Effect of physicochemical properties of solvents on clavulanic acid extraction from fermentation broth

L. M. Brites, J. H. Oliveira, , M. Barboza and C. O. Hokka

Department of Chemical Engineering,Universidade Federal de São Carlos. P.O. box: 67 Brazil
marlei@ufscar.br

Abstract— A study was made of the extraction of clavulanic acid (CA) from a fermented medium and its concentration and purification. The fermented broth was obtained by cultivation of Streptomyces clavuligerus in complex medium containing glycerol as the Carbon and energy source, and soybean protein isolate as the Nitrogen source. The extraction steps proposed here were evaluated from the standpoint of Yield, purification factor and concentration factor. The extractions were carried out with ethyl acetate, butyl acetate, 2-butanol, n-butanol and methyl isobutyl ketone. The distribution coefficient was related to the solvent properties and it was found that distribution coefficient is directly proportional to the product of the dielectric constant, with solubility. It was observed that extraction with butanol leads to a higher distribution coefficient than ethyl acetate; however, the higher solubility of water in butanol impairs the final precipitation of clavulanic acid. The optimal condition with ethyl acetate occurs at pH 1.63, yielding an extraction factor of 35.6% with a distribution coefficient of 0.73.

Keywords— Clavulanic Acid; Distribution Coefficient; Extraction; Downstream Process.

I. INTRODUCTION

The isolation of clavulanic acid -a highly unstable bioactive substance- from fermentative broth requires a series of steps due to the high contents of various contaminants that are typical in bioprocesses involving complex media. At the end of fermentation, the broth is clarified by filtration or centrifugation to eliminate mycelia, cell debris and insoluble particles, after which it is subjected to the most important step, the primary extraction, an operation that is based on several two-phase separation methods. One of these methods is the direct extraction from clarified broth with organic solvent, producing an organic phase containing CA, which is subsequently isolated (Butterworth, 1984, Nabais and Cardoso, 1995).

Thanks to its low cost and effectiveness, solvent extraction is by far the most widely used separation technique for antibiotics (Soto et al., 2005). Several patents describe the extraction of clavulanic acid onto organic solvents (Cole et al., 1978; Cardoso, 1998; Capuder, 1995; Capuder, 1998; Capuder, 2000; Cook and Nicola, 2001; Ruddick, 1996; Simon 2001). These patents explain that CA can be extracted from fermentation broth by different procedures, and that CA is isolated when its corresponding potassium salt is formed. This salt is produced by reacting CA with potassium 2-ethylhexanoate in the solvent-rich phase, i.e., after cell removal, the resulting clarified broth is acidified to reach pH values between 2 and 3. The CA is then extracted using organic solvents such as ethyl acetate, n-butanol, methyl isobutyl ketone (MIK), n-butyl acetate, and others, and the salt is formed. Care should be taken, since clavulanic acid is highly unstable depending on the pH of the medium (pH lower than 6.2 and higher than 7.0), the presence of ammonium compounds, and the temperature (Bersanetti et al., 2005). On the other hand, clavulanic acid forms unstable hygroscopic oil which cannot be used as a pharmaceutical compound (Cardoso, 1998); therefore, this purification step requires special attention to ensure an efficient process. Another process used for purification is ionic exchange; however, CA is also unstable under this condition, as reported by Mayer et al. (1996) and Barboza et al. (2002). Moreover, due to the very low concentration of CA in the fermentation broth in which it is produced, the separation stage of the overall antibiotic production process accounts for a large share of the total manufacturing cost. In solvent extraction methods, the organic molecules of interest such as antibiotics are usually distributed between an aqueous phase (heavy) and an antibiotic-rich organic phase (light), reaching the equilibrium state. The extraction of CA onto organic solvents is strongly affected by physicochemical properties such as solubility, polarity and dielectric constant.

This paper discusses the conditions, procedure and results obtained in the extraction of clavulanic acid from fermented broth using different organic solvents. The effect of physicochemical properties on the CA distribution coefficient, D, in the aqueous media-solvent system was determined with different organic solvents, namely, butyl acetate, ethyl acetate, methyl-isobutyl-ketone, 2-butanol and n-butanol. In addition, the effect of each solvent's physicochemical properties on the values of D was examined.

II. MATERIALS AND METHODS

A. Materials

Clavulanic acid was obtained from fermentation broth produced by Streptomyces clavuligerus ATCC 27064. The broths were obtained from fermentation runs carried out batch-wise in a 5L Bioflo III Fermentor (New Brunswick Scientific Co. Inc., Edison, NJ, USA) at 28°C, pH 6.8±0.1, 1 vvm aeration rate and agitation speed ranging from 500 to 1000 rpm. The culture me dium contained glycerol (15.0 g); malt extract (10.0 g); Samprosoy 90 NB^® (10.0 g); yeast extract (1.0 g) MgSO47H2O (0.75 g); K2HPO4 (0.80 g); soybean oil (1.0 g); trace element solution (1 mL) and distilled water (1 L), and was adjusted to pH 6.8 with NaOH/HCl (Maranesi et al., 2005; Rosa et al., 2005). The imidazole and salts used to prepare the phosphate buffer were reagents of analytical grade. The contaminants were analyzed in wavelength 280nm since most proteins are composed of tyrosine, triptophane, and phenylalanine which are aromatic amino acids with a benzene ring and conjugated double bounds with an absorption band at 280nm.

Before the extraction process, the fermentation broth was pretreated as follows: the broth was centrifuged to obtain a clear cell-free solution, whose pH was then lowered (3.5 - 4.0) to precipitate part of the soluble solids and proteins. Finally, the solution containing the product was centrifuged again and filtered through analytical filter papers to eliminate suspended impurities.

B. Analytical methods

The clavulanic acid was determined by high-performance liquid chromatography, as described by Foulstone and Reading (1982), after its reaction with imidazol, otherwise the presence of other substances could interfere with the UV analysis. The HPLC equipment with a Photodiode Array detector (Waters 996 PDA) was operated with a reverse-phase column (C-18 μ-Bondapak 3.9mm×300 mm) maintained at 28°C, with a flow rate of 2.5 mL/min, and calibrated against solutions of the pharmaceutical product Clavulin®. The mobile phase was a 0.1MKH2PO4 buffer solution with 6% methanol, adjusted to pH 3.2 with phosphoric acid.

C. Determination of contaminants

The concentration of contaminants was determined by spectrophotometric analysis at 280 nm in a Pharmacia Biotech UV/VIS spectrophotometer, disregarding the influence of clavulanic acid at that wavelength.

D. Determination of the distribution coefficient

Assuming that no CA molecule is charged during the extraction and that the extraction process takes place in a perfectly stirred tank, the distribution coefficient in equilibrium can be given by Eq. 1.

(1)

where C_OP1 is the solute concentration in the light phase, OP (organic phase), and C_AP is the solute concentration in the heavy phase, AP (aqueous phase).

The mass balance of the extraction process can yield:

(2)

(3)

where C_O is the clavulanic acid concentration in feed stream, C_AP1 is the solute concentration of the first aqueous phase, C_AP2 is the solute concentration of the second aqueous phase, C_OP1 is the solute concentration of the first organic phase, C_OP2: is the solute concentration of the second organic phase,V_AP1 is the volume of the first aqueous phase, V_OP1 is the volume of the first organic phase;V_AP2 is the volume of the second aqueous phase and V_OP2 is the volume of the second organic phase.

Assuming that all the CA molecules become charged during the back-extraction with neutral aqueous solution, the concentration in the organic phase becomes negligible (C_OP2=0). Hence, Eq. (3) becomes

(4)

Equation (1) can be combined with Eq. (4), leading to

(5)

The distribution coefficient (D) was calculated from Eq. (5) for different volume and concentration ratios. Figure 1 illustrates the extraction process schematically.

Fig. 1. Scheme of the batch extraction process for CA. A): extraction step, and B) back-extraction step.

The performance of the process can be analyzed in terms of such important parameters as yield Y, purification factor PF and concentration factor CF, defined by the following equations:

(6)

(7)

(8)

where C_To contaminants concentration in feed stream, C_T contaminants concentration in second aqueous phase, Y is the ratio of the mass of recovered CA to the mass of CA fed into the extractor. CF is the ratio of the CA concentration in the back-extraction to the CA concentration in the feed stream. PF is the ratio between the CF (C/C_o) of the CA and the CF of contaminants (C_T/C_To), according to Almeida et al. (2003) and Barboza et al. (2002).

At low pH, beta-lactam antibiotics, including clavulanic acid, usually have a protonated carboxylate group causing its low solubility in water and can be readily extracted into an organic solvent. Clavulanic acid is a weak acid, with pKa=2.5 (Mourã;o et al., 2006), and only the undissociated acid can be extracted by an organic solvent. At higher pH values the carboxylate group is deprotonated and charged, rendering the CA water soluble (clavulanate form). The neutral concentration can be expressed according to

(9)

where Cn is the concentration of the neutral form of CA and Co is the initial concentration. This equation allows for calculation of the clavulanic acid concentration in the neutral form, which remains in the heavy aqueous phase. The degree of extraction, E, can be evaluated using Eq. (10), as follows:

(10)

III. RESULTS AND DISCUSSION

The results obtained in each of the experiments allowed us to determine the distribution coefficient, D, for clavulanic acid in all the solvents utilized. Table 1 presents the experimental values together with each solvent's pertinent properties such as the dielectric constant, polarity, solubility in water and its molecular weight.

Table 1. Clavulanic acid distribution coefficient in different solvents and relevant properties of the solvents (Data from Perry and Green 1997 and Weast et al., 1983).

The effects of the physicochemical properties of the organic solvents on the CA extraction are listed in Table 1. The degree of extraction is intrinsically affected by these properties. The first influence observed was on the organic solvent molecular weights. Figure 2 shows that D is inversely related to the solvent's molecular weight, i.e., the higher the MW the lower the distribution coefficient. The extraction of CA with 2-butanol and n-butanol, representing the low MW solvents, was considerably higher than with the other solvents. As for the effect of the dielectric constant on D, there is no simple or theoretical function able to relate these two variables, as indicated in Fig. 3. However, if the data are divided into two regions, two directly proportional relationships are observed. The first region encompasses the results obtained with butyl and ethyl acetate, while the second region involves the other three solvents, with a dielectric constant higher than 10. The discontinuity of the tendency for the linear increase of D with the dielectric constant is intrinsically related to the molecular structure of each solvent, particularly with regard to the functional ketone group, and with the solubility of these solvents in water. CA contains two hydroxyl groups in its structure, allowing for interactions between molecules through hydrogen bonds. The dissolution of substances like CA is favored in polar solvents in which hydrogen bonds are formed, such as alcohols. Solvents like ethyl acetate and other esters present weaker intermolecular interaction (dipole-dipole) when compared with hydrogen bonds, leading to a low dielectric constant and a weak interaction between solute (CA) and solvent, despite their higher polarity.

Fig. 2. Relation between D and molecular weight

Fig. 3. Relation between D and the dielectric constant,d.

The composition of the medium strongly affects the thermodynamic properties, which in turn affect the intermolecular interaction between the solvent and CA, producing different yields and purification factors. Figure 4 illustrates the behavior of D with solubility, S, and their relationship. The third point shown in this figure, corresponding to ethyl acetate, deviates from the linear behavior shown by the other solvents, i.e., although the solubility of ethyl acetate is nearly the same as that of 2-butanol, the dielectric constant of 2-butanol is much higher than that of ethyl acetate. However, as Fig. 5 indicates, there is a well defined direct proportionality between D and the product of the dielectric constant and solubility, dxS, showing that this product can be used as a criterion to investigate other possible solvents to extract CA from aqueous media.

Fig. 4. Relation between D and S, solubility, of solvent in water.

Fig. 5. Linear relation between D and the product of solubility and the dielectric constant (dxS): Correlation obtained: D = 0.0053 d e + 0.2045 with r² = 0.9881

Table 2 lists the results of yield, purification factor and concentration factor for CA extraction from the clarified broth, while Table 3 presents these results for ethyl acetate and n-butanol.

Table 2. Extraction of CA with ethyl acetate

Table 3. Extraction of CA with n-butanol

* not determined

The solvent n-butanol produces better results than ethyl acetate in terms of its ability to concentrate the CA solution, because n-butanol is more water-soluble than ethyl acetate; CA, in turn, is highly water-soluble, improving the extraction. The values of the concentration factor were found to decrease when ethyl acetate was used to extract CA, due to its low solubility in water.

Although n-butanol presents a higher distribution coefficient for CA, the fact that it is more water-soluble actually hinders the purpose of the purification process, which is to trigger the precipitation reaction in this solvent. This is a decisive factor in choosing the most adequate solvent for the process. Therefore, ethyl acetate is potentially the solvent of choice to purify CA from fermentation broth.

In order to increase the value of the distribution constant, experiments were conducted to determine the pH at which all the CA was in the dissociated form. The results in Fig. 6 indicate the pH at which the maximum CA extraction is achieved, i.e., the pH corresponding to the lowest neutral CA (clavulanate form) concentration.

Fig. 6. Variation of neutral concentration and the degree of extraction, with pH, for ethyl acetatetion coefficient of 0.73.

The neutral concentration is zero at pH 1.63; in this condition the extraction factor is 35.6% with a distribution coefficient of 0.73. Fig. 7 presents the relation between D and CA in the neutral form (clavulanate), in the extraction with ethyl acetate. Therefore, at zero level of Cn, the maximum D value is 0.73.

Fig. 7. Distribution coefficient (D) clavulanic acid in ethyl acetate as a function of the neutral concentration (Cn).

Fig. 7 presents the relation between D and CA in the neutral form (clavulanate), in the extraction with ethyl acetate. Therefore, at zero level of Cn, the maximum D value is 0.73.

To select the best solvent for clavulanic acid extraction, a mass balance must be achieved along the process. The use of n-butanol is very attractive because it produces a high extraction yield, but in this case, the use of a desiccant substance or dehydrating material is mandatory in order to withdraw water solubilized in n-butanol . The cost of removing water from the process should be a decisive factor in the choice between n-butanol and ethyl acetate as the solvent.

During the extraction of clavulanic acid by organic solvents such as ethyl acetate and other esters, dipole-dipole type interactions occur. These are weak interactions, e.g., hydrogen bonding. Despite the higher polarity of some organic solvents, they usually have a low dielectric constant, which affects the CA extraction, and a weak solute-solvent interaction. These interactions may be stronger or weaker, depending on the components in the solution which are able to promote changes in the thermodynamic properties of the mixture, affecting the purification factor and yield.

IV. CONCLUSIONS

The extraction of clavulanic acid with n-butanol is better than with ethyl acetate, but the higher solubility of water in n-butanol impairs the final precipitation of CA.

The optimal condition for the extraction of clavulanic acid with ethyl acetate is at pH 1.63, which provides an extraction factor of 35.6% with a distribution coefficient of 0.73.

In addition, the distribution coefficient is linearly related to the product dxS, which could be used to investigate other solvents to extract CA or other beta-lactam antibiotics from fermentation broth.

ABBREVIATIONS

C_O: clavulanic acid concentration in feed stream (g.L^-1 );

C_AP1: solute concentration of the first aqueous phase (g.L^-1);

C_OP1: solute concentration of the first organic phase (g.L^-1);

Cn: neutral concentration of clavulanic acid in aqueous phase (mg.L^-1)

C_To: contaminants concentration in feed stream (absorbance unit- au),;

C_T: contaminants concentration in second aqueous phase (absorbance unit- au);

V_AP1 : volume of the first aqueous phase (L);

V_OP1: volume of the first organic phase (L);

C_AP2: solute concentration of the second aqueous phase (g.L^-1);

C_OP2: solute concentration of the second organic phase (g.L^-1);

V_AP2 : volume of the second aqueous phase (L);

V_{OP 2}: volume of the second organic phase (L);

Y: yield of extraction (%);

CF: concentration factor (-);

PF: purification factor(-);

D: distribution coefficient (-)

D: dielectric constant (e vacuum^-1)

E: extraction factor (%)

MIK:Methyl isobuthyl ketone

MW: Molecular weight

ε: dimensionless dielectric constant;

S: Solubility (g/100g H₂O)

ACKNOWLEDGEMENTS
The authors gratefully acknowledge the financial support of FAPESP (Grants 04/16056-6, 04/15540-1 and 05/55079-4) and the CNPq scholarship granted to L. M. Brites.

REFERENCES
1. Almeida, R.M.R.G; M. Barboza and C.O. Hokka, "Continuous clavulanic acid adsorption process," App. Bioch. and Biotech, 105, 867-879 (2003).         [ Links ]
2. Barboza, M.; R.M.R.G. Almeida and C.O. Hokka, "Kinetic studies of clavulanic acid recovery by ion-exchange chromatography," Biosep., 10, 221-227 (2002).         [ Links ]
3. Bersanetti, P.A., R.M.R.G. Almeida, M. Barboza, M.L.G.C. Aaraujo and C.O. Hokka, "Kinetics studies on clavulanic acid degradation," Biochem. Eng. J, 23, 31-36 (2005).         [ Links ]
4. Butterworth, D., "Clavulanic Acid: Properties, Bio-synthesis and Fermentation," In: Vandamme, E. J. - Biotechnology of Industrial Antibiotcs.- New York, Marcel Deckker, 22, 225-235(1984).         [ Links ]
5. Capuder, E., Novel Process for the Isolation of "Clavulanic Acid and of Pharmaceutically Acceptable Salts Thereof from the Fermentaion Broth of Streptomyces sp. P 6621 Ferm P2804," WO 95/23870 (1995).         [ Links ]
6. Capuder, E., Process for the isolation of clavulanic acid and of pharmaceutically acceptable saltes thereof from the fermentation broth of streptomyces sp. P6621 FERM P2804, USP 5.780.274 (1998).         [ Links ]
7. Capuder, E., Isolation of clavulanic acid from fermentation broth by ultrafiltration, U.S.P. 6127358 (2000).         [ Links ]
8. Cardoso, J.P., Process for the isolation of a pharmaceutically acceptable alckali metal salt of clavulanic acid, EP 0 867 515 A1 (1998),         [ Links ]
9. Cole, M., T.T. Howarth and C. Reading, Process of Production of Clavulanic Acid, USP 4110165 (1978).         [ Links ]
10. Cook; M.A. and M. Nicola, Process for the preparation of a metal salt of clavulanic acid, USP 6300495 (2001).         [ Links ]
11. Foulstone, M. and C. Reading, "ssay of Amoxicilin and Clavulanic Acid, the Components of Augmentin, in Biological Fluids with High-Performance Liquid Chromatography," Ant. Ag. and Chem., 22, 753-762 (1982).         [ Links ]
12. Mayer, A. F.; Anspach, F.B. & Deckwer, W. D. "Purification on Clavulanic acid by ion-pairing systems," Biosep., 6 25-39 (1996).         [ Links ]
13. Maranesi, G.L., A. Batista-Neto, C.O. Hokka and A.C. Badino, "Utilization of vegetable oil in the production of clavulanic acid by Streptomyces clavuligerus ATCC 27064," World J. of Microb. & Biotech., 21, 509-514 (2005).         [ Links ]
14. Morão, A.I.C., A.M.B. Alves, M.C. Costa and J.P Cardos, "Nanofiltration of a clarified fermentation broth," Chem. Eng. Sc., 61, 2418-2427 (2006).         [ Links ]
15. Nabais, A. M. A.; Cardoso, J. P. "Ultrafiltration of Fermented Broths and Solvent Extraction of Antibiotics,". Biop. Eng., 13 215-221 (1995).         [ Links ]
16. Perry, R.H. and D.W. Green, Perry's Chemical Engineers'Handbook; 7th edition, McGraw-Hill, (1997).         [ Links ]
17. Rosa, J.C., A. Baptista-Neto, C.O. Hokka and A.C. Badino, "Influence of dissolved oxygen and shear conditions on clavulanic acid production by Streptomyces clavuligerus," Biop. and Bios. Eng., 27, 99-104 (2005).         [ Links ]
18. Ruddick, S., Calvulanic acid extraction process, WO 96/22296 (1996).         [ Links ]
19. Simon, R., Clavulanic Acid Extraction Process, USP, 6172221 (2001).         [ Links ]
20. Soto, A., A. Arce and M.K. Khoshkbarchi, "Partitioning of antibiotics in a two-liquid phase systemformed by water and a room temperature ionic liquid," Sep. and Purif. Tech., 44, 242-246 (2005).         [ Links ]
21. Weast, R.C., Handbook of Chemistry and Physics; 64^th Edition, CRC Press Inc. (1983).         [ Links ]

Received: November 12, 2008.
Accepted: August 26, 2011.
Recommended by Subject Editor Ana Lea Cukierman.