IMPROVED DEMINERALIZATION OF FERMENTATION BROTHS AND PURIFICATION OF FINE CHEMICALS SUCH AS OLIGOSACCHARIDES

Abstract

The present invention relates to a method for improved demineralization of fermentation broths, including the steps of providing a solution comprising one or more oligosaccharides, carrying out a first membrane filtration and preferably being a microfiltration or ultrafiltration, a second membrane filtration of the permeate of the first membrane filtration and a first nanofiltration step; with a sub-step of concentration and/or a sub-step of diafiltration. Moreover, the present invention is directed to an improved purification of fine chemicals from a fermentation broth.

Claims

1-17. (canceled)

18. A method for the demineralization of a solution comprising one or more fine chemicals, wherein the method comprises the steps of: a. providing the solution comprising one or more oligosaccharides; b. optionally adjusting the pH to the desired value below 7 or less by adding at least one acid to the solution comprising at least one oligosaccharide, c. optional decolourisation step, d. an optional incubation step, e. carrying out a first membrane filtration, f. a second membrane filtration of the permeate of the first membrane filtration, g. optional decolourisation step of the permeate of the second membrane filtration, h. a first nanofiltration step; with a sub-step of concentration and/or a sub-step of diafiltration.

19. The method according to claim 18, wherein steps b) and c) are performed.

20. The method according to claim 18, wherein no cation exchange or anion ion exchange is carried out.

21. The method according to claim 18, wherein in step h), a sub-step of concentration and a sub-step of diafiltration is carried out.

22. The method according to claim 18, wherein the concentration sub-step of step h) is performed so that the concentration factor is at least 3 or more.

23. The method according to claim 18, wherein the diafiltration sub-step of step h) is performed so that the diafiltration factor is from 2.5 to 3.5.

24. The method according to claim 18, wherein the nanofiltration membrane has a NaCl retention between 5 to 30%.

25. A method for purification of one or more fine chemicals from a solution comprising biomass and one or more oligosaccharides, comprising the steps of i. providing a solution comprising biomass and one or more oligosaccharides, ii. setting the pH value of the solution below 7 or less by adding at least one acid to the solution comprising biomass and the at least one oligosaccharide, iii. decolourizing the solution at least in part by adding an adsorbing agent to the solution comprising biomass and one or more oligosaccharides, iv. optionally an incubation step, v. carrying out a membrane filtration also called herein the first membrane filtration and typically being a microfiltration or ultrafiltration so as to separate the biomass from the solution comprising the one or more oligosaccharides; vi. optionally carrying out at least one second or further membrane filtration with the permeate of the first membrane filtration; vii. optionally carrying out a decolourization step; viii. carrying out a nanofiltration with the permeate of the membrane filtration antedating this step viii, either with the permeate of the first or the second or any further membrane filtration; ix. optionally a decolourization step; x. optionally a second nanofiltration with the retentate of the nanofiltration of the previous nanofiltration step viii, wherein the nanofiltration membrane used is a different one to the one in the previous nanofiltration step; xi. optionally a third or further nanofiltration with a membrane differing from the one of the previous nanofiltration step.

26. The method of claim 25, with subsequent to step xi further processing of the retentate of the previous nanofiltration step by any of the following steps is conducted: a. carrying out a decolourization step, and / or b. carrying out a demineralization step, and / or c. carrying out an electrodialysis and / or reverse osmosis and / or concentration step and / or a decolourization step; and / or d. carrying out a simulated moving bed chromatography, and / or a solidification step creating a solid oligosaccharide product, and / or a spray drying of the oligosaccharide followed by drying as desired.

27. The method according to claim 26 wherein a demineralization step is conducted after the last one of the one or more nanofiltrations of steps viii, x and xi, wherein the demineralization is performed by ion exchange and wherein further the throughput of the demineralisation step is increased by a factor of at least 2 or more compared to the throughput of an identical ion exchange step without the one or more nanofiltrations of steps viii, x and xi preceding it.

28. The method according to claim 18, wherein the pH value of the solution is lowered to a pH value in the range of 3.0 to 5.5.

29. The method according to claim 18, wherein said adsorbing agent is added in an amount in the range of 0.5% to 3% by weight.

30. The method according to claim 18, wherein said first membrane filtration is carried out as cross-flow microfiltration or cross-flow ultrafiltration.

31. The method according to claim 18, wherein said first membrane filtration is carried out at a temperature of the solution in the range of 8° C. to 55° C.

32. The method according to claim 18, further comprising carrying out a second membrane filtration with the solution comprising oligosaccharides obtained by the first membrane filtration.

33. The method according to claim 31, wherein said second membrane filtration is carried out at a temperature of the solution being in the range of 5° C. to 15° C.

34. The method according to claim 18, wherein said at least one oligosaccharide comprises human milk oligosaccharide.

Description

DESCRIPTION OF FIGURES

[0208] FIG. 1 shows a block diagram of the method for purification of one or more fine chemicals from a solution comprising for example biomass and at least one fine chemical according to the present invention. Dashed outlines indicate optional parts, whereas, dotted outlines indicate parts which are, in case of demineralization according to claim 1, optional but otherwise non-optional. S10 denotes the provision of the solution comprising the fine chemical, preferably an HMO or an aroma compound; S12 is adjusting the pH value to the desired pH preferably below pH 7; S14 is a decolourization step; S15 is an optional incubation step with the adsorbing agent; S16 a first membrane filtration (MF) step; S18 a second membrane filtration step; S20 is a decolourization step; S22 a first nanofiltration (NF) step; S24 a second nanofiltration step,; S26 an optional demineralization step, shown typically as an ion exchange step (IEX); Decol.; Conc.; ED; RO indicates optional Decolourization, Concentration, Elektrodialysis and / or Reverse Osmosis steps in any order; SMB is short for simulated moving bed chromatography; Solidification indicates the step of producing solid particles of fine chemical if desired - some applications may prefer the fine chemical product to be in a purified solution instead. Perm. Stand for permeate; Ret. Stands for retentate; Reg. stands for regenerate of the ion exchanger; FT stand for the flow-through of the ion exchanger that largely comprises the desired fine chemical(s)

[0209] FIG. 2 shows in a block diagram a more preferred method for purification of oligosaccharides or other fine chemicals. Abbreviations, depictions and steps are as shown in FIG. 1, with the following changes: S10 is the provision of a solution comprising the fine chemical, preferably the oligosaccharide, and biomass in form of a fermentation broth. The first membrane filtration S16 is shown in three sub-steps (S16/1 to S16/3) of first membrane filtration being first diafiltration DF, concentrating C. and then optionally a second diafiltration. The second membrane filtration S18 is preferably an ultrafiltration (UF); Step 22 is shown in more detail as a first concentration sub-step (S22/1) of the nanofiltration followed by a second sub-step in diafiltration mode (S22/2). The demineralisation step by ion exchange is shown in two sub-steps, first (S26/1) a cation exchanger resin (CIEX) is used, preferably a strong one, and a subsequent sub-step S26/2 with an anion exchange resin (AlEX), preferably a weak one; Following step S26 in the preferred method, no further decolourisation is needed.

[0210] FIG. 3 shows the schematic set-up of a unit for testing the nanofiltration using spiral wound membrane elements. The unit is equipped with two pumps, one for feed pressure and one to generate the desired crossflow.

[0211] FIG. 4 shows as block diagram another preferred method for the purification of fine chemicals from fermentation broth. After the provision of the fermentation broth S10 it continuous in the same manner as in FIG. 1 but there is no demineralization step S26 included any longer, as the combination of biomass separation and nanofiltration to remove the ions results already in a purified solution that is suitable for use or the optional further processing steps as shown in FIG. 4.

[0212] FIG. 5 shows as block diagram another preferred method for a rapid purification of a fine chemical such as an oligosaccharide from a solution comprising said fine chemical, wherein the method begins with an optional step of adjusting the pH to the desired value below 7, then a first and an optional second membrane filtration step as described herein as S22 and S24 respectively, and optional further processing steps of the purified solution.

EXAMPLES

[0213] The method according to the present invention will be described in further detail below. Whatsoever, the Examples shall not be construed as limiting the scope of the invention.

Analytical Methods

[0214] The following analytical methods have been carried out. [0215] HPLC for the determination of the product, i.e. human milk oligosaccharides, disaccharides, monosaccharides, and secondary components [0216] Drying balances for measuring the dry content [0217] APHA for measuring the colour using standard methods, for example DIN EN ISO 6271 [0218] Bradford protein assay for measuring the concentration of protein.

Abbreviations and Symbols

[0219] Hereinafter, the following abbreviations are used: [0220] AC = Active Carbon [0221] UF = Ultrafiltration [0222] NF = Nanofiltration [0223] DP = Pressure drop along the module (p.sub.feed - p.sub.retentate) [0224] Cross-flow velocity = linear speed of the suspension in membrane channels (m/s) [0225] Membrane load = amount of permeate produced by 1m2 of membrane area (m3/ m2)

[0226] Further, regarding the liquid separation, the following symbols and explanations are used.

TABLE-US-00001 Symbol Meaning Unit Definition Letters CF Concentration factor - m.sub.R,t=0/ m.sub.R DF Diafiltration factor - m.sub.P/m.sub.R,t=0 J Flux LMH = L m.sup.-2 h.sup.-1 P Permeance, some-times also referred to as permeability L m.sup.-2 h.sup.-1 bar.sup.-1 m Mass kg p Pressure bar R Retention - 1 - c.sub.permeate/c.sub.retentate TMP Trans-membrane pressure bar (p.sub.feed + p.sub.retentate)/2 - p.sub.permeate

[0227] The retention for a specific compound i is calculated by:

[00001]$R_i=1-\frac{c_i^{\text{permeate}}}{c_i^{\text{retentate}}}$

i.e one minus the ratio of the concentration of a component i in the permeate to the concentration of a component i in the retentate.

[0228] When a mixture is diafiltrated, the concentration C of a component i decreases exponentially with the diafiltration factor DF according to the following relation:

[00002]$C_i=C_i^0\exp \left(-\left(1-R_i\right)\cdot DF\right)$

with C.sub.i.sup.0 being the concentration of the compound I at time 0.

Initial Purification Steps - Decolourization, Removal of Biomass and Initial Membrane Filtrations

[0229] A fermentation broth as a complex solution comprising biomass and at least one oligosaccharide has been prepared by standard methods. The pH value thereof has been lowered to 4 ± 0.1 by means of adding 10% sulfuric acid. Thereafter, about 100 g or more per 2.5 kg complex solution of a 30% suspension of active carbon Carbopal Gn-P-F (Donau Carbon GmbH, Gwinner-strasse 27-33, 60388 Frankfurt am Main, Germany), which is food safe, has been added and stirred for 20 min.

[0230] The thus prepared solution has been supplied to the process apparatus, a semi-automatic MF lab unit from Sartorius AG, Otto-Brenner-Str. 20, 37079 Goettingen, Germany, modified for the purpose, and heated to 37° C. in a circulating manner with closed permeate. For separation purposes, the process apparatus included a ceramic mono channel element (from Atech Innovations GmbH, Gladbeck, Germany) having an outer diameter of 10 mm, an inner diameter of 6 mm, a length of 1.2 m and a membrane made of Al.sub.2O.sub.3 having a pore size of 50 nm. As soon as the circulation of the solution is running and the solution comprises the target temperature of 37° C., the discharging of the permeate has been started and the control of the trans membrane pressure has been activated.

[0231] After terminating of the inventive method, the process apparatus has been stopped, the concentrate has been disposed and the process apparatus has been cleaned. Cleaning has been carried out by means of 0.5% to 1% NaOH at a temperature of 50° C. to 80° C., wherein the NaOH has been subsequently removed by purging.

[0232] i) With a fermentation broth containing inter alia 2-fucosyl lactose (2-FL), only a first diafiltration step with DF =1 and a concentrating step with CF = 2 were performed, each by means of a 50 nm Al.sub.2O.sub.3 membrane (available from Atech Innovations GmbH, Germany) and at a temperature of 40° C., a transmembrane pressure (TMP) of 1.2 bar and a cross-flow velocity of 4 m/s. Then, the first membrane filtration was stopped, the resulting solutions and remainder of the starting solutions were analyzed and the results compared.

[0233] Table A shows the analytical results depending on the pH value and active carbon. DC is the abbreviation for dry content. OD for the optical density.

TABLE-US-00002 pH Sample DC APHA OD 3.2-Di-Fl 2FL 2F-Lactulose Lactose Protein [%] [g/l] [g/l] [g/l] [g/l] [g/l] 7.0 Feed 17.8 138 3.43 62.07 0.6 4.28 0.478 Permeate 7.61 4196 1.99 34.54 0.43 2.54 0.124 Concentrate 18.5 136 1.882 7.0+1% AC Feed 18.3 119 3.29 62.22 0.33 3.93 0.964 Permeate 7.9 1467 2.14 37.69 0.26 2.25 0.073 Concentrate 17.3 237 1.89 31.39 0.54 0.12 1.41 4.0 Feed 17.3 150 2.83 54.98 0.59 0.89 0.76 Permeate 8.2 4784 1.90 35.00 0.37 2.57 0.019 Concentrate 17.7 434 0.026 4.0+1%AC Feed 16.8 151 2.83 55.52 0.34 3.65 0.760 Permeate 7.8 781 1.73 33.66 0.27 2.26 0.019 Concentrate 18.3 293 1.61 29.14 0.33 2.38 0.026

[0234] The following results are derivable from Table A: Adding 1% active carbon to the fermentation broth reduces the color value of the permeate. At a pH value of 7, 1% active carbon reduces the color value at approximately 65%. At a pH value of 4, 1% active carbon reduces the color value at approximately 84%. Thus. the color value is below the upper end of 1000 and a further decolorization is not necessary. Adding active carbon at a pH value of 7 reduces the concentration of protein within the permeate at approximately 40%. whereas no effect in this respect by adding active carbon can be derived at a pH value of 4 over the pH effect on protein concentration. Nevertheless. the concentration of protein within the permeate at a pH value of 4 and with adding 1% active carbon is smaller by a factor of 4 if compared to the concentration of protein within the permeate at a pH value of 7 and with adding of 1% active carbon. Adding active carbon has no significant influence on the concentration of the oligosaccharides 3.2-Di-fucosyllactose (3.2-Di-Fl). 2′Fucosyllactulose (2F-Lactulose) and 2′Fucosyllactose (2FL). within the permeate at both pH values. Thus. it can be derived that these components do not adhere to the active carbon in significant amounts. The disaccharide lactose shows in this experiment a small reduction in concentration when active carbon is used. yet the beneficial effect of lowered pH and active carbon allow for the application of the inventive method for this disaccharide.

[0235] ii) Several batches of fermentation broths produced with standard methods comprising 6′-sialyllactose, or Lacto-N-tetraose, have been subjected to the inventive methods. Lowering of the pH value and decolourization with an absorbing agent were the first steps.

[0236] First, the steps S10 to S18 were performed. Fermentation broths comprising Lacto-N-tetraose starting with a high concentration of colour components resulting in APHA values of 7000 or more in the fermentation broth, gave permeates after the first membrane filtration - by means of a 50 nm Al.sub.2O.sub.3 membrane (available from Atech Innovations GmbH, Germany) and at a temperature of 40° C., a transmembrane pressure (TMP) of 1.2 bar and a cross-flow velocity of 4 m/s -with an APHA value of below 1000, but typically below 300. The protein concentration was lowered from typically around 3 g/l to less than 0.01 g/l. The vast majority, typically above 95 %, of the Lacto-N-tetraose originally found in the fermentation broth was present in the combined permeate. Similarly, for other oligosaccharides present and also for the disaccharide lactose most was present in the combined permeate and only minor amounts found in the retentate at the end of the first membrane filtration. The applied DF values were below or equal to 3.5. For 6-SL and LNT, first a set-up with a diafiltration factor of 3 followed by a set-up with a concentration factor of 2 proved useful.

[0237] Also, fermentation broths comprising 6′-sialyllactose with APHA values of around 7000, after said first membrane filtration resulted in permeates with an APHA value of below 300. The protein concentration was lowered by a factor of at least 10 or more, even by more than 100, compared to the starting value in the fermentation broth, at DF values below 3. The vast majority, typically above 90% of the 6′-sialyllactose originally found in the fermentation broth was present in the combined permeate. Similarly, for other oligosaccharides present and also for the disaccharide lactose most was present in the combined permeate and only minor amounts found in the retentate at the end of the first membrane filtration.

[0238] It was also found that performing the methods with a pH of below 5.5 improved flux in the first membrane filtration compared to higher pH values (cross-flow speed 3.5 m/s, temperature 30° C.; DF = 3). This improved even further when the pH value of the solution comprising the biomass and the 6′- sialyllactose was pH 4.2. Compared to pH 6.3, the flux more than doubled when pH 5.2 was used and tripled when the pH value was pH 4.2.

[0239] The combined permeate of the first membrane filtration were submitted to an ultrafiltration as second membrane filtration.

[0240] For both 6′SL and LNT, the 4 kDa PES polymeric membrane (50 nm) UH004 of MICRODYN-NADIR GmbH, Kasteler Strasse 45, Gebaude D512, 65203 Wiesbaden/Germany gave a good performance with high performance and hardly any fouling. For 6′SL the pH of the fermentation broth was set to pH 4 with 10 to 20% sulphuric acid or phosphoric acid, 1.4% (w/w) active carbon were mixed in, and the first membrane filtration was conducted at 30-37° C., velocity 3.5 m/s: DF1=3 to 3.5 followed by CF=2 with a yield ≥95% at this first membrane filtration. Then as second membrane filtration an ultrafiltration was done with a 4 kDa PES Membrane at 8-12° C. and 10 bar, Cross-flow velocity of 1.5 m/s, with a CF1 >10 (up to 20), followed by a DF ≥ 3; the total yield was 95%.

[0241] For LNT, the pH of the fermentation broth was set to pH 4 with 10 to 20% sulphuric acid or phosphoric acid, 1.0% (w/w) active carbon were mixed in and stirred for 50 minutes, and the first membrane filtration was done with the Al.sub.2O.sub.3 membrane (50 nm) membrane was conducted at 30-37° C. with 3.5 m/s and a DF of 3, followed by a CF of 2; the yield was >97%. The following second membrane filtration was an ultrafiltration similar to the one for 6′SL at 8-12° C. and 10 bar with a concentration factor up to 80 and a subsequent DF = 3; total yield was over 99%.

[0242] The resulting permeates were stored refrigerated prior to use in the first nanofiltration or, if stored for longer times, frozen thawed and agitated before the first nanofiltration.

Nanofiltration

[0243] A number of nanofiltration membranes were tested with different oligosaccharides.

TABLE-US-00003 Overview of the used membranes and supplier’s cut-off and/or retention indications Membrane Supplier Retention / Cut-Off TS40 40% NaCl 99.0% MgSO.sub.4 TS80 80% NaCl 99.2% MgSO.sub.4 UA60 Microdyn-Nadir (Trisep) 10% NaCl 80% MgSO.sub.4 XN45 20% NaCl 96% MgSO.sub.4 NP030 80-95% Na.sub.2SO.sub.4 ESNA 3J 0.15-0.25 kD 7450 Nitto-Denko 50% NaCl 7470 70% NaCl Desal DL SUEZ 96% MgSO.sub.4 AS3014 AMS Technologies 0.4 kD >92% MgSO.sub.4 dNF40 NX Filtration 0.4 kD dNF80 0.8 kD

[0244] Most of the tested membranes showed very good retention of the oligosaccharide and often lactose as well. However, the tests also showed that some membranes are not well suited to let salts and other ions like phosphoric acid pass. Others, like UA60 or XN45 or AS3014 showed encouraging results indicated that separation of oligosaccharides from the ions like phosphoric acid can be achieved. Depending on the oligosaccharide and the set-up, less than 8%, less than 27% or less than 52% of the phosphoric acid was found in the retentate, respectively. The TS80-membrane shows a very high retention for LNT. In an experiment using the same set-up starting from the provision of the fermentation broth to a nanofiltration with the permeate of the ultrafiltration as second membrane filtration, but using a fermentation broth from bacterial strain producing the neutral HMO 2′-Fucosyllactose, the TS80 membranes showed retention values of >99% for the smaller 2′-FL molecule. However, TS80 also showed a strong retention of phosphoric acid as well.

[0245] The hollow fibre dNF 40 (not shown in the table above) was tested for 6′SL only and showed a retention of this HMO of 99.1 % while only 62.6 % of phosphoric acids was retained in this test.

Spiral Wound Elements

[0246] Spiral wound elements of nanofiltration membranes allow for a better scalability to large scale processes than for example experiments with flat-sheet membranes.

[0247] Experiments ran in crossflow set-up with spiral-wound elements, see Table 2. The first two experiments termed 002 and 003 used the UA60 membrane for concentrating with a CF of 10 and in case of experiment 003 for diafiltration with a DF of 2.9. Washing indicates that de-ionized water was added to the retentate continuously in the same amounts as permeate was removed. Experiments 006 and 007, were done similar to experiments 002 and 003 but to check the performance of the membrane at a lower concentration factor.

TABLE-US-00004 Overview of the LNT-experiments with spiral-wound elements Experiment Membrane Goal Process variant 002 UA60 Concentrating UF-permeate by CF = 10.0 NF before IEX 003 UA60 Concentrating UF-permeate by CF = 10.9, followed by washing with DF = 2.9 NF before IEX 006 UA60 Concentrating UF-permeate by CF = 7.6 NF 007 UA60 Concentrating UF-permeate by CF = 7.6, followed by washing with DF = 3.0 NF

[0248] Table 3 gives an overview of the purification by the nanofiltration steps. The purification is given in terms of product retentions for LNT, lactose and phosphoric acid (HPLC-analysis), since these parameters can be scaled for other concentration or diafiltration factors.

TABLE-US-00005 Overview of the membrane retentions and changes in conductivity Experiment.sup.1 (...) Membrane R.sub.LNT (%).sup.1 R.sub.lactose (%).sup.1 R.sub.H3PO4 (%).sup.1 Conductivity.sup.2 (mS/cm) 002 UA60 R: 92.4 R: 75.4 R: 19.6 6.4 .fwdarw. 8.02 P: 99.3 P: 80.0 P: 24.6 003-CF UA60 R: n/a P: 99.3 R: n/a P: 75.6 R: n/a P: 17.4 6.4 .fwdarw. 3.77 003-DF P: n/a n/a n/a 006 UA60 R: 86.1 P: 99.8 R: 87.6 P: 94.8 R: 8.0 P: 26.4 n/a 007-CF UA60 R: n/a P: 98.5 R: n/a P: 76.9 R: n/a P: 23.9 n/a 007-DF P: n/a P: n/a P: n/a .sup.1 R indicates calculation over the retentate, P calculation over the permeate .sup.2 Feed .fwdarw. Final concentrate (i.e. after diafiltration, when applicable)

[0249] As can be seen in Table 3, the retention of the UA60 membranes for LNT is generally high (>98.5% is measured for all data, based on the permeate, in many cases >99% was measured). Simultaneously, lactose retentions of 75-95% were recorded, washing led to lower lactose retention. The exact value varied depending on the conditions applied in this test. For the phosphoric acid, very low retentions in the range of 15-25% were measured. These data correspond to the measurement in the test cells, where similar values were recorded.

[0250] An overview of yields in the final retentates of the experiments in comparison with the feed for these experiments was as follows:

[0251] The experiments showed that with the UA60 membrane the yields of LNT in the retentate was good. Lactose yields in the retentate were nearly as high, but the phosphate was largely eliminated with the permeate. If concentration and diafiltration mode was used, the overall yield of LNT was good as well, but in contrast lactose yield was much lower. Hence, by choosing the set-up of the nanofiltration one can steer whether LNT and lactose are both retained, or if the HMO is preferably retained and lactose is reduced in comparison to the LNT. Phosphate was even better removed in this type of nanofiltration and only very small amounts of it were found in the retentates.

[0252] Experiments 002 and 003 were selected for demineralization after the nanofiltration step.

Test With Flat-sheet Membranes for 6′-SL

[0253] Three test cell experiments were performed with UF permeates of different pHs. All experiments ran a diafiltration with DF = 3 followed by a concentration step with CF = 10. This sequence is not fully optimized but should show the potential of removing salts and potentially smaller molecules using nanofiltration. Higher removal rates can possibly be reached when higher diafiltration factors are applied.

[0254] Table 4 presents an overview of the results. Independent of the feed pH, all experiments succeeded in the removal of acetic acid to below the detection limit. The phosphate levels relative to 6SL were reduced dramatically in the retentate at all measured pH values, and at pH5.56 there was an absolute reduction of phosphates in the retentate of significance as well.

[0255] One of the most interesting removals was the removal of lactose, since this could significantly ease the downstream crystallization or SMB step. In the experiments at pHs of 4.4 and 5.56, about half of the lactose was removed through the current way of running the process. Using a higher diafiltration factor, it can be expected that some more lactose could be removed. At a pH of 6.25, a surprisingly high amount of lactose was removed. Here, after the diafiltration, the ratio of lactose to 6′-SL was 0.30, turning to only 0.16 after the concentration step, a strong removal of the lactose. Even more, if a higher diafiltration factor would be implemented, an even lower amount of lactose could be obtained. Thus, the process of S10 to S22 can - if desired -be used to purify HMOs while reducing the amounts of lactose present due to its role in fermentation.

TABLE-US-00006 Overview of 3 test cell experiments performed with UF permeates with different pHs. All experiments ran a diafiltration with DF = 3 followed by a concentration step with CF = 10 pH 4.4 pH 5.56 pH 6.25 Feed Retentate Feed Retentate Feed Retentate Cation analysis ppm ppm ppm ppm ppm ppm NH.sub.4.sup.+ 290 270 270 810 260 990 Ca.sup.2+ <3 <9 <3 <9 <3 <9 Fe.sup.2+ <3 9 <3 9 <3 15 K.sup.+ 220 210 200 285 180 525 Mg.sup.2+ 6 39 <3 9 <3 9 Na.sup.+ 170 135 155 180 145 285 Anion analysis g/100 g g/100 g g/100 g g/100 g g/100 g g/100 g Cl.sup.- <0.001 <0.003 <0.001 <0.003 <0.001 <0.003 SO.sub.4.sup.2- 0.001 <0.003 <0.001 0.003 <0.001 0.003 H.sub.2PO.sub.4.sup.- / HPO.sub.4.sup.2- 0.20 0.18 0.082 0.018 0.025 n/a HPLC analysis g/L g/L g/L g/L g/L g/L NANA 0.15 0.87 0.05 0.79 0.08 0.87 6′-SL 5.2 45.16 4.4 35.69 4.94 46.66 Lactose 4.85 25.32 4 15.22 5.14 7.31 Phosphoric acid 2.12 0.27 4.84 0.23 0.3 0.08 Acetic acid 0.43 -.sup.1 0.67 -.sup.1 0.5 -.sup.1 .sup.1 - not detected

[0256] Crossflow nanofiltration experiments using a fermentation broth comprising 6′SL and the solution prepared by steps S10 to S18 thereof using a nanofiltration with a CF of up to 12.6 at a TMP of 28-30 bar and subsequent DF of 2.25, again at a TMP of 28-30 bar were performed.

[0257] The results demonstrated that in the nanofiltration of step S22 with a concentration and subsequent a diafiltration step, concentrations of 168 g/l 6′SL were achieved, while ions like chloride, sulphate, monovalent phosphoric acid and phosphate were all below 0.002 wt% in the final retentate. This demonstrates the potential of the method including steps S10 to S22 as an improved process that will allow for purification and concentration of 6′SL and other sialylated HMOS as well as other HMOs without the need for a demineralization step any longer. Lactose can be also retained by this process if desired - in the final retentate the lactose level was around half of the 6′Sl level in g/l.

[0258] For Experiments A to D in the following Table B, fermentation broths containing inter alia 2-fucosyl lactose (2-FL) were used. First, the biomass was removed from the broths which were, then, set to a given pH and, in case of Experiment A and D, treated with active carbon (AC). Subsequently, the broths were subjected to concentration using the nanofiltration membrane AMS AS3014 (available from AMS Technologies Ltd., Israel) having a cut-off of 0.4 kDa. In case of Experiment C, the concentrate obtained in Experiment A was diafiltrated using said membrane.

TABLE-US-00007 Nanofiltration Experiments Experiment Feed Average flux at TMP = 30 bar and 30° C. A AC-treated broth 10.3 kg/(m.sup.2h) Set to pH 5.05 using HCl B Broth (no AC) 4.8 kg/(m.sup.2h) Set to pH 5.03 C Concentrate of Exp A 3.7 kg/(m.sup.2h) D AC-treated broth 5.3 kg/(m.sup.2h) Set to pH 8.88 using NaOH

[0259] In the feeds used for and the concentrates resulting from nanofiltration, the concentrations of several components were analyzed via HPLC as can be seen in the following Table C:

TABLE-US-00008 Analysis Results. Values are concentrations in g/l Exp. A Exp. C Exp. B Exp. D Feed Concentrate After diafiltration concentrate of Exp. A Feed Concentrate Feed Concentrate 2-Fucosyl lactose (2-FL) 19.6 123.6 99.9 25.0 160.3 17.4 163.1 Lactose 12.9 87.4 64.1 12.3 84.7 12.1 84.9 Phosphoric acid 4.0 n.a. 2.3 3.5 n.a. 3.0 13.8 Pyruvic acid 0.3 1.2 0.2 0.2 0.8 0.2 0.1 Fucose 0.1 0.7 0.2 0.1 0.4 n.a. 0.1 Succinic acid 0.3 0.9 n.a. 0.5 1.1 0.4 0.8 Lactic acid 0.3 1.5 0.5 0.6 1.2 0.4 1.4 Formic acid 0.8 0.8 0.2 1.4 1.5 0.5 0.4 Acetic acid 2.6 1.7 n.a. 4.0 3.9 2.5 1.3 Ratio 2-FL / Phosphoric acid 4.9 n.a. 44.4 7.1 n.a. 5.8 11.8 Ratio 2-FL / Formic acid 25.0 162.8 463.8 18.1 108.1 33.1 465.1 Ratio 2-FL / Acetic acid 7.6 74.5 n.a. 6.3 41.4 6.9 122.8 n.a. = not available

[0260] As can clearly be seen from the increasing ratio of 2-FL to phosphoric, formic or acetic acid, respectively, after concentration or diafiltration, nanofiltration results in a removal of the according deprotonated acids, confirming effective demineralization of the fermentation broths.

Ion Exchange Experiments

[0261] Demineralization experiments in laboratory columns (inner diameter 20 mm) For initial testing of the demineralization procedures, two double-jacketed glass columns (Inner diameter 20 mm, height 1000 mm) were set up and filled with ca. 0.28 L Dowex Monosphere 88 H and 0.24 L Dowex Monosphere 77, respectively.

[0262] The demineralization experiments were carried out using the conditions shown in Table 5, with the cation exchange done before the anion exchange. The pre-rinsing, loading, product displacement and post-rinsing steps were carried out with the columns connected in series, first the cation exchanger and then the anion exchanger. The columns and the vessels for the feed and effluent solutions were cooled to ca. 10° C. The regeneration and the rinsing afterwards were carried out in countercurrent mode for each column separately. The resins were regenerated before first use to ensure that they were in a completely regenerated state.

[0263] During the experiments, fractions were collected and analysed to monitor the process.

TABLE-US-00009 Conditions for demineralization experiments in the 20 mm laboratory columns Step Medium Direction Amount [BV] Flow rate [BV/h] Pre-rinsing Sterile DI water .down-triangle-solid. Until conductivity <10 .Math.S/cm ca. 0.2 (relative to CEX) Loading Feed .down-triangle-solid. See experiment 0.8 (relative to CEX) Product displacement DI water .down-triangle-solid. 4 (relative to CEX) 0.8 (relative to CEX) Post-rinsing DI water .down-triangle-solid. 3 (relative to CEX) ca. 1.6 (relative to CEX) Regeneration CEX: H2SO4 5 wt% .triangle-solid. CIEX: 8 3 AEX: NaOH 4 wt% AIEX: 10 Rinsing DI water .triangle-solid. ca. 20 ca. 1.6

Demineralization of LNT Samples

[0264] Dowex Monosphere 88 H and Dowex Monosphere 77 were chosen as these have been used for oligosaccharides before. Their properties are shown in Table 6. The supplier has recently renamed these products and they are now being sold as AmberLite FPC88 UPS H and AmberLite FPA77 UPS, respectively.

TABLE-US-00010 Properties of the ion exchangers used for demineralization Cation exchanger Anion exchanger Supplier Dupont Dupont Name Dowex Monosphere 88 H (AmberLite FPC88 UPS H) Dowex Monosphere 77 (AmberLite FPA77 UPS) Type Strong acid cation Weak base anion Matrix Styrene-DVB, macroporous Styrene-DVB, macroporous Functional group Sulfonate Tertiary amine with some quarternary groups Delivery form H free base Total exchange capacity min 1.7 eq/L min 1.7 eq/L, min 1.5 eq/L as weak base Water content 50-56 % 40-50% Particle size distribution Median 500-600 .Math.m with 95% within 400-720 .Math.m Median diameter 475-600 .Math.m Swelling Na->H 5% Swelling FB->HCl 22% Whole beads min 95% min 95% Particle density 1.2 1.04 Shipping weight 770 g/L 640 g/L

[0265] Samples of UF permeate (i.e. permeates of the second membrane filtration step S18) and of NF retentates obtained from treating the UF permeate with two different methods of nanofiltration were (step S22) were analysed.

[0266] The final retentates from the experiments 002 and 003 (see above) where used for ion exchange experiments.

[0267] The results of the demineralization experiments are summarized in Table 7. As can be seen in the table, the amount of salt relative to the product was reduced considerably in nanofiltration, and with additional washing it could be reduced even further.

TABLE-US-00011 Properties of the UF permeate and NF retentate with and without additional washing Control sample (UF permeate) experiment 002 (NF retentate) experiment 003 (NF retentate, washed) Conductivity (mS/cm) 6.8 8.1 3.9 APHA 430 2760 2630 pH 3.7 3.8 4.0 g/L g/L Factor g/L Factor Lacto-N-tetraose 12.88 106.52 8.27 120.73 9.37 Lacto-N-triose 0.44 3.36 7.62 3.50 7.95 Lactose 0.96 5.25 5.46 3.79 3.94 ppm eq/kg ppm eq/kg ppm eq/kg NH.sub.4.sup.+ 140 0.008 130 0.007 10 0.001 Ca.sup.2+ 7 0.000 27 0.001 14 0.001 Fe.sup.2+ 4 0.000 14 0.001 10 0.000 K.sup.+ 1300 0.033 1300 0.033 75 0.002 Mg.sup.2+ 34 0.003 155 0.013 105 0.009 Na.sup.+ 290 0.013 310 0.013 20 0.001 SUM Cations 1775 0.057 1936 0.069 234 0.013 % eq/kg % eq/kg % eq/kg Cl.sup.- <0.001 0.000 0.001 0.000 <0.001 0.000 SO.sub.4.sup.2- 0.26 0.054 0.62 0.129 0.33 0069 H.sub.2PO.sub.4.sup.- 0.15 0.016 0.21 0.022 0.05 0.005 SUM Anions 0.41 0.070 0.83 0.151 0.38 0.074 g salt/g LNT 0.46 0.10 0.03 Cation/anion charge ratio 0.81 0.45 0.18

[0268] A control sample (no nanofiltration treatment after the ultrafiltration as second membrane filtration) was used as feed. The feed was loaded onto the ion exchange columns at a flow rate of 0.8 BV/h relative to the cation exchanger until a conductivity of 50 .Math.S/cm was reached in the effluent. This took place after approximately 18 BV, however a breakthrough in colour was observed already after 16 BV. The conductivity of the effluent throughout the process was approximately 15 .Math.S/cm indicating a small leakage of ions. Accordingly, the pH of the effluent was slightly alkaline since the salt leakage causes a small amount of hydroxide ions to be displaced from the anion exchanger. The reason for this leakage is not clear but may be that other components in the mixture can form complexes with some of the ions. The anion exchanger was observed to swell by approximately 5% during the loading and the cation exchange to shrink by a few percent.

[0269] The washed NF retentate (from experiment 003 above) was demineralized at a flow rate of 0.8 BV/h using the same columns as for the control samples after their regeneration. In this experiment, a predetermined amount of feed of the solution from experiment 003 was passed through the columns, 10 BV.

[0270] The elution of LNT was completed earlier than for the control sample.

[0271] As for the control sample, the breakthrough in colour came somewhat earlier, after 6.2 BV, and like previously with the control sample, also here a small leakage of ions was observed during the run. Also, in this case the anion exchanger was observed to swell by approximately 5% during the loading and the cation exchange to shrink by a few percent.

[0272] The colourless fractions of each demineralization experiment were combined per experiment. The combined fractions of the control sample and the combined fractions of the nanofiltrated sample from experiment 003 were then analysed. The results (see Table 8) showed that the pre-treatment with nanofiltration and washing resulted in lower residual levels of ions relative to the fine chemical LNT.

TABLE-US-00012 Analysis of demineralized products Colourless Fractions control sample Colourless Fractions washed NF retentate exp 003 Conductivity 16 .Math.S/cm 15 .Math.S/cm APHA 0.4 1.1 pH 8.2 8.1 Lacto-N-tetraose 12.75 g/L 92.22 g/L Lacto-N-triose 0.35 g/L 2.50 g/L Lactose 0.85 g/L 2.66 g/L NH.sub.4.sup.+ <10 ppm <10 ppm Ca.sup.2+ <1 ppm <1 ppm Fe.sup.2+ <1 ppm <1 ppm K.sup.+ 5 ppm 6 ppm Mg.sup.2+ <1 ppm <1 ppm Na.sup.+ 1 ppm <1 ppm Chloride <0.001% <0.001% Sulfate <0.001% <0.001% Phosphate <0.001% <0.001%

[0273] It was observed that the throughput was much higher when the NF retentate with reduced amounts of salts was used in the demineralization step, 167% more of LNT per cycle with a cycle time that was reduced by almost 60%, i.e. a total improvement by about 350%. Thus, when the step of nanofiltration including the washing of the samples was used, the throughput of the desired fine chemical in the demineralization step was improved by a factor of about 4.5 compared to the throughput of the demineralisation of the control sample that had not undergone any nanofiltration after the ultrafiltration as second membrane filtration. As shown in the experiments, the efficiency of the demineralization was also not compromised, and less residual ions relative to the fine chemical LNT were obtained when the washed NF retentate was demineralized compared to the control sample without NF treatment.

[0274] It was observed that the concentration step alone (experiment 002) does not change the ion concentrations at large; however, since the LNT concentration was increased by a factor 8.3, the relative concentration of ions to LNT was reduced significantly (see Table 7). For experiment 003, the extra diafiltration step results in a strong decrease in ion concentration, especially in the concentration of the monovalent K.sup.+ and H.sub.2PO.sub.4.sup.-. The divalent SO.sub.4.sup.2- ion is reduced to a lesser extent and the divalent Mg.sup.2+ is only reduced to a small extent.

[0275] As demonstrated nanofiltration before ion exchange can be used to remove nearly all the phosphoric acid and parts of the lactose from the solution or remove ions preferably but not the lactose or LNT. Furthermore, the broth can be concentrated by at least a factor 10, probably more, judging from the flow rates at the end of the concentration step. The currently employed concentration factors of 10 followed by a diafiltration factor of 3 allow for a removal of -65% of the lactose and >95% of the phosphoric acid. Using higher diafiltration factors, higher removal rates of lactose are achievable for the person skilled in the art.

Summary of the Results of the Demineralization Experiments

[0276] Carrying out NF before demineralization has been found to have several advantages: [0277] Considerably higher throughput during ion exchange, up to 350% higher [0278] If desired, lactose could be partially removed during the nanofiltration which demonstrated also improved purification of the product LNT [0279] Less residual salt after demineralization relative to the product were found when nanofiltration was employed

[0280] Overall, the inventive method to combine decolourization, biomass removal, purification by nanofiltration and ion exchange proved to be very efficient on resources and equipment while delivering fast recovery of a number of fine chemical products such as different HMO types.

CITED LITERATURE

[0281] WO 2015/032412 [0282] EP 2 379 708 [0283] CN 100 549 019 & CN 101 003 823 [0284] WO 2017/205705 [0285] EP 2 896 628 [0286] US 9 944 965 [0287] WO 2019/003133 [0288] WO2015106943