Method for in vivo production of glycosaminoglycans

Abstract

The present invention relates to a method for in vivo production of glycosaminoglycans (GAG), by metabolic engineering in a genetically modified cell. In a method according to the invention, said cell is genetically modified in order to express the genes coding for the enzymes that are suitable for synthesizing GAG from an exogenous precursor, preferably internalized by the cell. According to one specific feature, the present invention relates to a method for producing chondroitin or heparosan in a genetically modified bacterial cell, from an exogenous beta-galactoside precursor, preferably internalized by the cell. According to another specific feature, the present invention relates to the use of an Escherichia coli cell comprising at least the genes glcA-T, kfoC, kfiD and wbpP for the production of chondroitin. According to yet another specific feature, the present invention relates to the use of an Escherichia coli cell comprising at least the genes glcA-T, kfiA, kfiB, kfiC and kfiD for the production of heparosan. The present invention also relates to the use of the obtained glycosaminoglycans by implementing a method according to the invention in order to prepare a drug, a food composition or a cosmetic product.

Claims

1. A method for producing glycosaminoglycan (GAG) from an exogenous beta-galactoside substrate in a genetically modified cell, said method comprising: a) obtaining a cell comprising at least i) an exogenous gene encoding a beta-galactosyltransferase, and the elements enabling the expression of said gene in said cell and the synthesis of glycosyl beta-galactoside, ii) a gene encoding an enzyme capable of synthesizing said glycosaminoglycan from glycosyl beta-galactoside and the elements enabling the expression of said gene(s) in said cell, and iii) a gene encoding an enzyme capable of ensuring the internalization by the cell of exogenous beta-galactoside, and the elements enabling the expression of said gene in said cell; b) culturing said cell in the presence of exogenous beta-galactoside and under conditions compatible with the production of said glycosaminoglycan by said cell; and c) isolating said produced glycosaminoglycan.

2. The method according to claim 1, wherein said beta-galactosyltransferase is a mammalian glucuronyltransferase.

3. The method according to claim 1, wherein the beta-galactosyltransferase is a mouse glucuronyltransferase.

4. The method according claim 1, wherein said cell further comprises a gene encoding an enzyme capable of synthesizing UDP-GlcA from UDP-Glc, and the elements enabling the expression of said gene in said cell.

5. The method according to claim 4, wherein said enzyme capable of synthesizing UDP-GlcA from UDP-Glc is a UDP-glucose dehydrogenase.

6. The method according to claim 4, wherein the enzyme capable of synthesizing UDP-GlcA from UDP-Glc is a UDP-glucose dehydrogenase encoded by the kfiD gene of E. coli K5.

7. The method according to claim 1, wherein said cell further comprises a gene encoding an enzyme capable of synthesizing UDP-GaINAc, and the elements enabling the expression of said gene in said cell.

8. The method according to claim 1, wherein said cell is selected from bacteria and eukaryotic organisms.

9. The method according to claim 8, wherein said bacterium is an Escherichia coli nonpathogenic strain bacterium.

10. The method according to claim 1, wherein said glycosaminoglycan is chondroitin.

11. The method according to claim 1, wherein said glycosaminoglycan is heparosan.

Description

FIGURE LEGENDS

(1) FIG. 1: Overall diagram of the metabolic engineering of the genetically modified cells. After the entry of an acceptor in the bacterium: heterologous expression of the glycosyltransferase of interest (step 2), accumulation of sugar nucleotides, substrates of glycosyltransferases (step 3).

(2) FIG. 2: Diagram of the strategy for synthesizing chondroitin in E. coli from exogenous lactose (Lac), with galactose (Gal), glucose (Glc), uridine diphosphate (UDP), glucuronic acid (GlcA), N-acetylgalactosamine (GalNAc), and uridine diphosphate-glucose (UDP-Glc), uridine diphosphate-glucuronic acid (UDP-GlcA), uridine diphosphate-N-acetylgalactosamine (UDP-GalNac), uridine diphosphate-N-acetylglucosamine (UDP-GlcNac), the E. coli gene encoding a GlcNAc transferase (kfiA) the E. coli gene encoding a glcA transferase (kfiC), the E. coli gene encoding a UDP-glucose 6-dehydrogenase (kfiD) the gene encoding the UDP-N-acetylglucosamine epimerase from Pseudomonas aeruginosa (wbpP) the gene encoding glucuronyltransferase (glcA-T) the gene encoding the enzyme using UDP-GlcA for the synthesis of colanic acid (wcaJ) the E. coli gene encoding chondroitin polymerase (kfoC)

(3) FIG. 3: Measurement of uronic acids (black bars) and amino sugars (white bars) by colorimetric assays in intracellular fractions of the recombinant strains, with, from top to bottom, KCB2, KCB1 and SBP9-24. The amount of uronic acids (UA) and of amino sugars is indicated in g/l.

(4) FIG. 4: Measurement of uronic acids and amino sugars by colorimetric assays in intra and extracellular fractions of the recombinant strains, with, from top to bottom: extracellular fraction of KCO3, KCO1 and KCB2, intracellular fraction of KCO3, KCO1 and KCB2. The amount of uronic acids (UA, black bars) and of amino sugars (white bars) is indicated in g/1.

(5) FIGS. 5A and 5B: Spectra resulting from GPC analysis of chondroitin produced by the recombinant strains. FIG. 5A: Analysis of the chondroitin produced by the recombinant strain KCB2 (average measured size: 17,970 Da, polydispersity: 1.35). FIG. 5B: Analysis of the chondroitin produced by the recombinant strain KCO1 (average measured size: 73,510 Da, polydispersity: 1.42).

(6) FIGS. 6A to 6C: Spectra resulting from size-exclusion chromatography-multiangle light scattering (SEC-MALS) analysis of the SL (FIG. 6A), AL1 (FIG. 6B) and AL2 (FIG. 6C) fractions obtained during the production of heparosan.

(7) FIG. 7: Spectra resulting from GPC (SEC-MALS) analysis of the heparosan fraction produced in strain KHB3 from allyl-lactose. Peak 1, aggregates; peak 2, Mw 52,740 Da, Mw/Mn 1.17; peak 3, Mw 22,346 Da, Mw/Mn 1.09.

EXAMPLES

Example 1: Engineering of Glycosylation in Escherichia coli

(8) The E. coli K12 strain (called DJ) used to produce chondroitin from exogenous lactose is disclosed in Yavuz et al. (2008), this strain is an E. coli DH1 strain (reference ATCC: 68153) modified as indicated in the publication of Yavuz et al. The K12 strain used is transformed by a plasmid having the E. coli K4 kfoC gene. The kfoC enzyme uses UDP-GalNac which is provided by the expression of wbpP, the gene encoding a UDP-N-acetylglucosamine epimerase from Pseudomonas aeruginosa. The Pseudomonas aeruginosa wbpP gene, encoding UDP-Glc epimerase, was also introduced into the bacterium. For the synthesis of glucuronyl-lactose, the GlcA-T enzyme enables the addition of a glucuronic acid (GlcA) on the lactose from UDP GlcA. The gene encoding rat brain glucuronyltransferase (GlcA-T) is also introduced into the bacterium. It is important to have an upstream accumulation of UDP-GlcA. To that end, the strain is transformed with a plasmid containing the kfiD gene. This gene encodes a UDP glucose dehydrogenase present in K5 enabling the synthesis of UDP-GlcA from UDP-Glc. Moreover, the K12 strain was made LacZ- and wcaJ- by deletions. The lacZ gene is part of the lactose operon; it encodes a beta-galactosidase that hydrolyzes lactose to galactose and glucose. Its inactivation prevents the degradation of lactose and thus increases its availability for metabolic synthesis of the product of interest. The wcaJ gene, in turn, encodes an enzyme using UDP-GlcA in the synthesis of colanic acid. Its inactivation enables, in the present case, an accumulation of UDP-GlcA. Moreover, colanic acid being an acidic oligosaccharide, it can interfere during chondroitin purifications, the fact of inhibiting its synthesis is thus all the more beneficial. The lacZ-, wcaJ-K12 strain is called DJ.

(9) The various strains studied are SBP9-24, KCB1 and KCB2 (Table 1).

(10) TABLE-US-00001 TABLE 1 Characteristics of the strains studied Strain Host Plasmids present Resistances Comments SBP9-24 DJ pACT3-GlcA-T, pBS-kfoC, KATC Reference strain, pBBR-kfiD, pWKS-wbpP 4 plasmids present KCB1 DJ pBBR-kfiD, pBS-kfoC-GlcA-T, KAT 3 plasmids present, pWKS-wbpP GlcA-T gene inserted into a high-copy-number plasmid KCB2 DJ pBBR-GlcA-T-kfiD, KAT 3 plasmids present, pBC-kfoC, pWKS-wbpP GlcA-T inserted into an average-copy-number plasmid KCO1 DJ pBBR-kfiD, pBC-kfoC, KAT Control without GlcA-T pWKS-wbpP KCO3 DJ pBBR-GlcA-T-kfiD, KAT Control without KfoC pBS-empty, pWKS-wbpP

(11) Strain SBP9-24 is regarded as the reference construction with four plasmids, each carrying a gene of interest. KCB1 and KCB2 contain only three plasmids and differ by the plasmid that carries two genes (pBS-kfoC-GlcA-T for KCB1 and pBBR-GlcA-T-kfiD for KCB2), their difference being that KCB2 has the GlcA-T gene in a plasmid expressed in a lower level than that of KCB1. Two control strains are prepared: KCO1 and KCO3, in order to observe the impact of the expression of GlcA-T (KCO1) and KfoC (KCO3) on the production of chondroitin. The lacZ, wcaJ BL21 strain of E. coli is transformed by the plasmids: pBBR-glcAT-kfiD; pBS-kfoC; pWKS-wbpP, is also used.

Example 2: Production of Chondroitin in E. coli from Exogenous Lactose

(12) The transformations of E. coli K12 are carried out from the K12 strain DJ which is lacZ and wcaJ. The plasmids (200 ng for pBBR and pWKS, 100 ng for pBS) are added to 100 l of competent bacteria. After 15 minutes in ice and heat-shock for 45 seconds at 42 C., 400 pl of SOC medium is added and the whole is incubated for 1 hour at 37 C. After concentration in 100 A of supernatant, the mixture is plated on a plate containing LBA+glucose (0.1% v/v) and the corresponding antibiotics (0.1% v/v) and incubated at 30 C. overnight.

(13) Composition of the Media

(14) KH.sub.2PO.sub.4 (5 g/l), NH.sub.4PO.sub.4 (5 g/l), citric acid (0.5 g/l), NaOH (0.66 g/l), KOH (1.65 g/l), TMS (7.5 ml/l), glycerol (2.4 g/l). Glucose (17.5 g/l) and MgSO.sub.4 (1 g/l) are autoclaved separately. Thiamin (10 mM) and antibiotics (35 mM) are added after autoclaving.

(15) Composition of the Supply

(16) Glycerol (750 g/l), MgSO.sub.4 (18 g/l), TMS (37.50 ml/l). The supply is delivered at a flow rate of 9 ml/min during the second phase then at a flow rate of 6 ml/min during the third phase (see Culture preparation).

(17) Culture Preparation

(18) The cultures are grown in 3-liter Applikon fermentors with a total volume of medium of 1.5 liters. During these cultures, several parameters are controlled; oxygen at 40%, temperature at 33 or 28 C. according to the culture phase and pH at 6.80 with ammonia at 30% (10). Culture preparation is divided into several phases.

(19) The first phase corresponds to a growth phase with the provision of a carbon source, glucose. The latter enables catabolic repression, thus the lac operon, used as promoter in the plasmids carrying the genes of interest, is inhibited. This prevents any production before induction and thus promotes bacterial growth. During this step, the temperature of the culture is maintained at 33 C. This step is concluded when all the glucose and glycerol present in the initial culture medium has been consumed.

(20) The second phase is an induction phase with the provision of a supply containing glycerol as carbon source to replace the glucose of the initial medium. The supply is administered at a flow rate of 9 ml/min. After a test for reaction of the culture to the supply (stopping the supply and observing an increase in oxygen), the culture temperature is decreased to 28 C. After a second test for response to the supply at this temperature, induction of the expression of the genes of interest was achieved by injecting 50 mg/l IPTG. At that time, the exogenous precursor necessary to the synthesis chain, lactose, is also provided in the medium at a concentration of 1.6 g/l.

(21) The third and last phase is started 5 hours after induction by reducing the supply flow rate (6 ml/min) so as to slow bacterial growth while continuing to provide the energy supply necessary to the production of the polysaccharide of interest.

(22) During the cultures, 1.5 ml samples are taken 1 hour after induction (1 h), the day after induction (24 h), the third day one hour after the second lactose injection (48 h) and the last day before stopping the cultures (72 h). After centrifugation for 5 minutes at 13,000 g, the supernatant is set aside (extracellular fraction) while the pellet is taken up in 1 ml of distilled water then placed in a 100 C. water bath for 10 minutes. After resuspension and centrifugation for 3 minutes at 13,000 g, the pellet contains bacterial debris and the supernatant corresponds to the intracellular fraction. The fractions resulting from these samplings are analyzed on a thin layer of silica (TLC) by migration in n-butanol/acetic acid/water (2:1:1) buffer for 30 minutes. The standard used during these TLC is a GlcA-Lac/lactose mixture. The TLC are developed with a mixture of ethyl acetate (100 ml), diphenylamine (2 g), aniline (2 ml), 85% phosphoric acid (10 ml) and hydrochloric acid (1 ml).

(23) After stopping the fermentors, the culture is collected then centrifuged for 20 minutes at 7,000 rpm in order to separate the bacteria from the extracellular medium. The bacterial pellets, taken up in 500 ml of H.sub.2O, are subjected to heat treatment (20 minutes in an autoclave at 100 C.) so as to lyse the bacteria. To finish, the pellets are resuspended and ground then centrifuged for 20 minutes at 7,000 rpm so as to remove the bacterial debris and retain only the supernatant which corresponds to the intracellular fraction of the bacteria. The supernatants collected following the first centrifugation correspond to the extracellular fractions.

(24) Purification and Analysis of Sugars

(25) First, the pH of the fractions is adjusted to 3.5 by treatment with cation-exchange resin (Amberlite IR-120) which precipitates the proteins, the fractions are centrifuged for 20 minutes at 7,000 rpm in order to remove the protein precipitate and the supernatants had their pH adjusted to 6.5 by treatment with anion-exchange resin (Dowex 66). Second, 500 ml of the fractions is subjected to ethanol precipitation (3 volumes of 96% ethanol per 1 volume of fraction) so as to precipitate the chondroitin. After 3 hours at 4 C., the mixture is centrifuged for 20 minutes at 7,000 rpm and the pellet solubilized in 80 ml of water.

(26) The fractions thus purified are then assayed for uronic acid (UA) and for amino sugar by colorimetric assay, respectively by the method of Blumenkrantz (1973) with glucuronic acid (GlcA) as standard and that of Morgan (1955) with N-acetylgalactosamine (GalNAc) as standard. These assays made it possible to obtain a theoretical concentration of chondroitin by means of the following relationships:
Q.sub.chondroitin=(Q.sub.UA/M.sub.qIca)(M.sub.QalNAc18)+Q.sub.UA
Q.sub.chondroitin=(Q.sub.amino sugars/M.sub.GalNAc)(M.sub.GlcA18)+Q.sub.amino sugars

(27) The products thus obtained after purification were frozen in liquid nitrogen then lyophilized so as to obtain a dry mass weighed and used for the characterization of the product.

(28) The various culture products are analyzed by gel permeation chromatography (GPC-SEC). The GPC-SEC is carried out on Shodex 802/803 columns eluted with 0.1 M sodium nitrate (NaNO.sub.3) at a flow rate of 0.5 ml/min. Detection is triple with a Waters refractometer, a multiangle light scattering (MALS) detector with 18 angles (Wyatt) and a dn/dc of 0.142 (WO2012/089777). The data are processed with the Astra software (Wyatt).

(29) Results

(30) Comparison of the Various Recombinant Bacterial Strains

(31) Several recombinant strains are compared: SBP9-24, KCB1 and KCB2, in order to select the most advantageous strain in terms of quality and amount for producing the polysaccharide of interest. To that end, it was decided first to compare these strains in terms of synthesis of the chimeric acceptor (GlcA-Lac).

(32) Lactose must be correctly internalized in the bacterium to enable sufficient intracellular accumulation of GlcA-Lac. The latter could then be recognized as an acceptor by KfoC. Thus, during the cultures of the various recombinant strains, several samples were taken in order to evaluate by TLC the concentrations and distribution of lactose and GlcA-Lac.

(33) It is possible to differentiate the strains on certain points. First, for rate of lactose consumption, the three-plasmid strains, KCB1 and KCB2, consumed all the lactose from the first day after 24 to 48 hours, which is not the case for SBP9-24. In view of this result, it was decided to carry out a second lactose injection (1.6 g/l) at the end of 48 hours, which will also be the case for all the other cultures. This observation agrees with a second which is that lactose is internalized much more quickly by strains KCB1 and 2 than by strain SBP9-24 where extracellular lactose is observed longer after an injection. In the case of the three strains, lactose is observed to metabolize to GlcA-Lac, a sign that metabolic synthesis has begun. It seems, however, that the three-plasmid strains perform better in terms of GlcA-Lac production (high rate of lactose consumption, greater presence of GlcA-Lac at the end of the culture).

(34) Second, the strains are compared in terms of the production of chondroitin itself, by means of colorimetric assays of uronic acids and amino sugars, making it possible to quantify the production of chondroitin for each strain.

(35) Increased production of uronic acids and amino sugars is observed in the three-plasmid strains (FIG. 3). One possible reason for this difference is the fact that strain SBP9-24 grows with more difficulty than the two three-plasmid strains. Between KCB1 and KCB2, it seems that KCB2 produces more, a sign that GlcA-T expression does not need to be very high since this enzyme is encoded in an average-copy-number plasmid (pBBR) in KCB2 and not in a high-copy-number plasmid (pBS) in KCB1.

(36) It thus seems that the recombinant strain KCB2 is the most productive in chondroitin from the chimeric acceptor. It is thus this strain which was selected to carry out the production and characterization of the chondroitin.

(37) Analysis of Chondroitin Production by Strain KCB2

(38) It is important to show that in KCB2 chondroitin is produced from the chimeric acceptor, glucuronyl-lactose. To that end, a control strain, KCO1, is constructed, which does not have the gene encoding GlcA-T. Thus, this negative control cannot transform lactose into GlcA-Lac. That gives an indication of chondroitin production when glucuronyl-lactose is not present.

(39) The absence of glucuronyl-lactose in KCO1 results in a lower concentration of uronic acid and of amino sugar (FIG. 4), and thus of chondroitin (0.5 g/l), in the intracellular fraction. This amount is about three times lower than that produced in KCB2 (1.7 g/l). Thus, it appears that in the absence of glucuronyl-lactose, the synthesis of chondroitin occurs nevertheless, necessarily from an endogenous acceptor.

(40) The production of chondroitin in the recombinant strain KCB2 is also compared with that in another control: strain KCO3. This strain lacks the gene encoding KfoC. Its absence in strain KCO3 results in the virtual absence of uronic acid and amino sugar both in the homologous fraction and extracellularly (FIG. 4). From the TLC of the culture samples, it is observed that the production of glucuronyl-lactose indeed takes place. Thus, KfoC is essential to the synthesis of chondroitin, even if the chimeric acceptor is present.

(41) Characterization and Analysis of Chondroitin

(42) It was shown above that in the absence of glucuronyl-lactose, chondroitin is synthesized from an endogenous acceptor. That is consistent with the results of synthesis in vivo shown in various publications. However, the GPC analysis gives a size of chondroitin in KCB2 (produced from glucuronyl-lactose) of 17.5 kDa (FIG. 5A) whereas the publications report a size of synthesized chondroitin (in K4 or K5) in the vicinity of 100 to 600 kDa. Consequently, one can wonder whether the small size obtained in KCB2 is due to the properties of strain K12 or the fact of initiating the synthesis by a different acceptor. To answer this question, a GPC analysis is also carried out on the purified products of KCO1 (control strain without GlcA-T) and the measured size of the chondroitin produced is about 73 kDa (FIG. 5B). Thus, the presence of glucuronyl-lactose, as a precursor for the synthesis, results in the production of a different chondroitin, smaller than that produced from the endogenous acceptor.

(43) Another remarkable fact is that in the presence of glucuronyl-lactose, the 73 kDa chondroitin is completely absent (size between 15 and 20 kDa, polydispersity of 1.35) (FIG. 5A). Two hypotheses can be formulated to explain this observation. The first is that the production of glucuronyl-lactose results in a greater presence of substrate for KfoC than when the endogenous acceptor is used alone. Thus, the KfoC enzyme restarts synthesis more often from oligosaccharides and precursor instead of continuing polysaccharide elongation. The result is thus a decrease in the average size of the chondroitin produced.

(44) The second hypothesis is that the membrane environment stabilizes enzyme chondroitin coupling to the benefit of polysaccharide polymerization. Indeed, in the wild, chondroitin synthesis takes place in the perimembrane space, the endogenous acceptor being membranous. There, KfoC has membrane complex support for the synthesis of chondroitin whose roles are poorly understood (Whitfield, 2006). Glucuronyl-lactose being cytoplasmic, the synthesis, in the case of this study, thus takes place far from the membrane. This cytoplasmic positioning of the synthesis could have an impact on elongation of the chondroitin.

(45) The production of chondroitin in a nonpathogenic recombinant strain of E. coli is thus advantageous because it provides, compared to the other modes of production, ease of handling. Qualitatively, the chondroitin obtained is smaller in size than that obtained by other means (15 kDa versus around 100 kDa in culture and 20 kDa in vitro). This mode of production is also advantageous in terms of the amount produced, which reaches 1.7 g/l, whereas the highest productions in a fermentor in E. coli K4 are reported as having a production of 1.4 g/l (Schiraldi et al., 2010).

(46) The chondroitin produced was analyzed by NMR. The structure of the repeating unit [GlcAb-3GalNAcb-4] is confirmed by .sup.1H and .sup.13C NMR analysis, the results of which are consistent with those published in the literature for defructosylated K4 polysaccharide and chondroitin from Pasteurella multocida. The proton NMR spectrum shows signals at 4.55, 3.43 and 3.64 ppm, which correspond to H-1, H-3, and H-2 of the Glcp-UA residue. The signals at 4.18 and 2.08 ppm correspond to H-4 and H(Ac) of the Galp-NAc residue. The carbon NMR shows the assignments characteristic of the N-acetamido group at 23.91 and 176.22 ppm. One also finds the assignment characteristic of the C-6 of the Glcp-UA residue at 175.61 ppm.

Example 3: Production of Chondroitin in E. coli from Exogenous Beta-methylgalactoside

(47) The bacterial strains are prepared and grown according to the protocol disclosed in Example 2, except for the fact that the exogenous beta-galactoside consists in beta methylgalactoside, added in the amount of 4.2 grams in the culture medium. The amount of beta-methylgalactoside added to the medium is identical to the amount of lactose added in Example 2, taking into account the different molar masses of the two products.

(48) The chondroitin fraction produced has a size of 37-61 kDa, i.e. a size smaller than that of the chondroitin obtained in the absence of exogenous acceptor, but larger than that obtained in the presence of lactose.

Example 4: Production of Heparosan by an Escherichia coli K12 Strain in the Presence of Intracellular Glucuronyl-Lactose

(49) The production of recombinant heparosans is a major issue because they can be used as precursors of heparin, in particular if they have a small size of about 15-20 kDa (Manzoni et al., 1996). Heparosan is formed of a series of [-4)--GlcA-(1-4)--GlcNAc-(1]. Its biosynthetic pathway is similar to that of K4 polysaccharide and the genes associated with exportation are the same (Whitfield, 2006).

(50) An E. coli K12 strain was modified to produce heparosan (Barreteau et al., 2012). This strain expresses the E. coli kfiABCD genes (X77617 of 23 Oct. 2008), the presence of which is necessary and sufficient to produce heparosan intracellularly, having a size of about 80 kDa. This strain was now transformed by a plasmid expressing mouse glucuronyltransferase enabling the intracellular synthesis of glucuronyl-lactose in the presence of lactose. To follow the result of this study, a first expected clue of the handling of glucuronyl-lactose by the polymerases is the synthesis of a smaller polysaccharide, if one refers to the results obtained with chondroitin for which it is indeed the case.

(51) The same strain was grown in 2-liter fermentors in the presence (7.5 g of lactose added to the supply over 3 days) and in the absence of extracellular lactose.

(52) The results are as follows:

(53) Assay of uronic acids of the ethanol-precipitated fraction:

(54) Without addition of lactose: 825 mg total (about 0.8 g/l heparosan)

(55) With addition of lactose: 444 mg total (about 0.45 g/l heparosan)

(56) Size analysis of the recombinant heparosans:

(57) The fractions were analyzed by SEC-MAL in order to obtain the size and the dispersity of the heparosans produced. However, the large presence of aggregates necessitated a subsequent step of purification by ion-exchange chromatography.

(58) The without lactose fraction was passed through DEAE-A25 and a retained fraction representing 40% of the total deposited, called SL, was collected.

(59) The with lactose fraction was passed through DEAE-A25 and a retained fraction, called AL1, representing only 20% of the total deposited was collected, a difference in yield which could be due to the nature of the heparosans produced. It was decided to use a stronger exchanger of the QAE-A25 type. This made it possible to collect a second fraction, called AL2, representing 30% of the total of the initial fraction. Fractions SL, AL1 and AL2 were analyzed by SEC-MALS with no aggregate formation (FIGS. 6A with 6C). The characteristics of the analyzed peaks are as follows:

(60) TABLE-US-00002 TABLE 2 Characteristics of the heparosans produced Peak 1 Peak 2 SL % total (RI detection) 67 33 Mw (Da) 7.57 10.sup.4 2.05 10.sup.4 Mz (Da) 1.06 10.sup.5 2.26 10.sup.4 Polydispersity (Mw/Mn) 1.4 1.1 AL1 % 44 68 Mw (Da) 5.49 10.sup.4 1.87 10.sup.4 Mz (Da) 6.43 10.sup.4 2.17 10.sup.4 Polydispersity (Mw/Mn) 1.4 1.3 AL2 % 49 51 Mw (Da) 6.94 10.sup.4 2.72 10.sup.4 Mz (Da) 7.35 10.sup.4 2.89 10.sup.4 Polydispersity (Mw/Mn) 1.4 1.06

(61) Fraction SL is essentially composed of a polymer whose size has already been observed (Barreteau et al., 2012). However, it also has a fraction of smaller size which represents about 30% of the total. Fractions AL1 and AL2 are distinguished by an enrichment of the smallest fraction (peak 2), which represents at least 50-70% of the total based on refractometric quantification. Intracellular production in a recombinant E. coli K12 strain capable of synthesizing recombinant heparosan induced a change in the amount and the characteristics of the heparosans produced. The result is an enrichment of a small heparosan about 20 kDa in size.

Example 5: Production of Heparosan in an Escherichia coli K12 Strain in the Presence of Intracellular Glucuronyl-Lactose

(62) The genes specific to the synthesis of heparosan from E. coli belong to the cluster of operon K5 and are: kfiA, encoding a GlcNAc transferase; kfiB, whose role is unknown; kfiC, encoding a glcA transferase; kfiD, encoding a UDP-Glc dehydrogenase.

(63) The aim of this study is the metabolic engineering of E. coli K12 in order to enable a synthesis of heparosan from a beta-glucuronidated exogenous cytoplasmic acceptor coming from the action of mouse GlcA-T (encoded by the glcAT gene) on a beta-galactoside. Several strains are prepared, disclosed in the table below. Strain HB5 is disclosed in Barreteau et al. (2012).

(64) TABLE-US-00003 TABLE 3 Characteristics of the prepared strains Plasmids pBBR- pBS- pBBR- pSU- GlcAT- pBBR- kfiAC pSU- kfiAB kfiCD kfiD kfiABCD Amp, GlcAT Tet, low Cm, low Tet, low Tet, low high Cm, low Strain Host copy copy copy copy copy copy HB5 DJ + + KHB1 DJ + + KHB2 DJ + + KHB3 DJ + + +
A first strain, KHB1, is prepared, expressing the kfiA and kfiC genes encoding the two glycosyltransferases involved in the synthesis of heparosan in a high-copy-number pBluescript (pBS) plasmid, alongside the kfiD (synthesis of UDP-GlcA) and glcAT (mouse glucuronyltransferase) genes. This strain does not express KfiB, insofar as it was shown that this protein promotes the anchoring of KfiC to the membrane, where the endogenous membrane acceptor is located (Hodson et al., 2000). The aim here being to synthesize heparosan from glucuronyl-lactose, one can think that this membrane association is no longer necessary, or is even harmful, because the location of the endogenous acceptor is also membranous. Two cultures are prepared with or without added lactose. Lactose is added in the amount of 7.5 g (for 2 liters of culture) continuously over the 3 days of supply after induction. The results are obtained after ethanol precipitation of the intracellular fraction and the assay of uronic acids. No heparosan synthesis was obtained, whether in the presence or in the absence of lactose (Table 4). Nevertheless, these results corroborate those of the publication of Hodson et al. (2000), which mentions the crucial role of KfiB in the stability of the heparosan synthase enzyme complex.

(65) It is thus decided to express the set of kfiABCD genes, while adding thereto the expression of mouse glucuronyltransferase. To that end, the four kfiABCD genes are cloned into the same low-copy plasmid, in order to enable co-expression with the glcA-T gene in a second, also low-copy-number pSU plasmid. The strain, named KHB2, is grown in the presence and the absence of lactose. In the absence of lactose, this strain produces an amount of heparosan comparable with that obtained with strain HB5 published previously (Barreteau et al., 2012). This is consistent with the fact that these two strains are comparable, including in terms of the levels of expression of the four kfiABCD genes, all cloned into low-copy-number plasmids. The absence of lactose deprives the bacterium of the potential acceptor glucuronyl-lactose, which explains the similar results with the reference strain. In the presence of lactose, the result is a decrease in heparosan production, but the interesting fact is that one observes an enrichment of the small heparosan fraction, suggesting that the heparosan synthase complex has taken glucuronyl-lactose as polymerization acceptor, insofar as this phenomenon is observed with the synthesis of chondroitin from glucuronyl-lactose. These results show a low production yield of heparosan, and a polysaccharide fraction obtained composed of two fractions of different sizes.

(66) A new strain, named KHB3, is then prepared. KHB3 is in fact a strain for which the level of expression of the kfiA and kfiC genes is increased, compared to other genes, in order to increase the level of glycosyltransferase activities in relation to KfiB expression. The results are surprising: in the presence of lactose, heparosan production is doubled in relation to the reference strain, HB5, which is completely novel. This result was confirmed by growing this same strain in the presence of lactose-allyl. Without lactose, there is no increase in heparosan production, despite the higher expression of the heparosan synthase enzymes KfiA and KfiC. Also for chondroitin, we had observed that the addition of lactose increased the production yield. This suggests that the concentration of endogenous membrane acceptor is a factor limiting the production of chondroitin and heparosan in the Escherichia coli K-12 strain. The abundant presence of a cytoplasmic auxiliary acceptor would lift this limitation. The low expression of KfiB is apparently enough to stabilize the KfiA/KfiC enzyme complex despite the higher expression of the latter.

(67) The GPC analysis was carried out on the fraction prepared from allyl-lactose (FIG. 7). It is composed of two polysaccharide species, 53 kDa and 22 kDa in size, having very low polydispersity. These sizes are smaller than what is normally produced with the kfi genes, in our laboratory and also in the literature and existing patents. This phenomenon is consistent with what is observed with the production of chondroitin, where a decrease in size in the presence of glucuronyl-lactose is also observed.

(68) TABLE-US-00004 TABLE 4 Amount of heparosan produced (expressed as g/l of culture). Strain Without lactose With lactose With allyl lactose HB5 1.2 Not performed Not performed KHB1 Not detected Traces Not performed KHB2 0.9 0.45 Not performed KHB3 0.84 2.6 2.7

BIBLIOGRAPHICAL REFERENCES

(69) Cimini et al. (2010) Appl. Microbiol, and Biotechnol, 85(6):1779-1787. Jin et al. (2011) Biosci. Biotechnol. Biochem., 75(7), 1283-1289. Yamada and Sugahara (2008) Current Drug Discovery Technologies, 5, 289-301. Zanfardino et al. (2010) Microbiol. Cell Factories, 9:34. Wang et al. (2011) Bioengineered Bugs, 2:1, 63-67 (January/February) Schiraldi et al. (2010) Appl. Microbiol, and Biotechnol, 87:1209-1220. Inoue et al. (1990) Gene, 96:23-28. Yavuz et al. (2008) Glycobiology, 18(2):152-157. Ninomiya et al. (2002) J. Biol. Chem., June 14; 277(24):21567-75. Priem et al. (2002) Glycobiology. Vol. 12 no. 4 pp. 235-240. Krahulec et al. (2005) Molecular Biology, 30:129-134. Kakuda et al. (2004) J. Biol. Chem., 279 (21), pp 22693-22703. Whitfield (2006) Annual Review of Biochemistry, 75:39-68. Morgan et al. (1955) Biochem. Journal, 61(4):586-589. Blumenkrantz et al. (1973) Analytical biochemistry, 54:484-489. Manzoni et al. (1996) J. Bioactive and Compatible Polymers, Vol. 11, p 301 Barreteau et al. (2012) Carbohydrate Research, 1:360:19-24. Hodson et al. (2000) J. Biol. Chem., 275(35): 27311-5.