Effect of pH on biosynthesis of lovastatin and other secondary metabolites by Aspergillus terreus ATCC 20542

Journal of Biotechnology 162 (2012) 253–261

Contents lists available at SciVerse ScienceDirect

Journal of Biotechnology journal homepage: www.elsevier.com/locate/jbiotec

Effect of pH on biosynthesis of lovastatin and other secondary metabolites by Aspergillus terreus ATCC 20542 Marcin Bizukojc ∗ , Marta Pawlak, Tomasz Boruta, Joanna Gonciarz Technical University of Lodz, Faculty of Process and Environmental Engineering, Department of Bioprocess Engineering, ul. Wolczanska 213, 90-924 Lodz, Poland

a r t i c l e

i n f o

Article history: Received 17 July 2012 Received in revised form 30 August 2012 Accepted 5 September 2012 Available online 17 September 2012 Keywords: Lovastatin Mevinolinic acid pH Lactonisation Mass spectrometry Aspergillus terreus

a b s t r a c t The effect of the initial pH value of the cultivation medium on lovastatin (mevinolinic acid) biosynthesis by Aspergillus terreus ATCC20542 was studied. It was found that if the pH value of the broth is acidic, the direct chromatographic assay of mevinolinic acid leads to the underestimated values. Thus, the equilibrium curve was determined for the transformation of ␤-hydroxy acid form of lovastatin (mevinolinic acid) into lovastatin lactone. The calculation of the equilibrium constant shows that when the pH value of the solution is 4.98, concentrations of both forms of lovastatin are equal to each other. This finding was next used to study mevinolinic acid formation at the various initial pH values of the medium. It occurs that even at pH lower than 5.5 mevinolinic acid is still, although inefficiently, produced and its presence remains unnoticed, unless the samples of the broth are alkalised prior to the assay. Mevinolinic acid is efficiently produced at the initial pH value of the medium equal to 7.5 and 8.5 and it correlates with the rapid utilisation of lactose by A. terreus. Additionally, other secondary metabolites were sought at the various initial pH values of the medium with the use of mass spectrometry. (+)-Geodin is only formed at pH 6.5, while monacolin L is found at the highest amount at pH 7.5. © 2012 Elsevier B.V. All rights reserved.

1. Introduction Lovastatin is an important polyketide metabolite from Aspergillus terreus because of its capability, as the competitive inhibitor of 3-hydroxymethylglutaryl-CoA (3-HMG-CoA) reductase, to decrease the endogenous level of cholesterol in the human organism. A. terreus produces it in the form of mevinolinic acid (␤hydroxy acid form of lovastatin lactone). Since its discovery in the mid-seventies of the twentieth century research on the biosynthesis of this metabolite has been widely made with regard to the variety of factors like medium composition, aeration conditions, fungal morphology and broth rheology but the issue of the pH value of the medium has always been somewhat omitted. It is generally believed that the initial pH value of the medium for lovastatin production by A. terreus should be 6.5 and almost no discussion on this issue has been performed in literature. The effect of the initial pH value of the medium in the range from 5 to 9 has been recently studied in the limited way by Osman et al. (2011) only. Their results comprised only lovastatin titres after 8 days of cultivation without any thorough discussion and presentation of lovastatin formation kinetics. The control of pH on three different levels (5.5, 6.5 and 7.5) during the cultivation was studied by Lai

∗ Corresponding author. Tel.: +48 42 631 39 72; fax: +48 42 636 56 63. E-mail address: [email protected] (M. Bizukojc). 0168-1656/$ – see front matter © 2012 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.jbiotec.2012.09.007

et al. (2005). They concluded that it was ineffective at pH 6.5 and even deteriorating at 5.5 and 7.5. Lai et al. (2005) explained their results by the change of the activity of enzymes. The contradictory results of regarding pH control were obtained by Bizukojc and Ledakowicz (2008) and Pawlak et al. (2012), who showed that if a hydrogen carbonate solution was used to control pH level, lovastatin titres increased and formation of undesired (+)-geodin was repressed. The fact that the pH value of the medium may change the metabolism of any microorganism and so may in A. terreus was confirmed by the fact that A. terreus produces itaconic acid (another important metabolite of this fungus used in chemical industry) at the low pH from glucose as the sole carbon source, while lovastatin is biosynthesised at the neutral pH (Lai et al., 2007). It is also a premise to seek for other than lovastatin metabolites of A. terreus dependent on the pH value of the culture. The other issue concerning pH is lovastatin assay in the broth due to the presence of its two forms: ␤-hydroxy acid (excreted by fungi and water-soluble mevinolinic acid) and lactone (organics soluble). As lactonisation reactions are pH-dependent (Morrison and Boyd, 1992), it must be remembered to take this factor into account. This issue has not been discussed yet in detail in literature concerning lovastatin biosynthesis by A. terreus. The aims of this work are as follows. First of all, the effect of the pH value of the cultivation medium on the form of lovastatin, either ␤-hydroxy acid or lactone and reliability of its determination by the direct chromatographic measurement in the filtered broth is going to be evaluated. Secondly, the effect of the initial pH value

254

M. Bizukojc et al. / Journal of Biotechnology 162 (2012) 253–261

Nomenclature cH+ hydrogen ion concentration (mol l−1 ) clactone lovastatin lactone concentration (mg LOV l−1 ) cLOV total lovastatin concentration (mg LOV l−1 ) c␤-hydroxy acid , cMEV lovastatin ␤-hydroxy acid concentration (mg MEV l−1 ) LAC lactose LOV lovastatin m/z mass to electric charge ratio mevinolinic acid MEV rLAC volumetric lactose uptake rate (g LAC l−1 h−1 ) retention time in the chromatographic analysis tR (min) YMEV/LAC yield coefficient for mevinolinic acid formation on lactose (mg MEV g LAC−1 ) cLAC decrement of lactose concentration (g LAC l−1 ) equilibrium constant of reaction: ␤-hydroxy acid – Keq lactone cMEV increment of mevinolinic acid concentration (mg MEV l−1 ) (m/z) absolute error of determination of mass to electric charge ratio

of the cultivation medium on the formation of lovastatin and other secondary metabolites is going to be studied. 2. Materials and methods 2.1. Microorganism The strain A. terreus ATCC 20542 was employed in the experiments. The preculture was prepared from the spores grown on the 10-days malt extract slants. The spores were washed and suspended in the preculture medium to achieve approximately 109 spores per litre. The inoculation was performed with the 24h preculture, which was added in such an amount to hold the initial biomass concentration at the level of between 0.1 and 0.15 g l−1 . 2.2. Cultivation media The cultivation media contained the following mineral components and vitamins: KH2 PO4 : 1.51 g l−1 , MgSO4 ·7H2 O: 0.52 g l−1 , NaCl: 0.4 g l−1 , ZnSO4 ·7H2 O: 1 mg l−1 , Fe(NO)3 ·9H2 O: 2 mg l−1 , biotin: 0.04 mg l−1 and 1 ml solution of trace elements per 1 l of the medium. The solution of trace elements contained Na2 B4 O7 ·10H2 O: 100 mg l−1 , MnCl2 : 50 mg l−1 Na2 MoO4 ·2H2 O: 50 mg l−1 and CuSO4 ·5H2 O: 250 mg l−1 . Yeast extract (BD, USA) was used as the nitrogen source at the concentration of 4 g l−1 (8 g l−1 in the preculture). Lactose and glycerol (only in the bioreactor runs) were used as the carbon sources at concentrations from 10 to 20 g l−1 . In the preculture medium only lactose was used at 10 g l−1 (Pawlak and Bizukojc, 2012).

formic acid so that its concentration equalled 0.1%) designed as follows: • • • • •

step 1: 0.0–2.5 min 0:100 (v/v), step 2: 2.5–3.5 min 20:80 (v/v), step 3: 3.5–4.5 min 30:70 (v/v), step 4: 4.5–6.8 min 40:60 (v/v), step 5: 6.8–14.0 min 60:40 (v/v) was used.

Temperature of the column was 40 ◦ C and flow rate of the eluent was 0.2 ml min−1 . A Waters Mass Spectrometer SYNAPT G2 (quadrupole time-of-flight) was applied to measure the molecular masses of the metabolites and their fragments. Positive electrospray ionisation was used to generate ions. Its parameters were set as follows: capillary voltage at 3 kV, sampling and extraction cones at 40 and 4 V, respectively, source temperature at 120 ◦ C, desolvation temperature 200 ◦ C and flow of desolvation gas at 500 l h−1 . Upon the mass spectra and additional photodiode array spectra various metabolites were attempted to be recognized. For the standard quantitative assay of mevinolinic acid, the shortened elution gradient [0–3 min 40–60 (v/v); 3–7 min: 60:40 (v/v)] in the same column was used and this metabolite was detected by a photodiode array detector at  = 238 nm. All samples before the injection to the column were filtered through 0.2 ␮m syringe filters. Due to the reasons described further in Section 3 the samples of the low pH value were alkalised with a concentrated NaOH solution before filtration and chromatographic assay. The accuracy of this analysis was high and the subsequent three injections of the same sample did not deviate by more than 0.8 mg MEV l−1 , when a 100 mg MEV l−1 standard solution was used. and glycerol were simultaneously deterLactose mined on a Waters Acquity UPLC® BEH Amide column (2.1 mm × 150 mm × 1.7 ␮m). The compounds were eluted with 75% acetonitrile solution in deionised water enriched with 0.2% of triethylamine at flow rate of 0.29 ml min−1 , temperature of 35 ◦ C and detected by an evaporated light scatter (ELS) detector. Mevinolinic acid standard solution was prepared according to the method given by Casas Lopez et al. (2003). A weighed sample of lovastatin lactone of analytical grade was transformed into mevinolinic acid in the solution of 0.1 M NaOH and ethanol (50:50 v/v) at 50 ◦ C within 20 min. After cooling the solution down it was neutralised to pH 7 with concentrated hydrochloric acid. The solution of mevinolinic acid was stored at 4 ◦ C. In order to determine ␤-hydroxy acid vs. pH equilibrium curve a Britton–Robinson buffer solution was used to assure the appropriate pH value in the environment of the transformation reaction of ␤-hydroxy acid to lactone. The buffer was the mixture of 0.04 M solutions of boric (H3 BO3 ), phosphoric (H3 PO4 ) and acetic (CH3 COOH) acids. Its pH value was set to the desired value by titration with a 0.2 M sodium hydroxide (NaOH) solution. The possible range of pH values to be set with the use of this buffer is from 1.81 to 11.98. Mevinolinic acid solution was added to the buffer solutions of pH values from 3 to 10 to achieve its concentration of 20 mg MEV l−1 . In order to be sure that the equilibrium was achieved, the samples were incubated in the tightly sealed bottles overnight at 30 ◦ C (the temperature used for the cultivation of A. terreus) and next the chromatographic assay of mevinolinic acid was made.

2.3. Analytical methods 2.4. Shake flask and bioreactor runs Mevinolinic acid, (+)-geodin and other polyketide metabolites were simultaneously sought with the use of liquid chromatography coupled with mass spectrometry (Waters, USA). A Waters Acquity UPLC® BEH Shield RP18 column (2.1 mm × 100 mm × 1.7 ␮m) at the gradient elution CH3 CN:H2 O (both solvents acidified with

Shake flask runs were conducted in the flasks of 500 ml total and 150 ml working volume at 30 ◦ C. The speed of the rotary shaker was constant at 110 min−1 . The stirred tank bioreactor of 5.3 l working volume was also used at 30 ◦ C. Oxygen saturation level was

M. Bizukojc et al. / Journal of Biotechnology 162 (2012) 253–261

controlled at 20% by increasing of air flow rate and rotary speed of the impeller. The pH level was set constant from 24 h of the run at 7.6 and controlled by dosing a concentrated sodium and potassium hydrogen carbonate solution. 2.5. Calculation tools A non-linear curve fitting algorithm implemented in PTC Mathcad 14.0 software was used to approximate the equilibrium data: mevinolinic acid concentration versus pH value. 3. Results and discussion 3.1. Deviations in mevinolinic acid concentration in the cultivation broth In most bioprocesses, when a metabolite is formed by a microorganism, an increasing trend of its concentration in a batch culture is usually expected. If it is not the case, a sensible explanation is required why the desired metabolic product vanished from the broth because only then any measures to avoid this undesired effect could be undertaken. In many bioprocesses metabolic products may be reconsumed or hydrolysed, e.g. ethanol may be re-used as the carbon substrate by yeasts, if the anaerobic conditions in the culture were not hold, or penicillin can be hydrolysed by its producer Penicillium chrysogenum. In the latter case a hydrolysis term was often incorporated into the kinetic expression for product (penicillin) formation, when this process was modelled (Paul and Thomas, 1996). In lovastatin biosynthesis by A. terreus the decreasing concentration of this metabolite in batch cultures was sometimes observed and can be found in the figures in the articles by Novak et al. (1997), Liu et al. (2000), Hajjaj et al. (2001), Casas Lopez et al. (2003), Lai et al. (2005) and Rodriguez Porcel et al. (2008). The authors rather did not try to explain this phenomenon. Only Liu et al. (2000) for example assumed the enzymatic destruction of lovastatin in their structured model for lovastatin biosynthesis by A. terreus. Some authors claimed that mevinolinic

255

acid was reconsumed (Lai et al., 2005) or subjected to enzymatic hydrolysis. In Fig. 1 the present experimental data showing the decreasing trend of mevinolinic acid concentration are depicted. These more or less abrupt changes in mevinolinic acid concentration are difficult to explain only on the basis of the physiological reasons, i.e. connected with the metabolism of A. terreus. It is hardly possible that lovastatin vanished from the broth and returned suddenly being at the same time formed with so high rate as seen in Fig. 1b. Nevertheless, these data indicated that the most probable reason were the changes of the pH value of the broth. Looking for the explanation and, above all, a quantification of this phenomenon, it must be also mentioned that before chromatographic analysis the preparation of the sample can be different. Generally, two approaches were used. Either mevinolinic acid was directly determined in the cultivation broth (Casas Lopez et al., 2003, 2005; Rodriguez Porcel et al., 2007; Rollini and Manzoni, 2006; Hajjaj et al., 2001) or the broth was acidified and extracted by water miscible solvents like ethanol or methanol (Novak et al., 1997; Lai et al., 2002, 2003, 2005, 2007) or water non-miscible solvents like ethyl acetate, benzene, cyclohexane (Manzoni et al., 1998; Kumar et al., 2000; Jia et al., 2009; Osman et al., 2011). The latter approach gave different analytical results as after acidification and extraction mevinolinic acid was transformed into lactone and product concentration was determined as the sum of these two forms. Due to the replicable sample preparation process, this method seems to be more independent of the pH value of the sample. On the other hand due to its simplicity the direct determination is more convenient. However, none of the authors cited above explicitly discussed the effect of the initial or varying pH value of the broth during the cultivation on lovastatin assay. Only in the papers, whose subject was the analytical method to determine lovastatin, the existence of various forms of lovastatin was mentioned in the context of setting the pH values of the samples on the same level (Friedrich et al., 1995). So the following thesis was proposed. If lovastatin is directly analysed in the broth as mevinolinic acid, i.e. in its ␤-hydroxy acid

Fig. 1. Decrease of the pH value of the broth and accompanying changes in mevinolinic acid concentration during a batch bioreactor cultivation of A. terreus (a) without pH control and (b) with pH control control; lovastatin was assayed as mevinolinic acid; lactose and glycerol were the carbon sources (their changes in time not shown).

256

M. Bizukojc et al. / Journal of Biotechnology 162 (2012) 253–261

form, its decreasing concentration is correlated with the pH value of the culture (Fig. 1) due to the fact that the existence of two forms of lovastatin: lactone and ␤-hydroxy acid depends on pH. It is thus important to quantify this correlation. Furthermore, the solubility of lovastatin lactone in water is low (Sun et al., 2005) and thence being bound to the mycelium it may be lost during the filtration of the samples to remove biomass. As a consequence the results of the assay may be underestimated. 3.2. Lactone ˇ-hydroxy acid equilibrium for lovastatin Lovastatin can be found in two chemical forms either ␤-hydroxy acid or lactone dependent on pH. Lactonisation is an intramolecular esterification that takes place in those hydroxy acids, at which the carboxylic and hydroxyl groups are connected in the distance of at least 4 carbon atoms. It is called ␦-lactonisation (Morrison and Boyd, 1992). Mevinolinic acid is transformed into lovastatin lactone in accordance with the reaction shown in Fig. 2. Shortly, the reaction from Fig. 2 can be written as: ␤-hydroxy acid + H+  lactone

(1)

For this reaction the following equilibrium constant is defined: Keq =

clactone

In order to find the correlation between the concentration of ␤hydroxy acid form and pH Eq. (1) should be transformed to the form similar to Hasselbach–Henderson equation describing the equilibria for the various ionic forms of amino acids. Thus: c␤-hydroxy acid · cH+ 1 = Keq clactone

(3)

Applying the decimal logarithms on both sides of the equation one obtains:



log

1 Keq



= log

c

␤-hydroxy acid

clactone



+ log (cH+ )

(4)

Finally, using the definition of pH value, one gets: log

c

␤-hydroxy acid

clactone





= pH + log

1 Keq



(5)

Upon the balance of both forms of lovastatin in the sample the formula must be derived to find mevinolinic acid concentration cMEV (␤-hydroxy acid form), denoting cLOV as total lovastatin concentration (both forms taken into account):



cMEV = cLOV

1−

1 (c␤-hydroxy acid /clactone ) + 1

␤-hydroxy acid

clactone



(6)





= 10

pH+log



1 Keq

 (7)

is derived from Eq. (5). All in all, the equilibrium curve can be described by the following theoretical equation:



cMEV = cLOV

1−



1 10(pH+log (1/Keq )) + 1

(8)

In Fig. 3 the determination of two parameters Keq and cLOV of Eq. (8) is shown. It was made on purpose that the known concentration cLOV was set as a parameter and calculated to better check the validity of fitting. The results of fitting were as follows. Total lovastatin concentration cLOV was equal to 19.68 mg LOV l−1 , what deviates by not more than 1.6% from the real concentration used in the experiment (20 mg LOV l−1 ). Furthermore, the residual error of the fit did not deviate by more than ±1 mg l−1 for each experimental point (Fig. 3). Equilibrium constant Keq was found to be equal to 9.507·104 . For the convenient use it is worth transforming it to the form, which is in agreement with the definition of pH, so:



p (2)

c␤-hydroxy acid · cH+

where:

c

1 Keq



= − log



1 9.507 · 104



= 4.98

(9)

Thus, at pH 4.98 the concentrations of both forms of lovastatin: lactone and ␤-hydroxy acid are equal to each other. From this correlation it can be estimated that mevinolinic acid makes 95% of both forms of lovastatin when pH is equal to 6.26. Thus, the direct assay of mevinolinic acid described in Section 2 in the samples is reliable (5% of lactone) at pH not lower than 6.26. Mevinolinic acid makes 99% of both forms of lovastatin at pH equal to 6.99 and then the analysis is even more reliable (1% of lactone). It is clearly seen from Fig. 1 that due to the metabolic activity of A. terreus the pH value of the broth could decrease below these values. Therefore, in such conditions the applied direct mevinolinic acid assay led to the underestimated concentrations of this metabolite. Nevertheless, in this moment it is not the object of discussion, whether the decrease of the pH value of the broth had any, either positive or negative, effect on fungal physiology and secondary metabolites formation as it is going to be done in Section 3.3. To sum up, taking these considerations into account, the suspicious sample indicated by an arrow in Fig. 1b was alkalised and re-analysed. It was found that real mevinolinic concentration was equal to 75.243 mg MEV l−1 . If calculated upon the equilibrium data, mevinolinic acid concentration should be equal to 63.358 mg MEV l−1 . This discrepancy seems to be high but here the effect connected with physiology of A. terreus might have played a certain role. But it is impossible to evaluate how this abrupt change of the pH value of the broth influenced the metabolic pathways of lovastatin formation.

Fig. 2. Transformation of ␤-hydroxy acid into lactone for lovastatin.

M. Bizukojc et al. / Journal of Biotechnology 162 (2012) 253–261

257

Fig. 3. Determination of equilibrium parameters for mevinolinic acid lactonisation.

3.3. Formation of lovastatin at the various initial pH values In Section 3.1 it was shown that pH had the significant effect on the level of mevinolinic acid measured in the samples taken out of bioreactors. It is also well known that the pH value of the broth in the process of mevinolinic acid (lovastatin) biosynthesis by A. terreus does not remain constant (Fig. 1a), unless controlled at the given level (Fig. 1b). Furthermore, as mentioned in the Introduction in most scientific papers it is claimed that the optimum initial pH value of lovastatin production medium is 6.5 (Hajjaj et al., 2001; Lai et al., 2002, 2007; Casas Lopez et al., 2003, 2005; Bizukojc and Ledakowicz, 2008; Pawlak et al., 2012). Hajjaj et al. (2001) believed that at the lower pH values mevinolinic acid was not formed at all. Only limited data (no changes in time given) from Osman et al. (2011) on the influence of the initial pH value of the medium on mevinolinic acid formation by A. terreus in the wide range of pH are available. In Fig. 4 the changes of pH in time at the various initial pH values of the medium are shown. Despite the fact that the initial pH value

of the medium varied to the high extent at the start of the cultivation, after 24 h in each run the pH value was included in the range from 6.99 to 7.84. Thus, when the initial pH value was acidic and neutral from 3.5 to 7.5 the increase of pH was observed, while for the alkalic range the decrease of pH took place. It can be explained by the rapid excretion of carbon dioxide during the phase of the most intensive biomass growth, initially exponential then linear. Hydrogen carbonate ions formed contributed to the buffering of the broth. Later on, all pH curves showed the decreasing trend. The lower was the initial pH value of the medium, the strongest was the consequent decrease in pH. This trend was not unusual as it was previously observed by Lai et al. (2005), Bizukojc and Ledakowicz (2008) and Pawlak et al. (2012). The initial pH value of the medium exerted a strong effect on lactose (carbon substrate) uptake (Fig. 5). The shape of the curve for the runs with the initial pH values from 7.5 to 9.5 was concave and mean volumetric lactose uptake rates were high during the first 48 h (rLAC = 0.25–0.32 g LAC l−1 h−1 ). The curves for the initial

Fig. 4. Changes of pH of the broth during mevinolinic acid biosynthesis in the shake flask culture by A. terreus at the various initial pH values.

Fig. 5. Influence of the initial pH value of the medium on lactose utilisation during mevinolinic acid biosynthesis by A. terreus in the shake flask culture.

258

M. Bizukojc et al. / Journal of Biotechnology 162 (2012) 253–261

pH values ranging from 3.5 to 5.5 were convex and within the first 48 h lactose was slowly utilised with the approximately same rate equal to 0.04 g LAC l−1 h−1 . The curve for the initial pH value equal to 6.5 lay exactly between the curves described above (Fig. 5). It was partly convex and partly concave. It seems that the rapid utilisation of carbon substrate at the higher initial pH values was the evidence of the faster metabolism of A. terreus. Nevertheless, it did not have to be directed into lovastatin biosynthetic pathway resulting in the higher mevinolinic acid titres. That is why, lovastatin concentration assayed as mevinolinic acid was a decisive variable in the choice of the most suitable initial pH value of the medium for A. terreus. The influence of the initial pH value on mevinolinic acid production is depicted in Fig. 6. As expected high lovastatin titre was obtained for the initial pH value equal to 6.5, which is generally in agreement with literature. However, even higher mevinolinic acid concentration was achieved, when the initial pH value of the medium was at the higher level of 7.5. In the light of this result the previous suggestion of Bizukojc and Ledakowicz (2008) and Pawlak et al. (2012) to control pH level above 7 in the bioreactor runs seems to be justified. Furthermore, it was noticed that mevinolinic acid was also formed at the low initial pH values, even at pH 3.5. Of course, this formation was far less efficient than in the runs with the neutral or alkalic initial pH value. But if mevinolinic acid were directly determined in the broth, its presence would remain unnoticed due to the low pH values of the samples. In such case the appropriate assay of lovastatin required alkalisation of the samples. In Fig. 6 mevinolinic acid curves with the corrected amount of mevinolinic acid are also depicted. So are the points (open circles) indicating mevinolinic acid concentration estimated for the actual pH value of the sample from Eq. (8). Thus, the direct assay of mevinolinic acid did not reveal its real amount. What is more, even the correction with the use of the equilibrium equation may be insufficient, as there was a tendency to underestimate the amount of mevinolinic acid present in the broth. So only the alkalisation of the samples is the right approach to determine mevinolinic acid in the broth.

Table 1 Effect of the initial pH value of the medium and analytical procedure of mevinolinic acid assay on yield coefficient YMEV/LAC . pH

YMEV/LAC mg MEV/g LAC

YMEV/LAC a mg MEV/g LAC

3.5 4.5 5.5 6.5 7.5 8.5 9.5

<0.1 0.60 ± 0.21 0.94 ± 0.08 3.33 ± 0.08 6.09 ± 0.42 6.90 ± 0.32 6.53 ± 0.89

<0.3 1.16 ± 0.05 1.36 ± 0.11 3.37 ± 0.16

a

Samples were alkalised prior to chromatographic assay.

Additionally, in Table 1 the experimental yield coefficients of mevinolinic acid on lactose are collected. There were calculated from the following equation:

YMEV/LAC =

cMEV cLAC

(10)

and the corrected and uncorrected concentrations of mevinolinic acid were used. These results confirmed that the optimum initial pH value of the medium for lovastatin biosynthesis by A. terreus was higher than usually applied 6.5. Lactose was the most efficiently transformed into mevinolinic acid at the pH value equalled 8.5 (6.90 mg MEV g LAC−1 ), what was even higher pH value than the one (7.5) allowing for the maximum lovastatin titre obtained in this study (84.3 mg MEV l−1 ). The optimum initial pH value of the medium found here occurred to be in a good agreement with the data from Osman et al., 2011, who obtained the best lovastatin titres (60 and 66.69 mg LOV l−1 ) at pH equalled 8 and 8.5, respectively. Nevertheless, one need to be cautious comparing the aforementioned results, as those authors studied a local Egyptian strain of A. terreus and glucose, oat meal and sodium acetate were used as the carbon sources.

Fig. 6. Influence of the initial pH value of the medium on lovastatin (determined as mevinolinic acid) formation; in the right panel smaller symbols connected with dashed lines denounce the samples alkalised prior to assay; the open circles denounce mevinolinic acid concentration calculated from Eq. (8).

M. Bizukojc et al. / Journal of Biotechnology 162 (2012) 253–261

259

Fig. 7. Mass spectra of mevinolinic acid (a), monacolin L (b), monacolin L lactone (c) and (+)-geodin (d); the metabolites were found in the sample from 96 hour of the run at the initial pH value of the broth equal to 6.5; with the ovals the masses of the protonated forms of the aforementioned molecules were depicted.

3.4. Formation of other metabolites by A. terreus at the various initial pH values There are a variety of metabolites formed by A. terreus. The long list of them can be found on the official webpage of Aspergillus (www.aspergillus.org.uk). It must be however remembered that whether the given metabolite is formed depends mainly on the strain and culture conditions. The studied ATCC20542 strain is a lovastatin producer. Despite this, several other than lovastatin metabolites can be detected in the cultivation broth. Previously, Bizukojc and Ledakowicz (2007) described the formation of an octaketide (+)-geodin, which accompanied lovastatin. They also found the traces of demethylated (+)-geodin, namely (+)-erdin, when A. terreus was cultivated on the media containing two carbon sources glycerol and lactose (Pawlak and Bizukojc, 2012). Remembering about this potential of A. terreus, the analysis of metabolites upon mass spectra in the function of the initial pH value of the medium was also performed. The samples from 96 h of the run were used. In Fig. 7 the mass spectra of the recognised secondary metabolites of A. terreus are presented. These are respectively mevinolinic acid, monacolin L, monacolin L (lactone) and (+)-geodin. First, a well-known spectrum of mevinolinic acid (lovastatin) shall be discussed (Fig. 7a). This spectrum fits well with the fragmentation pathway for lovastatin presented in the study of Wang et al. (2001). The highest peak (M+H)+ at m/z = 405.2600 [(m/z) = +0.0041] was attributed to a protonated molecule of lovastatin lactone, which in this case may be treated as a fragment despite its dominating intensity as actually it was mevinolinic acid that eluted at tR = 8.818 min (Table 2). The addition of H2 O to this molecule resulted in the formation of the peak (M+H2 O+H)+ at m/z 423.2716 [(m/z) = −0.0031] forming a protonated molecule of mevinolinic acid. Elimination of the ester side-chain [diketide (2R)-2-methylbutyric acid] from mevinolinic acid led to a fragment with a corresponding peak at m/z = 321.2094 [(m/z) = +0.0028] and from lovastatin lactone led to a fragment with the corresponding peak at m/z = 303.1969 [(m/z) = +0.0009]. The latter ion was

subsequently subjected to further fragmentation giving rise to a number of ions at m/z: 285.1876, 267.1725, 243.1756, 225.1644, 201.1625, 199.1478 and 173.1322. All the peaks are neighboured by the minor M + 1 peaks contributed by the equivalent molecule structures with incorporated 13 C isotope. With regard to monacolin L the situation was more complicated (Fig. 7b and c). The group of peaks at m/z 287.2050, 305.2159 and 323.2245, which are the trace of this metabolite, were observed for two distinct retention time values, i.e. tR = 6.50 min and tR = 9.78 min (Table 2). It indicated the presence of isomers that differed in polarity. The peak at m/z = 323.2245 [(m/z) = +0.0023] detected at tR = 6.5 min could be attributed to the protonated ␤-hydroxy acid form of monacolin L (M+H)+ , while its derivatives (M-2H2 O + H)+ and (M-H2 O + H)+ are responsible for the peaks m/z = 287.2050 [(m/z) = +0.0039] and 305.2159 [(m/z) = +0.0042], respectively. The latter ion corresponds to monacolin L lactone, which could be called a fragment at this elution time. The peak at m/z = 305.2159 found at t = 9.8 min corresponded to the lactone form of monacolin L (M+H)+ , which gave rise to daughter ions (M-H2 O + H)+ at m/z = 287.2050 and (M+H2 O+H)+ at m/z = 323.2159. The latter is of course ␤-hydroxy acid form of monacolin L. Again, the presence of minor M + 1 peaks belonging to the ions containing 13 C isotope was observed. It is clear that the peak at m/z = 305.2050 contributed by monacolin L lactone was not equivalent to the peak at m/z = 303.1969, which corresponded to lovastatin lactone molecule upon the elimination of its ester side-chain. The fragment of lovastatin lactone was deprived of two hydrogen atoms when compared to monacolin L. Therefore, the resulting ions were different. They showed the distinct mass peaks and this way monacolin L was distinguished from lovastatin. The discrimination between lactone and acid form of monacolin L was performed upon their retention times. Lactones, as less polar compounds, always elute later, when more organic phase is added, from the chromatographic columns with octadecyl ligands. An octaketide (+)-geodin (Fig. 7d) is a chlorine-containing compound and one may expect a characteristic pattern in the

260

M. Bizukojc et al. / Journal of Biotechnology 162 (2012) 253–261

Table 2 Relative amounts of metabolites obtained at 96 h of the process at the various initial pH values of the medium. Initial pH of the medium

3.5 4.5 5.5 6.5 7.5 8.5 9.5

Mevinolinic acid tR = 8.818 min m/z = 423.2716

Monacolin L (acidic form) tR = 6.500 min m/z = 323.2245

Monacolin L (lactone form) tR = 9.780 min m/z = 305.2159

(+)-Geodin tR = 8.200 min m/z = 399.0057

Peak intensity

Relative amount

Peak intensity

Relative amount

Peak intensity

Relative amount

Peak intensity

Relative amount

12 299 568 7869 9935 7629 7446

0.001 0.038 0.072 1.00 1.26 0.96 0.95

3857 5147 2507 11366 12508 10569 8115

0.34 0.45 0.22 1.00 1.10 0.93 0.71

0 0 0 2964 3420 3406 3368

0 0 0 1.00 1.15 1.15 1.14

0 0 0 1077 87 0 0

0 0 0 1.00 0.081 0 0

mass spectrum resulting from the relative abundance of two stable chlorine isotopes, 35 Cl and 37 Cl. Indeed, the peak corresponding to a protonated (+)-geodin molecule at m/z 399.0057 [(m/z) = +0.0019] was accompanied by the peaks at m/z 401.0064 [(m/z) = +0.0055] and 403.0032 [(m/z) = +0.0053]. Furthermore, the other analysed peaks resulting from (+)-geodin fragmentation followed the same M, M + 2, M + 4 pattern, what illustrated the presence of two chlorine atoms in these fragments. Loss of COOCH3 group from (+)-geodin resulted in the abundant peak at m/z 339.9911 [(m/z) = +0.0006]. The peak at m/z 366.9729 could be associated with the elimination of OCH3 group together with one hydrogen atom from the molecule of (+)-geodin, leading to a loss of a molecule equivalent to methanol (CH3 OH). The peak at m/z 355.0138 corresponded to the elimination of CO2 from (+)-geodin, but the structure of the fragment was unknown. Under these conditions (lactose as the sole carbon source at various initial pH levels) only the traces of (+)-erdin (m/z = 384.9900) at pH 6.5 were found. It was not enough to discuss the spectrum of this metabolite here. In Table 2 the relative amounts of the aforementioned metabolites were collected too. It was impossible to determine their concentrations due to lack of standard of monacolin L. So it was assumed that the amount of mevinolinic acid and other metabolites formed at pH 6.5, believed to be optimum for lovastatin biosynthesis, was equal to 1. It is clearly seen in Table 2 that monacolin L in the acidic form was present in the relatively higher amounts at the pH value below 5.5. But at pH 9.5 relatively less monacolin L is formed than mevinolinic acid. The initial pH value had little impact on the presence of monacolin L lactone, nevertheless it was surprisingly not detected at pH lower than 6.5. The reason that monacolin L was revealed in the broth must be shortly discussed. This metabolite occurs in the first step of post-PKS tailoring of 4a,5-dihydromonacoline L, which is the final product of the action of lovastatin nonaketide synthase (LNKS). 4a,5-Dihydromonacoline L is first hydroxylated (oxidised) to monacoline L. Monacoline L should be then again oxidised to monacoline J, which finally is transesterified with (2R)-2-methylbutyric acid to form mevinolinic acid. In this study probably the second oxidation step must have failed and a certain pool of monacolin L was released from the cells. Nevertheless, some scientists claim that in A. terreus monacolin L should not be excreted (Treiber et al., 1989). (+)-Geodin formation was strongly pH-dependent. Practically, only at the initial pH value of the medium equal to 6.5 it was found in the broth at low amounts. Nevertheless, it must be mentioned that the used medium with lactose only (without glycerol) and cultivation in shake flasks did not favour the formation of (+)-geodin (Bizukojc and Pecyna, 2011).

4. Conclusions On the basis of the presented data the following conclusions can be drawn:

1. It is very important to measure the pH value of the cultivation broth in the biosynthesis of mevinolinic acid (lovastatin) by A. terreus, especially if one uses a convenient direct method of its assay in the broth. If the pH value of the sample is acidic, mevinolinic acid concentration can be underestimated. 2. The pH value is a strong factor influencing substrate utilisation by A. terreus. Rapid lactose uptake is profitable for lovastatin production, therefore the neutral and alkalic pH give better lovastatin titres than the acidic one. 3. Although inefficient, the production of lovastatin at the low pH value is still observed. It is detectable provided the samples are alkalised prior to the assay. 4. Although the initial pH value of the medium at 6.5 is generally believed to be optimum for lovastatin production by A. terreus, it is worth considering its increase to 7.5 or even higher as it is possible to achieve higher lovastatin titres. It also justifies pH control in the culture at the levels above 7.0. 5. The initial pH value of the medium influences biosynthesis of other polyketide metabolites of A. terreus, nevertheless this effect is weaker compared to mevinolinic acid.

Acknowledgement The authors wish to acknowledge National Science Centre (Republic of Poland) for the partial financial support of this work, project no. N N209 765240.

References Bizukojc, M., Ledakowicz, S., 2007. Simultaneous biosynthesis of (+)-geodin by a lovastatin-producing fungus Aspergillus terreus. Journal of Biotechnology 132, 453–460. Bizukojc, M., Ledakowicz, S., 2008. Biosynthesis of lovastatin and (+)-geodin by Aspergillus terreus in batch and fed-batch culture in the stirred tank bioreactor. Biochemical Engineering Journal 42, 198–207. Bizukojc, M., Pecyna, M., 2011. Lovastatin and (+)-geodin formation by Aspergillus terreus ATCC 20542 in a batch culture with the simultaneous use of lactose and glycerol as carbon sources. Engineering in Life Sciences 11, 272–282. Casas Lopez, J.L., Sanchez Perez, J.A., Fernandez Sevilla, J.M., Acien Fernandez, F.G., Molina Grima, E., Chisti, Y., 2003. Production of lovastatin by Aspergillus terreus: effects of the C N ratio and the principal nutrients on growth and metabolite production. Enzyme and Microbial Technology 33, 270–277. Casas Lopez, J.L., Sanchez Perez, J.A., Fernandez Sevilla, J.M., Rodriguez Porcel, E.M., Chisti, Y., 2005. Pellet morphology, culture rheology and lovastatin production in cultures of Aspergillus terreus. Journal of Biotechnology 116, 61–77. Friedrich, J., Zuzek, M., Bencina, M., Cimerman, A., Strancar, A., Radez, I., 1995. Highperformance liquid chromatographic analysis of mevinolin as mevinolinic acid in fermentation broths. Journal of Chromatography A 704, 363–367. Hajjaj, H., Niederberger, P., Duboc, P., 2001. Lovastatin biosynthesis by Aspergillus terreus in a chemically defined medium. Applied and Environmental Microbiology 67, 2596–2602. Jia, Z., Zhang, X., Zhao, Y., Cao, X., 2009. Effects of divalent metal cations on lovastatin biosynthesis from Aspergillus terreus in chemically defined medium. World Journal of Microbiology and Biotechnology 25, 1235–1241. Kumar, M.S., Jana, S.K., Senthil, V., Shashanka, V., Kumar, S.V., Sadhukhan, A.K., 2000. Repeated fed-batch process for improving lovastatin production. Process Biochemistry 36, 363–368.

M. Bizukojc et al. / Journal of Biotechnology 162 (2012) 253–261 Lai, L.-S.T., Hung, C.S., Lo, C.C., 2007. Effects of lactose and glucose on production of itaconic acid and lovastatin by Aspergillus terreus ATCC20542. Journal of Bioscience and Bioengineering 104, 9–13. Lai, L.-S.T., Pan, C.C., Tzeng, B.K., 2003. The influence of medium design on lovastatin production and pellet formation with a high producing mutant of Aspergillus terreus in submerged cultures. Process Biochemistry 38, 1317–1326. Lai, L.-S.T., Tsai, T.-H., Wang, T.C., 2002. Application of oxygen vectors to Aspergillus terreus cultivation. Journal of Bioscience and Bioengineering 94, 453–459. Lai, L.-S.T., Tsai, T.-H., Wang, T.C., Cheng, T.Y., 2005. The influence of culturing environments on lovastatin production by Aspergillus terreus in submerged cultures. Enzyme and Microbial Technology 36, 737–748. Liu, G., Xu, Z., Cen, P., 2000. A morphologically structured model for mycelial growth and secondary metabolite formation. Chinese Journal of Chemical Engineering 8, 46–51. Manzoni, M., Rollini, M., Bergomi, S., Cavazzoni, V., 1998. Production and purification of statins from Aspergillus terreus strains. Biotechnology Techniques 12, 529–532. Morrison, R.T., Boyd, R.N., 1992. Organic Chemistry, sixth ed. Prentice Hall Inc., Englewood Cliffs, New Jersey. Novak, N., Gerdin, S., Berovic, M., 1997. Increased lovastatin formation by Aspergillus terreus using repeated fed-batch process. Biotechnology Letters 19, 947–948. Osman, M.E., Khattab, O.H., Zaghlol, G.M., Abd El-Hameed, R.M., 2011. Optimization of some physical and chemical factors for lovastatin productivity by local strain of Aspergillus terreus. Australian Journal of Basic and Applied Sciences 5 (6), 718–732.

261

Paul, G.C., Thomas, C.R., 1996. A structured model for hyphal differentiation and penicillin production using Penicillium chrysogenum. Biotechnology and Bioengineering 51, 558–572. Pawlak, M., Bizukojc, M., 2012. Intensification of lovastatin biosynthesis with the use of lactose and glycerol as carbon sources in a batch and in discontinuous fed-batch and continuous fed-batch culture. (in Polish). Chemical Engineering and Equipment (Inzynieria i Aparatura Chemiczna) 51 (4), 164–166. Pawlak, M., Bizukojc, M., Ledakowicz, S., 2012. Impact of bioreactor scale on lovastatin biosynthesis by Aspergillus terreus ATCC 20542 in a batch culture. Chemical and Process Engineering 33 (1), 71–84. Rodriguez Porcel, E.M., Casas Lopez, J.L., Sanchez Perez, J.A., Chisti, Y., 2007. Enhanced production of lovastatin in a bubble column by Aspergillus terreus using a twostage feeding strategy. Journal of Chemical Technology and Biotechnology 82, 58–64. Rodriguez Porcel, E.M., Casas Lopez, J.L., Sanchez Perez, J.A., Chisti, Y., 2008. Lovastatin production in a two-staged feeding operation. Journal of Chemical Technology and Biotechnology 83, 1236–1243. Rollini, M., Manzoni, M., 2006. Influence of medium design on lovastatin and mevastatin production by Aspergillus terreus strains. Annals of Microbiology 56 (1), 47–51. Sun, H., Gong, J.-B., Wang, J.K., 2005. Solubility of lovastatin in acetone, methanol, ethanol, ethyl acetate, and butyl acetate between 283 K and 323 K. Journal of Chemical and Engineering Data 50 (4), 1389–1391. Treiber, L.R., Reamer, R.A., Rooney, C.S., Ramjit, H.G., 1989. Origin of monacolin L from Aspergillus terreus cultures. Journal of Antibiotics 42 (1), 30–36. Wang, H., Wu, Y., Zhao, Z., 2001. Fragmentation study of simvastatin and lovastatin using electrospray ionization tandem mass spectrometry. Journal of Mass Spectrometry 36, 58–70.