NANOFILTRATION ASSISTED METHODS OF CHEMICAL SYNTHESIS

Abstract

A method and apparatus for conducting chemical reactions under controlled conditions, where one compound is in stochiometric excess over the other, using membrane filtration.

Claims

1-15. (canceled)

16. A method for reacting a first compound and a second compound, thereby forming a third compound, the method comprising the following steps: a) providing a first solution S1 comprising the first compound and a second solution S2 comprising the second compound; b) providing, inside a reaction vessel, a solution S3 comprising the second compound in a liquid reaction medium, which comprises an organic solvent; and c) concomitantly adding the first solution S1 and the second solution S2 over an extended period of time into the solution S3 while mixing the contents of the reaction vessel, thereby forming a reaction mixture comprising the third compound; wherein: d) the second compound is in excess over the first compound inside the reaction vessel during step c); and e) during step c), the contents of the reaction vessel are subjected to membrane filtration so as to essentially retain the first compound, the second compound, and the third compound inside the reaction vessel.

17. The method of claim 16, wherein the excess of the second compound over the first compound inside the reaction vessel during step c) is essentially constant over time.

18. The method of claim 16, further wherein a liquid, which is miscible with the reaction mixture, is added into the reaction vessel while subjecting the contents of the reaction vessel to membrane filtration.

19. The method of claim 16, further wherein the volume of liquid inside the reaction vessel is kept essentially constant over the duration of step c).

20. The method of claim 18, wherein the contents of the reaction vessel are subjected to diafiltration after completion of step c).

21. The method of claim 16, wherein the third compound is metastable.

22. The method of claim 21, wherein the metastable third compound undergoes a spontaneous cyclization reaction, thereby forming a cyclic fourth compound.

23. The method of claim 16, further comprising a step of incubating the third compound with a fifth compound, thereby forming a sixth compound.

24. The method of claim 16, wherein at least one of the first and second compounds comprises a carboxyl group or a hydroxyl group.

25. The method of claim 16, wherein the equivalent concentration of the second compound inside the reaction vessel is higher than the equivalent concentration of the first compound inside the reaction vessel by a factor of at least 1.5.

26. The method of claim 16, wherein the contents of the reaction vessel are conducted to a cross-flow membrane filtration unit and a retentate from the cross-flow membrane filtration unit is conducted back into the reaction vessel.

27. The method of claim 16, wherein a rejection rate of a membrane of the membrane filtration employed is at least 90% for each of compound 1, compound 2, and compound 3.

28. The method of claim 22, wherein the fourth compound is a cyclic peptide.

29. The method of claim 16, wherein the first compound is a peptide and the second compound is a carboxyl group activating reagent.

30. The method of claim 16, further wherein a cyclic peptide or peptide carrier conjugate produced is subjected to a step of purification or isolation selected from the group consisting of: diafiltration, a concentration step, crystallization, lyophilization, precipitation, dialysis, chromatography, and reversed phase high performance liquid chromatography.

Description

DETAILED DESCRIPTION OF THE FIGURES

[0143] FIG. 1: Analytical chromatogram of the sample from example 1, experiment no 1. The peaks quantitated as monocyclic peptide (arrows), cyclic dimer (star) and other side products (dots) were considered for the analysis of the cyclisation reaction performance.

[0144] FIG. 2: Theoretical concentration profiles of reactants and products inside the reaction vessel during the cyclization of a model peptide, as detailed in example 9. Panel A: case study 9.1 (addition of peptide only into excess of coupling reagent inside the reaction vessel; no membrane filtration). Panel B: case study 9.2 (addition of peptide only into excess of coupling reagent inside the reaction vessel; concomitant membrane filtration). Panel C: case study 9.3 (addition of peptide and coupling reagent into excess of coupling reagent inside the reaction vessel; concomitant membrane filtration; high, constant excess of coupling reagent). Panel D: case study 9.4 (addition of peptide and coupling reagent into excess of coupling reagent inside the reaction vessel; concomitant membrane filtration; lower, constant excess of coupling reagent).

[0145] FIG. 3: Example of instrument setup with two membrane filtration units arranged in parallel inside the same retentate loop. A reaction vessel 1 comprising two liquid inlets 2,3 is connected via liquid lines to two upstream feed tanks 4,5. A liquid line connected to an outlet 11 of the reaction vessel 1 branches to serve two membrane filtration units 6, 7. The two liquid lines emerging from the retentate side 8 of each of the membrane filtration units 6,7 merge to one single line recycling retentate back to the reaction vessel 1. Likewise, the two liquid lines emerging from the permeate side 9 of the two membrane filtration units 6,7 merge to one single drain line. Multiway valves positioned at the nodes of the retentate loop (not shown) may be used to channel the liquid to either of the two membrane filtration units.

[0146] FIG. 4: Example of instrument setup with two membrane filtration units arranged in independent retentate loops. A reaction vessel 1 comprising two liquid inlets 2,3 is connected via liquid lines to two upstream feed tanks 4,5. Each of two membrane filtration units 6,7 is connected to the reaction vessel 1 via liquid lines directing liquid from a dedicated outlet 11, 17 of the reaction vessel to the membrane filtration unit and from the retentate side 8 of the membrane filtration unit back to the reaction vessel 1.

[0147] FIG. 5: Example of instrument setup with two membrane filtration units arranged in parallel inside the same retentate loop. A reaction vessel 1 comprising two liquid inlets 2,3 is connected via liquid lines to two upstream feed tanks 4,5. Pumps 12 drive and control liquid flow from the feed tanks 4,5 to the reaction vessel 1. A liquid line connected to an outlet 11 of the reaction vessel 1 branches to serve two membrane filtration units 6, 7. A pump 12 driving liquid circulation through the retentate loop is integrated into said line. The two liquid lines emerging from the retentate side 8 of each of the membrane filtration units 6,7 merge to one single line recycling retentate back to the reaction vessel 1. Likewise, the two liquid lines emerging from the permeate side 9 of the two membrane filtration units 6,7 merge to one single drain line leading to a container 18. A pressure control valve 13 and pressure sensors 14 are integrated into the retentate loop. Multiway valves positioned at the nodes of the retentate loop (not shown) may be used to channel the liquid to either of the two membrane filtration units.

[0148] FIG. 6: Example of instrument setup with two membrane filtration units arranged in parallel inside the same retentate loop. A reaction vessel 1 comprising two liquid inlets 2,3 is connected via liquid lines to two upstream feed tanks 4,5. A third upstream feed tank 10 is connected to the liquid inlet 2 via the liquid line connecting feed tank 5 with liquid inlet 2. Pumps 12 drive and control liquid flow from the feed tanks 4,5,10 to the reaction vessel 1. A liquid line connected to an outlet 11 of the reaction vessel 1 branches to serve two membrane filtration units 6, 7. A pump 12 driving liquid circulation through the retentate loop is integrated into said line. The two liquid lines emerging from the retentate side 8 of each of the membrane filtration units 6,7 merge to one single line recycling retentate back to the reaction vessel 1. Likewise, the two liquid lines emerging from the permeate side 9 of the two membrane filtration units 6,7 merge to one single drain line leading to a container 18. A pressure control valve 13 and pressure sensors 14 are integrated into the retentate loop. Multiway valves positioned at the nodes of the retentate loop (not shown) may be used to channel the liquid to either of the two membrane filtration units. A level sensor 16 determines the volume of liquid inside the reaction vessel 1. A central control unit 15 receives input from at least the pumps 12, the pressure sensors 14, and the level probe. The input is used to control the actions of the pumps 12 and the pressure valve 13.

[0149]

TABLE-US-00001 List of Reference signs 1 reaction vessel 2 liquid inlet 3 liquid inlet 4 feed tank 5 feed tank 6 membrane filtration unit 7 membrane filtration unit 8 retentate side of membrane 9 permeate side of membrane 10 feed tank 11 liquid outlet 12 pump 13 pressure control valve 14 pressure sensor 15 central control unit 16 level sensor 17 liquid outlet 19 container

EXAMPLES

General Procedure

[0150] Peptide 1 [H-Phg-D-Trp(Boc)-Lys(Boc)-Tyr(Bzl)-Phe-Hyp(2-Boc-aminoethyl-carbamoyl)-OH] was synthesized by SPPS using standard methods and Fmoc-amino acid derivatives. Peptide cleavage was performed using 30/70 HFIP/DCM. The raw peptide was precipitated from the cleavage cocktail using diisopropyl ether as anti-solvent. No further purification steps were applied.

[0151] The amounts of linear and cyclic peptide, as well as oligomers in the reaction mixture were determined by means of reversed-phase high-performance liquid chromatography (RP-HPLC). The samples were diluted to a concentration of approximately 1 g/l with respect to peptide and eluted using a gradient of 0.05% trifluoroacetic acid (TFA) in 1/99 v/v acetonitrile/water and 100% acetonitrile. The cyclisation performance was determined based on the relative amount of side products with respect to monocyclic peptide. LC-MS analysis was carried out to identify the peaks corresponding to peptidic products, such as monocyclic peptide and cyclic dimer. All the other peaks in the chromatogram of the reaction mixture, which did not appear in the pure solvent (baseline) or in the linear peptide samples, were assumed to correspond to other side products, including oligomers.

Example 1: Cyclisation of Peptide

[0152] Raw peptide 1 produced according to the general procedure was dissolved in DMF to a final peptide concentration of 125 g/l (91.6 mM). A round-bottomed flask equipped with a magnetic stirrer was charged with 20 ml of DMF solution containing either 119.4 mM or 59.7 mM PyOxim and a constant concentration of N,N-Diisopropylethylamine (DIPEA), proportional to the concentration of linear peptide used for the cyclisation. Subsequently, 10 ml of the peptide solution was gradually added at a constant flow rate of 2 ml/min using a liquid dosing pump. Samples were taken at the end of peptide addition and subjected to HPLC analysis. Experimental conditions and results are reported in Table 1.

TABLE-US-00002 TABLE 1 V(peptide Cyclic Exp. c(PyOxim) V(DIPEA) sol. added) monomer Dimer Other side No. [g/l (mM)] [ml] [ml] [%] [%] products [%] 1 62.95 (119.4) 0.625 10 93.38 0.85 5.74 2 31.48 (59.7) 0.625 10 91.89 0.96 7.16

Example 2: Cyclisation of Peptide with Decreasing Excess of Coupling Reagent

[0153] The results from example 1 point to an advantageous effect of an excess of coupling reagent on the reaction performance. In the reaction scheme applied in example 1, the concentration of coupling reagent in the reaction mixture is expected to drop over time due to a) the dilution of the coupling reagent with the addition of new starting material in solution and b) its consumption during the coupling reaction. In order to assess the effect of this drop on the quality of the final product, experiments were carried out to compare the reaction mixture at the beginning of the reaction under the same reaction conditions as in example 1 (during the first 0.8 min of peptide addition: experiments 3 and 5 in Table 2) with the reaction mixture under the conditions at the end of the reaction as in example 1 (during the last 0.8 min of peptide addition: experiments 4 and 6 in Table 2). The experimental setup was essentially as in example 1. Experimental conditions and results are reported in Table 2.

TABLE-US-00003 TABLE 2 V(peptide Cyclic Exp. c(PyOxim) V(DIPEA) sol. added) monomer Dimer Other side No. [g/l (mM)] [ml] [ml] [%] [%] products [%] 3 62.95 (119.4) 0.1 1.6 98.9 0.53 0.57 4 36.50 (69.2) 0.1 1.6 98.51 0.7 0.79 5 31.48 (59.7) 0.1 1.6 98.19 0.7 1.09 6 5.04 (9.6) 0.1 1.6 92.95 1.24 5.79

[0154] The results from Table 2 demonstrate that the formation of by-products increases with the decrease in concentration of coupling reagent (i.e. with the reaction time, compare experiment 3 with 4 and experiment 5 with 6). This effect appears particularly pronounced if starting with low concentrations of coupling reagent (compare experiment 4 and 6).

Example 3: Cyclisation of Peptide with Constant Concentration of Coupling Reagent

[0155] To evaluate whether the formation of by-products could be suppressed by keeping the concentration of coupling reagent constant over time, experiments 7 and 8 (see Table 3) were carried out to be compared with experiments 1 and 2, respectively (see Table 1). In experiments 7 and 8, 20 ml of a highly concentrated solution of the coupling reagent PyOxim in DMF was added simultaneously to peptide addition, using a second KNF SIMDOS 10 liquid dosing pump at a constant flow rate of 4 ml/min. The remaining experimental conditions were as in experiments 1 and 2, respectively. The concentration of the coupling reagent solution was selected so as to compensate for the drop in the concentration of coupling reagent by consumption and dilution. Experimental conditions and results are reported in Table 3.

TABLE-US-00004 TABLE 3 c(PyOxim) Other V(peptide in solution Cyclic side Exp. c(PyOxim) V(DIPEA) sol. added) added monomer Dimer products No. [g/l (mM)] [ml] [ml] [g/l (mM)] [%] [%] [%] 7 62.95 (119.4) 0.625 10 125.90 (238.7) 95.99 0.68 3.30 8 31.48 (59.7) 0.625 10 78.69 (149.2) 94.98 0.55 4.49

[0156] The results from Table 3 demonstrate that keeping the concentration of coupling reagent constant at the expense of having to increase the reaction volume helps to control the amount of dimers formed, but is not satisfactory with respect to the formation of other side products.

Example 4: Cyclisation of Peptide with Increasing Reaction Volume

[0157] In order to assess the effect of increasing the reaction volume on the reaction performance, experiments 9 and 10 (see Table 4) were carried out to be compared with experiments 7 and 8, respectively (see Table 3). In experiments 9 and 10, instead of 20 ml of PyOxim in DMF, 20 ml of the solvent DMF only were added. Experimental conditions and results are reported in Table 4.

TABLE-US-00005 TABLE 4 c(PyOxim) Other V V(peptide in solution Cyclic side Exp. c(PyOxim) (DIPEA) sol. added) added monomer Dimer products No. [g/l (mM)] [ml] [ml] [g/l (mM)] [%] [%] [%] 9 62.95 (119.4) 0.625 10 0 (0) 93.80 0.85 5.34 10 31.48 (59.7) 0.625 10 0 (0) 92.11 0.84 7.03

[0158] The results from Table 4 confirm that an increase in reaction volume is deleterious to the yield and purity of the desired cyclic monomer.

Example 5: Retention of Coupling Reagent by Nanofiltration

[0159] 2.5 l of PyOxim solution in DMF at a concentration of 1 g/l (1.9 mM) were loaded into a stirred reaction vessel, which also functioned as the feed tank of a nanofiltration system. Said system comprised a retentate loop with pressure sensors and a pressure control valve, a recirculation pump, and a nanofiltration unit with a flat-sheet polymeric membrane (filtration area of 54 cm.sup.2). Commercially available membranes made from modified polyimide with a molecular weight cut-off of 150 g/mol or 200 g/mol were used during this experiment.

[0160] In order to measure the retention of PyOxim, the liquid was circulated within the nanofiltration system with a flow rate of 1.2 I/min. The pressure in the retentate loop was set at 20 bar. Samples of permeate and retentate were taken at various time points and the concentration of PyOxim was determined by means of UV-Vis spectroscopy. The results are presented in Table 5.

TABLE-US-00006 TABLE 5 Time [min] 50 120 180 250 330 390 450 540 PyOxim retention 97.3 98.7 99.0 99.8 99.8 99.8 99.7 99.6 Duramem 150 [%] PyOxim retention 90.4 95.1 95.1 98.8 99.1 99.2 98.9 99.0 Duramem 200 [%]

[0161] As is illustrated in Table 5, the apparatus according to the present invention allows retaining the coupling reagent within the reaction vessel.

Example 6: Cyclisation of Peptide with Increasing Reaction Volume

[0162] Raw peptide 1 will be dissolved in 1.25 l of DMF at a concentration of 80 g/l (58.7 mM). In a separate container, 2.5 l of a 40.3 g/l (76.4 mM) solution of coupling reagent (PyOxim) will be prepared. The first solution will be transferred to a first feed tank, whereas the other one will be placed directly in the stirred reaction vessel. The first feed tank will be connected to the stirred reaction vessel. The reaction will be started by gradually adding the solution from the first feed tank into the reaction vessel. The liquid present in the reaction vessel will be circulated within the nanofiltration system without applying any pressure. Samples of reaction mixture will be taken at regular time points and then analyzed by analytical RP-HPLC. After the completion of the reaction, the nanofiltration system will be depressurized and drained to recover the retentate, to quantify final purity and yield.

Example 7: Nanofiltration Assisted Peptide Cyclization with Constant Reaction Volume

[0163] Raw peptide 1 will be dissolved in 1.25 l of DMF at a concentration of 80 g/l (58.7 M). In a separate container, 2.5 l of a 40.3 g/l (76.4 mM) solution of coupling reagent (PyOxim) will be prepared. The first solution will be transferred to a first feed tank, whereas the other one will be placed directly in the stirred reaction vessel. The first feed tank will be connected to the stirred reaction vessel. The reaction will be started by gradually adding the solution from the first feed tank to the reaction vessel. The volumetric flow of peptide solution will be adjusted so that it is equal to the volumetric flow rate of permeate. The liquid present in the reaction vessel will be circulated within the nanofiltration system at 20 bar before starting the reaction and during the whole time of the reagent addition from the storage vessel. Samples of permeate and retentate will be taken at regular time points and then analyzed by analytical RP-HPLC. After the completion of the reaction, the nanofiltration system will be depressurized and drained to recover the retentate.

Example 8: Nanofiltration Assisted Peptide Cyclization with Constant Volume and Constant Addition of Coupling Reagent

[0164] Raw peptide 1 will be dissolved in 0.75 l of DMF at a concentration of 133.33 g/l (97.7 mM). In a separate container, 0.5 l of a 100.7 g/l (191.0 mM) solution of coupling reagent (PyOxim) will be prepared. These solutions will be transferred to a first and second feed tank, respectively. The reaction vessel is filled with 2.5 l of 40.3 g/l (76.4 mM) coupling reagent solution. The feed tanks will be connected via locally separated inlets to the stirred reaction vessel. The reaction will be started by gradually adding the solution from the feed tanks to the reaction vessel. The liquid present in the reaction vessel will be circulated within the nanofiltration system at 20 bar before starting the reaction and during the whole time of the reagent addition from the storage vessel. Samples of permeate and retentate will be taken at regular time points and then analyzed by analytical RP-HPLC. After the completion of the reaction, the nanofiltration system will be depressurized and drained to recover the retentate.

Example 9: Simulation of Concentration Profiles During the Cyclisation Reaction

[0165] The theoretical concentration profiles of reactants and products during the cyclisation reaction were simulated numerically using Matlab software. Molecular weights of reactants and products as well as general operating conditions and system parameters used for the simulations are listed in Table 6.

TABLE-US-00007 TABLE 6 Description Value Unit Molecular weight linear peptide 1364 g/mol Molecular weight coupling reagent 527 g/mol Molecular weight cyclic peptide 1346 g/mol Permeance of the membrane 3.0 l/(m.sup.2 bar h) Filtration area 0.0108 m.sup.2 Initial volume of coupling reagent 2.0 l solution in reaction vessel Cyclisation reaction rate constant 104 l/(mol h) Rejection of linear and cyclic peptides 100 %

[0166] The simulation was carried out for four different cases:

[0167] 9.1) Solution of coupling reagent is loaded into the reaction vessel and peptide solution is fed from the feed tank. During the whole process no pressure is applied. The simulated setup corresponds to the experimental setup as in Example 6.

[0168] 9.2) Solution of coupling reagent is loaded to the reaction vessel and peptide solution is fed from the feed tank. During the whole process a constant pressure is applied. The simulated setup corresponds to the experimental setup as in Example 7.

[0169] 9.3) Solution of coupling reagent is loaded to the reaction vessel. Both peptide and coupling reagent solutions are fed from separate feed tanks. During the whole process a constant pressure is applied. The simulated setup corresponds to the experimental setup as in Example 8.

[0170] 9.4) Solution of coupling reagent is loaded to the reaction vessel, with half the concentration of case 9.3. Both peptide and coupling reagent solutions are fed from the feed tanks. During the whole process a constant pressure is applied. The simulated setup corresponds to the experimental setup as in Example 8.

[0171] The simulated concentration profiles for case studies 9.1 to 9.4 are shown in FIG. 2. Clearly, in case study 9.1 the concentration profile of the cyclic peptide is approaching a plateau due to intrinsic dilution caused by continuous addition of peptide solution. Concentration of the cyclic product during reaction is possible only to a certain extent, as the overall process efficiency is limited by the self-dilution effect. On the other hand, when nanofiltration is applied (case studies 9.2-9.4), the concentration of cyclic peptide increases linearly and can be theoretically increased up to the solubility limit of the cyclic peptide. In case study 9.2 the concentration of the coupling reagent decreases monotonically, due to its consumption by the cyclisation reaction. This phenomenon is compensated by the addition of coupling reagent solution in case studies 9.3 and 9.4, where the concentration of coupling reagent remains constant. The optimal starting concentration of coupling reagent in the reaction vessel must be found for the specific case study and may be different from the optimal starting concentration for the classical batch process as in case 9.1 (case 9.4 vs 9.3). All case studies simulate instantaneous consumption of linear peptide, which determines the very low concentration of linear peptide in the reaction mixture throughout the whole reaction time. The specific parameters used for each case study are listed in Table 7.

TABLE-US-00008 TABLE 7 Case Case Case Case Parameter study 9.1 study 9.2 study 9.3 study 9.4 Peptide concentration in storage vessel [mmol/l] 9.5 9.5 13.6 13.6 Coupling reagent concentration in storage vessel [mmol/l] 0.0 0.0 31.8 31.8 Initial coupling reagent concentration in reaction vessel [mmol/l] 9.5 9.5 9.5 4.8 Initial volume of peptide solution in storage vessel [l] 1.0 1.0 0.7 0.7 Initial volume of coupling reagent solution in storage vessel [l] 0.0 0.0 0.3 0.3 Volumetric flow of peptide solution [l/h] 0.648 0.648 0.454 0.454 Volumetric flow of coupling reagent solution [l/h] 0.0 0.0 0.195 0.195 Differential pressure across the membrane [bar] 0.0 20.0 20.0 20.0

Example 10: Nanofiltration Assisted Peptide Cyclization

General Procedure

[0172] Raw peptide 1 was dissolved in DMF at variable concentration (see Tables 8-10, line 1) and 3 eq of DIPEA were added. In a separate container, a solution of coupling reagent (PyOxim) in DMF at variable concentration (see Tables 8-10, line 2) was prepared. These solutions were transferred to a first and second feed tank, respectively. The reaction vessel was filled with a solution of coupling reagent in DMF at variable concentration (see Tables 8-10 line 3). The feed tanks were connected via locally separated inlets to the stirred reaction vessel. The reaction was started by gradually adding the solution from the feed tanks to the reaction vessel. The liquid present in the reaction vessel was circulated within the nanofiltration system at variable pressure (see Tables 8-10 line 11) before starting the reaction and during the whole time of the reagent addition from the feed tank. The pressure in the system was adjusted so that the volumetric flow of the permeate was equal to the sum of volumetric flows of feed solutions 1 and 2. Samples of permeate and retentate were taken at regular time points and then analyzed by analytical RP-HPLC as described above. The sum of the areas of all significant peaks, which were not identified as cyclic monomer and had a retention time higher than the cyclic monomer, was used to quantify the amount of side products. After the completion of the reaction, i.e. at the end of reagent addition, the nanofiltration system was depressurized and drained to recover the retentate. The reaction performance was assessed based on the relative amount of peptidic side products in the retentate (Tables 8-10, line 15).

Example 10.1: Effect of Constant Reaction Volume and Continuous Addition of Coupling Reagent

[0173] In the method according to the present invention, the reaction conditions may be controlled, e.g., by keeping the reaction volume and consequently the concentration of the linear peptide in the reaction mixture constant and by continuously adding coupling reagent in order to balance its consumption by the cyclisation reaction and loss across the membrane. The effect of these measures on the amount of side products formed is shown in Table 8. If no membrane filtration was performed and the peptide solution was merely added into an excess of coupling reagent inside the reactor (experiment T1), 5.48% of side products were observed. Using membrane filtration to keep the volume of the reaction mixture constant over time (experiment T2) resulted in a reduction to 4.40% of side products, and using membrane filtration plus addition of coupling reagent (experiment T3A) gave a further reduction to 3.90% of side products. Hence an improvement of approx. 30% is observed when membrane filtration is applied and coupling reagent is topped-up so as to keep its concentration within the reaction vessel constant.

TABLE-US-00009 TABLE 8 Experiment No. No. Parameter T1 T2 T3A 1 Peptide concentration in feed tank 1 [mmol/l] 91.64 91.64 91.64 2 Coupling reagent concentration in feed tank 2 [mmol/l] 0 0 278.54 3 Initial coupling reagent concentration in reaction vessel [mmol/l] 69.62 69.62 69.62 4 Initial volume of peptide solution in feed tank 1 [l] 0.014 0.014 0.014 5 Initial volume of coupling reagent solution in feed tank 2 [l] 0.005 0.005 0.005 6 Initial volume of coupling reagent solution in reaction vessel [l] 0.050 0.050 0.050 7 Total permeate volume [l] 0 0.019 0.025 8 Volumetric flow of peptide solution [l/h] 0.027 0.027 0.027 9 Volumetric flow of coupling reagent solution [l/h] 0.009 0.009 0.009 10 Volumetric flow of permeate stream [l/h] 0 0.036 0.036 11 Differential pressure across the membrane [bar] 0 40 44 12 MWCO of Duramem membrane [g/mol] 150 150 150 13 Membrane area [m.sup.2] 0 0.0066 0.0066 14 Operating time [min] 31 31 31 15 Side products [%] 5.48 4.40 3.90

Example 10.2: Effect of Diafiltration During the Reaction

[0174] Performing membrane filtration during the course of the reaction to keep the reaction volume constant may give an inline diafiltration effect: this may result in the removal of (usually small) molecular species, which pass through the membrane filtration unit more easily than the reactants and the desired products. Such species may be introduced together with the reagents or may be formed during the reaction. To assess whether the removal of such molecular species from the reaction mixture influences product purity, variations in the total volume of liquid flux through the reaction vessel during the reaction time and in the molecular weight cut-off (MWCO) of the membrane filtration unit were tested.

[0175] In experiment T3E vs. T3D, the total volume of permeate formed was 5 times higher (see Table 9, experiments T3D and T3E, line 7), but the liquid volume inside the reaction vessel was kept constant. This was achieved by using a more dilute solution of coupling reagent in feed tank 2, i.e. a larger volume of coupling reagent solution was added to the reaction vessel in order to deliver the same amount of coupling reagent. To counter this influx, the volumetric flow of permeate out of the reaction vessel was increased by a factor of five (see Table 9, experiments T3D and T3E, line 10). As a result of this increased inline diafiltration, the percentage of side products formed dropped from 5.39% to 2.47%, which is a reduction by 54%.

[0176] In experiment T3B vs. T3J, two membrane filtration units with different MWCOs (200 vs. 500) were compared. As the membrane used in T3J was found to have a rejection of 90% for PyOxim, the loss of coupling reagent needed to be compensated by adding more coupling reagent in order to keep its concentration within the reaction vessel constant. This is reflected in the larger concentration of coupling reagent in feed tank 2 (see Table 9, experiments T3B and T3J, line 2).

[0177] For practical reasons, the whole system (reaction vessel volume, amount of peptide cyclized, membrane area etc.) was scaled down by a factor of 2 in experiment T3J vs T3B, but no other parameters were changed. It was found that using a membrane with a higher MWCO resulted in a reduction of side products from 2.71% to 2.16%.

TABLE-US-00010 TABLE 9 Experiment No. No. Parameter T3D T3E T3B T3J 1 Peptide concentration in feed tank 1 [mmol/l] 18.33 18.33 3.67 3.67 2 Coupling reagent concentration in feed tank 2 [mmol/l] 55.69 3.28 11.13 39.00 3 Initial coupling reagent concentration in reaction vessel [mmol/l] 69.62 69.62 69.62 69.62 4 Initial volume of peptide solution in feed tank 1 [l] 0.067 0.067 0.336 0.168 5 Initial volume of coupling reagent solution in feed tank 2 [l] 0.023 0.381 0.112 0.056 6 Initial volume of coupling reagent solution in reaction vessel [l] 0.050 0.050 0.050 0.025 7 Total permeate volume [l] 0.090 0.448 0.448 0.224 8 Volumetric flow of peptide solution [l/h] 0.013 0.013 0.067 0.034 9 Volumetric flow of coupling reagent solution [l/h] 0.005 0.076 0.022 0.011 10 Volumetric flow of permeate stream [l/h] 0.018 0.090 0.090 0.045 11 Differential pressure across the membrane [bar] 9 23 38 7 12 MWCO of Duramem membrane [g/mol] 200 200 200 500 13 Membrane area [m.sup.2] 0.0066 0.0066 0.0066 0.0033 14 Operating time [min] 300 300 300 300 15 Side products [%] 5.39 2.47 2.71 2.16

Example 10.3: Effect of Concentration of Coupling Reagent

[0178] The present setup allows keeping the concentration of coupling reagent constant over the entire reaction time. It was investigated how the concentration of the coupling reagent inside the reaction vessel influences product purity. For practical reasons, the whole system (reaction vessel volume, amount of peptide cyclized, membrane area etc.) was scaled down by a factor of 2 in experiment T3F and T3G vs T3B, but no parameters other than the initial concentration of coupling reagent were changed. It was found that decreasing the excess of coupling reagent in the reaction vessel (2 vs. 1.25 vs 0.5 molar equivalents of the total amount of peptide added) while keeping the same linear peptide and coupling reagent concentrations in the feed solutions (Table 10, lines 1-3) and the same rates of reagent addition causes a decrease in product purity (Table 10, line 15). In other words: a higher excess of the coupling reagent (second compound) over the linear peptide (first compound) corresponds to a lower amount of by-products.

TABLE-US-00011 TABLE 10 Experiment No. No. Parameter T3B T3G T3F 1 Peptide concentration in feed tank 1 [mmol/l] 3.67 3.67 3.67 2 Coupling reagent concentration in feed tank 2 [mmol/l] 11.13 11.13 11.13 3 Initial coupling reagent concentration in reaction vessel [mmol/l] 69.62 43.52 17.40 4 Initial volume of peptide solution in feed tank 1 [l] 0.336 0.168 0.168 5 Initial volume of coupling reagent solution in feed tank 2 [l] 0.112 0.056 0.056 6 Initial volume of coupling reagent solution in reaction vessel [l] 0.050 0.025 0.025 7 Total permeate volume [l] 0.448 0.224 0.224 8 Volumetric flow of peptide solution [l/h] 0.067 0.034 0.034 9 Volumetric flow of coupling reagent solution [l/h] 0.022 0.011 0.011 10 Volumetric flow of permeate stream [l/h] 0.090 0.045 0.045 11 Differential pressure across the membrane [bar] 38 38 33 12 MWCO of Duramem membrane [g/mol] 200 200 200 13 Membrane area [m.sup.2] 0.0066 0.0033 0.0033 14 Operating time [min] 300 300 300 15 Side products [%] 2.71 4.77 5.29

Example 10.4: Effect of Solvent Recycling (Permeate Solution from Previous Experiments as a Solvent)

Comparative Example T3C

[0179] The same reaction conditions as in T3B were applied (see Table 9). The only difference was the solvent used to prepare the coupling reagent solution in reaction vessel. In T3C the permeate collected during T3B was used (without further purification), whereas in T3B fresh DMF was used for the same purpose.

Comparative Example T3H

[0180] The same reaction conditions as in T3B were applied (see Table 9). The only difference was the solvent used to prepare the peptide solution in feed tank 1. The solvent was prepared as follows: the nanofiltration system was cleaned after performing T3F and then the permeate collected during T3F was transferred to the clean retentate loop. This permeate solution was subjected to concentration using the same membrane as in T3B and T3H (MWCO of 200 g/mol). During concentration, the volume of the solution in the retentate loop was reduced as much as possible, and subsequently discarded. The retentate loop was then cleaned. The permeate was collected in a separate vessel and then transferred back to the clean retentate loop. The concentration was repeated 2 more times. The final permeate solution from the 3.sup.rd concentration was mixed with fresh DMF (62.5 vol % permeate, 37.5 vol % DMF) and used to dissolve the peptide used in feed tank 1.

[0181] A quantitative comparison of overall reaction performance between tests mentioned above is presented in Table 11.

TABLE-US-00012 TABLE 11 Experiment No. T3B T3C T3H Side products [%] 2.71 5.63 3.73

[0182] When recycled solvent was used instead of fresh solvent, the level of impurities roughly doubled (5.63% vs. 2.71%). The results suggest that recycling used solvent leads to reintroduction of molecular species, which may enhance the formation of peptidic side products. Such species could not be satisfactorily removed even by repeated membrane filtration, as is shown in experiment T3H.