Engineering mannitol biosynthesis in Escherichia coli 1 and Synechococcus sp. PCC 7002 using a green algal 2 fusion protein 3

10 The genetic engineering of microbial cell factories is a sustainable alternative to the chemical 11 synthesis of organic compounds. Successful metabolic engineering often depends on 12 manipulating several enzymes, requiring multiple transformation steps and selection markers, 13 as well as protein assembly and efficient substrate channeling. Naturally occurring fusion 14 genes encoding two or more enzymatic functions may offer an opportunity to simplify the 15 engineering process and to generate ready-made protein modules, but their functionality in 16 heterologous systems remains to be tested. Here we show that heterologous expression of a 17 fusion enzyme from the marine alga Micromonas pusilla, comprising a mannitol-1-phosphate 18 dehydrogenase and a mannitol-1-phosphatase, leads to synthesis of mannitol by Escherichia 19 coli and by the cyanobacterium Synechococcus sp. PCC 7002. Neither of the heterologous 20 systems naturally produces this sugar alcohol, which is widely used in food, pharmaceutical, 21 medical and chemical industries. While the mannitol production rates obtained by single-gene 22 manipulation were lower than those previously achieved after pathway optimization with 23 multiple genes, our findings show that naturally occurring fusion proteins can offer simple building blocks for the assembly and optimization of recombinant metabolic pathways.


Reuse
Items deposited in White Rose Research Online are protected by copyright, with all rights reserved unless indicated otherwise. They may be downloaded and/or printed for private study, or other acts as permitted by national copyright laws. The publisher or other rights holders may allow further reproduction and re-use of the full text version. This is indicated by the licence information on the White Rose Research Online record for the item.

Takedown
If you consider content in White Rose Research Online to be in breach of UK law, please notify us by emailing eprints@whiterose.ac.uk including the URL of the record and the reason for the withdrawal request.
Microbial cell factories, particularly photosynthetic chassis that consume carbon dioxide as 29 their sole carbon source, are an attractive alternative to chemical synthesis 1 . They present a 30 sustainable approach to producing an array of substances with usages in food, 31 pharmaceutical, nutraceutical and cosmetic industries. 32 Biological production platforms struggle to compete economically, particularly for the 33 production of low value commodity chemicals and materials 2 . A recurring problem is 34 performance, which is measured as titer, yield and productivity. Artificial metabolic pathways 35 inherently generate imbalances in pathway flux resulting in (toxic) intermediate and side 36 product accumulation, growth inhibition, and ultimately low product yield 3 . Therefore, 37 individual modifications (e.g. the introduction of a single enzyme or pathway) are not sufficient 38 to achieve industrially relevant titers. As a result, cell factory development requires the 39 introduction of several genes to synthesize the desired product, the overproduction of 40 precursors, and the deletion of competing pathways 4-5 . 41 The use of fusion genes, generated from previously separate genes, can help to simplify the 42 biological engineering process 6  Mannitol can be produced by extraction, chemical synthesis or biosynthesis. Traditionally, 59 mannitol was harvested from plant material which was seasonal and yields highly variable 9 . 60 To stabilize mannitol supplies, commercial production now mostly occurs through chemical 61 hydrogenation of fructose-glucose syrups. However, poor selectivity of the nickel catalyst 62 results in a mixture of mannitol and sorbitol which is relatively difficult to separate and thus 63 costly 10 . Chemical synthesis can be improved by altering the substrate, e.g. isomerizing 64 glucose to fructose by enzymatic conversion; however, enzyme availability and added costs 65 of additional steps prevent this from being economical. 66 The chemical industry is now looking to bio-based production methods to reduce costs and 67 environmental impact. The most successful approach to mannitol biosynthesis currently uses 68 heterofermentative lactic acid bacteria. Under anaerobic conditions, these organisms reduce 69 fructose using the native enzyme mannitol dehydrogenase. This approach requires an 70 external sugar supply that is a) predominantly obtained from traditional crops, e.g. corn and b) 71 a major cost for biosynthesis of commodities such as mannitol. External carbon sources and 72 associated costs can be eliminated from the bio-production pipeline by using photosynthetic 73 organisms, which assimilate atmospheric carbon dioxide into sugars via the Calvin cycle 11 . 74 Cyanobacteria represent an incredibly diverse phylum of phototrophic prokaryotes that are 75 being developed for photosynthetic bio-production 12-14 . One particularly attractive chassis is 76 the unicellular euryhaline cyanobacterium Synechococcus sp. PCC 7002 due to its reported 77 fast growth and tolerance of high salt, light and temperature [15][16][17]  found that both E. coli and Synechococcus produced mannitol when transformed with the 105 fusion gene. The one-step mannitol production pathway provides an excellent starting point 106 for further optimization of sustainable mannitol production in cyanobacteria. 107

108
Mpusfus is functional in E. coli 109 6 sequence mpusfus. To test expression and function in heterologous systems, mpusfus was 112 first introduced into E. coli using the pFO4 vector that carries an IPTG-inducible expression 113 system. A six-histidine tag was added to the 5' end of the gene. Western-blot of protein 114 extracts from IPTG-induced cells revealed a band of the expected size while no band was 115 detected in the controls without IPTG (Figure 1). The size of the primary band was close to 116 the value of 94.8 kDa calculated for the full-length predicted amino acid sequence. This 117 showed that a single fusion protein was produced from the fusion gene. 118 Western-blot analysis of His-tagged M1PDH/M1Pase in soluble fractions of protein extracts from E. coli transformed with mpusfus. Lanes marked 1, 2, and 3 contain protein extracts from three independent clones after culture in absence (-) and in presence (+) of IPTG. Numbers at the left of the panel indicate size (kDa) of proteins in the ladder.

119
Mannitol was measured both in the culture media and in the cells after 20 hours of incubation 120 in absence and in presence of IPTG. In 5 ml cultures with a final OD600 of 2.5-3.0 we found 1 121 ± 0.05 mg of mannitol in the media and 0.08 ± 0.008 mg of mannitol inside the cells (n = 3 122 clones) after incubation with IPTG. Extrapolated to a 1-litre culture, total mannitol production 123 was 218 ± 11.9 mg/L, of which 202 ± 10.7 mg (93 %) were exported into the medium and 16 124 ± 1.6 mg (7 %) were retained inside the cells. No mannitol was detected in cultures grown 125 without IPTG. The results showed that the algal fusion gene produces a single protein in E. 126 coli that catalyses the biosynthesis of mannitol. 127

New molecular tools and protocols to engineer Synechococcus sp. PCC 7002 128
BioBricks represent the largest collection of standardized parts for genetic engineering 29

152
Ribosome binding sites (RBSs) play an essential role in translation initiation during protein 153 synthesis. Software have been developed to design synthetic RBSs considering factors such 154 as secondary mRNA structure influenced by flanking nucleotide sequences, the Shine-155 Dalgarno (SD) sequence complementary to the 16S ribosomal RNA, spacing between the SD 156 and start codon, and the sequence of the start codon itself 32 . We used a web-based RBS 157 calculator 33 to design RBS for Synechococcus (sequences in Table S2) and assessed the 158 suitability of three potential RBS for producing green fluorescent protein (GFP). RBS-GFP 159 were cloned into pAQ1BB and used to transform Synechococcus. Similar to a previous report 160 21 we observed a moderate correlation between predicted and actual translation rates ( Figure  161 3). Both the RBS calculator and the GFP-assay identified RBS3 as the strongest RBS and we 162 therefore used RBS3 to express mpusfus. 163 Grey bars represent translation initiation rates predicted in silico using RBS calculator. Black bars represent GFP fluorescence normalised to cell density (optical density at 730 nm, OD730). Each RBS-GFP construct was expressed in Synechococcus under the control of the PcpcB594 promoter and GFP fluorescence was measured when culture density reached an OD730 of 1. Data are presented as means ± S.E.M. of three independent cultures. BBa_0030 is a standard RBS from the Registry of Standardised Parts. RBS1, RBS2 and RBS3 were forward engineered using RBS calculator. We first expressed the mpusfus fusion gene in E. coli. This organism has been previously 214 engineered to develop whole-cell biotransformation systems for mannitol synthesis from 215 glucose [43][44] , resulting in mannitol molar yield of 80% 43 and 87% 44 . We found that E. coli 216 expressing the single mpusfus fusion gene produced a single recombinant protein of 217 approximately 100 kDa and produced mannitol, most of which was exported into the medium. 218 This shows that the fusion protein is functional in E. coli. However, the achieved mannitol 219 concentration of 218 mg/L under our experimental conditions (molar yield of 2% on glucose) 220 was considerably lower than in the previously engineered strains [43][44] . Additional 221 manipulations addressing codon usage, metabolic flux towards substrate, substrate transport 222 and feeding could now be attempted to increase mannitol titres in E. coli. 223 Photosynthetic bacteria provide an opportunity to produce organic compounds from CO2 224 without the need of feeding sugars. We were therefore interested to test whether the mpusfus 225 fusion gene can be used to produce mannitol in a photo-autotrophic system. Synechococcus 226 sp. PCC 7002 is one of the model systems for metabolic engineering of cyanobacteria and 227 has previously been engineered to produce mannitol from F6P using two separate genes 228 (M1PDH from E. coli and M1Pase from Eimeria tenella) 23  Successful functional expression of mpusfus in Synechococcus was proven by the produced was exported into the media where it can easily be harvested. Since 237 Synechococcus does not naturally produce mannitol it is likely that the export occurs through 238 non-specific transport proteins for other compounds. Synechococcus and other cyanobacteria 239 have been shown to release low-molecular-weight metabolites when subjected to hypo-240 osmotic stress 45-46 , but the exact transport pathways remain to be identified. 241 The total amount of mannitol produced in this report is considerably lower than in the previous 242 study 23 , namely around 0.1 g/L compared to 0.6 g/L. Usage of a glycogen-deficient strain 243 helped to increase titres 23 , but the main differences between the two studies lies in the 244 growth rate of the cultures. The highest mannitol-producing strain reported previously 23 245 reached a maximal OD730 of around 10 within 150 h and achieved the aforementioned 246 mannitol concentration in 300 h. By contrast, our strain grew much more slowly and required 247 50 days to produce 0.1 g/L mannitol albeit reaching a higher OD730 of 25. It is likely that 248 protein turnover over such a long period of time prevents the accumulation and maintenance 249 of substantial amounts of recombinant protein. Protein synthesis then becomes very sensitive 250 to promoter activity, which was maximal during early culture growth (see Fig. 4). A 251 combination of low promoter activity and protein turnover would explain why production rates 252 were very low and decreased even before the cultures entered stationary phase (see Fig.  253 5C). It can therefore be expected that usage of new promoter(s), which are active during the 254 late stages of growth, and changes in growth conditions, e.g. fed-batch cultivation to keep 255 cultures in the production stage for longer, could increase mannitol productivity by engineered 256 Synechococcus sp. PCC 7002. 257 Despite the ability of cyanobacteria to use atmospheric carbon for industrial bioproduction, 258 carbon availability is a key limiting factor in polyol production. The use of fast-growing strains 259 and the development of efficient photo-bioreactors will be crucial to move toward industrial 260 scale production systems that can compete with the traditional sugar-fed cultures. Our 261 demonstration that an algal fusion gene is functional in cyanobacteria presents an important 262 step towards simplifying the generation of recombinant metabolic pathways, and can now be 263 combined with the usual metabolic engineering strategies to overcome metabolic bottlenecks. insertion into pAQ1EX using EcoRI/NcoI and NdeI/BamHI restriction sites respectively (Table  273   S1). To prevent interference with the BioBrick assembly, an XbaI restriction site at the 5' end 274 of the spectinomycin resistance gene aadA was replaced with an XhoI restriction site by site-275 directed mutagenesis using primers listed in Table S1 Details of codon optimization, native and optimized sequences can be found in 300 Supplementary data file 1. 301

Generation of transgenic Synechococcus sp. PCC 7002 strains 302
Synthetic ribosome binding sites were designed using the Salis lab RBS calculator 33 and 303 added directly upstream of the transgene during PCR amplification (primer sequences can be 304 found in Table S2). Following sequence confirmation in the pGEM-T® Easy (Promega, UK) 305 vector, the amplified DNA (RBS + gene) was cloned into the pAQ1BB vector, downstream of 306 the PcpcB594 promoter. The synthetic expression constructs were integrated into the 307 Synechococcus genome by natural transformation. Transformation efficiency was optimised 308 by varying either amount of DNA (1-25 µg) or incubation time (1-3 days) prior to plating on 309 selective media ( Figure S1) and the following optimised transformation protocol was used: 1.5 310 mL culture (OD730 1) was combined with 10 µg circular plasmid DNA and incubated for 72 h 311 under standard growth conditions with minimal sparging. Cells were plated on solid A+ 312 medium with 1.5% w/v agar and 50 µg/ml spectinomycin. Single colonies appeared after 5-7 313 days. Individual colonies were isolated and grown for characterization. Genomic DNA was 314 isolated using phenol-chloroform extraction 48 , and the correct insertion of the synthetic 315 expression constructs were verified by PCR amplification using primers pAQ1BB-seq-F (5'-316 CACATGAGAATTTGTCCAG-3') and pAQ1BB-seq-R (5'-CCTTTCGGGCTTTGTTAG-3') and 317 sequencing.  interpretation, or the decision to submit the work for publication.