Development of an engineered probiotic for the treatment of branched chain amino acid related metabolic diseases

1 Metabolic dysfunction arising from missing or impaired enzymes comprising the 2 branched chain amino acid (BCAA) degradation pathway, especially those involving 3 leucine, can result in the accumulation of toxic metabolic intermediates and cause 4 severe metabolic disease. Removal of dietary BCAAs via their degradation by 5 engineered microbes could be a viable approach to prevent BCAA-mediated disease 6 sequelae. In this article, we describe the design and construction of an engineered 7 leucine degrading strain of E. coli Nissle 1917, the improvement of the degradation 8 pathway through high throughput screening, and the demonstration of strain activity in 9 animal models monitored by disease and strain-specific biomarkers. This work provides 10 a path for the development of engineered probiotic bacterial strains as a treatment for 11 BCAA-related metabolic diseases and disorders in humans. 12


Introduction
Human proteins are comprised of 20 amino acids, 9 of which are deemed essential 14 because they cannot be synthesized by the human body. Essential amino acids must 15 therefore be obtained from the diet. The branched-chain amino acids (BCAAs) leucine 16 (Leu), isoleucine, and valine belong to this group [1]. BCAAs are reported to promote 17 protein synthesis, improve muscle mass production, stimulate post-exercise recovery, 18 enhance immune function, and improve insulin secretion, most likely through the 19 rapamycin (mTOR) signaling pathway ( Figure 1A) [2][3][4][5][6]. BCAAs from dietary sources are 20 largely absorbed via the gut intestinal tract, bypassing the liver, and delivered to the 21 peripheral tissues [7][8][9][10][11]. The human pathways for BCAA catabolism involve many 22 highly regulated enzymes, and impaired BCAA catabolism caused by genetic defects in 23 pathway enzymes may result in metabolic diseases such as maple syrup urine disease 24 (MSUD), propionic Acidemia (PA), methylmalonic acidemia (MMA), and isovaleryl 25 acidemia (IVA), among others ( Figure 1B) [7,8,[12][13][14]. 26 Frontline treatment for the aforementioned rare metabolic diseases includes dietary 27 protein restriction [15][16][17][18]. Patients consume artificially formulated prescription protein 28 food that has been depleted of one or more BCAAs. However, this treatment option is 29 not without drawbacks. Education, poor availability of the proper food or inconvenience 30 all result in poor compliance with the treatment diet, which is even more challenging in 31 less-developed countries or regions. Therefore, treatments that maintain the ability to 32 consume natural protein foods as part of daily dietary intake are a desired option to 33 improve a patient's quality of life [19][20][21][22]. 34 Advances in synthetic biology techniques have enabled the engineering of microbes as 35 potential therapies for a number of different diseases. For example, the probiotic strain 36 E. coli Nissle 1917 (EcN) has been engineered to express multiple enzymes aimed at 37 the metabolism of phenylalanine as a potential treatment for PKU [23]. Similarly, oral 38 delivery of BCAA-degrading engineered microbes could be used as a therapeutic 39 treatment option for BCAA-related metabolic diseases or disorders. Among the BCAAs, 40 leucine is thought to play the most critical role in disease pathogenesis and therefore is 41 targeted for dietary removal in patients with MSUD, IVA, 3-methylglutaconic aciduria 42 type I and 3-hydroxy-3-methylglutaryl-CoA lyase deficiency [14]. In this article, we 43 describe the genetic engineering of a leucine degradation pathway in the widely used 44 probiotic strain, E. coli Nissle 1917. Following demonstration of its activity in vitro, we 45 applied high throughput screening techniques to optimize the leucine degradation 46 activity of the strain. We show that pathway optimization translates to increased in vivo 47 leucine degradation activity in naïve mouse and non-human primate pharmacokinetic 48 models as measured by tracking blood metabolites and a non-invasive, strain-specific 49 urinary biomarker identified during the course of this study. In this work we present a 50 process for the development of the engineered probiotic EcN with a multi-enzyme 51 pathway and its optimization using high throughput screening for a broad therapeutic 52 application targeting BCAA-related human diseases and disorders.

54
Leucine-degradation pathways engineered into E. coli Nissle 1917 55 To engineer EcN for leucine degradation, we designed a three-enzyme pathway to 56 catabolize leucine to isopentanol via ketoisocaproate and isopentanal intermediates. 57 This pathway was composed of leucine dehydrogenase (LeuDH, from Bacillus cereus), 58 ketoacid decarboxylase (KivD, from Lactococcus lactis), and alcohol dehydrogenase 59 (Adh, from Saccharomyces cerevisiae) (Figure 2A). To enhance the transport of leucine 60 into EcN, the gene encoding the E. coli BCAA transporter BrnQ was also included in the 61 design [24]. The genes encoding the three catabolic pathway enzymes and the BrnQ 62 transporter were assembled as a multi-cistronic operon in a low copy plasmid which 63 was used to transform a genetically modified EcN strain (SYN469) with the following 64 chromosomal changes: 1) genes encoding a second copy of the endogenous high 65 affinity BCAA transporter LivKHMGF to facilitate additional import of leucine 2) deletion 66 of the leucine biosynthetic gene ilvC to prevent BCAA production [25], and 3) deletion of 67 the amino acid export gene, leuE, to prevent leucine export[26] ( Figure 2B, SYN1980). 68 The same plasmid was also used to transform wild type EcN resulting in strain 69 SYN6034 ( Figure 2C). 70 Following construction, leucine consumption by both strains was assessed in vitro. 71 SYN1980 was able to consume leucine in vitro ( Figure 2D), and the rates of leucine 72 consumption between SYN1980 and SYN6034 were not significantly different, 73 indicating that the accessory chromosomal modifications in the SYN1980 chassis did 74 not affect the leucine consumption rate of the heterologous pathway ( Figure 2D). 75 However, the BrnQ transporter was required for optimal activity of the heterologous 76 pathway, as elimination of the additional copy of BrnQ resulted in a >55% decrease in 77 the leucine consumption rate ( Figure 2D, SYN1992 vs SYN1980). Since the accessory 78 modifications did not improve strain activity, focus was shifted to the three pathway 79 enzymes, LeuDH, KivD, or Adh, to explore the potential to improve the rate of leucine  (Table 1). To enable the required 91 screening throughput, we developed spectrophotometric enzymatic assays for direct or 92 indirect measurement of LeuDH, KivD, and Adh activities in E. coli cell lysates ( Figure   93 S2), which are amenable to automation for high throughput screening. A total of 1,175 94 candidate LeuDH enzymes were screened for the ability to deaminate leucine. The 95 initial round of screening identified 110 enzymes with activity on leucine. These 110 96 enzymes were further analyzed in a second screen, and 43 LeuDH enzymes had mean 97 activity higher than the B. cereus LeuDH expressed in the prototype strain SYN1980 98 ( Figure 3A). The best-performing enzyme, LeuDH from Cetobacterium ceti, exhibited a 99 3.4x greater activity than the B. cereus LeuDH (Table S2). 100 Similarly, to identify a superior ketoisovalerate decarboxylase (KivD), a total of 1,296 101 candidate KivD enzymes were screened for decarboxylase activity on ketoisocaproate.

102
Interestingly, the L. lactis KivD enzyme in the SYN1980 pathway did not have 103 measurable activity when screened in the high throughput assay, so KivD activity is 104 reported relative to the non-zero activity of the lysate-only negative control. The initial 105 round of screening identified 55 enzymes with decarboxylase activity. The second round 106 of screening demonstrated that >40 KivD enzymes had at least 6-to 8-fold increase in 107 KivD activity relative to the background lysate activity ( Figure 3B, Table S3).

108
For Adh enzyme screening, a library of 1,177 candidates was screened for the ability to 109 reduce isopentanal to isopentanol. Similar to KivD in the decarboxylase assay, the S. 110 cerevisiae ADH2 enzyme in the SYN1980 prototype strain did not exhibit activity in the 111 Adh assay above the non-zero lysate-only background control. The Equus caballus Adh 112 was found to have the desired activity on isopentanal during assay development and 113 was used as a positive control for the Adh screens. We identified 55 Adh enzymes with 114 Adh activity in our first round of screening. The second screening round identified 5 Adh 115 enzymes with at least a 20-fold increase in Adh activity relative to the non-zero 116 background lysate activity, and >10-fold higher than the positive control E. caballus Adh 117 ( Figure 3C, Table S4).

118
The top hits from each individual enzyme screening were assembled into a library of 119 operons. These Leu catabolic operons were synthesized (leuDH-kivD-adh) and cloned 120 into the same plasmid backbone conferring Leu catabolism in SYN1980, replacing the 121 original catabolic genes upstream of brnQ ( Figure 2B). Complete pathways encoded 122 four-gene operons (leuDH-kivD-adh-brnQ) in which the LeuDH, KivD, and Adh enzymes 123 and their cognate ribosome binding sites (RBSs) were varied while the location and 124 identity of the brnQ RBS and coding sequence were held constant ( Figure S3) Figure 3D, strains were ranked based on leucine consumption, and 108 pathways 133 demonstrated mean leucine consumption equivalent to or higher than the prototype 134 pathway in SYN1980. Enzyme origin and RBS information are listed in Table 2.

135
The 3 strains with the highest leucine consumption rates (SYN5721, SYN5722 and 136 SYN5729, all in SYN469 background) were further characterized. As observed in the 137 pathway library screen, all 3 strains consumed leucine at a faster rate than the 138 prototype strain SYN1980 ( Figure S1A). Notably, the 3 strains with optimized pathways 139 had drastically reduced levels of ketoisocaproate in the supernatant when compared to 140 the control strains, indicating that a KivD bottleneck was relieved through the pathway 141 optimization campaign ( Figure S1B).

142
To understand the impact of leucine-specific host strain chromosomal modifications on 143 the improved leucine-consuming pathways (the liv operon, ilvC, and leuE), plasmids from the top three strains were used to transform wild type EcN, resulting in strains 145 SYN5941, SYN5942 and SYN5943, and the in vitro leucine consumption activity was 146 evaluated for each strain (Figure 4). Similar to the prototype pathway (SYN1980 vs. 147 SYN6034, Figure 4), the accessory gene modifications carried in SYN1980-derivative 148 strains did not enhance activity over the wild type chassis strains. We reasoned that 149 reduced host cell engineering would result in better strain fitness; therefore, we chose to 150 move forward with characterization of a single optimized leucine catabolizing pathway in 151 the wild type EcN background (SYN5941). 153 To enable the assessment of strain activity in vivo, biomarkers of leucine consumption 154 activity that could be identified in urine were developed. Glucuronidation reaction 155 catalyzed by UDP-glucuronosyltransferase (UGT) is the major pathway for foreign 156 chemical removal in humans as well as other animals [27,28]. Therefore, we 157 hypothesized that the end product of the leucine degradation pathway in SYN5941, 158 isopentanol, may be converted by UGT into isopentyl glucuronide (IPG) and excreted in 159 urine ( Figure 5A) [27,29,30]. To test this hypothesis, isopentanol was orally 160 administered to mice at 100, 250 or 500 mg/kg and urine was collected 4 hours later for 161 detection of IPG. Isopentanol administration resulted in a dose-dependent recovery of 162 IPG, with concentrations reaching 2.06 ± 0.33 µmol at the 500 mg/kg dose ( Figure 5B).

163
In naïve mice, a single dose of SYN5941 (5.62 x 10 10 CFU) resulted in a significant 164 increase in urinary IPG compared to treatment with vehicle or WT EcN ( Figure 5C), 165 demonstrating that the strain degraded leucine in vivo. 166 To further investigate the potential of using IPG as a strain-specific biomarker for the 167 evaluation of SYN5941 activity in humans, the baseline urinary IPG levels were 168 analyzed from ten healthy human volunteers. All ten subjects were found to have very 169 low background IPG levels (below the limit of detection ~ 0.16 µg/mL) ( Figure S4). This 170 suggests that IPG may be useful as a clinical biomarker, as the background signal for 171 IPG would not be expected to confound the assessment of EcN leucine degradation 172 activity in humans. 174 We next examined the activity of leucine-consuming EcN strains in non-human primates 175 (NHPs), as these animals have GI physiology and dietary patterns similar to humans.  [20,31]. Even when carefully followed, dietary restriction can cause 190 significant pressure and create negative psychological issues for patients [19][20][21].   In addition to inborn errors of metabolism, chronically elevated levels of BCAAs are 207 associated with cardiometabolic diseases such as type 2 diabetes (T2D), obesity, and 208 cardiovascular disease ( Figure 1A) [5,[35][36][37]. Current evidence suggests that elevated 209 BCAAs may not only be biomarkers for cardiometabolic diseases but may also play a 210 role in the pathogenesis of these diseases. Further, reduced BCAA intake from daily 211 food may serve to reduce the long-term risk of cardiometabolic diseases [38][39][40]. BCAA 212 consuming strains could provide a positive effect on overall metabolic fitness or even 213 make intriguing prophylactic treatments for cardiometabolic diseases [41]. 214 EcN was selected as the chassis to develop a strain capable of degrading BCAAs in 215 this work. EcN's safety, in its un-engineered form, has been validated by long-term use 216 in humans, and some trials have been done safely in infants [42][43][44]. In its engineered

Strain construction
Refer below to Table S1  The deletion of the ilvC and leuE gene was conducted by using P1 transduction from the corresponding BW25113 strains and the removal of antibiotic resistance cassettes via pCP20 and subsequent pCP20 removal. Insertion of Ptet-livKHPMF cassette at lacZ locus was carried out by the lambda red recombineering method via pKD46 and subsequent pKD46 removal, followed by the removal of antibiotic resistance cassettes via pCP20 and subsequent pCP20 removal. For constructing plasmids, DNA fragments containing the BCAA-consuming operon under the control of various promoters were synthesized at GENEWIZ, IDT DNA, Ginkgo, or PCR-amplified from corresponding vectors from the Synlogic collection and cloned into a Synlogic vector containing an ampicillin resistance gene, a low copy number origin of replication (pSC101) and the regulatory control of the Ptet promoter by Gibson assembly method. These plasmids were electroporated into SYN001 or other host strains. After electroporation (Eporator, Eppendorf, 1.8-kV pulse, 1-mm gap length electro-cuvettes), the transformed cells were selected as colonies on LB agar (Sigma, L2897) containing proper antibiotics.

Preparation of strains in flask
For analysis of leucine or other BCAA metabolism, cells were inoculated in 4 mL of 2xYT containing appropriate antibiotics in a 14-mL round bottom tube and incubated at 37 °C with shaking at 250 RPM overnight. The next day, cell cultures were diluted 1:100 in 50 mL of FM1 medium in 125-mL baffled flasks at 37 °C with shaking (250 RPM) for 2h, also with appropriate antibiotics. Cells were induced with the addition of anhydrous tetracycline (Sigma, ATC, 100 ng/mL final concentration) or IPTG (Sigma, 1 mM final concentration). All induction proceeded for 4h. Following induction, cells were centrifuged for 5 minutes at 8000 x g at 4 °C, washed once with an ice-cold formulation buffer (KH2PO4 (2.28 g/L) and K2HPO4 (14.5 g/L)) containing 15% of glycerol (pH 7.5).
Finally, cells were concentrated 20-fold in the formulation buffer and stored at 80°C until the days of testing.

Preparation of strains in fermenter
A seed flask fermentation was started from a scraping of the frozen MCB culture in a cryovial with an inoculum loop into a 500-mL baffled flask with 50 mL FM1 media supplemented with 25 g/L glucose or FM2 media supplemented with 40 g/L glycerol and 100 µg/mL carbenicillin. This culture was grown overnight for ~15-16h at 37 °C and shaken at 350 RPM. Next day, a seed culture of ~30-40 OD600 was used to inoculate a fermentation vessel with corresponding media and 100 µg/mL carbenicillin to a starting OD600 of 0.15. The fermentation was grown at 37 °C and pH 7 with a dissolved oxygen setpoint of 60% for ~6h to achieve final biomass production. The fermentation growth phase was allowed to multiply for 2h until the OD600 reached 1.5-3. At the target OD, the culture was induced by ATC at a final concentration of 600 ng/mL to activate the cells.
The induction of cells continued for 4h until the generation of final biomass reached between 20-30 OD600. Fermentation was harvested at 4h post-induction endpoint and spun down by centrifuging culture broth for 30 min at ~ 5000 g at 4 °C. Cells were finally resuspended in glycerol/phosphate buffer, aliquoted and stored at -80°C.

Cultivation conditions for enzyme assays
For each of the enzyme libraries screened, strains harboring library plasmids were transformed into E. coli T7 expression host cells. 5 µL of thawed glycerol stocks were stamped into 500 µL/well of LB + 100 µg/mL carbenicillin (LB-Carb100) in half-height deepwell plates, which were sealed with AeraSeals. Samples were incubated at 37 °C and shaken at 1000 RPM in 80% humidity overnight. 50 µL/well of the resulting precultures were stamped into 450 µL/well of LB-Carb100 + 1 mM IPTG in half-height deepwell plates, which were sealed with AeraSeals. Samples were incubated at 30˚C and shaken at 1000 RPM in 80% humidity overnight. 250 µL/well of the resulting production cultures were stamped into deepwell plates containing 500 µL of phosphate buffered saline (PBS) and centrifuged for 10 minutes at 4000 x g. Supernatant was removed and the resulting cell pellet was resuspended in 200 µL of BugBuster Protein Extraction Reagent + 1 µL/mL purified benzonase + 1 µL/6 mL purified Lysozyme.
Samples were incubated for 10 minutes at room temperature to generate the cell lysates used in in vitro enzyme assays.  isovaleraldehyde, pH 7.0). Optical absorbance measurements were taken on a plate reader at 340 nm for 10 minutes. The resulting kinetic data was used to resolve the maximum rate of NADH oxidation, a proxy for ADH activity.

Pathway enzyme selection and operon library assembly
Selected top enzymes were incorporated in the final operon designs with various RBS strength to balance flux of the leucine-consuming pathway. Among the 3 RBSs, two were designed using the RBS Calculator to have translation initiation rates of approximately 5,000 au and approximately 50,000 au [45][46][47][48]. In addition to designing two conventional RBSs for each enzyme, we also included an RBS that was cotranslationally coupled to a short leader peptide, in what is termed a bicistronic design [49]. RBS-enzyme pairs were assembled into a partial combinatorial library of 462 pathways. Each pathway contains either conventional RBSs or BCD-type RBSs, but not both. The gene order was held constant (leuDH-kivD-adh).
These leucine catabolic pathways were synthesized (leuDH-kivD-adh) and cloned into the same plasmid backbone conferring leucine catabolism in SYN1980, replacing the

Operon screening
The successful library transformants were screened in a high throughput leucine consumption assay to identify operons conferring greater leucine consumption compared to SYN1980. All pathway-containing strains were cultured and screened for leucine consumption in 96-well plates. For plate-to-plate standardization, each plate included control strains which harbored the prototype pathway (SYN1980) or variants lacking LeuDH (SYN1980 ΔleuDH) or BrnQ (SYN1980 ΔbrnQ).

Ethical statement
All procedures performed on animals were in accordance with the humane guidelines

Non-human primate (NHP) study
A total of twelve male cynomolgus non-human primates (NHP), approximately 2-5 years of age, were housed at the Charles River Laboratories (CRL) where studies were conducted by experienced staff members. Samples were collected at CRL and shipped to Synlogic where they were plated for analysis. Due to the limited number of animals available per study, three separate studies were conducted, and the results were combined for analysis. In each study NHP's were separated into two groups. Each group was orally dosed into the stomach with three compounds; 14 mL of peptone (500 g/L), followed by 7.8 mL of vehicle (PBS + 15% glycerol) or bacteria (SYN1980 or SYN5941), followed by 5 mL of 0.36 M bicarbonate, and then finally a small amount of water to rinse the material from the gavage tube. The animals were then placed individually into cages with clean urine collection pans. After a period of 6 hours the urine pans were removed, and the urine was collected into 50-mL tubes and weighed.

Statistical analysis
Raw data generated was entered in a Microsoft Excel (Microsoft, Seattle, WA) spreadsheet and transferred to GraphPad Prism 8.0 (GraphPad Software, San Diego, CA). Statistical analysis was performed using GraphPad Prism and the statistical analysis was performed using two-way ANOVA followed by Tukey's multiple comparison test. Significance was set as p < 0.05. Samples were extracted with 9 parts 2:1 acetonitrile:water containing 1 µg/mL leucine-d3 as an internal standard, vortexed, and centrifuged. Supernatants were diluted with 9 parts 0.1% formic acid and analyzed concurrently with standards processed as above from 0.8 to 1000 µg/mL. Samples were separated on a Phenominex Synergi 4 µm Hydro-RP 80A, 75 x 2 mm using a 0.1% formic acid (A), 0.1% formic acid/acetonitrile (B) at 0.3 mL/min and 50 °C. After a 2 µL injection and an initial 5% B hold from 0 to 0.5 minutes, analytes were gradient eluted from 5 to 90% B over 0.5 to 1.5 minutes followed by high organic wash and aqueous equilibration steps. Analytes were detected using Selected Reaction Monitoring (SRM) of compound specific collision induced fragments in electrospray positive ion mode (leucine: 132>86). SRM chromatograms were integrated and the unknown/internal standard peak area ratios were used to calculate concentrations against the standard curve.

LC-MS/MS quantitation of isopentyl glucuronide
Isopentyl glucuronide was quantified using a liquid chromatography triple quadrupole tandem mass spectrometry (LC-MS/MS) Thermo TSQ Altis system. In a conical bottom plate, 10 µL of standards and samples were diluted with 90 µL water containing 5 µg/mL of internal standard (Isopentyl-glucuronide-d11).
Chromatographic gradient separation was carried out using an Accucore aQ C18 column 2.6 µm 100 Ǻ, 100 x 2.1 mm column at 40 °C with mobile phases 10 mM ammonium acetate in water (A) and 10 mM ammonium acetate in acetonitrile (B).
Multiple reaction monitoring in negative mode was used for tandem MS analysis. The following mass transitions were monitored for quantitation: isopentyl glucuronide (263.1/75) and internal standard isopentyl glucuronide-d11 (274.1/75).

LC-MS/MS method for leucine quantification
Leucine was quantified in bacterial supernatant by liquid chromatography coupled to tandem mass spectrometry (LC-MS/MS) using either an Ultimate 3000 UHPLC-TSQ Quantum or a Vanquish UHPLC-TSQ Altis system. Samples were extracted with 9 parts of acetonitrile : water = 2 :1 containing 1 µg/mL leucine-d3 as an internal standard, vortexed, and centrifuged. Supernatants were diluted with 9 parts 0.1% formic acid and analyzed concurrently with standards processed as above from 0.8 to 1000 µg/mL. Samples were separated on a Phenominex Synergi 4 µm Hydro-RP 80A, 75 x 2 mm using a 0.1% formic acid (A), 0.1% formic acid/acetonitrile (B) at 0.3 mL/min and 50°C.
After a 2 µL injection and an initial 5% B hold from 0 to 0.5 minutes, analytes were gradient eluted from 5 to 90% B over 0.5 to 1.5 minutes followed by high organic wash and aqueous equilibration steps. Analytes were detected using Selected Reaction Monitoring (SRM) of compound specific collision induced fragments in electrospray positive ion mode (leucine: 132>86, isoleucine: leucine-d3: 135>89). SRM chromatograms were integrated, and the unknown/internal standard peak area ratios were used to calculate concentrations against the standard curve.

LC-MS methods for pathway intermediates
Leucine (Leu), ketoisocaproate (KIC), and isovaleraldehyde (IVA) were detected by LCMS analysis performed on a Thermo Ultimate 3000 UPLC system with a Thermo Q-Exactive quadrupole-orbitrap mass detector and a Thermo Accucore PFP column (2.1 x 100 mm, 2.6 µm packing) using the following elution solvents: A=0.1% formic acid and 0.1% TFA in water; B=0.1% formic acid in acetonitrile. The gradient was at 0.5 mL/min of 1% B in A for 60 seconds, followed by a linear ramp from 1% to 40% Bover 270 seconds. The column is then flushed with 95% B in A for 60 seconds, and reequilibrated with 1% B in A for 180 seconds. MS acquisition was from 0.8 to 5.3

minutes. Column effluent is introduced into the mass spectrometer via a standard
Thermo ESI source with positive mode ionization at +3800V, vaporizer temperature of 400 °C, and ion transfer tube temperature of 375 °C. Thermo reports gas flow rates in arbitrary units probably approximating L/min at STP. Set points were: sheath gas, 60; aux gas, 30; sweep gas, 1. To increase data acquisition rate, orbitrap resolution was set              TCCGAGGAGGCGGCCATCGAGGACGCACTTCGTCTTGCTAAGGGTATGACCTATAAGAACGCGGCAGCCGGTCTGAATCTGGGGGGTGCTA  AGACTGTAATCATCGGTGATCCACGCAAGGATAAGAGTGAAGCAATGTTTCGCGCTTTAGGGCGCTATATTCAGGGCTTGAACGGCCGCTACA  TTACCGCAGAAGACGTAGGGACAACAGTAGACGACATGGACATCATCCATGAGGAAACTGATTTCGTGACCGGTATTTCACCTTCATTCGGGT  CATCCGGTAACCCTTCCCCCGTAACAGCCTATGGGGTTTATCGCGGAATGAAGGCCGCAGCCAAGGAGGCATTTGGCACTGACAATTTAGAA  GGAAAAGTAATTGCTGTCCAAGGCGTGGGCAATGTGGCCTACCATTTGTGTAAACACCTTCACGCGGAAGGTGCAAAATTGATCGTTACGGAT  ATTAACAAGGAGGCAGTCCAGCGCGCTGTAGAGGAATTTGGAGCATCGGCTGTGGAACCAAATGAGATCTACGGTGTAGAATGTGACATTTAC  GCTCCATGCGCACTTGGTGCCACGGTGAATGACGAGACCATCCCCCAACTTAAGGCGAAGGTAATCGCTGGTTCAGCTAACAACCAATTAAAA  GAGGACCGTCACGGAGATATCATCCACGAAATGGGTATCGTGTACGCCCCCGATTATGTTATCAACGCGGGCGGCGTAATCAACGTAGCCGA  TGAGCTTTATGGATACAACCGCGAACGTGCGCTGAAACGCGTGGAAAGCATTTATGACACGATCGCAAAGGTAATCGAGATCAGTAAGCGCG  ACGGCATTGCGACATACGTGGCAGCGGACCGTCTGGCCGAAGAACGCATCGCGAGTTTGAAGAATAGCCGTAGTACCTACTTGCGCAACGG  GCACGATATTATCAGCCGTCGCtgataagaaggagatatacatatgtatacagtaggaGATTACTTATTGGACCGGTTGCACGAACTTGGAATTGAGGAAATTTT  TGGAGTTCCGGGTGACTACAACCTGCAGTTCCTTGACCAAATCATCTCCCATAAGGACATGAAATGGGTCGGCAATGCCAATGAGCTGAACGC  ATCATATATGGCAGACGGGTATGCTCGGACCAAAAAGGCTGCAGCATTCCTTACCACGTTTGGCGTGGGGGAATTAAGTGCTGTAAATGGACT  GGCAGGATCCTATGCGGAGAATTTACCGGTAGTCGAAATTGTTGGCTCGCCTACGTCCAAGGTGCAGAATGAGGGGAAATTCGTCCATCACA  CACTTGCAGACGGTGATTTTAAGCACTTTATGAAGATGCATGAGCCGGTAACGGCTGCGCGGACGCTTCTTACTGCGGAAAACGCAACAGTAG  AGATTGATCGCGTTCTGAGCGCACTGCTTAAGGAACGGAAGCCCGTCTATATTAACTTACCGGTAGACGTGGCCGCAGCCAAAGCCGAAAAA  CCAAGCCTGCCTCTTAAGAAGGAGAATTCCACGTCCAACACCAGTGACCAAGAGATTTTGAACAAAATTCAAGAGTCTTTGAAGAACGCGAAG  AAGCCCATCGTAATTACAGGACATGAGATTATCTCGTTTGGCCTGGAGAAAACGGTTACACAGTTTATTTCCAAAACGAAGTTACCTATAACGA  CGTTAAACTTTGGAAAGAGCTCTGTGGATGAGGCACTTCCTAGTTTCTTAGGAATCTATAATGGGACCCTTTCAGAGCCAAACTTAAAGGAATT  CGTTGAAAGTGCGGATTTTATCTTAATGCTTGGGGTTAAATTGACTGATTCCAGCACCGGAGCTTTTACGCACCATTTAAACGAGAACAAAATG  ATCTCTTTGAATATCGACGAAGGCAAAATTTTTAATGAAAGAATTCAGAACTTTGATTTTGAATCCCTTATTAGTTCACTTTTAGATTTAAGTGAAA  TAGAGTATAAGGGAAAGTATATAGACAAGAAGCAAGAGGATTTCGTTCCGTCTAATGCTCTTTTAAGTCAAGACAGACTTTGGCAGGCGGTTGA  GAACCTTACACAATCCAATGAAACGATAGTCGCCGAACAAGGGACCAGTTTCTTCGGCGCTTCTTCCATATTCCTGAAGTCTAAGTCTCATTTC  ATTGGACAGCCCCTGTGGGGGTCTATAGGATATACGTTTCCCGCAGCTCTTGGAAGCCAGATCGCCGATAAGGAGAGCAGACACCTGTTGTT  CATCGGGGACGGCTCGTTGCAGCTGACTGTTCAGGAACTGGGGTTGGCGATCAGAGAGAAGATTAATCCCATTTGCTTTATCATAAATAATGA  TGGTTATACCGTAGAACGTGAGATTCATGGACCTAATCAGAGCTATAATGACATTCCTATGTGGAACTATTCAAAATTGCCAGAGAGTTTTGGT  GCAACTGAGGATCGCGTTGTTAGTAAAATAGTCCGCACGGAGAACGAGTTTGTCAGCGTAATGAAGGAGGCCCAAGCGGACCCTAATCGGAT  GTACTGGATCGAACTTATTCTGGCTAAAGAAGGAGCACCTAAAGTTTTAAAGAAGATGGGAAAACTTTTTgctgaacaaaataaatcataataagaaggagata  tacatatgtctattccaGAAACGCAGAAAGCCATCATATTTTATGAATCGAACGGAAAACTTGAGCACAAGGACATCCCCGTCCCGAAGCCAAAACCTAA  TGAGTTGCTTATCAACGTTAAGTATTCGGGCGTATGCCACACAGACTTGCACGCATGGCACGGGGATTGGCCCTTACCGACTAAGTTGCCGTT  AGTGGGCGGACATGAGGGGGCGGGAGTCGTAGTGGGAATGGGAGAGAACGTGAAGGGTTGGAAGATTGGAGATTATGCTGGGATTAAGTGG  TTGAATGGGAGCTGCATGGCCTGCGAATATTGTGAACTTGGAAATGAGAGCAATTGCCCACATGCTGACTTGTCCGGTTACACACATGACGGT  TCATTCCAGGAATATGCTACGGCTGATGCAGTCCAAGCAGCGCATATCCCGCAAGGGACGGACTTAGCAGAAGTAGCGCCCATTCTTTGCGC  TGGGATCACCGTATATAAAGCGTTAAAGAGCGCAAATTTACGGGCCGGACATTGGGCGGCGATCAGCGGGGCCGCAGGGGGGCTGGGCAG  CTTGGCCGTCCAGTACGCTAAAGCTATGGGTTATCGGGTTTTGGGCATTGACGGAGGACCGGGAAAGGAGGAATTATTCACGTCCTTGGGAG  GAGAGGTATTCATTGACTTTACCAAGGAAAAAGATATCGTCTCTGCTGTAGTAAAGGCTACCAATGGCGGTGCCCACGGAATCATAAATGTTTC  AGTTTCTGAAGCGGCGATCGAAGCGTCCACTAGATATTGCCGTGCAAATGGGACAGTCGTACTTGTAGGACTTCCGGCTGGCGCCAAATGCA  GCTCCGATGTATTTAATCATGTCGTGAAGTCAATCTCTATCGTTGGTTCATATGTAGGAAACCGCGCCGATACTCGTGAGGCTCTTGACTTTTTT  GCCAGAGGCCTGGTTAAGTCCCCCATAAAAGTTGTTGGCTTATCCAGCTTACCCGAAATATACGAGAAGATGGAGAAGGGCCAGATCGCGGG  GAGAtacgttgttgacacttctaaataataagaaggagatatacatatgacccatcaattaagatcgcgcgatatcatcgctctgggctttatgacatttgcgttgttcgtcggcgcaggtaacattattttccctcca  atggtcggcttgcaggcaggcgaacacgtctggactgcggcattcggcttcctcattactgccgttggcctaccggtattaacggtagtggcgctggcaaaagttggcggcggtgttgacagtctcagcacgcca  attggtaaagtcgctggcgtactgctggcaacagtttgttacctggcggtggggccgctttttgctacgccgcgtacagctaccgtttcttttgaagtgggcattgcgccgctgacgggtgattccgcgctgccgctgttt  atttacagcctggtctatttcgctatcgttattctggtttcgctctatccgggcaagctgctggataccgtgggcaacttccttgcgccgctgaaaattatcgcgctggtcatcctgtctgttgccgcaattatctggccggc  gggttctatcagtacggcgactgaggcttatcaaaacgctgcgttttctaacggcttcgtcaacggctatctgaccatggatacgctgggcgcaatggtgtttggtatcgttattgttaacgcggcgcgttctcgtggcg  ttaccgaagcgcgtctgctgacccgttataccgtctgggctggcctgatggcgggtgttggtctgactctgctgtacctggcgctgttccgtctgggttcagacagcgcgtcgctggtcgatcagtctgcaaacggtgc  ggcgatcctgcatgcttacgttcagcatacctttggcggcggcggtagcttcctgctggcggcgttaatcttcatcgcctgcctggtcacggcggttggcctgacctgtgcttgtgcagaattcttcgcccagtacgtac  cgctctcttatcgtacgctggtgtttatcctcggcggcttctcgatggtggtgtctaacctcggcttgagccagctgattcagatctctgtaccggtgctgaccgccatttatccgccgtgtatcgcactggttgtattaagttt  tacacgctcatggtggcataattcgtcccgcgtgattgctccgccgatgtttatcagcctgctttttggtattctcgacgggatcaaggcatctgcattcagcgatatcttaccgtcctgggcgcagcgtttaccgctggc Figure 1 Branched chain amino acids and human diseases. A) Branched chain amino acids (BCAAs) are essential amino acids which humans must obtain from protein food. They bene t human health but also are believed to relate to various diseases and disorders. B) Abnormal degradation of branched chain amino acids due to defective enzymes in human will result in severe disorders such as maple syrup urine disease (MSUD), propionic acidemia (PA), methylmalonic acidemia (MMA), isovaleryl acidemia (IVA), or 3-methylcrotonyl-CoA carboxylase de ciency (3-MCC).  Adh enzymes from the primary screen were re-screened for activity (n = 4). Since the strain expressing S. cerevisiae ADH2 had no measurable activity in this assay, activities are reported relative to E. caballus Adh (green dot, green dashed line). D) Pathway operons with optimized enzymes were screened for leucine consumption. Strains were assayed in biological replicates (n = 2 or 3, depending on the number of successful transformants). Data points are shown as dots, and the average for each strain is shown as a horizontal blue line. Control and reference strains are indicated with colored labels.  In vivo biomarker validation and strain activity in mice. A) Proposed pathway of in vivo isopentyl glucuronide (IPG) formation. B) IPG urinary recovery in response to oral administration of isopentanol in mice (n = 5 for each group). Statistical analysis was performed using one-way ANOVA followed by Tukey's multiple 3comparison test (****p < 0.0001, ***p < 0.0007 versus vehicle). C) IPG urinary recovery in response to orally administered leucine consuming strain SYN5941 in mice (n = 5 for each group, bacterial dosed at 0 h, 1 h and 2 h and totaled at 5.62 e10 cells). All vehicle & SYN094 samples are below limit of quanti cation = 0.00003 μmol. Statistical analysis was performed using two-way ANOVA analysis followed by Tukey's multiple comparison test (****p < 0.0001 versus vehicle).

Figure 6
E cacy of SYN1980 or SYN5941 in healthy nonhuman primates. A) Effect of SYN1980 and SYN5941 on plasma leucine levels following oral administration of leucine consuming strains. Statistical analysis was performed using two-way ANOVA analysis followed by Tukey's multiple comparison test (* p < 0.05 versus vehicle). B) Quanti cation of area under the curves (AUC) from plot A in this gure. Statistical analysis was performed using ordinary one-way ANOVA analysis (**p < 0.0051 versus vehicle) followed by Tukey's multiple comparison test. C) Urinary IPG recovery following oral administration of leucine consuming strains. Statistical analysis was performed using one-way ANOVA analysis followed by Tukey's multiple comparison test (**p < 0.005 versus vehicle).