With the development of biotechnologies, academics and industrialists have achieved the exploitation of microorganisms in a rational manner to bioproduce a great diversity of compounds. This principle could be used in many fields such as health, textiles, cosmetics or even the food industry. To increase the yields of the production, one can optimize culture conditions, invest in bioreactors or enhance extraction methods. However there remains one limiting factor: the intrinsic activity of the enzymes involved in the bioprocess. We are introducing a biosensor to easily screen a library of variants on one enzyme used in a bioproduction process (Screening tool description) and we will use this system to improve the bioproduction of a rare sugar: D-Psicose.
This sugar is found in plants in very low quantities and is expensive, costing 1000 $ per gram from Sigma Aldrich and between 35 and 60 $ from All-u-Lose®, a company which proposes a product composed of 100 % of psicose (Click here to find out more). This sugar is interesting because the human body does not metabolize it . It goes through the enterocytes with the GLUT5 and GLUT2 glucose transporters  and is eliminated in the urine in 24 hours. Therefore, this sugar could be useful for people suffering from diabetes by maintaining their postprandial blood sugar level and avoiding hyperglycemia.
Since D-psicose is a C3 epimer of D-fructose, its production can be achieved by a D-psicose 3-epimerase (EC: 18.104.22.168) or D-tagatose 3-epimerase (EC: 22.214.171.124) in the presence of fructose as a substrate (figure 1). Bioproduction of psicose has already been presented in the literature but the yields are still too low for cheap industrial production (Rare Sugar Bioproduction). This is why we want to improve the bioproduction by selecting the best epimerase.
For that we must set up our bioproduction conditions to ensure that we are able to produce psicose. This part will focus on the insertion of an epimerase in a strain of Escherichia coli and then optimization and quantification of the bioproduction.
The D-Psicose 3-Epimerase
Since we wanted to use a D-psicose 3-epimerase (Dpe) to produce psicose we had to learn about its structure and kinetic properties. For this, we collaborated with iGEM Aix-Marseille who have a team of skilled bio-informaticians who were able to characterize our enzyme.
The Dpe from Agrobacterium tumefaciens (UniProt ID A9CH28) is the enzyme that converts fructose into psicose and the reverse by changing the configuration of the C-3 stereogenic center . This enzyme requires cofactors such as Mn2+ or Co2+ for optimal activity, although it has a low basal activity without ions. Dpe from other bacteria can have different cofactors. This Dpe shows a tetrameric arrangement of 132 kDa, and each subunit has 289 amino acids with a molecular weight of 33 kDa . The topology of each subunit is a TIM-barrel fold with a cluster of eight β-strands surrounded by twelve α-helices (figure 2).
The active site is located in the central cleft of the subunit, where there is a metal-binding site. As shown in figure 3, a Mn2+ ion is bound with octahedral coordination by four residues (E150, D183, H209 and E244) and two water molecules. The conformational change between the apo- and holo-enzymes is minimal (RMSD of 0,6 Å for the α-backbone after superimposing the backbone of the tetramers), so the binding of the substrate does not produce any large structural changes. However, some small alterations are noticeable, for example, residues Y6, W14, W112 and F246 move towards the bound D-fructose.
The binding of D-fructose shifts the loops between β4 and α4 and between α1’ and α1. This permits the enclosure of the substrate in the active site. With the concerted action of the D-fructose binding and the loop shift, all the water molecules leave the active site in the ”closed” conformation, even those implicated in Mn2+ ion coordination, which are replaced by O2 and O3 from D-Fructose.
In the Dpe-fructose complex, the O2 and O3 of the sugar are in an eclipsed conformation that mimics the intermediate cis-enediolate of the epimerization reaction. Three residues, E156, H186 and R215 are involved in a putative hydrogen bond with O1 of D-fructose. The Mn2+ ion is within 2,4 Å of the O2 and O3. O2 might be involved in hydrogen bonds with H186 or E244. O3 is also possibly stabilized by interactions with E150 or E244. For O4 there is a possible interaction with E150. O5 had no possible interaction within 4.0 Å. To finish, O6 seems to interact with A107 or I66 with hydrogen bonds to the backbone.
The putative catalytic mechanism of isomerisation involves two glutamate acid residues coordinated with the Mn2+ ion: E150 and E244. One of the two residues removes a proton from C3 to generate a cis-enediolate intermediate, and the second one helps to protonate C3 from the other side, in order to form the epimer. In the case of the epimerization from D-fructose to D-psicose, E244 removes the proton and E150 gives it back.
All this information can be crucial for us to investigate a directed mutagenesis on these residues that interact with fructose or the cofactor to improve the activity of the enzyme.
Clostridium cellulolyticum Dpe (UniProt ID B8I944) has its enzymatic activity  greatly enhanced in the presence of Mn2+ or Co2+. The bioconversion level of D-psicose is 32 % with a reaction buffer containing 0.5 μM of enzyme and 50 g/L of D-Fructose at 55°C pH 8.0 for 30 minutes. When the enzyme is at 55°C and with a pH of 8.0, it reaches its maximum activity and can produce D-psicose with a concentration of 218 g/L. The equilibrium ratio at 30°C between D-psicose and D-fructose is 32:68 [ref]. Dpe shows a Michaelis Menten constant (KM) of 17.4 ± 0.2 mM for D-psicose and 62.7 ± 1.5 mM for D-fructose. Which means that the enzyme has a better affinity for D-psicose than D-fructose. Moreover, the turnover number (kcat) are equivalent with 3243.5 ± 56.5 min-1 for D-psicose and 3354.5 ± 47.2 min-1 for D-fructose. Therefore, the catalytic efficiency (kcat/KM) values are 62.7 ± 1.5 mM-1min-1 for D-fructose and 186.4 ± 1.9 mM-1min-1 for D-psicose. The conversion rate is better with D-psicose. Finally, the thermostability of Dpe has been measured and the half-life of the enzyme is about 9.5 h at 55°C.
Cloning of a D-Psicose 3-Epimerase in Escherichia coli
After investigating the characteristics of Dpe, we searched for a range of Dpe from different species. We ordered the gBlocks of Dpe from Clostridium cellulolyticum (UniProt ID B8I944), Flavonifractor plautii (UniProt ID G9YVF8) and Pseudomonas cichorii (UniProt ID O50580) preceded by a RBS and a pTacI promoter. We also designed primers to amplify by PCR two epimerases from the Agrobacterium tumefaciens C58 genome. These epimerases showed great yields in the literature. We wanted to compare the efficiencies of these epimerases and to see which one was the best for our bioproduction techniques.
To express these Dpe, we decided to insert them into pSB1C3. We readily produced clones of the Dpe from C. cellulolyticum with RBS and pTacI promoter (BBa_K2448033) in pSB1C3 and one of the A. tumefaciens Dpe with RBS (BBa_K2448034). Knowing that the A. tumefaciens Dpe was already well characterized we decided to start our production with the C. cellulolyticum Dpe.
Bioproduction of Psicose
To bioproduce psicose from fructose we had to choose from many bioproduction methods (Click here to find out more). Since our goal was to test if we were able to produce psicose with our cloned Dpe we chose the fastest and cheapest method: Batch cultures for a whole cell production. Indeed, this technique requires little material and offers strong resistance to environmental perturbations. It also limits the purification steps compared to in vitro or cell lysate. This is essential for us since we want to directly measure our psicose with an HPLC. This method has still some limitations such as culture conditions that do not match those of the enzyme, therefore our yields may be limited.
Study of the Strains and Culture Conditions
Having a plasmid containing a gene coding for the Dpe from C. cellulolyticum, under the control of a pTacI promoter, we first investigated in which strain we would perform our psicose bioproduction. We were interested in two strain of E. coli: Top10, a strain often used for cloning and biosensor characterization, with a fast growth and a high plasmid replication; and BL21-AI, a strain used for expression of recombinant proteins. We compared the growth of these strains.
The cultures were made in Synthetic Medium to ensure its exact composition and avoid column deterioration in further HPLC measurements. We also wanted to investigate the impact of the carbon source on the growth of these two strains. We used four kinds of carbon sources: Glucose, Fructose, Glucose + Fructose and Glycerol + Fructose.
As shown in figure 4, the BL21-AI strains had high growth rates regardless of the sugar, with a stationary phase reached in five hours. The carbon source has no impact on this strain. The growth of the Top10 strain is slower compared to BL21-AI. The sugar used in the medium has a great impact on the fitness of the strain, with a maximal optical density for the fructose condition. Therefore, we decided to do all further bioproduction assays in BL21-AI.
Impact of the Concentration of Fructose on the Psicose Production
Knowing that the sugar used had no impact we decided to use a standard concentration of 2 g/L of glucose for every culture and vary the fructose concentration. To test whether or not our strains were producing psicose we prepared a negative control, a BL21-AI strain transformed with pSB1C3 containing BBa_J04450, an RFP under the control of a constitutive promoter (BL21-AI-RFP). Cultures have been made for 24 hours, with IPTG induction when the culture reached an OD600 of 0.6. Samples were taken throughout the bioprocess and centrifuged to purify the samples. Then the psicose contained in each fraction was measured with an HPLC previously calibrated to quantify sugars such as glucose, fructose, glycerol and psicose.
As shown in figure 5, we had similar growths for every strain except for BL21AI-Dpe with 2g/L of fructose which has a delayed growth. The HPLC analysis showed us that BL21-AI-Dpe is able to produce up to 9 g/L of psicose from 50 g/L and 80 g/L of fructose in the medium and that there is a proportional relationship between the fructose concentration and the psicose measured. No psicose was produced with our negative control which was expressing RFP.
In these results we did not reach saturation of the enzyme with 80 g/L of fructose. Therefore, we wanted to test the highest concentration of fructose and see if the yield could be enhanced. So, we used higher concentrations of fructose and we observed that only the cultures with 50 g/L of fructose showed a high growth rate.
As we can see in figure 6, the increase in the fructose concentration is inversely proportional to the growth of this strain. The high concentrations might create an osmotic stress on the cultures. We decided to use a concentration of 50 g/L of fructose to ensure high levels of growth and psicose production.
Impact of the Induction Intensity on Psicose Production
Finally, we wanted to investigate if we could enhance the yield of the bioproduction by increasing the IPTG induction, thus increasing the quantity of Dpe in the cells.
As we can observe from the results presented in figure 7, we have the same growth rate regardless of the amount of inducer (IPTG) added to the media and also the production of psicose is similar, around 8 g/L. The intensity of the induction of our Dpe did not optimize our bioproduction.
We achieved the bioproduction of D-Psicose using recombinant E. coli cells expressing the D-Psicose 3-Epimerase from Clostridium cellulolyticum (UniProt ID B8I944). The strain was capable of producing 9 g/L of psicose for 50 g/L of fructose in the culturing media after 24 h of incubation at 37°C, which represents a yield of 18%. This conversion rate is comparable to the biocatalysis yields described in the literature (Click here to find out more). However, we used non-optimal temperature (37°C the optimal E. coli growth temperature and not 55°C, the optimal activity temperature of Dpe from C. cellulolyticum), cells were not permeabilised and the enzyme was not purified. To the best of our knowledge our study is the first attempt to bioproduce psicose using living microorganisms in culturing conditions.
Multiple aspects of the bioproduction could be improved on, but one key element is the activity of the Dpe. To improve it, we developed a directed evolution system for Dpe and we screened around 400 mutants with our biosensor to select variants with improved activity (Click here to find out more). The fluorescence intensity generated by our biosensor can be correlated to the activity of the mutated Dpe, allowing an easy and quick screen of the mutant bank.
Our bioproduction method was also interesting for us since the biosensor will evaluate the activity of the enzyme inside a living cell. Therefore, our production and our screening techniques are carried out in the same conditions.
Short Term Perspectives
Going forward, we should recover the mutated plasmids to sequence the mutant enzymes, extract the sequences by PCR to put them in an expression vector and then implement them in a BL21-AI strain to produce psicose with better yields. There could be multiple rounds of directed evolution on these selected variants with higher activity to enhance the production. This would allow the prices to decrease so that psicose could have a greater societal impact.
Longer Term Perspectives
Apart from more rounds of directed evolution there are many factors that could allow us to improve our production to a global scale level. First of all, we could still focus on an in vivo technique by integrating our Dpe in Bacillus subtilis. This strain is “generally recognized as safe” (GRAS), so it can be used to produce compounds for human consumption. B. subtilis is also able to efficiently export compounds of its cytosol, thus avoiding cell lysis to recover products.
In vitro production offers more versatility for bioproduction. By purifying an enzyme, one can avoid the limitations of the bacteria, therefore using the optimal conditions for the enzyme activity. There are new methods such as the Simulated Moving Bed Chromatography which seems to be promising for the production and purification of rare sugars like D-psicose . This industrial process is used to purify reaction products catalyzed by epimerases, which ensures the enzyme never reaches it’s equilibrium.
-  Iida T, Hayashi N, Yamada T, Yoshikawa Y, Miyazato S, Kishimoto Y, Okuma K, Tokuda M, Izumori K. Failure of D-psicose absorbed in the small intestine to metabolize into energy and its low large intestinal fermentability in humans. Metabolism (2010) 59, 206-214.
-  Hishiike T, Ogawa M, Hayakawa S, Nakajima D, O'Charoen S, Ooshima H, Sun Y. Transepithelial transports of rare sugar D-psicose in human intestine. J Agric Food Chem (2013) 61, 7381-7386.
-  Kim HJ, Hyun EK, Kim YS, Lee YJ, Oh DK. Characterization of an Agrobacterium tumefaciens D-psicose 3-epimerase that converts D-fructose to D-psicose. Appl Environ Microbiol (2006) 72, 981-985.
-  Kim K, Kim HJ, Oh DK, Cha SS, Rhee S. Crystal structure of D-psicose 3-epimerase from Agrobacterium tumefaciens and its complex with true substrate D-fructose: a pivotal role of metal in catalysis, an active site for the non-phosphorylated substrate, and its conformational changes. J Mol Biol (2006) 361, 920-931.
-  Mu W, Chu F, Xing Q, Yu S, Zhou L, Jiang B. Cloning, expression, and characterization of a D-psicose 3-epimerase from Clostridium cellulolyticum H10. J Agric Food Chem (2011) 59, 7785-7792.
-  Wagner N, Bosshart A, Failmezger J, Bechtold M, Panke S. A separation-integrated cascade reaction to overcome thermodynamic limitations in rare-sugar synthesis. Angew Chem Int Ed Engl (2015) 54, 4182-4186.