Continuous extraction of β-d-galactosidase from Escherichia coli in an aqueous two-phase system: effects of biomass concentration on partitioning and mass transfer

Continuous extraction of / -D-galactosidase from Escherichia coli in an aqueous two-phase system: effects of biomass concentration on partitioning and mass transfer Andres Veide, Torgny Lindb~ick and Sven-Olof Enfors Department o f Biochemistry and Biotechnology, The Royal Institute o f Technology, S-100 44 Stockholm, Sweden

(Received 9 January 1984) High concentrations o f Escherichia colt disintegrate move the binodial o f a poly(ethylene glycol) (PEG) 4000/potassium phosphate aqueous two-phase system towards lower concentrations. It has also been shown that the yield and purification factor o f (J-D-galactosidase (fl-D-galactoside galactohydrolase, EC3.2.1.23) in the PEG phase was gradually improved by moving the experimental system to a composition closer to the binodial. The mass transfer rates o f cell debris, total protein, (3-D-galactosidase and DNA have been studied and were f o u n d to be fast enough to reach equilibrium between the phases after 1.9 s o f mixing in a static mixer with 24 mixing elements. A continuous extraction process f o r (J-D-galactosidase from E. colt has been designed on the basis o f these studies with a mean residence time o f 6.3 rain from the disintegrator inlet to the (3-D-galactosidase containing PEG-phase outlet o f the centrifuge. This PEG phase contained 83.5% o f the total ~-D-galactosidase with a purification factor o f 13.6, and only 2.8% o f the total protease activity o f the disintegrate. All cell debris and almost all DNA were confined to the potassium phosphate phase. Keywords: Mass transfer; aqueous two-phase system; continuous extraction; EscheHchia colt;/3-D-galactosidase

Introduction Proteins are the product type made available by the new developments in genetic engineering. Currently, Escherichia colt is the main host organism for this production, although efforts are being made to use other organisms. From the genetic point of view, however, /~: colt is a superior host. The two main objections to this choice are the restrictions caused because the organism is not classified as 'generally recognized as safe' (GRAS) in food legislation, as are Saccharomyces cerevisiae and Bacillus subtilis, and the fact that E. colt does not excrete its proteins as does, for example. B. subtilis. If the technical problems involved in the large-scale isolation of intracellular proteins of/_:: colt can be solved, the genetic advantages of this organism may lead to greater industrial benefits. One of the major problems in large-scale isolation of intraceUular proteins is the separation of cell debris and DNA from the protein fraction. Another major problem is caused by the proliferation of protease activity during cell disintegration. This phenomenon emphasizes the need for rapid operations in all steps from disintegration to protease separation. The use of extraction in aqueous two-phase systems overcomes many of the problems in the primary purification procedure, l-s in a previous report we described a 0141--0229/841070325--06$03.00 © 1984 Butterworth & Co. (Publishers) Ltd

batch process for large-scale isolation of/3-D-galactosidase (/3-D-galactoside galactohydrolase, EC3.2.1.23) from disintegrated E. colt ceUs.6 In this paper the effects of biomass on the phase system and on the mass transfer of total protein, /3-D-galactosidase, and DNA were investigated. Based on these data a process for fast isolation of a/3-Dgalactosidase-rich protein fraction of E. colt from cell debris, DNA and proteases by means of continuous extraction in an aqueous two-phase system was designed. Special emphasis was placed on the application of conventional unit operations and equipment that could be scaled up to production capacity.

Materials and methods Materials

Poly(ethylene glycol) (PEG) 4000 was obtained from KeboGrave, Sweden. /3-D-Galactosidase, grade VI, from L: colt and DNase (EC 3.1.21.1) DN-25 from bovine pancreas were purchased from Sigma Chemical Company (St Louis, Missouri). Coomassie Brilliant Blue G 250 was a product of Carl Roth (FRG). Tritium labelled protein was provided by Pharmacia Fine Chemicals, Sweden. The salts and other chemicals used were of analytical grade.

Enzyme Microb. Technol., 1984, vol. 6, July

325

Papers

Cultivation a n d cell harvest kL coil ATCC 15224, which is constitutive with respect to /3-D-galactosidase, was cultivated at 37°C in a 50 1 conventional fermenter. A glycerol-mineral salts medium 6 was used. A pH of 7 was maintained by additions of 4 M NaOH, and the dissolved oxygen tension was kept above 10% air saturation by aeration at 1 w m and a stepwise increase in stirrer speed. Ceils were harvested batchwise in the late logarithmic phase, when the cell concentration was 5 . 7 g l -~, dry weight, in a 6 1 tubular bowl centrifuge (CEPA Z61, Carl Padberg, FRG), yielding a product containing 2 1 2 g l -t, dry weight (=100% wet weight). The specific/3-t)-galactosidase activity of the cells was 5.6 U nag-z, dry weight. The cells were suspended to 50% wet weight in a 50 mM Tris IICI buffer, ptt 7.2, containing 100ram NaCI and 10raM MgC12. Since all investigations were not performed immediately, this suspension was divided into several parts, frozen in liquid nitrogen, and stored at -20°C.

15,

0 0 0

I

cn

I

2

A5 A4

>. "5 0-

0

Cell disintegration After thawing, the cell suspension was further diluted to 25c~ wet weight with the same buffer and supplemented with 2-mercaptoethanol to 100 mM. This suspension was then fed through a glass bead disintegrator (Dyno-Mill KDL, Willy A Bachhofen Maschinenfabrik, Switzerland) with 0.2 mm q~ glass beads stirred at 3000 rev rain -~. The mean residence time of the cells in the 600 ml disintegrator was 4.5 rain.

10 15 Polosslum phosphate (°/o,W/W)

5

Figure1 Binodial for a phase system consisting potassium phosphate in water at 25°C and pH 4000 and potassium phosphate; B, same as A with 26.3 mM T r i s - H C I . A, Compositions used in

of PEG 4000 and 7.0. A, Pure PEG but supplemented the investigation

StQtlC mixer

Phase s y s t e m s The basic composition of the phase system was PEG 4000, potassium phosphate and water with a K2HPO4/ KH2PO4 molar ratio of 1.42. giving the phase system a pH of 7. The binodials of the phase systems A and B presented in Figure 1 were determined at 25°C according to Albertsson. 2 A was pure PEG 4000 and potassium phosphate, and B was the same but supplemented with 26.3 mM Tris-HC1. The exact compositions used in the phase systems in this investigation are referred to as points 1 to 4 in t,'igure 1.

*~

~

~

Centrifuge E

(.9 ~

c

E

z~ w

Figure2

Experimental equipment used for studies on mass transfer

l~'r tra c t i o n For all studies except the continuous extraction, the phase system components and sample were mixed in graded 10 ml centrifuge tubes by rotation at 25 rev rain -t for 10 rain. The phases were then separated at 1000g for 2 rain in a Wifug Doctor. Samples were withdrawn for analysis, phase volumes were measured by means of the tube graduation, and cell debris partitioning was visually estimated from the turbidity of the two phases.

Mass transfer studies The experimental system is shown in Figure 2. Three solutions containing PEG 4000, potassium phosphate and disintegrate were pumped at various flow rates to give different concentrations. The liquids were mixed by passing through a static mixer with an inner diameter of 6.25 mm (Kenics static mixer, Kenics Corp., USA), containing 24 or 12 mixing elements, corresponding to the lengths 15 or 7.5 cm, respectively. 7 The partitioning at the exit of the mixer was determined by connecting the mixer outlet to a running centrifuge (Wifug Doctor, 3700revmin -l) via a flexible silicon tube which ended in one of the centrifl~ge

326

Enzyme Microb. Technol., 1984, vol. 6, July

tubes. The state of partition in fhis tube under different experimental conditions was compared to the corresponding values in samples that were mixed and separated as described above (i.e. equilibrium samples). Studies on pure /3-D-galactosidase were performed with 1.0 U ml-I enzyme solutions in 50 mM Tris HCI, pll 7.2, instead of disintegrate solutions. The phase system used is given by point 4 in Figure 1 and the effects of liquid flow rate and number of mixing elements were tested. In experiments with disintegrate the liquid flow rate through the mixer was kept at 150 ml rain -t and 24 mixing elements were used, giving a mean residence time of 1.9 s. The effects of concentration of biological material were studied. The phase composition is given by point 2 in Figure 1.

Con tin uous extraction A 25% (wet weight) cell suspension was pumped at 75 ml ruin -~ through the disintegrator. "l'he disintegrate continued into the static mixer, and was complemented with 35 ml rain -~ of a 40% (w/w) potassium phosphatc solution and 40 inl rain -t 20.3% (w/w) PEG 4000 solution.

Continuous extraction in aqueous two-phase system: A Veide et al.

before entrance into a Kenics static mixer with inner diameter 8.1 mm, length 35.6 cm, and 27 mixing elements. The effluent of the mixer was fed into a 0.25 1 tubular bowl centrifuge (CEPA Labor GLE, Carl Padberg, FRG) at 2400g, which continuously separated the light PEG phase from the heavy potassium phosphate phase. Samples were drawn at 5 rain intervals from the mixer outlet and the two centrifuge outlets, respectively. Flow measurements were performed at the centrifuge outlets.

A nalyses ~-D-Galactosidase activity was determined by the rate of 2-nitrophenyl-~-n-galactopyranoside hydrolysis, as described elsewhere.6 Protein concentration was determined at 595 nm after treatment with Coomassie Brilliant Blue. 8 The partitioning of DNA was followed by analysis of the viscosity in a rotation viscometer (Rotovisko RV2, Gebrt~der Haake, FRG) with system NV and sample volume 8 ml. To determine the viscosity contributed by components other than DNA, 0.2 m g m l -~ of DNase was added and a further viscbsity determination performed-after an 8 h incubation. The protease activity was analysed by means of a tritium labelled protein. The tritium content after digestion and trichloroacetic acid precipitation was used as a measure of proteolytic activity, according to a method developed by Pharmacia Fine Chemicals (personal communication). One unit of proteolytic activity corresponded to a change in radioactivity of 1 dpm/min.

Temperature All experiments, unless stated otherwise, were performed at ambient temperature (about 20-23°C). The two CEPA centrifuges and disintegrator were fitted with cooling coils and jacket, respectively. Results

Effects o f biomass c(mtent in the phase system In Table 1 the effect of increasing biomass load on phase separation at three different points, 1, 2 and 3, in the phase diagram (Figure 1) is shown. A phase system composition of 4.6% (w/w) PEG 4000 and l l.0~ (w/w) potassium phosphate (which is on the one-phase side of the binodial) gives no phase separation. Increasing the biomass load to 16% (wet w/w) changed the conditions so that the two phases could separate. As the cell disintegrate contained biomass and Tris-HCl buffer, the effect of this component on the binodial was studied. Curve B in Figure 1 shows that addition of 26.3 mM T r i s HC1 to the PEG 4000/potassium phosphate system (curve A in Figure 1) moves the binodial towards lower concentrations.

Table 2 Partitioning of cell debris at different biomass loads and phase system compositions % Biomass added (wet w / w ) Phase system composition a

6.9

1 1.5

16.0

1 2 3

Single phase C CFr

Single phase C T

C C T

a See Figure 1 C, Clear top phase with no cell debris T, Turbid top phase with cell debris Table 3 Partitioning of /3-D-galactosidase and total protein at different biomass load and phase system composition

% Biomass added (wet w/w) Phase system composition b

Phase

6.9

1 1.5

16.0

~-D-Galactosidaseactivity(U ml-~) a 1 2 3

Top Bottom T

826

B

9

T B

636 9

1212 14 955 30

2322 32 1587 34 1109 81

Total protein concentration (mg mt ~)c 1

T B

2

T B T B

3

-

-

2.1 8.8 2.7 9.2

4.0 15.1 7.3 14.0

6.7 19.1 5.7 20.3 12.8 18.3

a/3-D-Galactosidase activity in disintegrated cells was 492 U ml -~ bSee Figure I c Protein concentration in disintegrated cells was 24.6 mg ml -~

The position of the system used, with reference to the phase diagram, influences the partitioning of the cell debris between the top and the bottom phases as shown in Table 2. In most cases all visible cell debris was partitioned to the bottom phase, but with the composition given as point 3 in Figure 1, increasing biomass concentration caused a considerable part of the debris to accumulate in the top phase. At the phase compositions ( 1 - 3 ) shown in Figure 1, and for three different biomass loads at each point, the /3-1>galactosidase activity and the total protein concentration were analysed in bottom and in top phases. These data are given in Table 3. From these data the partition coefficient, yield and purification factor for /3-Dgalactosidase in the top phase were calculated for the different conditions. The results are given in Table 4.

Mass transfer studies Table 1 Top phase volume (as percentage of total volume) at differ-

ent biomass loads and compositions of the PEG 4000/potassium phosphate aqueous two-phase system % 8iomass added (wet w/w) Phase system composition a

0

1 2 3

Single phase 16.1 21.4

6.9 Single phase 18.0 22.4

11.5

16.0

Single phase 18.4 22.5

13 18.0 22.2

aSee Figure I. All systems contained 26.3 mM Tris--HCI irrespective of biomass concentration

The mass transfer studies performed using ~-D-galactosidase without other biological material are summarized in Table 5. When 24 mixing elements were used equilibrium for transfer of enzyme to the PEG phase had been reached within 1.4 s. However, when the mean residence time was reduced to 0.95 s (with only 12 mixing elements) equilibrium had not been obtained. The deviation of the top phase volume then obtained shows that mixing of the three liquids was limited. Mass transfer studies on /3-D-galactosidase and total protein when whole cell disintegrate was added to the system showed that equilibrium could be obtained in

E n z y m e M i c r o b . T e c h n o l . , 1984, vol. 6, J u l y

327

Papers 1,9 s when 24 mixing elements were used. This was the case even when the highest biomass load of 16.0% (wet w/w) was used. Figure 3 shows the activities and concentrations obtained in the top phases of samples taken from the static mixer effluent and after equilibrium had been attained. Corresponding analysis of the DNA transport was made by viscosity analysis. Figure 4 shows the viscosity in top and bottom phases at different shear rates. There is no obvious difference in the viscosity of the phases at the mixer outlet and after equilibrium. Addition of DNase drastically reduced the viscosity of the bottom phase and changed its rheology from pseudoplastic to Newtonian. Corresponding treatment of the top phase did not influence the viscosity.

800

/ 600

60

~_ 400

40

g Z~

20 ~-

2OO

Continttous extraction

c

The results of continuous extraction of 13-D-galactosidase from the 1;: coil disintegrate are shown in Table 6. Relatively large variations were observed between the O ii t O 6

Table 4 Partition coefficient, yield in t o p phase, and purification factor for B-D-galaetosidase at different biomass loads and phase system compositions

Phase system composition a

% Biomass added (wet w / w )

6.9 1

Partition coefficient b

Yield (%)

Purification factor

88

17

I 9

I I 12 15 B~0rnass load t%, we1 w/w)

'0 18

F i g u r e 3 Comparison o f equilibrium samples and samples taken at the static mixer outlet after 1.9 s mean residence time. - - , Equilibrium samples; ---, samples at the mixer outlet, o, 13-o-Galactosidase; z~, total protein

-

11.5 16.0

. . 74

.

.

2

6.9 11.5 16.0

91 84 47

99 90 83

20 15 14

3

6.9 11.5 16.0

75 32 14

95 86 71

12 7 4

samples drawn at 5 rain intervals. These are included in Table 6 as coefficients of variance. During the 40 rain of analysis the top phase flow rate was 28.0 ml rain-~ containing 916 U ml -t ~4)-galactosidase with a specific activity of 233 U mg -t. The total yield was 83.5% (of the content in the disintegrate) and the purification factor was 13.6. Samples withdrawn at the mixer outlet showed that equilibrium had been achieved.

.

a See Figure I b p a r t i t i o n coefficient K = concn in t o p phase/concn in b o t t o m phase Table 5 Effects of mean residence time and mixer length on mass transfer of ~-D-cjalactosidase from a pure buffer solution of the enzyme to the PEG 4 0 0 0 phase o f a system corresponding to p o i n t 4 in Figure I (curve B)

Protease partitioning At a phase composition given by point 2 in Figure 1 the proteasc activity was 94 U ml -~ in the PEG (top) phase and 412 Uml -~ in potassium phosphate (bottom)phase. The total amount of cell desintegrate in the system was I 1.5% (wet w/w) with an activity of 1320 U ml 'l.

N u m b e r of mixing elements

Flow rate (cm s ')

Mean residence time (s)

~0"

a/a*

Discussion

24 24 24 12

5.3 7.9 10.5 7.9

2.9 1.9 1.4 0.95

0.98 1.01 0.99 1.21

1.02 0.95 0.97 0.88

When scaling up the unit operations inw~lved in extraction of proteins with aqueous two-phase systems it is important to know the exact phase diagram. No systematic studies have been carried out in this area, but the results we obtained (Table 1 and Figure 1) showed that low molecular weight (Tris H('I) as well as other components of lhc cell disintegrate, int]uenced the position of the binodial. Thus.

~¢~

~, Top phase volume fraction at the mixer outlet ~p*, T o p phase v o l u m e fraction in e q u i l i b r i u m samples a,/3-D-Galactosidase activity in t o p phase at the mixer outlet a*,/3-D-Galactosidase activity in e q u i l i b r i u m samples

Table 6 Results from a 40 min continuous process run w i t h disintegration and extraction of ~bD-galactosidase f r o m E. coil Samples were analysed at 5 rain intervals. The phase system used is given by point 2 in Figure 1 and contained 11.5% (wet w / w ) of disintegrated cells. A corresponding f l o w diagram is shown in Figure 5

Disintegrate T o p phase B o t t o m phase

328

Flow rate (ml min -~)

~-D-Galactosidase activity (U m l - ' )

Total protein ( m g m l 1)

Specific activity (U mg ')

Yield (%)

Purification factor

28.0 -+ 9.5% 1 1 0 . , 6.3%

4 5 2 , 5.6% 916 t 7.8% 42.5 ± 7.7%

26.4 ~ 12% 3.9 +- 5.9% -

17.1 z 7.7% 233 t 4.5% -

100 83.5 ~ 6.9% 14.9 ~: 15%

1.0 13.6 --

Enzyme Microb. Technol., 1984, vol. 6, July

Continuous extraction in aqueous two-phase system." A Veide et aL

a two-phase system could be formed with high biological load at point 1 in Figure 1. The importance of such a change is emphasized by the results shown in Tables 2- 4. Changing the experimental point away from the binodial reduced the value of important parameters such as the partition coefficient of the enzyme, the yield of enzyme, the purification factor, and the partition of cell debris to the bottom phase. Our results also showed that increased biomass load on the system had the same effects on these parameters. This is probably caused by a movement of the binodial (Figure 1) towards lower concentrations when the biological material is added. This means that the experimental point is moved away from the binodial. The effects of increased biomass load on the partition coefficient and the yield for an enzyme agree with results obtained by other workers. 3'9 Thus, considering the experimental results in general, it is important to choose a position in the phase diagram close to the binodial, but when selecting the system the effects of the biological material on the phase diagram must be taken into consideration. When selecting the exact position in the phase diagram it is also important to take into consideration the effect of the top to bottom phase

Medium

55

DW : g I-I : 260 s

r

I

Cultrvohon

Dtsintegrahon

(batch)

(botch)

DW -" 24 g I-I r = 109 s

DW = 24 g I"t "r = 8.0 s

Phase separahon

J_

]~

1

Mixing

I: ® F©

Top phase : .8 - o- Golactosidase Yield: 8 3 5 % Purification: 13.6 x Specific activity: 255 U mg "~

1

i II PFG 4 0 0 0

I Potassium phosphate

Bottom phase : Proteoses DNA Cell debris F i g u r e 5 Continuous extraction of/3-o-galactosidase in an aqueous two-phase system. DW, d r y cell weight; r, residence time

I0 la--

0

0

I

0

1

O

I

o E

w

>

~0

5

b

0 ~1 0

I 400

I 600 Shear rate (s-I)

I 800

I000

Figure 4 Rheological studies on equilibrium samples and ~amples taken at the static mixer outlet after 1 . 9 s mean residence time. (a), T o p phase; (b), b o t t o m phase. , E q u i l i b r i u m samples; , samples at the mixer outlet; . . . . . , e q u i l i b r i u m samples after incubation w i t h DNase. ±, 16%; ~, 11.5%; and c 0% (wet w / w ) biomass

volume ratio on the purification. Thus, as pointed out in earlier work 6 by applying a small top to bottom volume ratio it is possible to obtain considerable purification of a protein with a high partition coefficient, despite the fact that most of the other proteins partition only partly into the bottom phase. The mass transfer studies showed that a static mixer can be conveniently used for extraction of the various components to respective phases (Figures 3-4). Very short residence times in the order of a few seconds could be used, which means that the unit operation's contribution to the total residence time of a continuous extraction process would be insignificant. Figure 5 shows a flow diagram of a fast continuous extraction procedure based on our results. The relatively large variations in the parameters analysed each 5 min during the continuous operation (Table 6) were mainly caused by the difficulties in keeping a constant ratio between the three liquid flows. Better metering equipment would improve this situation. The relatively low yield of i3-D-galactosidase (83.5%), compared to the much higher values obtained in most cases shown in Table 4, was caused by the difficulties in obtaining a sharp point of separation between the two phases in the continuous liquid phase-separating centrifuge. This point can be improved by choosing a more suitable separator. 3 A fast primary separation of a product-containing fraction is a frequently stated priority in discussions on downstream processing of proteins. Our process fulfils this demand. It is also important that proteases are inactivated or eliminated during this procedure. The extraction procedure used in this study eliminated most of the protease activity from the PEG phase. Only 2.8% of the total activity found in the disintegrate was detected in the PEG phase after separation. However, as only 53% of the initial activity was recovered in the phase system the lost part of the proteases could have been inactivated. Another possibility is that the environment interacted with the assay. Thus, further studies on the fate of proteases during extraction with aqueous two-phase systems are required.

Enzyme Microb. Technol., 1984, vol. 6, July

329

Papers 3

Acknowledgement This investigation was s u p p o r t e d b y a grant B i o t e c h n o l o g y Research F o u n d a t i o n , S w e d e n .

f r o m the

4 5

6

References 1 Beijerinck, M. N. Zentralbl. Bakteriol. 1869, 2, 697 699 2 Alberts.son, P. ,~. Partition o f Cell Particles and Macromolecules 2nd ed., Wiley-lnterscience, New York, 1971

330

Enzyme Microb. Technol., 1984, vol. 6, July

7 8 9

Kula, M.-R., Kroner, K. 11. and Hustedt, H. Adv. Biochem. Eng. 1982.24, 73-118 Kopperschlfger, G. and Johansson, G. Anal. Biochem. 1982, 124, 117 124 Quirk, A. V., Woodrow, J. R., Hammond, P. M. and Scawcn, M. D. in Abstracts o f 3rd hzternational Conference on Partitioning h~ Two-Polymer Systems July 1983, University ~t" British Columbia. Canada Veide. A., Smeds. A.-L. and Enfors S.-O. Biotechnol. Bioeng. 1983,25, 1789. 1800 (;odfrey, J.C. andSlater, M.J. ChenLhld. 1978,745 748 Bradford, M. M. Anal, Biochem. 1976, 72. 248 254 Kroner, K. H., Hustedt, H.. Granda, S., and Kula. M.-R. BiotechnoL Bioeng. 1978, 20, 1967-- 1988