or activity derived from a source other than the referenced species whereas "homologous" refers to a molecule or activity derived from the host microbial organism. Accordingly, exogenous expression of an encoding nucleic acid of the invention can utilize either or both a heterologous or homologous encoding nucleic acid.

It is understood that when more than one exogenous nucleic acid is included in a microbial organism that the more than one exogenous nucleic acids refers to the referenced encoding nucleic acid or biosynthetic activity, as discussed above. Thus, the number of referenced exogenous nucleic acids or biosynthetic activities refers to the number of encoding nucleic acids or the number of biosynthetic activities, not the number of separate nucleic acids introduced into the host organism.

Pathway 1 (via (i?)-3.3-dimethylmalate)

Pathway 1 converts DHIV to isoprene via the sequence of steps described in Figure 5. In this embodiment DHIV is first dehydrated to 2-ketoisovalerate (KIV). KIV undergoes an NAD(P)H- dependent carboxylation to (i?)-3,3-dimethylmalate, which then undergoes ATP-dependent decarboxylation and dehydration to 3-methylbut-2-enoic acid. 3-methylbut-2-enoic acid is subsequently transformed to (3-methylbut-2-enoyl)-CoA in an ATP-dependent reaction. (3- methylbut-2-enoyl)-CoA undergoes two sequential NAD(P)H-dependent reductions to 3- methylbut-2-enal and 3-methylbut-2-en-l-ol, which is then dehydrated to produce isoprene. From DHIV, the net equation to isoprene in this embodiment is: (1) DHIV + 3 NAD(P)H + 3 H + 2 ATP - isoprene + 3 NAD(P) + 2 ADP + 2 Pi + 2 H ₂ 0

To calculate the yield of isoprene from glucose:

(2) C ₆ H ₁₂ 0 ₆ + 2 NAD ⁺ + 2 ADP + 2 Pi - 2 PYR + 2 ATP + 2 NADH + 2 H ⁺ + 2 H ₂ 0

(3) 2 PYR + NAD(P)H + 2 H ⁺ - DHIV + C0 ₂ + NAD ⁺

(4) 1/6 C ₆ Hi ₂ 0 ₆ + H ₂ 0 + 2 NAD ⁺ - C0 ₂ + 2 NADH + 2 H ⁺ and the net equation from glucose to isoprene is:

(5) 7/6 C ₆ Hi ₂ 0 ₆ - isoprene + 2 C0 ₂ + 3 H ₂ 0

The theoretical yield of isoprene from glucose in this embodiment is 0.324 g g ^"1 .

The first reaction in this sequence (step 1.1 in Fig 5) is catalyzed by the enzyme dihydroxyacid dehydratase (EC 4.2.1.9), which acts on DHIV as its natural substrate, and is a key enzyme in the biosynthesis of branched amino acids (L-valine, L-isoleucine, L-leucine). This enzyme is endogenous to industrially relevant hosts such as E. coli (gene ilvD), S. cerevisiae and

Corynebacterium glutamicum (gene ilvD). The E. coli dihydroxyacid dehydratase contains a catalytically active [4Fe-4S] cluster, which is unstable in the presence of dioxygen. Since the proposed pathway is designed to operate under anaerobic conditions, the E. coli enzyme should be compatible with the envisioned process. Step 1.2 is the reverse of the reaction catalyzed by dimethylmalate dehydrogenase (DMMD; EC 1.1.1.84) on its native substrate, (i?)-3,3-dimethylmalate. The endogenous reaction catalyzed by DMMD is the decarboxylation of (i?)-3,3-dimethylmalate to KIV. It is notable that this direction is less thermodynamically favorable than the reverse (http://equilibrator.weizmann.ac.il/, pH = 7.0):

(R)-3 ,3-dimethylmalate + NAD ⁺ <=> KIV + NADH + C0 ₂ K _eq = 0.19

KIV + C0 ₂ + NADH <=> (R)-3,3-dimethylmalate + NAD ⁺ K _eq = 5.2

DMMD is a 4 subunit, NADH-dependent enzyme, whose activity on (i?)-3,3-dimethylmalate, n- propyl malate, and malate has been identified in purified protein from Pseudomonas fluorescens and other Pseudomonas species. ¹⁰ Since the reverse reaction shown in step 1.2 is

thermodynamically favorable, and DMMD has been reported to accept multiple substrates, this enzymatic transformation should be a feasible step. The sequence of Pseudomonas DMMD has not been elucidated. Therefore, candidate genes for suitable DMMD activity can be found by searching the Pseudomonas genome database (http://wvvw.pseudomonas.coni/index.isp).

Candidate enzymes will be screened for activity on (i?)-3,3-dimethylmalate and KIV production, and engineered by directed evolution for higher affinity for KIV.

Step 1.3 describes the decarboxylation and dehydration of a 3-hydroxy acid. Mevalonate diphosphate decarboxylase (MDD; EC 4.1.1.33) catalyzes the ATP-dependent decarboxylation of mevalonate diphosphate to IPP, as part of the MVA pathway. This enzyme (gene MDD, MVD1) is endogenous to many organisms, including the industrially relevant host S. cerevisiae. MDD from S. cerevisiae has been reported to accept at least one nonnative substrate, 3-hydroxy- 3-methylbutyrate. ¹¹ In order to find a decarboxylase that is active on (i?)-3,3-dimethylmalate, i) S. cerevisiae MDD can be engineered towards (i?)-3,3-dimethylmalate or ii) several MDD homologs can be identified through a BLAST search and codon-optimized for E. coli for activity on (i?)-3,3-dimethylmalate. Taken together, steps 1.2 and 1.3 convert KlV to 3-methyl-2- butenoic acid, with the overall effect of a reduction and a dehydration.

Step 1.4 utilizes a CoA synthetase to make (3-methylbut-2-enoyl)-CoA, which can then be reduced to the aldehyde by a CoA-dependent semialdehyde dehydrogenase in step 1.5. ADP- forming succinate-CoA ligase (EC 6.2.1.5) is a candidate enzyme for catalyzing step 1.4, or for serving as a starting point for directed evolution experiments to create a (3-methylbut-2-enoyl)- CoA ligase. This enzyme is endogenous to many organisms, including E. coli, Acetobacter aceti, Advenella mimigardefordensis, Alcaligenes faecalis, Bacillus megaterium, Blastocystis sp., Brevibacterium linens, Calliphoridae, Columba livia, Drosophila melanogaster, Gallus gallus, Geobacillus stearothermophilus, Glycine max, Haloferax volcanii, Homo sapiens, Klebsiella aerogenes, Kurthia zopfii, natronobacterium gregoryi, natroconococcus occultus, Natronomonas pharaonis, Neocallimastix frontalis, Neocallimastix patricarium, Nitrosomonas europaea, paracoccus denitrificans, Pigeon, Pimelobacter simplex, Pseudomonas aeruginosa,

Pseudomonas fluorescens, Psedomonas stutzeri, Rattus norvegicus, Rhodobacter sphaeroides, Salmonella enterica subsp. Arizonae, Serratia marcescens, Spinacia oleracea, Sulfolobus acidocaldarius, Sulfolobus solfataricus, Sus scrofa, Thermococcus kodakarensis, Theroplasma acidophilum, Thermus aquaticus, thermos thermophilus, Toxoplasma gondii, and Trichomonas vaginalis. Acylating succinate semialdehyde dehydrogenase (EC 1.2.1..76), an NAD(P)H- dependent enzyme, is found in Clostridium kluyveri and Metallosphaera sedula. Purified enzyme from Clostridium kluyveri is air-sensitive, and presumably sensitive to oxygen. Since the proposed pathway is designed to operate under anaerobic conditions, this enzyme should be compatible with the envisioned process. The net result of steps 1.4 and 1.5 is the NAD(P)H- and ATP-dependent reduction of an unsaturated carboxylic acid to an α,β-unsaturated aldehyde. Theoretically, a single enzyme, aldehyde dehydrogenase (EC 1.2.1.3), could catalyze this reduction using NAD(P)H. However, such reactions are thermodynamically unfavorable and require ATP hydrolysis to shift the position of equilibrium towards the aldehyde. This is illustrated below using 2-butenoic acid as an example of an unsaturated carboxylic acid (htt : // equilibrator .weizmann . ac . i V, pH = 7.0):

2-butenoic acid + NADPH <=> 2-butenal + NADP ⁺ + H ₂ 0 K _eq = 7.4 x 10 ^"8

2-butenoic acid + NADPH + ATP <=> 2-butenal + NADP ⁺ + ADP + Pi K _eq = 0.013

2-butenoic acid + NADPH + ATP <=> 2-butenal + NADP ⁺ + AMP + PPi K _eq = 0.17

Aldehyde dehydrogenases that catalyze these ATP-dependent reactions have not been described. Notably, carboxylate reductase from Nocardia sp. (EC 1.2.99.6) catalyzes the Mg ²⁺ -, ATP-, and

NADPH-dependent reduction of carboxylic acids to aldehydes, forming AMP. Because the enzyme is post-translationally phosphopantetheinylated, heterologous expression in E. coli requires co-expression with the phosphopantetheine transferase from Nocardia for maximum activity. ¹² Carboxylate reductase is a 128 kDa monomeric protein comprising three domains: adenylating, phosphopantetheine attachment and reductase domains. Bound carboxylic acids are first converted to acyl adenylate intermediates in the adenylating domain and are subsequently reduced to aldehydes by NADPH in the reductase domain. Carboxylate reductase exhibits broad substrate acceptance, including various aromatic and aliphatic carboxylic acids and 2-oxoacids (e.g. benzoate, vanilic acid, ferulic acid, a-ketoglutarate, aconitate, citrate, maleate). ¹³ At least two strategies can be implemented in order to find a carboxylate reductase that is active on 3- methyl-2-butenoic acid: i) engineer the Nocardia enzyme towards 3-methyl-2-butenoic acid or ii) screen several homologous enzymes, identified through a protein database search such as BLAST for activity on 3-methyl-2-butenoic acid. These candidate enzymes could be codon- optimized for expression in a recombinant host such as E. coli, and engineered for improved activity on 3-methyl-2-butenoic acid. Replacing steps 1.4 and 1.5 with a carboxylate reductase step is less favorable for the overall transformation described in this embodiment. Carboxylate reductase produces AMP as a result of ATP hydrolysis, instead of the ADP that is used during glycolysis. This results in an adenosine phosphate imbalance that is rectified by adenylate kinase, an enzyme endogenous to E. coli. Adenylate kinase catalyzes the following reaction:

ATP + AMP --> 2 ADP which restores adenosine phosphate balance at the expense of an additional ATP hydrolysis step. Therefore, the transformation of glucose to isoprene using carboxylate reductase in pathway 1 would be ATP deficient. We believe that the use of CoA ligase and semialdehyde dehydrogenase to reduce carboxylic acids to aldehydes is the superior route for this embodiment.

Step 1.6, aldehyde reduction to an alcohol, can be achieved by alcohol dehydrogenases (EC 1.1.1.1 and 1.1.1.2). These enzymes typically exhibit broad substrate specificity. ¹⁴ Furthermore, E. coli has a large number of genes annotated as alcohol dehydrogenases: E. coli K-12 has about 100 dehydrogenase genes (10% of all enzymes). ¹⁵ Accordingly, others have reported that native E. coli enzymes are often capable of achieving the desired reduction. For instance, Lamm et al. used carboxylate reductase to reduce vanillic acid to vanillin and found that alcohol

dehydrogenases endogenous to E. coli further reduced vanillin to vanillyl alcohol. ¹⁶ Likewise, Yim et al. reported that in an artificial pathway for 1 ,4-butanediol, an alcohol dehydrogenase native to E. coli catalyzes the final step in the pathway, converting 4-hydroxybutyraldehyde to

17

1 ,4-butanediol. It is noteworthy that 3-methylbutanal reductase (EC 1.1.1.265) from S.

cerevisiae has demonstrated activity on a number of straight chain and branched aldehydes, and

18

would thus be a good candidate for catalyzing step 1.5. Therefore, we anticipate that step 1.6 is not likely to be a significant bottleneck in the overall effort to engineer the proposed artificial isoprene pathway.

3-methylbut-2-en-l-ol, the product of step 1.6, is at the same redox level as isoprene, such that a single dehydration (step 1.7) can convert this intermediate into the final product. Hydro lyases, the class of enzymes catalyzing water addition to a C=C bond and also the reverse dehydration, usually require a carbonyl group conjugated to the C=C bond. ¹⁹ An exception is oleate hydratase (EC 4.2.1.53), which occurs in many organisms but was only recently purified from

Elizabethkingia meningoseptica. The gene has been cloned and expressed in E. coli. ²⁰ The enzyme is a monomer of 73 KDa and contains a nonessential calcium ion. The reaction mechanism remains to be elucidated. Unfortunately, oleate hydratase has a very narrow substrate

21

scope.

Marliere et al. cloned and expressed in E. coli some of the 165 homologues to the E.

20

meningoseptica sequence (that had been traced by Bevers et al. ) and found dehydration activity on several isoprene precursors. ²² Information about the types of enzyme or amounts of isoprene were not disclosed. A drawback of this approach is that the oleate hydratase required a

"cofactor" (e.g. octanoic acid) for steric and/or electronic complementation in its catalytic site since the alcohol used as the substrate was significantly shorter than the fatty acid which functions as the natural substrate (e.g. oleate). We propose the directed evolution of oleate hydratase towards 3-methylbut-2-en-l-ol via a substrate walk approach, where the enzyme is progressively evolved to accept shorter and shorter alcohols, will be able to eliminate the need for the "cofactor" whilst maintain (maintaining? retain or retaining?) wild-type-like catalytic parameters.

Prenyl isoflavanoid hydratases can be used instead of oleate hydratase to catalyze step 1.7. These fungal enzymes are involved in secondary metabolism (defense mechanism) and are thus thought to be more accommodating in terms of substrate acceptance. ¹⁹ Marliere et al. have described using kievitone hydratase (EC 4.2.1.95) and phaseollidin hydratase (EC 4.2.1.97) for making isoprene, but again this approach has the added complication of requiring added "cofactors" to fill the active site. ²³

We propose to use linalool dehydratase-isomerase (EC 4.2.1.127) to catalyze step 1.7. Linalool dehydratase-isomerase from Castellaniella defragrans is a 40 kDa enzyme with no cofactors that catalyzes two reactions: i) the isomerization of geraniol to linalool and ii) the dehydration of linalool to myrcene (Figure 6). The enzyme has been overexpressed in E. coli and is well- characterized biochemically. As shown in Figure 6, the target product isoprene has the same geometry as myrcene but is 5 carbons shorter (i.e. it is a hemiterpene instead of a monoterpene). We thus propose to engineer linalool dehydratase-isomerase to accept 3-methylbut-2-en-l-ol as a substrate.

Pathway 2 (via (3-methyl-2-oxobutanoyl)-CoA)

Pathway 2 converts DHIV to isoprene via the sequence of steps described in Figures 7 and 8. In this embodiment DHIV is dehydrated to KIV, which is then converted to (3-methyl-2- oxobutanoyl)-CoA in an ATP-dependent reaction. (3-methyl-2-oxobutanoyl)-CoA undergoes three sequential NAD(P)H-dependent reductions to 3-methyl-2-oxobutanal, 1 -hydroxy- 3 -methyl- butan-2-one and 3-methylbutane-l ,2-diol. Two subsequent dehydrations convert 3- methylbutane-l ,2-diol to 3-methylbut-2-en-l-ol and finally to isoprene. From DHIV, the net equation to isoprene in this embodiment is:

(6) DHIV + 3 NADH + 3 H ⁺ + ATP isoprene + 3 NAD ⁺ + ADP + Pi + 3 H ₂ 0

To calculate the yield of isoprene from glucose: (2) C ₆ Hi ₂ 0 ₆ + 2 NAD ⁺ + 2 ADP + 2 Pi ^ 2 PYR + 2 ATP + 2 NADH + 2 H ⁺ + 2 H ₂ 0 (3) 2 PYR + NAD(P)H + 2 H ⁺ - DHIV + C0 ₂ + NAD ⁺ (4) 1/6 C ₆ Hi ₂ 0 ₆ + H ₂ 0 + 2 NAD - C0 ₂ + 2 NADH + 2 H and the net equation from glucose to isoprene is: (7) 7/6 C ₆ Hi ₂ 0 ₆ + ADP + Pi - isoprene + 2 C0 ₂ + ATP + 4 H ₂ 0

The theoretical yield of isoprene from glucose in this embodiment is 0.324 g g ¹ . The overall transformation from glucose to isoprene in Pathway 2 yields 1 equivalent of ATP per isoprene formed, which will allow pathway optimization by metabolic evolution. ⁹

The first reaction in this sequence (step 2.1 in Fig 7) is catalyzed by the enzyme dihydroxyacid dehydratase (EC 4.2.1.9), which acts on DHIV as its natural substrate. Dihydroxyacid dehydratase is endogenous to industrially relevant hosts and is suitable to operate under the anaerobic conditions described in the envisioned process.

Steps 2.2 and 2.3 represent the NAD(P)H and ATP-dependent net reduction of a 2-oxoacid to a 2-oxoaldehyde. ADP-forming succinate-CoA ligase (EC 6.2.1.5) and acylating succinate semialdehyde dehydrogenase (EC 1.2.1.76) are candidate enzymes for catalyzing these steps, or for serving as a starting point for directed evolution experiments. The CoA-dependent succinate semialdehyde dehydrogenase from Clostridium kluyveri or Metallosphaera sedula can potentially catalyze the reduction of (3-methylbut-2-oxobutanoyl)-CoA to 3-methyl-2- oxobutanal. Although Clostridium kluyveri acylating succinate semialdehyde dehydrogenase may be sensitive to oxygen, this is compatible with the anaerobic process described herein. The AMP-forming enzyme carboxylate reductase from Nocardia sp. (EC 1.2.99.6) or its homologs may be engineered for activity on KIV to produce 3-methyl-2-oxobutanal in one step. This route is not ideal for the overall biotransformation of glucose to isoprene, however, because the ATP hydrolysis required for AMP phosphorylation lowers the net ATP yield to 0.

Steps 2.4 and 2.5, aldehyde and ketone reductions to alcohols, can be achieved by alcohol dehydrogenases (EC 1.1.1.1 and 1.1.1.2), or 3-methylbutanal reductase (EC 1.1.1.265).

The product of step 2.5, 3-methylbutane-l ,2-diol, is already at the same redox level as isoprene, such that two sequential dehydrations (steps 2.6 and 2.7) can convert this intermediate into the final product. Hydro lyases, including oleate hydratase (EC 4.2.1.53), kievitone hydratase (EC 4.2.1.95), phaseollidin hydratase (EC 4.2.1.97), and their homologs may be engineered towards activity on 3-methylbutane-l,2-diol and 3-methylbut-2-en-l-ol. Linalool dehydratase-isomerase (EC 4.2.1.127) may be ideal to catalyze step 2.7, given its biochemical characterization in E. coli and lack of cofactor requirement. We thus propose to engineer linalool dehydratase-isomerase to accept 3-methylbut-2-en-l-ol as a substrate.

Figure 8 shows several alternative versions for pathway 2 that convert KIV to isoprene via different permutations of reductions and dehydrations. Proposed enzymes for each step are summarized in Table 2. All versions of pathway 2 yield 1 equivalent of ATP per isoprene formed. Table 2. Enzymes for alternative syntheses of isoprene from KIV.

2.18 2 -hydroxy- 3- (2-hydroxy-3- ADP-forming succinate various methylbutanoic acid methylbutanoyl)-CoA succinate-CoA

ligase (EC 6.2.1.5)

Pathway 3 (via (2,3-dihydroxy-3-methylbutanoyl)-CoA)

Pathway 3 converts DHIV to isoprene via the sequence of steps described in Figure 9. In this embodiment DHIV is first converted to (2,3-dihydroxy-3-methylbutanoyl)-CoA in an ATP- dependent reaction. (2,3-dihydroxy-3-methylbutanoyl)-CoA undergoes an NAD(P)H-dependent reduction to 2,3-dihydroxy-3-methylbutanal, which is dehydrated to 3-methyl-2-oxobutanal. 3- methyl-2-oxobutanal undergoes two sequential NAD(P)H-dependent reductions to 2-hydroxy-3- methylbutanal and 3-methylbutane-l ,2-diol. Two subsequent dehydrations convert 3- methylbutane-l ,2-diol to 3-methylbut-2-en-l-ol and finally to isoprene. From DHIV, the net equation to isoprene in this embodiment is:

(6) DHIV + 3 NADH + 3 H ⁺ + ATP isoprene + 3 NAD ⁺ + ADP + Pi + 3 H ₂ 0 and the net equation from glucose to isoprene is:

(7) 7/6 C ₆ Hi ₂ 0 ₆ + ADP + Pi isoprene + 2 C0 ₂ + ATP + 4 H ₂ 0

The theoretical yield of isoprene from glucose in this embodiment is 0.324 g g ^"1 . The overall transformation from glucose to isoprene in Pathway 3 yields 1 equivalent of ATP per isoprene formed, which will allow pathway optimization by metabolic evolution. ⁹

Steps 3.1 and 3.2 are catalyzed by the enzymes ADP-forming succinate-CoA ligase (EC 6.2.1.5), acylating succinate semialdehyde dehydrogenase (EC 1.2.1.76), or their homologs, engineered for activity on DHIV and (2,3-dihydroxy-3-methylbutanoyl)-CoA, respectively. 2,3-dihydroxy- 3-methylbutanal may be formed directly from DHIV by the AMP-forming enzyme carboxylate reductase from Nocardia sp. (EC 1.2.99.6) or its homologs. This route is not ideal for the overall biotransformation of glucose to isoprene, however, because the ATP hydrolysis required for AMP phosphorylation lowers the net ATP yield to 0.

Step 3.3 is the dehydration of an a-hydroxy aldehyde to a dialdehyde, which may be catalyzed by the enzymes dihydroxyacid dehydratase (EC 4.2.1.9) or dioldehydratase (EC 4.2.1.28). 2,3- dihydroxy-3-methylbutanal, the substrate for step 3.3, is very similar to DHIV, the natural dihydroxyacid dehydratase substrate. Therefore, it is likely that this enzyme can be engineered for activity on 2,3-dihydroxy-3-methylbutanal. This is particularly advantageous, as

dihydroxyacid dehydratase is present in industrially relevant hosts such as E. coli and S.

cerevisiae, and is suitable to operate under the anaerobic conditions described in the envisioned process. Dioldehydratases are coenzyme B12-dependent, and have been identified in many organisms, including Acetobacterium sp., Citrobacter freundii, Clostridium glycolicum,

Flavobacterium sp., Klebsiella pneumonia, Lactobacillus brevis, Lactobacillus buchneri,

Lactobacillus collinoides, Lactobacillus reuteri, Proprionibacterium freudenreichii, Salmonella enterica, and Salmonella enterica subsp. enterica serovar Typhimurium. Dioldehydratases have been expressed in E. coli, and have been reported to accept a range of substrates, including glycerol, 1 ,2 propanediol, 1,2-butanediol, and 2,3-butanediol. ²⁴ Dioldehydratase has been reported to be sensitive to oxygen. Since the proposed pathway is designed to operate under anaerobic conditions, this enzyme should be compatible with the envisioned process.

Steps 3.4 and 3.5, aldehyde and ketone reductions to alcohols, can be achieved by alcohol dehydrogenases (EC 1.1.1.1 and 1.1.1.2) or 3-methylbutanal reductase (EC 1.1.1.265).

Hydrolyases, including oleate hydratase (EC 4.2.1.53), kievitone hydratase (EC 4.2.1.95), phaseollidin hydratase (EC 4.2.1.97), and linalool dehydratase-isomerase (EC 4.2.1.127) may be used to catalyze the sequential dehydration reactions of steps 3.6 and 3.7. Pathway 4 (via 2-hydroxy-3-methyl-3-butenoic acid)

Pathway 4 converts DHIV to isoprene via the sequence of steps described in Figure 10. In this embodiment DHIV is first dehydrated to 2-hydroxy-3-methyl-3-butenoic acid (step 4.1), which is then converted to (2-hydroxy-3-methylbut-3-enoyl)-CoA in an ATP and CoA-dependent reaction (step 4.2). (2-hydroxy-3-methylbut-3-enoyl)-CoA undergoes two sequential NAD(P)H- dependent reductions to 2-hydroxy-3-methyl-3-butenal (step 4.3) and 3-methyl-3-buten-l ,2-diol (step 4.4). 3-methyl-3-buten-l ,2-diol is dehydrated to 3-methyl-3-butenal (step 4.5), which is then reduced in an NAD(P)H-dependent reaction to 3-methylbut-3-en-l-ol (step 4.6). 3- methylbut-3-en-l-ol is dehydrated to produce isoprene (step 4.7). The enzymes used for transformation of identified substrates to products include: 4.1) hydrolyases, including oleate hydratase (EC 4.2.1.53), kievitone hydratase (EC 4.2.1.95), phaseollidin hydratase (EC 4.2.1.97), or linalool dehydratase-isomerase (EC 4.2.1.127); 4.2) ADP-forming succinate-CoA ligase (EC 6.2.1.5); 4.3) acylating succinate semialdehyde dehydrogenase (EC 1.2.1.76); 4.4) alcohol dehydrogenase (EC 1.1.1.1 and 1.1.1.2) or 3-methylbutanal reductase (1.1.1.265); 4.5) dihydroxyacid dehydratase (4.2.1.9) or dioldehydratase (EC 4.2.1.28); 4.6) alcohol

dehydrogenase (EC 1.1.1.1 and 1.1.1.2) or 3-methylbutanal reductase (1.1.1.265); and 4.7) hydrolyases, including oleate hydratase (EC 4.21.53), kievitone hydratase (EC 4.2.1.95), phaseollidin hydratase (EC 4.2.1.97), or linalool dehydratase-isomerase (EC 4.2.1.127). From DHIV, the net equation to isoprene in this embodiment is: (6) DHIV + 3 NADH + 3 H ⁺ + ATP isoprene + 3 NAD ⁺ + ADP + Pi + 3 H ₂ 0 and the net equation from glucose to isoprene is:

(7) 7/6 C ₆ Hi ₂ 0 ₆ + ADP + Pi isoprene + 2 C0 ₂ + ATP + 4 H ₂ 0

The theoretical yield of isoprene from glucose in this embodiment is 0.324 g g ^"1 . The overall transformation from glucose to isoprene in Pathway 4 yields 1 equivalent of ATP per isoprene formed, which will allow pathway optimization by metabolic evolution. ⁹ Figure 11 shows several alternative versions for pathway 4 that convert DHIV to isoprene via different permutations of reductions and dehydrations. Proposed enzymes for each step are summarized in Table 3. All versions of pathway 4 yield 1 equivalent of ATP per isoprene formed, which will allow pathway optimization by metabolic evolution. ⁹ Table 3. Enzymes for alternative syntheses of isoprene from DHIV.

References

1. Whited, G. M.; Feher, F. J.; Benko, D. A.; Cervin, M. A.; Chotani, G. K.; McAuliffe, J. C; La Duca, R. J.; Ben-Shoshan, E. A.; Sanford, K. J., TECHNOLOGY UPDATE:

Development of a gas-phase bioprocess for isoprene-monomer production using metabolic pathway engineering. Industrial Biotechnology 2010, 6 (3), 152-163.

2. Julsing, M. K.; Rijpkema, M.; Woerdenbag, H. J.; Quax, W. J.; Kayser, O., Functional analysis of genes involved in the biosynthesis of isoprene in Bacillus subtilis. Applied

Microbiology and Biotechnology 2007 ', 75 (6), 1377-1384.

3. Cervin, M. A.; Whited, G. M.; Miasnikov, A.; Peres, C. M.; Weyler, W.; Wells, D. H.; Bott, R. R. Plant isoprene synthase variants for improved microbial production of isoprene. 2009-US41581

2009132220, 20090423., 2009.

4. Beck, Z. Q.; Estell, D. A.; Rife, C. L.; Wells, D. H.; Miller, J. V. Isoprene synthase variants for improved production of isoprene. 2011-US58188

2012058494, 20111027., 2012.

5. (a) Miller, B.; Oschinski, C; Zimmer, W., First isolation of an isoprene synthase gene from poplar and successful expression of the gene in Escherichia coli. Planta 2001, 213 (3), 483- 487; (b) Lindberg, P.; Park, S.; Melis, A., Engineering a platform for photosynthetic isoprene production in cyanobacteria, using Synechocystis as the model organism. Metabolic Engineering 2010, 12 (1), 70-79; (c) Hong, S.-Y.; Zurbriggen Andreas, S.; Melis, A., Isoprene hydrocarbons production upon heterologous transformation of Saccharomyces cerevisiae. Journal of applied microbiology 2012.

6. (a) Rohdich, F.; Kis, K.; Bacher, A.; Eisenreich, W., The non-mevalonate pathway of isoprenoids: genes, enzymes and intermediates. Current Opinion in Chemical Biology 2001, 5

(5), 535-540; (b) Eisenreich, W.; Bacher, A.; Arigoni, D.; Rohdich, F., Biosynthesis of isoprenoids via the non-mevalonate pathway. Cellular and Molecular Life Sciences 2004, 61 (12), 1401-1426.

7. Martin, V. J. J.; Pitera, D. J.; Withers, S. T.; Newman, J. D.; Keasling, J. D., Engineering a mevalonate pathway in Escherichia coli for production of terpenoids. Nature Biotechnology

2003, 21 (7), 796-802.

8. Westfall, P. J.; Pitera, D. J.; Lenihan, J. R.; Eng, D.; Woolard, F. X.; Regentin, R.;

Horning, T.; Tsuruta, H.; Melis, D. J.; Owens, A.; Fickes, S.; Diola, D.; Benjamin, K. R.;

Keasling, J. D.; Leavell, M. D.; McPhee, D. J.; Renninger, N. S.; Newman, J. D.; Paddon, C. J., Production of amorphadiene in yeast, and its conversion to dihydroartemisinic acid, precursor to the antimalarial agent artemisinin. Proceedings of the National Academy of Sciences of the United States of ^' America 2012, 109 (3), E11 1-E118, SE11 1/1-SEl 11/13.

9. (a) Zhang, X.; Jantama, K.; Moore, J. C; Shanmugam, K. T.; Ingram, L. O., Production of L-alanine by metabolically engineered Escherichia coli. Applied Microbiology and

Biotechnology 2007, 77 (2), 355-366; (b) Zhang, X.; Jantama, K.; Moore, J. C; Jarboe, L. R.; Shanmugam, K. T.; Ingram, L. O., Metabolic evolution of energy-conserving pathways for succinate production in Escherichia coli. Proceedings of the National Academy of Sciences of the United States of America 2009, 106 (48), 20180-20185, S20180/1-S20180/7.

10. (a) Lahdesmaki, M.; Mantsala, P., Comparison of D-malate and (E<,(E<-dimethylmalate dehydrogenases from Pseudomonas fluorescens UK- 1. Biochimica et Biophysica Acta,

Enzymology 1980, 613 (2), 266-74; (b) Mantsala, P., Purification of pantoate and dimethylmalate dehydrogenase from Pseudomonas fluorescens UK-1. Biochimica et Biophysica Acta,

Enzymology 1978, 526 (1), 25-33; (c) Magee, P. T.; Snell, E. E., The bacterial degradation of pantothenic acid. IV. Enzymic conversion of aldopantoate to (E±-ketoisovalerate. Biochemistry 1966, 5 (2), 409-16.

1 1. Gogerty, D. S.; Bobik, T. A., Formation of isobutene from 3-hydroxy-3-methylbutyrate by diphosphomevalonate decarboxylase. Applied and Environmental Microbiology 2010, 76 (24), 8004-8010.

12. Venkitasubramanian, P.; Daniels, L.; Rosazza, J. P. N., Reduction of Carboxylic Acids by Nocardia Aldehyde Oxidoreductase Requires a Phosphopantetheinylated Enzyme. Journal of Biological Chemistry 2007, 282 (1), 478-485.

13. Venkitasubramanian, P.; Daniels, L.; Rosazza, J. P. N., Biocatalytic reduction of carboxylic acids: mechanism and applications. Biocatalysis in the Pharmaceutical and

Biotechnology Industries 2007, 425-440.

14. Hirano, J.-i.; Miyamoto, K.; Ohta, H., Purification and characterization of the alcohol dehydrogenase with a broad substrate specificity originated from 2-phenylethanol-assimilating

Brevibacterium sp. KU 1309. Journal of Bioscience and Bioengineering 2005, 100 (3), 318-322.

15. Riley, M.; Abe, T.; Arnaud, M. B.; Berlyn, M. K. B.; Blattner, F. R.; Chaudhuri, R. R.; Glasner, J. D.; Horiuchi, T.; Keseler, I. M.; Kosuge, T.; Mori, H.; Perna, N. T.; Plunkett, G., Ill; Rudd, K. E.; Serres, M. H.; Thomas, G. H.; Thomson, N. R.; Wishart, D.; Wanner, B. L., Escherichia coli K- 12: a cooperatively developed annotation snapshot-2005. Nucleic Acids Research 2006, 34 (1), 1 -9.

16. Lamm, A. S.; Khare, A.; Rosazza, P. N., Reduction of carboxylic acids by carboxylic acid reductase heterologously expressed in Escherichia coli. Practical Methods for Biocatalysis and Biotransformations 2010, 295-298.

17. Yim, H.; Haselbeck, R.; Niu, W.; Pujol-Baxley, C; Burgard, A.; Boldt, J.; Khandurina, J.; Trawick, J. D.; Osterhout, R. E.; Stephen, R.; Estadilla, J.; Teisan, S.; Schreyer, H. B.;

Andrae, S.; Yang, T. H.; Lee, S. Y.; Burk, M. J.; Van Dien, S., Metabolic engineering of Escherichia coli for direct production of 1 ,4-butanediol. Nat Chem Biol 2011, 7 (7), 445-452.

18. van Iersel, M. F. M.; Eppink, M. H. M.; van Berkel, W. J. FL; Rombouts, F. M.; Abee, T., Purification and characterization of a novel NADP-dependent branched-chain alcohol dehydrogenase from Saccharomyces cerevisiae. Applied and Environmental Microbiology 1997, 63 (10), 4079-4082.

19. Jin, J.-F.; Hanefeld, U., The selective addition of water to C:C bonds; enzymes are the best chemists. Chemical Communications (Cambridge, United Kingdom) 2011, 47 (9), 2502- 2510.

20. Bevers, L. E.; Pinkse, M. W. FL; Verhaert, P. D. E. M.; Hagen, W. R., Oleate hydratase catalyzes the hydration of a nonactivated carbon-carbon bond. Journal of Bacteriology 2009, 191 (15), 5010-5012.

21. Kisic, A.; Miura, Y.; Schroepfer, G. L, Jr., Oleate hydratase. Substrate specificity. Lipids 1971, 6 (8), 541-5.

22. Marliere, P. Alkene production from an alcohol catalyzed by oleate hydratase. 2010- EP70103

201 1076691 , 20101217., 201 1.

23. Marliere, P. Alkene production from an alcohol catalyzed by a prenyl isoflavanoid hydratase. 2010-EP70095

201 1076689, 20101217., 201 1. 24. (a) Toraya, T.; Fukui, S., Coenzyme B 12-dependent propanediol dehydratase system. Ternary complex between apoenzyme, coenzyme, and substrate analog. Biochimica et

Biophysica Acta, Enzymology 1972, 284 (2), 536-48; (b) Toraya, T.; Shirakashi, T.; Kosuga, T.; Fukui, S., Substrate specificity of coenzyme B 12-dependent diol dehydrase: glycerol as both a good substrate and a potent inactivator. Biochem Biophys Res Commun 1976, 69 (2), 475-80; (c) Sauvageot, N.; Pichereau, V.; Louarme, L.; Hartke, A.; Auffray, Y.; Laplace, J.-M., Purification, characterization and subunits identification of the diol dehydratase of Lactobacillus collinoides. European Journal of Biochemistry 2002, 269 (22), 5731 -5737.

25. (a) Schramm, E.; Schink, B., Ether-cleaving enzyme and diol dehydratase involved in anaerobic polyethylene glycol degradation by a new Acetobacterium sp. Biodegradation 1991, 2

(2), 71 -9; (b) Hartmanis, M. G. N.; Stadtman, T. C, Solubilization of a membrane -bound diol dehydratase with retention of EPR g = 2.02 signal by using 2-(N- cyclohexylamino)ethanesulfonic acid buffer. Proceedings of the National Academy of Sciences of the United States of America 1987, 84 (1), 76-9; (c) Kinoshita, K.; Kawata, M.; Ogura, K.-i.; Yamasaki, A.; Watanabe, T.; Komoto, N.; Hieda, N.; Yamanishi, M.; Tobimatsu, T.; Toraya, T., Histidine-(E±143 Assists 1 ,2-Hydroxyl Group Migration and Protects Radical Intermediates in Coenzyme B12-Dependent Diol Dehydratase. Biochemistry 2008, 47 (10), 3162-3173; (d) Moore, K. W.; Bachovchin, W. W.; Gunter, J. B.; Richards, J. FL, Hydrogen transfer in catalysis by adenosylcobalamin-dependent diol dehydratase. Biochemistry 1979, 18 (13), 2776-82.