- Email: [email protected]
A new NMR airlift bioreactor used in 31P-NMR studies of itaconic acid producing Aspergillus terreus
Journal of Biochemical and Biophysical Methods, 27 (1993) 105-116 Elsevier Science Publishers B.V.
105
JBBM01011
A new NMR airlift bioreactor used in 31p_NMR studies of itaconic acid producing Aspergillus terreus Magnus Lyngstad 1 and Hans Grasdalen
Department of Biotechnology, University of Trondheim, Trondheirn (Norway) (Received 11 January 1993) (Accepted 31 March 1993)
Summary An airlift bioreactor for in-vivo NMR studies of cells is described. The 10-mm diameter airlift reactor was constructed for studies of mycelial/pellet forming organisms, grown in suspension. With this device 161 MHz 31p-NMR spectra of living Aspergillus terreus cells, producing itaconic acid, have been obtained. Signals were observed for intra- and extracellular orthophosphate, glycerol-3-phosphorylethanolamine (GPE), glycerol-3-phosphorylcholine(GPC), sugar phosphates and polyphosphate. The spectra also showed broad overlapping resonances in the shift range of NAD(H) and NADP(H). Polyphosphate disappeared when the respiratory gas was exchanged for pure N2. The intracellular pH was estimated at 6.2. In spectra of cell extracts approx. 60 peaks were observed in the range of 20 to - 22 ppm, and they confirmed the appearance of the metabolites observed in living cells. Key words: NMR bioreactor; NMR, in vivo; Fungus; Itaconic acid; Polyphosphate
Introduction T h e complex physiology of fungi creates p r o b l e m s in studies of their b i o c h e m istry. T h e use of N M R might solve some of these problems. A t its best, a n N M R s p e c t r u m of living cells provides a n o n - i n v a s i v e analysis, yielding m u c h i n f o r m a tion. T o date, the n u c l e u s s t u d i e d most in in-vivo m e t a b o l i c studies is 31p. It has good receptivity for N M R studies, n o expensive e n r i c h m e n t is n e e d e d , a n d since t h e r e
Correspondence address: M. Lyngstad, Nycomed Pharma AS, FoU 5, Nycovegen 2, 0410 Oslo, Norway. 1 Present address: Nycomed Pharma AS, FoU 5, Nycovegen 2, 0410 Oslo, Norway.
106 are relatively few phosphorylated metabolites present in cells in mM concentrations, it usually is relatively easy to assign the observed resonances to specific metabolites [1]. Many phosphorus metabolites carry charged phosphate groups and as a result the chemical shifts of their phosphorus nuclei are usually a function of pH in the physiological range. 3~P-NMR studies have, during the last decade, provided much information on living cells of all kinds. Matsunaga et al. [2] obtained 31P-NMR spectra of intact cells and HC104-extracted cells of a filamentous fungus, Penicillium ochro-chloron, at a frequency of 40.3 MHz. The spectra showed resonances assigned to sugar phosphates (SP), orthophosphate (Pi), glycerol-3-phosphorylethanolamine (GPE), glyeerol-3-phosphorylcholine (GPC), ATPPa, U D P G and polyphosphates. The intracellular p H was estimated to be 6.4 in the logarithmic phase of growth. The 31p spectra reflected the age of the fungus and showed the accumulation of polyphosphate in the stationary phase. Legisa and Kidric [3] used 31p-NMR to measure pH in citric acid-producing Aspergillus niger. They were able to detect a drop in intracellular pH from 7.1 to about 6.5, during the early stages of growth. The main problem with in-vivo N M R is low sensitivity due to limited native concentrations of metabolites, which means that high cell densities or long sampling times are needed. This makes in-vivo studies of aerobic cultures difficult, and for this reason, biological NMR studies are most often done on resting cells or cell extracts. Some of the unique advantages of N M R are thereby lost. For many years perfusion N M R bioreactors have been used. These are efficient tools, but costly and difficult to produce. Recently Santos and Turner [4] designed an airlift bioreactor with three major advantages: (1) it is easy to make and to use; (2) the effective relaxation times are reduced because of the fluid flow, which improves the optimum sensitivity; and (3) it provides an efficient supply of respiratory gas. For studies of fungi the small dimensions (10-mm diameter) of Santos-Turner reactor is a problem. A large pellet can be 2-3 mm wide, which is the same dimension as the riser in their airlift. Kramer and Baily [5] built an airlift reactor geometrically similar to that of Santos and Turner, but in a 20-mm N M R tube for which a wide bore spectrometer system is needed. This is also the case for the N M R bioreactor recently described by D e G r a a f et al. [6] for use in in-vivo studies of microbial cell suspensions with low biomass concentrations. Our airlift bioreactor is a simplification of the Santos-Turner 10-mm reactor; it can be used in a narrow bore magnet, but at the same time it gives more space. It was designed for studies of itaconic acid production by a strain of Aspergillus terreus as reported in this study. Itaconic acid is an unsaturated dicarboxylic acid (HEC=C(CO2H)-CH2-CO2 H) used as a precursor for synthetic polymers and greases. The production in 1985 was estimated at a few thousand tons [7]. Even though its production has been studied since the early 1940s, its biosynthesis and the conditions required for it are not well understood. Itaconic acid is produced under conditions of stress by high sugar concentrations and low pH. Many studies also indicate the importance of specific trace-metal concentrations for optimal production [7,8]. Our unpublished results show a positive cooperative effect between zinc, iron and probably man-
107 ganese. Further Larsen [10] and others [7] showed that even a short stop in aeration of the A. terreus culture would lead to a complete cessation of itaconic acid production.
Materials and Methods
Microorganism Aspergillus terreus NRRL 1960 was maintained on Czapek-Dox agar plates [11]. Spore inocula were prepared by harvesting fresh spores from 30-day-old plates and suspending them in sterile 0.002% Tween 20 solution in distilled water (approx. 5 ml) contained in a Potter-Elvehjem homogeniser. Portions (0.2 ml) of spore suspension, containing 2-6" 107 spores/ml, were transferred to shake-flasks containing 100 ml medium. The flasks were incubated in a reciprocating shaking machine at 32.5°C and 160 rpm. The culture was grown for 65-75 h, giving a concentration of 4-10 g itaconic acid/1 just before the culture was used. Medium A modification of the medium described by Larsen [10] was used. The composition in g / l was: sucrose, 100; .(NH4)2SO4, 3.00; MgSOn.7H20, 0.50; KHzPO 4, 0.60 and CaSO 4 (P.A. salt, ignited at 530°C for 18 h) 0.3, in tap water with no adjustment of pH. To the medium were also added the following trace-metals, in mg/l: Z n S O 4 . 7 H 2 0 , 0.44; F e S O 4 . 7 H 2 0 , 0.30 and MnSO4.7H20, 0.025. Portions (100 ml) of the medium in 500 ml shake-flasks were sterilized at 121°C for 20 min. The phosphorus concentration was six times as high as normal, giving a reduced production speed. This condition gave pellet diameters of 0.2-2.0 mm, and a final concentration of 32 + 5 g itaconic acid/1 and cell concentration of approx. 10 g dry cell wt./l. The airlift reactor The reactor consisted of a 10-mm NMR tube divided into two vertical compartments by a glass plate 0.9 mm thick (Fig. 1). The glass plate was concave in the bottom to facilitate the flow of ceils. The liquid height was 100 mm. The respiratory gas was supplied from a RENA 301 membrane-pump through an Assistent 10-/xl micropipette at a rate of 50-85 ml/min, approx. 10 mm above the detector coil. The flow thereby generated in the sample resulted in a mean residence time of cells in the sensitive volume of the detector coil of about 0.3 s, every 8-10 s. For each fermentation, about a half-drop of Dow Corning 1510 silicon antifoam was added to the reactor, and then approx. 5 ml of culture in early production phase was transferred from a shaker-flask. To confirm that adequate oxygen transfer was taking place in the bioreactor, the production of itaconic acid was compared with that in a shake-culture. A portion
108
180 mm
A
100 mrn Flow pattern
80rnm
40ram
4.5 mm
Fig. 1. The NMR airlift bioreactor consists of the following parts: (A) 10-ram NMR tube with perforated top, (B) glass plate curved upwards at the bottom, (C) 10-/zl micropipette, (D) plastic tubing. The stream of bubbles rises at one side of the glass plate and generates a flow of liquid in the direction indicated by the arrows.
(5 ml) of itaconic acid producing culture was transferred to the bioreactor in the constant-temperature room where the shake-culture was grown. The entire process took about 30 s. The shaker-flask was immediately put back in the reciprocating shaking machine, and the bioreactor was kept in the same constant t e m p e r a t u r e room. Even with a relative humidity of 80-90%, water evaporated from the airlift, and had to be replenished from time to time. After 12 h both cultures were harvested, the cells were removed, and the culture filtrates were diluted appropriately (0.1-0.5 m M / l ) and analyzed by HPLC. In the N M R experiments the culture was transferred to the bioreactor in the N M R room. The flask was shaken continuously until transfer, and the air supply was connected immediately afterwards. T h e cells were kept at appropriate temperature through body heating during transfer, taking 2 - 3 min, and by the thermostat of the spectrometer during the N M R measurements. Prior to the N M R experiments about 0.8 ml D 2 0 was added to each N M R reactor to give a field-frequency lock-signal. W h e n pure oxygen was supplied to the bioreactor to increase oxygen transfer, the gas was bubbled through a water bath, to decrease subsequent evaporation
109 from the bioreactor. The same method was used when N z was introduced to kill the cells.
NMR analysis All 31p-NMR spectra were recorded at 23°C with proton noise decoupling on a J E O L JNM-EX400 F T N M R spectrometer operating at 161 M H z for 31p. For in-vivo spectra the aquisition parameters were: 60 ° pulse angle, 1.1 s pulse repetition time, 14000 scans and 16 K data points. For cell extract spectra the pulse repetition time was 5 s and 65 K data points were used. The spectral width was 8 kHz in both types of spectra. Chemical shifts were given by using methylenediphosphonate at 16.2 p p m as an internal reference in the cell extract. The living cells contained G P C with a resonance at - 0 . 1 3 p p m [12], which served as a standard for the chemical shift measurement. In order to use the chemical shift of Pi as a p H indicator, the dependence of Pi shift upon p H in the production medium was measured. It contained approx. 20 g itaconic acid/l, giving an initial p H of 3 after addition of K 2 H P O 4 to 5 m M concentration. The medium p H was adjusted by adding K O H and measured with a pH-sensitive glass electrode before and after determination of the Pi chemical shift with NMR.
HPLC analysis Itaconic acid was estimated by H P L C with a Shimadzu LC-9A pump, SIL-9A autoinjector, C T O - 6 A column oven, SPD-6AV U V / V I S spectrophotometric detector, and C - R 4 A integrator. The column was a H P X - 8 7 H from B I O - R A D . Sulphuric acid (5 m M ) was used as buffer with a flow of 0.6 m l / m i n . The oven was set at 45°C. Portions (10 /~1) of sample were injected and the absorbance of the effluent was measured at 210 nm. At this wavelength the C = C bond absorbs more strongly than the C = O bond, and the detection limit is low compared to most other acids. The optimum concentration for detection was 0.01-0.5 mM, and all samples were diluted to this level. Standards of 0.1 and 0.5 m M were analyzed for every 5 - 6 samples. From six identical standards the standard deviation was estimated to be + 15%.
Cell extraction 20 ml of cell culture was transferred to a 35-ml centrifuge tube to which pure trichloroacetic acid (TCA) had previously been added to 5% (w/v), to give a final p H of 0.5. The c u l t u r e / T C A solution was immediately sonicated continuously for 15 min on an MSE-sonicator (Mk 2). The sample was cooled in wet ice, so that the t e m p e r a t u r e at the end of sonication was approx. 25°C. The sonicated extract was centrifuged at 34000 x g for 10 min, at 4°C, then E D T A (Titriplex III) was added to the soluble part to a final concentration of 100 mM. The solution was then neutralized with 5 M K O H to p H 7.4, and concentrated 10-fold on a Biichi rotavapor-R, at 25°C. The cellular extract was analyzed in a 5-mm N M R tube.
110 TABLE 1 Itaconic acid produced in the NMR airlift bioreactor and in the reference shaker-flask during a 12-h period. Itaconic acid produced (mM) a Case: Reference shaker-flask N M R airlift bioreactor
1
2
3
19.2 11.6 b
31.0 22.9 c
26.0 23.6 d
a All cultures were 66-h old when transferred. The itaeonic acid concentration when transferred varied from 50 to 70 mM. b Air was supplied to the N M R tube 3 cm from the bottom. c Air was supplied to the N M R tube at the bottom. d Pure 0 2 was supplied to the N M R tube 3 cm from the bottom.
Results
Production Table 1 compares the production of itaconic acid in the N M R airlift bioreactor outside the spectrometer with that in a shaker-flask over a 12-h period. Itaconic acid production is highly sensitive to anaerobic conditions. When aeration was stopped for 10-15 min and then re-started no increase in itaconic acid concentration could be detected within the next 24 h (results not shown). NMR airlift bioreactor In a 10 mM Pi and 3 mM A T P aqueous solution the spectral lines were 11 Hz wide. A pulse repetition time of 1.024 s seemed to be optimal, while a pulse repetition time of 0.624 s gave a slight decrease in the Pi/ATPa ratio (results not shown), indicating a saturation of the Pi which relaxes more slowly. With cells in the airlift bioreactor there was a considerable broadening of the resonances because of non-homogeneous samples and cationic interference. In the best spectra, line widths of about 40-50 Hz for intra- and extracellular Pi were observed, while the sharpest polyphosphate resonance had a width of 75 Hz. In the spectra of cellular extracts, the Pi resonance had a width of 8 Hz, measured in a spinning 5-mm N M R tube. To sharpen the lines, E D T A was added to a final concentration of 100 mM, as this prevented divalent cations from interacting with the various phosphates. In extracts without addition of E D T A most phosphorus resonances were hardly detectable because of line-broadening. 31p spectra Fig. 2 shows a 31p-NMR spectrum of living A. terreus cells in the beginning of itaconic acid production phase, and with a stream of pure 0 2 bubbles in the reactor. Several broad signals are observed in the chemical shift range from 5 to - 24 ppm. The spectrum is not identical but most of the signals are very similar to those in the spectrum of Penicillium ochro-chloron cells reported by Matsunaga et
111 pe×
GPE~/GPC SP
~-
i
i
,
i
i
51
Pt~
PoI"-P
1
_12
)
r
r-i
i
~
i
i
i
i _~
,
)
r
,
1_01 ,
,
i
F
~5
,
,
i
i
20
1
,
,
PPM , _~
Fig. 2. In-vivo 31p-NMR spectrum of itaconic acid producing .,4. t e r r e u s cells at 23°C in the presence of 02 . The spectrum is the sum of 15916 scans obtained with 60 ° pulses during an accumulation time of about 5 h. The spectral region from - 2 to 6 ppm is expanded to visualize the overlapped line pattern. The chemical shifts are referenced to glycerol-3-phosphorylcholine (GPC) at - 0 . 1 3 ppm.
al. [2]. The resonances were assigned on a basis of their chemical shifts and by comparing the spectra of the two types of fungal cells. According to the identification of resonances made by Matsunaga et al., the composite signal in the range 1 to - 0 . 2 ppm is due to GPC, Pi in the medium, GPE, and intracellular Pi at - 0 . 1 , 0.25, 0.5 and 0.8 ppm, respectively. The chemical shift range covered by the broad unresolved signals between - 1 0 and - 1 1 . 5 ppm, typical of NAD(H) and NADP(H), is more narrow than that observed in the spectrum of P. ochro-chloron which displayed two resonances between - 10 and - 13 ppm. The strong, well-defined signal at - 2 2 . 5 ppm arises from inorganic polyphosphate. This resonance was significantly detectable after approx. 200 scans (2.5 rain). When air or oxygen were exchanged for pure N 2 the polyphosphate peak disappeared completely after approx. 10 min (see Fig. 3). In an attempt to confirm the assignment, an analysis was made of the spectrum (Fig. 4) of cellular extract after addition of 100 mM EDTA. Approx. 60 resonances are observed in the chemical shift range from 20 to - 2 2 ppm. To check if any resonances arose from extracellular compounds, filtered and centrifuged growth media from cultures in the production phase were also examined. Only traces of inorganic phosphate could be detected (spectra not shown). The first extractions were done on cells concentrated before extraction and without EDTA. In the spectra of these, only Pi, GPC and G P E were detected; this was also the case with cells killed by supplying N 2. The glycerol-3-phosphoryl compounds, GPC and GPE, are known to exhibit resonances independent of the ions present in solution, and GPC is recommended as a good internal chemical shift standard for in-vivo 31p-NMR because its shift ( - 0 . 1 3 ppm) is also independent of pH down to pH 3
112
Pt
F~N -t5
-~
-~
Poly-P
iO
5
0
-5
-tO
-i5
-k~
-~
Fig. 3. 161-MHz 31P-NMR spectra of A. terreus cells during oxygenation (full spectrum) and after supply of N 2 for approx. 10 min, when the resonance of polyphosphate had disappeared (inset).
[13]. GPE has a pK a close to 9.7 and by increasing the pH to 11 the chemical shift separation between GPC and GPE increased by approx. 0.5 ppm; this is consistent with literature data on related phospholipids [14]. In the well resolved spectrum (Fig. 4), the resonances of ATP, GPC, AMP and GIc-6P were identical to those of added compounds (Sigma). The eight absorptions between - 1 1 and -13.5 ppm arise from two UDP sugars. In the high field region a strong resonance at -21.6 ppm comes from middle (or interior) phosphates of long-chain polyphosphates. The weaker and nearly equally intense lines at - 5 and -20.2 ppm arise from terminal phosphate groups (doublet) and penultimate phosphates (triplet), respectively [15]. In the low-field region of the spectrum, an absorption at 19.2 ppm indicates that a relatively large amount of a compound containing a 5-membered phosphate ring is present in the cellular extract [16]. In a proton coupled spectrum this resonance split into a 1 : 3 : 3 : 1 quartet pattern of separation approx. 15 Hz caused by 1H-alP spin-spin coupling, as shown in inset C, Fig. 4. In the spectrum of living cells this resonance was not observed. Discussion
The N M R airlift reactor The production experiment showed a decreased production when air was introduced 3 cm above the bottom of the NMR tube. When air was supplied at the
113
ATP /
C)
,
B)
AA
"Y
IMP+ AMP
~IV~
.... ,jV '~,4t,~...~W%"
II
• ~ I UDP-S
,
I / I,
'
'
I
2o
~
,
is
'
P-c i l~cyC
GIc-6P -
Ref.~
,
I
'
'
'
20
'
J,l'l.
GPC
~
I
'
5
r
l 11 terminal], _
,
,
,
,
{
0
'
'
'
'
I
-5
I
I,
,
,
,
I
-t0
'
penultimate
,
L
,
E
]
-i5
,
,
,
,
I
t
I
-20
Fig. 4. 161-MHz 31p-NMR spectrum of TCA/sonication extract of .4. terreus cells (pH 7.4). (B) After addition of ATP. (C) An expanded spectral region of the cyclic phosphate compound as obtained in a proton-coupled spectrum. The positions of the resonances from ATP, Glc-6P, AMP, and GPC were confirmed by addition of the authentic compounds (Sigma). The chemical shifts are referenced to internal trisodium-methylenediphosphonateat - 16.2 ppm.
bottom the production increased considerably. In order to optimize the spectral resolution it is essential to avoid gas bobbling through the receiver coil and this prohibited introduction of gas at the bottom of the bioreactor. Therefore, to obtain sufficient oxygen transfer, pure 0 2 was introduced while performing the N M R experiments as shown in Fig. 1. During the N M R experiments it was possible to maintain the polyphosphate for at least 24 h. The rapid disappearance of the polyphosphate resonance when 0 2 was exchanged for N 2, killing the cells, also indicated that the culture was alive during sampling.
The 31p spectra The similarity of the spectra of intact cells of P. ochro-chloron obtained by Matsunaga et al. [2] and of A. terreus cells obtained in this work, is striking. All signals correspond well except that the one from U D P G at -12.7 ppm is missing in the latter spectrum. The reason why the UDP sugars were not observed in intact ceils, while being clearly visible in the extract spectrum in this chemical shift region, was probably due to binding of divalent cations. The broadening of the resonances in the living cell spectrum was probably due to interactions with Ca 2+ and small amounts of paramagnetic ions such as Fe z+ and Mn 2+. In fact several expected resonances are not seen or are scarcely visible. The broadening of the
114 signals in the living cell spectrum is difficult to avoid, but a lower concentration of trace-metals should probably make an improvement. The appearance of GPC and GPE may indicate a trend. GPC is known to be an osmolyte in different mammalian cells, together with betaine, polyols and some amino acids [17,18]. In fungi, glycerol and other polyols are most abundant as osmolytes. The importance of GPC (and GPE) as osmolyte(s) in mycelial fungi is not elucidated to. the authors' knowledge. The role of polyphosphate as a metabolite is still under some discussion. Gottschalk [19] stated that polyphosphate is a phosphate reserve, while Roberts [20] stated that it has an important role in regulating and maintaining the energy charge under stress conditions. The rapid disappearance of polyphosphate in our study when 0 2 was exchanged for N 2 probably supports the second theory. From the extract spectrum an average chain length of approx. 20 for the polyphosphate was estimated by using intensities of signals of middle, penultimate and terminal phosphate units. The many resonances in the mono- and di-ester regions are typical of a 31p cell extract spectrum. AMP and IMP overlap independently of pH, but IMP rather than AMP is the major product from ATP breakdown. The relatively large ADP signals compared with those from ATP indicate that this has occurred during the cell extraction. The signal at 19.2 ppm probably arises from a 5-membered phosphate ring in a compound formed during cell extraction in the acid medium. The biologically expected 2',3' cyclic nucleotides are all known to give a doublet in a 1H-coupled 31p-NMR spectrum [16] and can be ruled out. The glycerol 1,2(cyclic) phosphate is in agreement with the chemical shift [16], and the quartet pattern in the 1H-coupled spectrum is compatible with a planar phosphate ring conformation. The observed coupling (15 Hz) is, however, larger than expected for an anti-clinal arrangement of the three protons, - C H 2 - O - P - O - C H - , in glycerol 1,2(cyclic) phosphate (6.3 Hz), as estimated from a Karplus-like relationship reported for a three-bond 1H-31P coupling [21]. Uncertainties about the~phosphate ring conformation and in the equation used to relate coupling constant to dihedral angle make it impossible to draw any conclusion about the identity of the cyclic phosphate compound from NMR data. The splitting of the signal centred at approx. 0.8 ppm in the living cell spectrum is probably caused by noise, and may represent a single resonance from intracellular Pi. The intracellular pH was estimated to be 6.2 using the chemical shift-pH titration curve. The pH value measured, 6.2, is quite acidic in biological terms and will probably affect the metabolism considerably. The accuracy in this type of pH evaluation is normally not better than ___0.1 pH unit unless a calibration is performed under specific cellular conditions [22]. However, due to the rather low signal-to-noise ratio and the strongly overlapping lines, the accuracy may be somewhat lower here. Nevertheless, the measured pH value corresponds well with those given by Matsunaga et al. [2] and by Legisa and Kidric [3]. The external Pi gave a signal at approx. 0.25 ppm indicating a pH of about 2 in the medium in which the pH measured with a pH-meter was 1.7. The small change in chemical shift induced by pH variation at
115
low pH, is another reason leading to inaccurate pH determination by 31p-NMR in this region.
Simplified description of the method and its applications An airlift bioreactor for in-vivo NMR studies of cells in suspension is described. With this device 31p-NMR spectra of living Aspergillus terreus cells, producing itaconic acid, have been measured. The bioreaetor is a draft-tube reactor driven by a respiratory gas. This gave good oxygen supply and culture mixing when pure 0 2 was applied. The reactor is well suited for NMR studies of mycelial and pellet forming organisms. Because of its simplicity it is also well suited for studies of bacteria, algae and tissue cells.
References 1 Vogel, H.J., Brodelius, P., Lilja, H and Lohmeier-Vogel, E.M. (1987) Nuclear magnetic resonance studies of immobilized cells. Methods Enzymol. 135, 512-528. 2 Matsunaga, T., Okubo, A., Yamazaki, S. and Toda, S. (1982) 31p Nuclear magnetic resonance analysis of Penicillium ochro-chloron. Biochem. Biophys. Res. Commun. 106, 596-601. 3 Legisa, M. and Kidric, J. (1989) Initiation of citric acid accumulation in the early stages of Aspergillus niger growth. Appl. Microbiol. Biotechnol. 31,453-457. 4 Santos, H. and Turner, D.L. (1986) Characterization of the improved sensitivity obtained using a flow method for oxygenating and mixing cell suspensions in NMR. J. Magn. Reson. 68, 345-349. 5 Kramer, H.W. and Bailey, J.E. (1991) Mass transfer characterization of an airlift probe for oxygenating and mixing cell suspensions in an NMR spectrometer. Biotechnol. Bioeng. 37, 205-209. 6 DeGraaf, A.A., Wittig, R.M., Probst, U., Strohhaecker, J., Schoberth, S.M., and Sahm, H. (1992) Continuous-flow NMR bioreactor for in-vivo studies of microbial cell suspensions with low biomass concentrations. J. Magn. Reson. 98, 654-659. 7 Milsom, P.E. and Meets J.L. (1985) in Comprehensive Biotechnology (Moo-Young, M., ed.), Vol. 3, pp. 681-700, Pergamon Press. 8 Kautulla, H., Rymowics, W., Linko, Y. and Linko, P. (1991) Itaconic acid production by immobilized Aspergillus terreus with varied metal additions. Appl. Microbiol. Biotechnol. 35, 154-158. 9 Lockwood, L.B. (1979) Production of Organic Acids by Fermentation. In Microbiol. Technology (Peppier, H.J. and Perlman, I., eds.), Vol. 1, pp. 367-373, Academic Press, NewYork. 10 Larsen, H. (1957) Itaconic acid and related compounds. In Methods in Enzymology (Colowick, S.P. and Kaplan, N.O., eds.) Vol. 3, pp. 437-444, Academic Press, New York. l l Thom, C. and Raper, K.B (1945) A Manual of the Aspargilli, 32 pp., The Williams & Wilkins Co., Baltimore. 12 Burt, C.T., Gionek, T. and Barany, M. (1976) Phosphorus-31 nuclear magnetic resonance detection o f unexpected phosphodiesters in muscle. Biochemistry 15, 4850-4853. 13 Barany, M. and Glonek, T. (1982) Phosphorns-31 nuclear magnetic resonance of contractile systems. Methods Enzyrnol. 85, 624-676. 14 London, E. and Feigenson, G.W. (1979) Phosphorus-31 NMR analysis of phospholipids in detergents. J. Lipid Res., 20, 408-412. 15 Navon, G., Schulman, R.G., Yamane, T., Eccleshall, T.R., Ram, K.-B., Baronofsky, J.J. and Marmur, J. (1979) Phosphorus-31 nuclear magnetic resonance studies of wild-type glycolytic pathway mutants of Saccharmyces cerevisia. Biochemistry 18, 4487-4499. 16 Brown, T.R., Graham, R.A., Szwergold, B.S., Thoma, W.J. and Meyer R.A. (1987) Phosphorylated Metabolites in Tumors, Tissues, and Cell Lines. Ann. N.Y. Acad. Sci. 508, 229-240. 17 Yancey, P.H. and Burg, M.B. (1989) Distribution of major organic osmolytes in rabbit kidneys in diuresis and antidiuresis. Am. J. Physiol. 257, 602-607.
116 18 Nakanishi, T., Balban, B.S. and Burg, M.B. (1988) Survey of osmolytes in renal cell lines. Am. J. Physiol. 255, 181-191. 19 Gottschalk, G. (t988) in Bacterial Metabolism 2nd edn., pp. 130-131, Springer-Verlag, New York. 20 Roberts, M.F. (1987) Polyphosphates. In Phosphorus NMR in Biology (Burt C.T.,ed.), pp. 85-94, CRC Press, Boca Raton, Florida. 21 Blackburn, B.J., Lapper, R.D. and Smith, I.C.P. (1973) A Proton Magnetic Resonance Study of the Conformations of 3',5'-Cyclic Nucleotides. J. Am. Chem. Soc. 95, 2873-2878. 22 Gadian, D.G., Radda, G.K., Richards, R.E. and Seeley, P.J. (1979) 31p NMR in living tissue: the road from a promising to an important tool in biology. In Biological Applications of Magnetic Resonance (Shulman, R.G., ed.), pp. 463-529. Academic Press.
105
JBBM01011
A new NMR airlift bioreactor used in 31p_NMR studies of itaconic acid producing Aspergillus terreus Magnus Lyngstad 1 and Hans Grasdalen
Department of Biotechnology, University of Trondheim, Trondheirn (Norway) (Received 11 January 1993) (Accepted 31 March 1993)
Summary An airlift bioreactor for in-vivo NMR studies of cells is described. The 10-mm diameter airlift reactor was constructed for studies of mycelial/pellet forming organisms, grown in suspension. With this device 161 MHz 31p-NMR spectra of living Aspergillus terreus cells, producing itaconic acid, have been obtained. Signals were observed for intra- and extracellular orthophosphate, glycerol-3-phosphorylethanolamine (GPE), glycerol-3-phosphorylcholine(GPC), sugar phosphates and polyphosphate. The spectra also showed broad overlapping resonances in the shift range of NAD(H) and NADP(H). Polyphosphate disappeared when the respiratory gas was exchanged for pure N2. The intracellular pH was estimated at 6.2. In spectra of cell extracts approx. 60 peaks were observed in the range of 20 to - 22 ppm, and they confirmed the appearance of the metabolites observed in living cells. Key words: NMR bioreactor; NMR, in vivo; Fungus; Itaconic acid; Polyphosphate
Introduction T h e complex physiology of fungi creates p r o b l e m s in studies of their b i o c h e m istry. T h e use of N M R might solve some of these problems. A t its best, a n N M R s p e c t r u m of living cells provides a n o n - i n v a s i v e analysis, yielding m u c h i n f o r m a tion. T o date, the n u c l e u s s t u d i e d most in in-vivo m e t a b o l i c studies is 31p. It has good receptivity for N M R studies, n o expensive e n r i c h m e n t is n e e d e d , a n d since t h e r e
Correspondence address: M. Lyngstad, Nycomed Pharma AS, FoU 5, Nycovegen 2, 0410 Oslo, Norway. 1 Present address: Nycomed Pharma AS, FoU 5, Nycovegen 2, 0410 Oslo, Norway.
106 are relatively few phosphorylated metabolites present in cells in mM concentrations, it usually is relatively easy to assign the observed resonances to specific metabolites [1]. Many phosphorus metabolites carry charged phosphate groups and as a result the chemical shifts of their phosphorus nuclei are usually a function of pH in the physiological range. 3~P-NMR studies have, during the last decade, provided much information on living cells of all kinds. Matsunaga et al. [2] obtained 31P-NMR spectra of intact cells and HC104-extracted cells of a filamentous fungus, Penicillium ochro-chloron, at a frequency of 40.3 MHz. The spectra showed resonances assigned to sugar phosphates (SP), orthophosphate (Pi), glycerol-3-phosphorylethanolamine (GPE), glyeerol-3-phosphorylcholine (GPC), ATPPa, U D P G and polyphosphates. The intracellular p H was estimated to be 6.4 in the logarithmic phase of growth. The 31p spectra reflected the age of the fungus and showed the accumulation of polyphosphate in the stationary phase. Legisa and Kidric [3] used 31p-NMR to measure pH in citric acid-producing Aspergillus niger. They were able to detect a drop in intracellular pH from 7.1 to about 6.5, during the early stages of growth. The main problem with in-vivo N M R is low sensitivity due to limited native concentrations of metabolites, which means that high cell densities or long sampling times are needed. This makes in-vivo studies of aerobic cultures difficult, and for this reason, biological NMR studies are most often done on resting cells or cell extracts. Some of the unique advantages of N M R are thereby lost. For many years perfusion N M R bioreactors have been used. These are efficient tools, but costly and difficult to produce. Recently Santos and Turner [4] designed an airlift bioreactor with three major advantages: (1) it is easy to make and to use; (2) the effective relaxation times are reduced because of the fluid flow, which improves the optimum sensitivity; and (3) it provides an efficient supply of respiratory gas. For studies of fungi the small dimensions (10-mm diameter) of Santos-Turner reactor is a problem. A large pellet can be 2-3 mm wide, which is the same dimension as the riser in their airlift. Kramer and Baily [5] built an airlift reactor geometrically similar to that of Santos and Turner, but in a 20-mm N M R tube for which a wide bore spectrometer system is needed. This is also the case for the N M R bioreactor recently described by D e G r a a f et al. [6] for use in in-vivo studies of microbial cell suspensions with low biomass concentrations. Our airlift bioreactor is a simplification of the Santos-Turner 10-mm reactor; it can be used in a narrow bore magnet, but at the same time it gives more space. It was designed for studies of itaconic acid production by a strain of Aspergillus terreus as reported in this study. Itaconic acid is an unsaturated dicarboxylic acid (HEC=C(CO2H)-CH2-CO2 H) used as a precursor for synthetic polymers and greases. The production in 1985 was estimated at a few thousand tons [7]. Even though its production has been studied since the early 1940s, its biosynthesis and the conditions required for it are not well understood. Itaconic acid is produced under conditions of stress by high sugar concentrations and low pH. Many studies also indicate the importance of specific trace-metal concentrations for optimal production [7,8]. Our unpublished results show a positive cooperative effect between zinc, iron and probably man-
107 ganese. Further Larsen [10] and others [7] showed that even a short stop in aeration of the A. terreus culture would lead to a complete cessation of itaconic acid production.
Materials and Methods
Microorganism Aspergillus terreus NRRL 1960 was maintained on Czapek-Dox agar plates [11]. Spore inocula were prepared by harvesting fresh spores from 30-day-old plates and suspending them in sterile 0.002% Tween 20 solution in distilled water (approx. 5 ml) contained in a Potter-Elvehjem homogeniser. Portions (0.2 ml) of spore suspension, containing 2-6" 107 spores/ml, were transferred to shake-flasks containing 100 ml medium. The flasks were incubated in a reciprocating shaking machine at 32.5°C and 160 rpm. The culture was grown for 65-75 h, giving a concentration of 4-10 g itaconic acid/1 just before the culture was used. Medium A modification of the medium described by Larsen [10] was used. The composition in g / l was: sucrose, 100; .(NH4)2SO4, 3.00; MgSOn.7H20, 0.50; KHzPO 4, 0.60 and CaSO 4 (P.A. salt, ignited at 530°C for 18 h) 0.3, in tap water with no adjustment of pH. To the medium were also added the following trace-metals, in mg/l: Z n S O 4 . 7 H 2 0 , 0.44; F e S O 4 . 7 H 2 0 , 0.30 and MnSO4.7H20, 0.025. Portions (100 ml) of the medium in 500 ml shake-flasks were sterilized at 121°C for 20 min. The phosphorus concentration was six times as high as normal, giving a reduced production speed. This condition gave pellet diameters of 0.2-2.0 mm, and a final concentration of 32 + 5 g itaconic acid/1 and cell concentration of approx. 10 g dry cell wt./l. The airlift reactor The reactor consisted of a 10-mm NMR tube divided into two vertical compartments by a glass plate 0.9 mm thick (Fig. 1). The glass plate was concave in the bottom to facilitate the flow of ceils. The liquid height was 100 mm. The respiratory gas was supplied from a RENA 301 membrane-pump through an Assistent 10-/xl micropipette at a rate of 50-85 ml/min, approx. 10 mm above the detector coil. The flow thereby generated in the sample resulted in a mean residence time of cells in the sensitive volume of the detector coil of about 0.3 s, every 8-10 s. For each fermentation, about a half-drop of Dow Corning 1510 silicon antifoam was added to the reactor, and then approx. 5 ml of culture in early production phase was transferred from a shaker-flask. To confirm that adequate oxygen transfer was taking place in the bioreactor, the production of itaconic acid was compared with that in a shake-culture. A portion
108
180 mm
A
100 mrn Flow pattern
80rnm
40ram
4.5 mm
Fig. 1. The NMR airlift bioreactor consists of the following parts: (A) 10-ram NMR tube with perforated top, (B) glass plate curved upwards at the bottom, (C) 10-/zl micropipette, (D) plastic tubing. The stream of bubbles rises at one side of the glass plate and generates a flow of liquid in the direction indicated by the arrows.
(5 ml) of itaconic acid producing culture was transferred to the bioreactor in the constant-temperature room where the shake-culture was grown. The entire process took about 30 s. The shaker-flask was immediately put back in the reciprocating shaking machine, and the bioreactor was kept in the same constant t e m p e r a t u r e room. Even with a relative humidity of 80-90%, water evaporated from the airlift, and had to be replenished from time to time. After 12 h both cultures were harvested, the cells were removed, and the culture filtrates were diluted appropriately (0.1-0.5 m M / l ) and analyzed by HPLC. In the N M R experiments the culture was transferred to the bioreactor in the N M R room. The flask was shaken continuously until transfer, and the air supply was connected immediately afterwards. T h e cells were kept at appropriate temperature through body heating during transfer, taking 2 - 3 min, and by the thermostat of the spectrometer during the N M R measurements. Prior to the N M R experiments about 0.8 ml D 2 0 was added to each N M R reactor to give a field-frequency lock-signal. W h e n pure oxygen was supplied to the bioreactor to increase oxygen transfer, the gas was bubbled through a water bath, to decrease subsequent evaporation
109 from the bioreactor. The same method was used when N z was introduced to kill the cells.
NMR analysis All 31p-NMR spectra were recorded at 23°C with proton noise decoupling on a J E O L JNM-EX400 F T N M R spectrometer operating at 161 M H z for 31p. For in-vivo spectra the aquisition parameters were: 60 ° pulse angle, 1.1 s pulse repetition time, 14000 scans and 16 K data points. For cell extract spectra the pulse repetition time was 5 s and 65 K data points were used. The spectral width was 8 kHz in both types of spectra. Chemical shifts were given by using methylenediphosphonate at 16.2 p p m as an internal reference in the cell extract. The living cells contained G P C with a resonance at - 0 . 1 3 p p m [12], which served as a standard for the chemical shift measurement. In order to use the chemical shift of Pi as a p H indicator, the dependence of Pi shift upon p H in the production medium was measured. It contained approx. 20 g itaconic acid/l, giving an initial p H of 3 after addition of K 2 H P O 4 to 5 m M concentration. The medium p H was adjusted by adding K O H and measured with a pH-sensitive glass electrode before and after determination of the Pi chemical shift with NMR.
HPLC analysis Itaconic acid was estimated by H P L C with a Shimadzu LC-9A pump, SIL-9A autoinjector, C T O - 6 A column oven, SPD-6AV U V / V I S spectrophotometric detector, and C - R 4 A integrator. The column was a H P X - 8 7 H from B I O - R A D . Sulphuric acid (5 m M ) was used as buffer with a flow of 0.6 m l / m i n . The oven was set at 45°C. Portions (10 /~1) of sample were injected and the absorbance of the effluent was measured at 210 nm. At this wavelength the C = C bond absorbs more strongly than the C = O bond, and the detection limit is low compared to most other acids. The optimum concentration for detection was 0.01-0.5 mM, and all samples were diluted to this level. Standards of 0.1 and 0.5 m M were analyzed for every 5 - 6 samples. From six identical standards the standard deviation was estimated to be + 15%.
Cell extraction 20 ml of cell culture was transferred to a 35-ml centrifuge tube to which pure trichloroacetic acid (TCA) had previously been added to 5% (w/v), to give a final p H of 0.5. The c u l t u r e / T C A solution was immediately sonicated continuously for 15 min on an MSE-sonicator (Mk 2). The sample was cooled in wet ice, so that the t e m p e r a t u r e at the end of sonication was approx. 25°C. The sonicated extract was centrifuged at 34000 x g for 10 min, at 4°C, then E D T A (Titriplex III) was added to the soluble part to a final concentration of 100 mM. The solution was then neutralized with 5 M K O H to p H 7.4, and concentrated 10-fold on a Biichi rotavapor-R, at 25°C. The cellular extract was analyzed in a 5-mm N M R tube.
110 TABLE 1 Itaconic acid produced in the NMR airlift bioreactor and in the reference shaker-flask during a 12-h period. Itaconic acid produced (mM) a Case: Reference shaker-flask N M R airlift bioreactor
1
2
3
19.2 11.6 b
31.0 22.9 c
26.0 23.6 d
a All cultures were 66-h old when transferred. The itaeonic acid concentration when transferred varied from 50 to 70 mM. b Air was supplied to the N M R tube 3 cm from the bottom. c Air was supplied to the N M R tube at the bottom. d Pure 0 2 was supplied to the N M R tube 3 cm from the bottom.
Results
Production Table 1 compares the production of itaconic acid in the N M R airlift bioreactor outside the spectrometer with that in a shaker-flask over a 12-h period. Itaconic acid production is highly sensitive to anaerobic conditions. When aeration was stopped for 10-15 min and then re-started no increase in itaconic acid concentration could be detected within the next 24 h (results not shown). NMR airlift bioreactor In a 10 mM Pi and 3 mM A T P aqueous solution the spectral lines were 11 Hz wide. A pulse repetition time of 1.024 s seemed to be optimal, while a pulse repetition time of 0.624 s gave a slight decrease in the Pi/ATPa ratio (results not shown), indicating a saturation of the Pi which relaxes more slowly. With cells in the airlift bioreactor there was a considerable broadening of the resonances because of non-homogeneous samples and cationic interference. In the best spectra, line widths of about 40-50 Hz for intra- and extracellular Pi were observed, while the sharpest polyphosphate resonance had a width of 75 Hz. In the spectra of cellular extracts, the Pi resonance had a width of 8 Hz, measured in a spinning 5-mm N M R tube. To sharpen the lines, E D T A was added to a final concentration of 100 mM, as this prevented divalent cations from interacting with the various phosphates. In extracts without addition of E D T A most phosphorus resonances were hardly detectable because of line-broadening. 31p spectra Fig. 2 shows a 31p-NMR spectrum of living A. terreus cells in the beginning of itaconic acid production phase, and with a stream of pure 0 2 bubbles in the reactor. Several broad signals are observed in the chemical shift range from 5 to - 24 ppm. The spectrum is not identical but most of the signals are very similar to those in the spectrum of Penicillium ochro-chloron cells reported by Matsunaga et
111 pe×
GPE~/GPC SP
~-
i
i
,
i
i
51
Pt~
PoI"-P
1
_12
)
r
r-i
i
~
i
i
i
i _~
,
)
r
,
1_01 ,
,
i
F
~5
,
,
i
i
20
1
,
,
PPM , _~
Fig. 2. In-vivo 31p-NMR spectrum of itaconic acid producing .,4. t e r r e u s cells at 23°C in the presence of 02 . The spectrum is the sum of 15916 scans obtained with 60 ° pulses during an accumulation time of about 5 h. The spectral region from - 2 to 6 ppm is expanded to visualize the overlapped line pattern. The chemical shifts are referenced to glycerol-3-phosphorylcholine (GPC) at - 0 . 1 3 ppm.
al. [2]. The resonances were assigned on a basis of their chemical shifts and by comparing the spectra of the two types of fungal cells. According to the identification of resonances made by Matsunaga et al., the composite signal in the range 1 to - 0 . 2 ppm is due to GPC, Pi in the medium, GPE, and intracellular Pi at - 0 . 1 , 0.25, 0.5 and 0.8 ppm, respectively. The chemical shift range covered by the broad unresolved signals between - 1 0 and - 1 1 . 5 ppm, typical of NAD(H) and NADP(H), is more narrow than that observed in the spectrum of P. ochro-chloron which displayed two resonances between - 10 and - 13 ppm. The strong, well-defined signal at - 2 2 . 5 ppm arises from inorganic polyphosphate. This resonance was significantly detectable after approx. 200 scans (2.5 rain). When air or oxygen were exchanged for pure N 2 the polyphosphate peak disappeared completely after approx. 10 min (see Fig. 3). In an attempt to confirm the assignment, an analysis was made of the spectrum (Fig. 4) of cellular extract after addition of 100 mM EDTA. Approx. 60 resonances are observed in the chemical shift range from 20 to - 2 2 ppm. To check if any resonances arose from extracellular compounds, filtered and centrifuged growth media from cultures in the production phase were also examined. Only traces of inorganic phosphate could be detected (spectra not shown). The first extractions were done on cells concentrated before extraction and without EDTA. In the spectra of these, only Pi, GPC and G P E were detected; this was also the case with cells killed by supplying N 2. The glycerol-3-phosphoryl compounds, GPC and GPE, are known to exhibit resonances independent of the ions present in solution, and GPC is recommended as a good internal chemical shift standard for in-vivo 31p-NMR because its shift ( - 0 . 1 3 ppm) is also independent of pH down to pH 3
112
Pt
F~N -t5
-~
-~
Poly-P
iO
5
0
-5
-tO
-i5
-k~
-~
Fig. 3. 161-MHz 31P-NMR spectra of A. terreus cells during oxygenation (full spectrum) and after supply of N 2 for approx. 10 min, when the resonance of polyphosphate had disappeared (inset).
[13]. GPE has a pK a close to 9.7 and by increasing the pH to 11 the chemical shift separation between GPC and GPE increased by approx. 0.5 ppm; this is consistent with literature data on related phospholipids [14]. In the well resolved spectrum (Fig. 4), the resonances of ATP, GPC, AMP and GIc-6P were identical to those of added compounds (Sigma). The eight absorptions between - 1 1 and -13.5 ppm arise from two UDP sugars. In the high field region a strong resonance at -21.6 ppm comes from middle (or interior) phosphates of long-chain polyphosphates. The weaker and nearly equally intense lines at - 5 and -20.2 ppm arise from terminal phosphate groups (doublet) and penultimate phosphates (triplet), respectively [15]. In the low-field region of the spectrum, an absorption at 19.2 ppm indicates that a relatively large amount of a compound containing a 5-membered phosphate ring is present in the cellular extract [16]. In a proton coupled spectrum this resonance split into a 1 : 3 : 3 : 1 quartet pattern of separation approx. 15 Hz caused by 1H-alP spin-spin coupling, as shown in inset C, Fig. 4. In the spectrum of living cells this resonance was not observed. Discussion
The N M R airlift reactor The production experiment showed a decreased production when air was introduced 3 cm above the bottom of the NMR tube. When air was supplied at the
113
ATP /
C)
,
B)
AA
"Y
IMP+ AMP
~IV~
.... ,jV '~,4t,~...~W%"
II
• ~ I UDP-S
,
I / I,
'
'
I
2o
~
,
is
'
P-c i l~cyC
GIc-6P -
Ref.~
,
I
'
'
'
20
'
J,l'l.
GPC
~
I
'
5
r
l 11 terminal], _
,
,
,
,
{
0
'
'
'
'
I
-5
I
I,
,
,
,
I
-t0
'
penultimate
,
L
,
E
]
-i5
,
,
,
,
I
t
I
-20
Fig. 4. 161-MHz 31p-NMR spectrum of TCA/sonication extract of .4. terreus cells (pH 7.4). (B) After addition of ATP. (C) An expanded spectral region of the cyclic phosphate compound as obtained in a proton-coupled spectrum. The positions of the resonances from ATP, Glc-6P, AMP, and GPC were confirmed by addition of the authentic compounds (Sigma). The chemical shifts are referenced to internal trisodium-methylenediphosphonateat - 16.2 ppm.
bottom the production increased considerably. In order to optimize the spectral resolution it is essential to avoid gas bobbling through the receiver coil and this prohibited introduction of gas at the bottom of the bioreactor. Therefore, to obtain sufficient oxygen transfer, pure 0 2 was introduced while performing the N M R experiments as shown in Fig. 1. During the N M R experiments it was possible to maintain the polyphosphate for at least 24 h. The rapid disappearance of the polyphosphate resonance when 0 2 was exchanged for N 2, killing the cells, also indicated that the culture was alive during sampling.
The 31p spectra The similarity of the spectra of intact cells of P. ochro-chloron obtained by Matsunaga et al. [2] and of A. terreus cells obtained in this work, is striking. All signals correspond well except that the one from U D P G at -12.7 ppm is missing in the latter spectrum. The reason why the UDP sugars were not observed in intact ceils, while being clearly visible in the extract spectrum in this chemical shift region, was probably due to binding of divalent cations. The broadening of the resonances in the living cell spectrum was probably due to interactions with Ca 2+ and small amounts of paramagnetic ions such as Fe z+ and Mn 2+. In fact several expected resonances are not seen or are scarcely visible. The broadening of the
114 signals in the living cell spectrum is difficult to avoid, but a lower concentration of trace-metals should probably make an improvement. The appearance of GPC and GPE may indicate a trend. GPC is known to be an osmolyte in different mammalian cells, together with betaine, polyols and some amino acids [17,18]. In fungi, glycerol and other polyols are most abundant as osmolytes. The importance of GPC (and GPE) as osmolyte(s) in mycelial fungi is not elucidated to. the authors' knowledge. The role of polyphosphate as a metabolite is still under some discussion. Gottschalk [19] stated that polyphosphate is a phosphate reserve, while Roberts [20] stated that it has an important role in regulating and maintaining the energy charge under stress conditions. The rapid disappearance of polyphosphate in our study when 0 2 was exchanged for N 2 probably supports the second theory. From the extract spectrum an average chain length of approx. 20 for the polyphosphate was estimated by using intensities of signals of middle, penultimate and terminal phosphate units. The many resonances in the mono- and di-ester regions are typical of a 31p cell extract spectrum. AMP and IMP overlap independently of pH, but IMP rather than AMP is the major product from ATP breakdown. The relatively large ADP signals compared with those from ATP indicate that this has occurred during the cell extraction. The signal at 19.2 ppm probably arises from a 5-membered phosphate ring in a compound formed during cell extraction in the acid medium. The biologically expected 2',3' cyclic nucleotides are all known to give a doublet in a 1H-coupled 31p-NMR spectrum [16] and can be ruled out. The glycerol 1,2(cyclic) phosphate is in agreement with the chemical shift [16], and the quartet pattern in the 1H-coupled spectrum is compatible with a planar phosphate ring conformation. The observed coupling (15 Hz) is, however, larger than expected for an anti-clinal arrangement of the three protons, - C H 2 - O - P - O - C H - , in glycerol 1,2(cyclic) phosphate (6.3 Hz), as estimated from a Karplus-like relationship reported for a three-bond 1H-31P coupling [21]. Uncertainties about the~phosphate ring conformation and in the equation used to relate coupling constant to dihedral angle make it impossible to draw any conclusion about the identity of the cyclic phosphate compound from NMR data. The splitting of the signal centred at approx. 0.8 ppm in the living cell spectrum is probably caused by noise, and may represent a single resonance from intracellular Pi. The intracellular pH was estimated to be 6.2 using the chemical shift-pH titration curve. The pH value measured, 6.2, is quite acidic in biological terms and will probably affect the metabolism considerably. The accuracy in this type of pH evaluation is normally not better than ___0.1 pH unit unless a calibration is performed under specific cellular conditions [22]. However, due to the rather low signal-to-noise ratio and the strongly overlapping lines, the accuracy may be somewhat lower here. Nevertheless, the measured pH value corresponds well with those given by Matsunaga et al. [2] and by Legisa and Kidric [3]. The external Pi gave a signal at approx. 0.25 ppm indicating a pH of about 2 in the medium in which the pH measured with a pH-meter was 1.7. The small change in chemical shift induced by pH variation at
115
low pH, is another reason leading to inaccurate pH determination by 31p-NMR in this region.
Simplified description of the method and its applications An airlift bioreactor for in-vivo NMR studies of cells in suspension is described. With this device 31p-NMR spectra of living Aspergillus terreus cells, producing itaconic acid, have been measured. The bioreaetor is a draft-tube reactor driven by a respiratory gas. This gave good oxygen supply and culture mixing when pure 0 2 was applied. The reactor is well suited for NMR studies of mycelial and pellet forming organisms. Because of its simplicity it is also well suited for studies of bacteria, algae and tissue cells.
References 1 Vogel, H.J., Brodelius, P., Lilja, H and Lohmeier-Vogel, E.M. (1987) Nuclear magnetic resonance studies of immobilized cells. Methods Enzymol. 135, 512-528. 2 Matsunaga, T., Okubo, A., Yamazaki, S. and Toda, S. (1982) 31p Nuclear magnetic resonance analysis of Penicillium ochro-chloron. Biochem. Biophys. Res. Commun. 106, 596-601. 3 Legisa, M. and Kidric, J. (1989) Initiation of citric acid accumulation in the early stages of Aspergillus niger growth. Appl. Microbiol. Biotechnol. 31,453-457. 4 Santos, H. and Turner, D.L. (1986) Characterization of the improved sensitivity obtained using a flow method for oxygenating and mixing cell suspensions in NMR. J. Magn. Reson. 68, 345-349. 5 Kramer, H.W. and Bailey, J.E. (1991) Mass transfer characterization of an airlift probe for oxygenating and mixing cell suspensions in an NMR spectrometer. Biotechnol. Bioeng. 37, 205-209. 6 DeGraaf, A.A., Wittig, R.M., Probst, U., Strohhaecker, J., Schoberth, S.M., and Sahm, H. (1992) Continuous-flow NMR bioreactor for in-vivo studies of microbial cell suspensions with low biomass concentrations. J. Magn. Reson. 98, 654-659. 7 Milsom, P.E. and Meets J.L. (1985) in Comprehensive Biotechnology (Moo-Young, M., ed.), Vol. 3, pp. 681-700, Pergamon Press. 8 Kautulla, H., Rymowics, W., Linko, Y. and Linko, P. (1991) Itaconic acid production by immobilized Aspergillus terreus with varied metal additions. Appl. Microbiol. Biotechnol. 35, 154-158. 9 Lockwood, L.B. (1979) Production of Organic Acids by Fermentation. In Microbiol. Technology (Peppier, H.J. and Perlman, I., eds.), Vol. 1, pp. 367-373, Academic Press, NewYork. 10 Larsen, H. (1957) Itaconic acid and related compounds. In Methods in Enzymology (Colowick, S.P. and Kaplan, N.O., eds.) Vol. 3, pp. 437-444, Academic Press, New York. l l Thom, C. and Raper, K.B (1945) A Manual of the Aspargilli, 32 pp., The Williams & Wilkins Co., Baltimore. 12 Burt, C.T., Gionek, T. and Barany, M. (1976) Phosphorus-31 nuclear magnetic resonance detection o f unexpected phosphodiesters in muscle. Biochemistry 15, 4850-4853. 13 Barany, M. and Glonek, T. (1982) Phosphorns-31 nuclear magnetic resonance of contractile systems. Methods Enzyrnol. 85, 624-676. 14 London, E. and Feigenson, G.W. (1979) Phosphorus-31 NMR analysis of phospholipids in detergents. J. Lipid Res., 20, 408-412. 15 Navon, G., Schulman, R.G., Yamane, T., Eccleshall, T.R., Ram, K.-B., Baronofsky, J.J. and Marmur, J. (1979) Phosphorus-31 nuclear magnetic resonance studies of wild-type glycolytic pathway mutants of Saccharmyces cerevisia. Biochemistry 18, 4487-4499. 16 Brown, T.R., Graham, R.A., Szwergold, B.S., Thoma, W.J. and Meyer R.A. (1987) Phosphorylated Metabolites in Tumors, Tissues, and Cell Lines. Ann. N.Y. Acad. Sci. 508, 229-240. 17 Yancey, P.H. and Burg, M.B. (1989) Distribution of major organic osmolytes in rabbit kidneys in diuresis and antidiuresis. Am. J. Physiol. 257, 602-607.
116 18 Nakanishi, T., Balban, B.S. and Burg, M.B. (1988) Survey of osmolytes in renal cell lines. Am. J. Physiol. 255, 181-191. 19 Gottschalk, G. (t988) in Bacterial Metabolism 2nd edn., pp. 130-131, Springer-Verlag, New York. 20 Roberts, M.F. (1987) Polyphosphates. In Phosphorus NMR in Biology (Burt C.T.,ed.), pp. 85-94, CRC Press, Boca Raton, Florida. 21 Blackburn, B.J., Lapper, R.D. and Smith, I.C.P. (1973) A Proton Magnetic Resonance Study of the Conformations of 3',5'-Cyclic Nucleotides. J. Am. Chem. Soc. 95, 2873-2878. 22 Gadian, D.G., Radda, G.K., Richards, R.E. and Seeley, P.J. (1979) 31p NMR in living tissue: the road from a promising to an important tool in biology. In Biological Applications of Magnetic Resonance (Shulman, R.G., ed.), pp. 463-529. Academic Press.