Process Biochemistry 45 (2010) 196–202

Optimization of osmotic shock process variables for enhancement of the release ofperiplasmic interferon-a2b from Escherichia coli using response surface method

Ramakrishnan Nagasundara Ramanan a, Joo Shun Tan a, Mohd Shamzi Mohamed b, Tau Chuan Ling c,Beng Ti Tey a,d, Arbakariya B. Ariff a,b,*a Institute of Bioscience, Universiti Putra Malaysia, 43400 Serdang, Selangor, Malaysiab Department of Bioprocess Technology, Faculty of Biotechnology and Biomolecular Sciences, Universiti Putra Malaysia, 43400 UPM Serdang, Selangor, Malaysiac Department of Process and Food Engineering, Faculty of Engineering, Universiti Putra Malaysia, 43400 UPM Serdang, Selangor, Malaysiad Department of Chemical and Environmental Engineering, Faculty of Engineering, Universiti Putra Malaysia, 43400 UPM Serdang, Selangor, Malaysia

A R T I C L E I N F O

Article history:

Received 3 August 2009

Accepted 7 September 2009

Keywords:

Interferon-a2b

Periplasm

Osmotic shock

Process optimization

Response surface methodology

Escherichia coli

A B S T R A C T

The osmotic shock process for the release of periplasmic recombinant human interferon-a2b from

Escherichia coli was optimized using response surface method (RSM). The process parameters such as pH,

buffer concentration and sucrose concentration in hypertonic solution, cell concentration to hypertonic

solution, contact time of cells with hypertonic solution, temperature of hypertonic solution, cell

concentration to hypotonic solution, contact time of cells with hypotonic solution and temperature of

hypotonic solution were initially screened using Plackett Burman design. Further optimization was

carried out using central composite design (one of the design in RSM) for sucrose concentration in

hypertonic solution as well as cell concentration to hypertonic and hypotonic solutions. The optimal cell

concentration was 0.05 g/mL in hypertonic solution and 0.2 g/mL in hypotonic solution. The use of

hypertonic solution containing 18% sucrose with a combination of 100 mM Tris and 2.5 mM EDTA buffer

(pH 8.0 and 25 8C) and cold water (4 8C) as a hypotonic solution gave the optimum release of interferon-

a2b. Increased product concentration in the final solution resulted from the optimized process would

reduce the downstream steps during purification. The concept of reuse of hypertonic solution was also

demonstrated.

� 2009 Elsevier Ltd. All rights reserved.

Contents lists available at ScienceDirect

Process Biochemistry

journal homepage: www.e lsev ier .com/ locate /procbio

1. Introduction

Human interferon-a2b (IFN-a2b) is a physiologically activeprotein falls under the cytokine family. It has excellent antiviral,antimicrobial and antitumor actions. As a result, it is being usedas one of the bio-pharmaceutical drug for the diseases like hairycell leukemia, chronic hepatitis C, chronic hepatitis B [1–5].Recombinant expression in periplasmic space of Escherichia coli

provides excellent room for correct disulphide formation andproper folding. Reduced protein contamination and proteaseactivity could be obtained in periplasm compared to cytoplasm[6,7].

Mechanical and physical cell disruption methods have beenassessed for the release of periplasmic proteins. Mechanicalmethods such as high pressure homogenizer [8,9], hydrodynamiccavitation [10,11], bead mill [12] are not selective in releasing the

* Corresponding author at: Department of Bioprocess Technology, Faculty of

Biotechnology and Biomolecular Sciences, Universiti Putra Malaysia, 43400 UPM

Serdang, Selangor, Malaysia. Tel.: +603 8946 7516; fax: +603 8946 7510.

E-mail address: arbarif@biotech.upm.edu.my (A.B. Ariff).

1359-5113/$ – see front matter � 2009 Elsevier Ltd. All rights reserved.

doi:10.1016/j.procbio.2009.09.007

individual periplasmic proteins. On the other hand, physicalmethods such as osmotic shock and a combined osmotic shock–lysozyme treatment could provide high selectivity and yield. Theuse of lysozyme in osmotic shock–lysozyme treatment would addprotein impurities, which in turn, increase the subsequentpurification steps and the operation cost. Osmotic shock alonecould be considered as a mild treatment with low operation costwhich also has the ability to release periplasmic proteins with veryhigh selectivity.

The effect of osmotic shock, particularly on E. coli, has beenextensively studied by Heppel et al. [13–16] and Bayer [17]. Briefly,osmotic shock process involves of suspending the cells in the highosmolarity (hypertonic) solution and then to low osmolarity(hypotonic) solution. Due to the rapid change in the externalenvironment, the cell wall loses its intactness and proteins arereleased into the solution. The osmotic shock technique has beenapplied to several cell systems for the release of periplasmicproteins [14–16,18–22]. However, osmotic shock process requireshigh volume of liquid. As a result the product concentration fromthis process is low and large scale downstream equipment isrequired to handle high volume. In addition, high cost of effluenttreatment is also required by the osmotic shock process.

R.N. Ramanan et al. / Process Biochemistry 45 (2010) 196–202 197

Optimization of osmotic shock procedure for the specific studyhas been conducted by several researchers [18–22]. Chen et al. [18]claimed that the pretreatment of cells with divalent cations ofcalcium and magnesium prior to the introduction of hypertonicsolution would have chelated the lipopolysaccharide and increasedthe recovery of creatinase from 60% to 75%. They also found thattheir modified method could work for the cell concentration up to1.2%. On the other hand, the addition of magnesium chloride inhypotonic solution reduced the cytoplasmic contaminants in themedium by reducing the chelation of plasma membrane [16,19].However, this strategy would also reduce the release ofperiplasmic protein for the cells harvested during the stationarygrowth phase [16].

The effect of cell concentration (0.8–1.8%), sucrose concen-tration (15–25%) and ethylene diamine tetraacetic acid (EDTA)concentration (1.0–1.74%) for the extraction of penicillin acylasewas studied by Rodriguez et al. [22]. Box–Wilson and user-defined design were used for the two stages optimization. Theyobserved that EDTA has negligible effect on the extraction, whichmight be due to the usage of phosphate buffer. The effect of EDTAfor the chelation of cell membrane was high in tris (hydro-xymethyl) aminomethane (Tris) buffer but not significant inphosphate and 4-(2 hydroxyethyl)piperazine-1-ethane-sulfonicacid (HEPES) buffer [23]. The use of high sucrose concentrationalso reduced the purification factor and the use of buffer insteadof water as hypotonic solution reduced the yield by about 60%[22].

A slightly different procedure of osmotic shock (instead ofseparating hypertonic solution from the cells, the cells suspendedin the hypertonic solution was directly diluted with hypotonicsolution) was optimized using conventional method (varied onefactor at one time) for large scale application [20]. The mixingmode, buffer concentration, equilibration time, pH of hypertonicsolution and variants of osmotic supplements have significanteffect on the recovery, while cell slurry concentration, concentra-tion of sucrose solution, temperature and ratio of dilution did notsignificantly affect the recovery of periplasmic proteins.

Application of optimized hypotonic solution at 100 8C toincrease the purity of human parathyroid hormone 1–84 coupledwith thermostable thioredoxin fusion partner has been reported[21]. Using this technique, thermostable proteins were notaggregated and can be separated with high purity.

The effects of many process variables on the performance ofosmotic shock as cell disruption technique for the release ofperiplasmic proteins are obvious and have been reported in theliterature. However, no optimization work was carried out byconsidering the hypertonic solution and hypotonic solution asseparate factors which will ease the subsequent downstreamoperations. Further more, the use of response surface method(RSM) for optimization of osmotic shock process variables torelease the periplasmic proteins from recombinant bacteria hasnot been well documented in the literature. RSM is one of thestrong tools to know the effect of many factors with less number ofexperiments and can also be used to refine the optimization of theprocess.

The objective of this study was to investigate the significantfactors that affect the osmotic shock process performance inreleasing periplasmic IFN-a2b from the disrupted cells of E. coli.The major factors were then optimized using RSM to improvefurther the extraction performance. The concept of reuse ofhypertonic solution was also investigated.

2. Materials and methods

2.1. Feed stock preparation

IFN-a2b producing recombinant bacterium, E. coli strain Rosetta-gami 2(DE3),

was used in this study. This recombinant strain harbors pET-26b-IFN plasmid which

contains T7lac promoter and pelB signal sequence to aid the transfer of IFN-a2b to

periplasmic area. pET-26b-IFN was prepared by subcloning the coding sequence of

IFN-a2b from the source plasmid pALCA1SIFN (ATCC 53369) to the target plasmid

pET-26b (Cat. No. 69862-3, Novagen). Inoculum was prepared by adding 1% of stock

culture in TB along with antibiotics (34 mg/L of chloramphenicol and 30 mg/L of

kanamycin) and incubated overnight in a rotary shaker (Certomat1 BS-1 B. Braun,

Germany) operating at 250 rpm and 37 8C. The bacterium was grown in a medium,

optimized in-house, for high expression of IFN-a2b. The medium consisted of 5.5 g/

L glucose, 10 mL glycerol, 55 g/L yeast extract, 42 g/L peptone and 100 mL of 0.1 M

phosphate buffer at pH 7.2. The optimized medium was inoculated with 8% of

overnight culture and incubated for 4 h at 37 8C. Subsequently it was induced with

the addition of 1 mM of IPTG and maintained at 30 8C for 4 h. The cells were

harvested by centrifugation (rotor model 1619, Universal 32R centrifuge, Hettich

AG, Switzerland) at 2860 � g for 15 min at 25 8C and the collected cells were stored

at �20 8C.

2.2. Osmotic shock

Frozen cell pellets were washed with water twice and appropriate cell

concentration (0.02 g of wet cell/mL to 0.2 g of wet cell/mL) was mixed with

respective hypertonic solution and shaken at different times (10 min and 30 min).

The cell suspension was centrifuged at 8000 � g (rotor model 1189, Universal 22R

centrifuge, Hettich AG, Switzerland for 1 mL and rotor model SS-34 RX and SLA-

1500, Sorvall Evolution RC, Thermo scientific, USA for 10 mL and 100 mL) for 10 min

to harvest the shrunk cells and then re-suspended in particular hypotonic solution

(water maintained at different temperature) with a concentration ranging from

0.02 g/mL to 0.4 g/mL. The cell suspension was shaken at different times (10 min

and 30 min) and the periplasmic protein was recovered after separating the cells

from solution by centrifuging at 8000 � g for 10 min. Optimization study was

performed in 1 mL cell suspension and scaled-up to 10 mL and 100 mL with respect

to hypertonic solution.

2.3. Statistical design

Statistical design and analysis was performed using design expert software

(version 6.0.6, Stat-Ease, Inc., Minneapolis, MN, USA). Plackett Burman design was

used to screen important factors that affect the osmotic shock process. Twelve

experimental runs were run in triplicates with two levels of each factor. The nine

factors considered were hypertonic buffer pH (pH 7.0 and pH 9.0), buffer

concentration in hypertonic solution (20 mM, Tris, 0.5 mM EDTA and 200 mM

Tris, 5.0 mM EDTA), sucrose concentration in hypertonic solution (2.5% and 25%),

hypertonic solution temperature (4 8C and 60 8C), cell concentration to hypertonic

solution (0.05 g/mL and 0.2 g/mL), contact time with hypertonic solution (10 min

and 30 min), hypotonic solution temperature (4 8C and 60 8C), cell concentration to

hypotonic solution (0.05 g/mL and 0.2 g/mL), contact time with hypotonic solution

(10 min and 30 min).

Refinement of optimization was performed using central composite design

(CCD) of RSM. Three factors (sucrose concentration (10% and 20%), cell

concentration to hypertonic solution (0.05 g/mL and 0.15 g/mL) and cell

concentration to hypotonic solution (0.1 g/mL and 0.3 g/mL)), were varied over

three levels and additional two alpha levels with approximately 2 as its factor to

evaluate their effect on the release of IFN-a2b and selective product release from

the disrupted cells.

2.4. Analytical procedures

Total protein content was analyzed by Bradford method [24] with bovine serum

albumin (BSA) as a standard and calculated with respect to wet cell weight. Samples

and standards were placed in the 96-well plate flat bottom and the absorbance was

measured at 595 nm using microplate reader (Tecan sunrise Absorbance reader).

Both samples and standards were analyzed in triplicates. Percentage of coefficient

variance was found to be less than 8%.

The quantity of IFN-a2b was determined using fully automated Surface Plasmon

Resonance Detection System (BIAcore 3000, GE HealthCare). A CM-5 sensor chip

(GE HealthCare, Uppsala, Sweden) was immobilized with anti-a-IFN mouse

monoclonal antibody (MMHA-2, Cat No. 407290, Merck, San Diego, USA). After

stabilizing the chip with few pre-runs, 5 mL of standards and samples were run

through chip and the output was measured as response unit. The standards were

calibrated using ELISA Kit (catalog No. RPN2759, GE Healthcare, UK). The quantity of

IFN-a2b in the samples was estimated from the linear standard curve and reported

with respect to wet cell weight.

2.5. Calculation of selective product release

The selective product release was calculated as follows:

selective product release ð10�6 mg=mgÞ

¼ amount of IFN-a2bð10�6 mg=g of cellÞamount of total protein releaseðmg=g of cellÞ (1)

Ta

ble

1P

lack

ett

Bu

rma

nd

esi

gn

an

dit

sre

spo

nse

.

Std

IDFa

cto

r1

A:

pH

Fact

or

2

B:

bu

ffe

r

con

cen

tra

tio

n

(mM

)

Fact

or

3

C:

sucr

ose

con

cen

tra

tio

n

(%)

Fact

or

4

D:

hy

pe

rto

nic

solu

tio

n

tem

pe

ratu

re(8

C)

Fact

or

5

E:

cell

con

cen

tra

tio

n

toh

yp

ert

on

ic

solu

tio

n(g

/mL)

Fact

or

6

F:co

nta

ctti

me

wit

hh

yp

ert

on

ic

solu

tio

n(m

in)

Fact

or

7

G:

hy

po

ton

ic

solu

tio

n

tem

pe

ratu

re(8

C)

Fact

or

8

H:

cell

con

cen

tra

tio

n

toh

yp

oto

nic

solu

tio

n(g

/mL)

Fact

or

9

J:co

nta

ct

tim

ew

ith

hy

po

ton

ic

solu

tio

n(m

in)

Re

spo

nse

1

IFN

- a2

b

rele

ase

(10�

6m

g/g

)

Re

spo

nse

2

Se

lect

ive

pro

du

ctre

lea

se

(10�

6m

g/m

g)

19

20

25

40

.05

10

60

0.2

30

69

9.7

12

7.4

29

20

02

.56

00

.05

10

40

.23

04

32

.57

3.2

37

20

02

54

0.2

10

40

.05

30

39

00

.12

65

.0

49

20

25

60

0.0

53

04

0.0

51

03

82

7.1

28

5.8

59

20

02

.56

00

.21

06

00

.05

10

45

2.9

71

.9

69

20

02

54

0.2

30

40

.21

02

47

2.5

29

8.4

77

20

02

56

00

.05

30

60

0.0

53

01

64

2.1

27

1.4

87

20

25

60

0.2

10

60

0.2

10

62

2.9

11

3.2

97

20

2.5

60

0.2

30

40

.23

09

3.3

61

.2

10

92

02

.54

0.2

30

60

0.0

53

09

6.8

49

.8

11

72

00

2.5

40

.05

30

60

0.2

10

29

39

.92

28

.3

12

72

02

.54

0.0

51

04

0.0

51

02

21

3.3

27

4.9

Eff

ect

on

IFN

- a2

b

rele

ase

++

++

++

xx

xx

xx

Eff

ect

on

sele

ctiv

e

pro

du

ctre

lea

se

xx

x+

++

xx

xx

xx

xx

xx

xx

Se

lect

ed

fact

ors

for

CC

D8

10

0y

yy

25

yy

y1

04

yy

y1

0

++

+a

nd

xx

xsi

gn

ify

po

siti

ve

an

dn

eg

ati

ve

inte

ract

ion

wit

hre

spe

ctto

the

fact

ors

an

dy

yy

sig

nifi

es

the

sele

cte

dfa

cto

rsfo

rC

CD

.

R.N. Ramanan et al. / Process Biochemistry 45 (2010) 196–202198

3. Results and discussion

3.1. Screening of important factors for osmotic shock by Plackett

Burman design

The Plackett Burman experimental design and the result areshown in Table 1. The response of IFN-a2b release and selectiveproduct release were transformed using inverse square root andnatural log to get the significant models, respectively. Bufferconcentration, sucrose concentration, cell concentration to hyper-tonic solution and contact time with hypotonic solution wereobserved significant for the release of IFN-a2b from the disruptedcells. Apart from the above factors, buffer pH, hypertonic solutiontemperature and hypotonic solution temperature were also foundsignificant for selective release of IFN-a2b. Increased in contacttime with hypotonic solution resulted to the reduction in theamount of IFN-a2b release and this may be due to the interactingfactors (alias) of buffer concentration with respect to cellconcentration. At 10 min of contact time with hypotonic solution,low buffer concentration was used for low cell concentration (StdID 11 and 12). On the other hand, low buffer concentration wasused for high cell concentration (Std ID 9 and 10) for contact timewith hypotonic solution of 30 min. This result suggests that thebuffer concentration was the important factor particularly at highcell concentrations. At low buffer concentration, the amount ofEDTA might not be sufficient to chelate the lipopolysaccharidebridges (a component of outer membrane) of all cells at highconcentrations.

The temperature of the solutions had negative effect on therelease of the product and this is in agreement with Neu andHeppel [14]. However, this is in contrary to Guo et al. [21] whoclaimed that the host proteins aggregated at high temperature butnot the product due to the thermostable fusing partner with theproduct. In this study both IFN-a2b and host proteins aggregatedin a similar fashion which could be seen from the Std ID 3 and 7.The parameters of the six factors (pH and buffer strength ofhypertonic solution, temperature and contact time for bothhypertonic and hypotonic solutions) were identified (Table 1)while the other factors were considered for further optimization.

3.2. Optimization of important factors for osmotic shock by CCD

Variation in sucrose concentration in the hypertonic solution(10–20%), cell concentration to hypertonic solution (0.05–0.15 g/mL) and cell concentration to hypotonic solution (0.1–0.3 g/mL)were considered in CCD. The levels of these parameters were fixedto incorporate the alpha levels (extreme points) to the PlackettBurman design level except for the low level of cell concentrationto hypertonic solution and high level of cell concentration tohypotonic solution. These levels were changed in order to reducethe downstream process for concentrating the proteins. The CCDconsisted of two replicates of star and axial points and fourreplicates of center points. Total of 32 runs were conducted as perthe design of experiment and it is presented in Table 2.

The value of IFN-a2b release could be best fitted using second-order polynomial equation. On the other hand, the value ofselective product release could be best fitted using third orderpolynomial equation. Both the models were verified using ANOVA(Tables 3a and 3b). The large F value (49.97 for IFN-a2b and 19.55for selective product release) and very low P value (<0.0001 forboth IFN-a2b and selective product release) suggested that themodels were significant at high confidence level. The lack of fitvalues (1.28 for IFN-a2b and 1.13 for selective product release)were not significant with respect to their corresponding pure error.This insignificant lack of fit proved that the models could be fittedto evaluate the responses. Further, the fitness of the models was

Table 2Central composite design and its response.

Std Factor 1

A: sucrose

concentration (%)

Factor 2

B: cell concentration

to hypertonic

solution (g/mL)

Factor 3

C: cell concentration

to hypotonic solution

(g/mL)

Response 1

IFN-a2b release

(10�6 mg/g)

Response 2

Selective product

release (10�6 mg/mg)

1 10 0.05 0.1 2553.5 283.9

2 10 0.05 0.1 2811.3 290.3

3 20 0.05 0.1 2308.3 286.2

4 20 0.05 0.1 2919.5 344.7

5 10 0.15 0.1 1894 251.1

6 10 0.15 0.1 1715.1 219.8

7 20 0.15 0.1 2007.1 245.2

8 20 0.15 0.1 2236.3 264.8

9 10 0.05 0.3 1233.8 172.1

10 10 0.05 0.3 1322.8 167.1

11 20 0.05 0.3 1722.5 244.6

12 20 0.05 0.3 2024.7 301.8

13 10 0.15 0.3 960.9 142.2

14 10 0.15 0.3 1050.9 164.8

15 20 0.15 0.3 1209.4 149.5

16 20 0.15 0.3 899.3 116.4

17 5 0.1 0.2 1365.6 179.9

18 5 0.1 0.2 1031.4 135.3

19 25 0.1 0.2 2045.4 243.9

20 25 0.1 0.2 1900.3 240.9

21 15 0.02 0.2 2310.2 316.3

22 15 0.02 0.2 2589 307.4

23 15 0.2 0.2 1116.3 146.2

24 15 0.2 0.2 1604.4 199.5

25 15 0.1 0.02 2637.1 219.5

26 15 0.1 0.02 2739.4 299.1

27 15 0.1 0.4 1042.2 206.4

28 15 0.1 0.4 1303.6 211.6

29 15 0.1 0.2 2043 254

30 15 0.1 0.2 1974.6 256.1

31 15 0.1 0.2 2054.8 259.7

32 15 0.1 0.2 2478 294.6

Table 3aANOVA for IFN-a2b release in central composite design.

Source Sum of squares DF Mean square F value Prob> F Significant term based

on prob> F value

Model 1.02E+07 4 2.55E+06 49.97 <0.0001 Significant

A 7.45E+05 1 7.45E+05 14.61 0.0007 Significant

B 2.72E+06 1 2.72E+06 53.26 <0.0001 Significant

C 6.24E+06 1 6.24E+06 122.42 <0.0001 Significant

A2 5.83E+05 1 5.83E+05 11.43 0.0022 Significant

Residual 1.38E+06 27 50984.24

Lack of fit 5.91E+05 10 59060.18 1.28 0.3158 Not significant

Pure Error 7.86E+05 17 46233.69

Cor total 1.16E+07 31

Std. dev. 225.8 R-squared 0.881

Mean 1847.02 Adj R-squared 0.8634

C.V. 12.22 Pred R-squared 0.8353

Press 1.91E+06 Adeq precision 20.566

R.N. Ramanan et al. / Process Biochemistry 45 (2010) 196–202 199

assessed by determination coefficient (R2). Adjusted R2 (0.8634 forIFN-a2b and 0.8073 for selective product release) suggested thatmore than 80% of the variation was due to the variables present inthe models which was in reasonable agreement with the predictedR2 (0.8353 for IFN-a2b and 0.7075). High R2 (0.8810 for IFN-a2band 0.8508 for selective product release) indicated a goodagreement between predicted and experimental values.

An adequate precision measures the signal to noise ratio whichshould be greater than 4. High adequate precision (20.566 for IFN-a2b and 15.093 for selective product release) proved that themodel could be used to predict the response. The model equationapplicable for IFN-a2b release was b0 + b1A + b2B + b3C + b4A2

while for selective product release was g0 + g1A + g2B + g3C + -g4A2 + g5C2 + g6AB + g7C3. The significance of each term in themodel was evaluated by their corresponding P value. The value

lesser than 0.05 indicates that the terms are significant while thevalue greater than 0.1 indicates that the terms are not significant.As shown in Tables 3a and 3b, the P values of all the terms for boththe models were less than 0.1 and in particular all the terms exceptC2 (square term of cell concentration to hypotonic solution) forselective product release were less than 0.05.

The two-dimensional contour plots and three-dimensionalresponse curve plots described by the models were assessed toknow the interaction between the terms for the response of IFN-a2b and the selective product release. Three contour plots weregenerated by keeping one variable at a constant position andassessing the interaction of other two variables for the response ofIFN-a2b. Each curve inside the contour plot shows the interactionof the variables for a particular response of IFN-a2b. As shown inFigs. 1 and 2, the interaction between the sucrose concentration

Table 3bANOVA for selective product release in central composite design.

Source Sum of squares DF Mean square F value Prob> F Significant term based

on prob> F value

Model 97030.53 7 13861.5 19.55 < 0.0001 Significant

A 11290.11 1 11290.11 15.92 0.0005 Significant

B 37571.95 1 37571.95 52.99 < 0.0001 Significant

C 26737.94 1 26737.94 37.71 < 0.0001 Significant

A2 8980.75 1 8980.75 12.67 0.0016 Significant

C2 2649.45 1 2649.45 3.74 0.0651

AB 4413.86 1 4413.86 6.23 0.0199 Significant

C3 8816.29 1 8816.29 12.44 0.0017 Significant

Residual 17015.67 24 708.99

Lack of fit 5411.95 7 773.14 1.13 0.3887 Not significant

Pure error 11603.73 17 682.57

Cor total 1.14E+05 31

Std. dev. 26.63 R-squared 0.8508

Mean 231.71 Adj R-squared 0.8073

C.V. 11.49 Pred R-squared 0.7075

Press 33362.63 Adeq precision 15.093

R.N. Ramanan et al. / Process Biochemistry 45 (2010) 196–202200

and the cell concentrations (both with respect to hypertonic andhypotonic solution) reached a maximum at sucrose concentrationof above 17.5%. This result shows that the concentration of sucrosein the hypertonic solution could be maintained as low as 17.5% formaximum product extraction. On the other hand, the interactionbetween the cell concentration to hypertonic solution and the cellconcentration to hypotonic solution (Fig. 3) was in negativedirection, indicating that increased in cell concentration in bothsolutions would reduce the recovery efficiency. The plots for theselective product release (not shown here) were similar to theplots for IFN-a2b.

The criterion for the numerical solution was evaluated bysetting the goals for IFN-a2b release and selective product releaseas maximum while the other variables were in the appropriaterange. The observed solutions are shown in Table 4. In order toconcentrate the proteins in the hypotonic solution, the cellconcentration to hypotonic solution was set to the target value,which was 0.2 g/mL and the observed solutions are shown inTable 5. The sucrose concentration, cell concentration to hyper-tonic solution and cell concentration to hypotonic solution wereset at 18%, 0.05 g/mL and 0.20 g/mL, respectively.

Fig. 1. Contour plot described by the model for IFN-a2b release (10�6 mg/g) with

respect to sucrose concentration and cell concentration to hypertonic solution at

0.2 g of cell/mL of hypotonic solution.

3.3. Effect of scale-up on optimized osmotic shock

The scale-up experiments were conducted using optimizedparameters (sucrose concentration at 18%) up to the cellsuspension volume of 100 mL. In addition, the cell concentrationof 0.05 g/mL with respect to both hypertonic and hypotonicsolution was also performed and the results are shown in Table 6. Itwas observed that the protein solution could be concentrated up to4 times without much reduction in the yield and this process couldbe used at least up to 100 mL (equivalent to 250 mL of the originalculture).

3.4. Recycling of hypertonic solution

Middelberg [23] pointed out that osmotic shock had thelimitation of high cost of additives and increased biological oxygendemand of process waste. The optimized process reduced thehypotonic solution to 4 times which also reduces the processwaste. In order to reduce the consumption of hypertonic solution,the solution was reused for 5 times in 10 mL volume and the resultis presented in Fig. 4. Total protein release and IFN-a2b release

Fig. 2. Contour plot described by the model for IFN-a2b release (10�6 mg/g) with

respect to sucrose concentration and cell concentration to hypotonic solution at

0.05 g of cell/mL of hypertonic solution.

Fig. 3. Contour plot described by the model for IFN-a2b release (10�6 mg/g) with

respect to cell concentration to hypertonic solution and cell concentration to

hypotonic solution at 18% sucrose concentration.

Fig. 4. Percentage of IFN-a2b release and total protein release during the reuse of

hypertonic solution. The error bar shows the standard error among the duplicates.

The experiments were conducted in 10 mL vol. of hypertonic solution using

optimized process.

Table 4Constraints targeting for IFN-a2b release and its solutions according to the model.

Name Goal Lower limit Upper limit Lower weight Upper weight Importance

Sucrose concentration (%) Is in range 15 20 1 1 3

Cell concentration to hypertonic

solution (g/mL)

Is in range 0.05 0.15 1 1 3

Cell concentration to hypotonic

solution (g/mL)

Is in range 0.1 0.3 1 1 3

IFN-a2b release (10�6 mg/g) Maximize 899.344 2919.49 1 1 3

Selective product release

(10�6 mg/mg)

Maximize 250 344.681 1 1 3

Number Sucrose

concentration (%)

Cell concentration

to hypertonic

solution (g/mL)

Cell concentration

to hypotonic

solution (g/mL)

IFN-a2b release

(10�6 mg/g)

Selective

product release

(10�6 mg/mg)

Desirability

1 18.51 0.05 0.10 2781.2 345.0 0.965

2 18.72 0.05 0.10 2781.1 345.6 0.965

3 18.43 0.05 0.10 2778.4 344.7 0.964

4 19.57 0.05 0.10 2776.1 347.6 0.964

5 19.87 0.05 0.10 2753.6 347.9 0.958

6 17.97 0.05 0.10 2780.4 343.3 0.958

7 16.62 0.05 0.10 2766.5 337.3 0.923

Table 5Constraints targeting for both cell concentration to hypotonic solution and IFN-a2b release and its solutions according to the model.

Name Goal Lower limit Upper limit Lower weight Upper weight Importance

Sucrose concentration (%) Is in range 15 20 1 1 3

Cell concentration to hypertonic solution (g/mL) Is in range 0.05 0.15 1 1 3

Cell concentration to hypotonic solution (g/mL) Is target = 0.20 0.1 0.3 1 1 3

IFN-a2b release (10�6 mg/g) Maximize 899.344 2919.49 1 1 3

Selective product release (10�6 mg/mg) Maximize 250 344.681 1 1 3

Number Sucrose

concentration (%)

Cell concentration

to hypertonic

solution (g/mL)

Cell concentration

to hypotonic

solution (g/mL)

IFN-a2b

release (10�6 mg/g)

Selective

product release

(10�6 mg/mg)

Desirability

1 19.85 0.05 0.19 2377.7 318.6 0.774

2 20.00 0.05 0.19 2378.1 319.1 0.773

3 17.67 0.05 0.18 2424.9 317.7 0.750

4 17.89 0.05 0.17 2461.0 322.5 0.747

5 17.20 0.05 0.19 2386.1 311.6 0.743

6 17.08 0.05 0.17 2443.4 317.9 0.736

R.N. Ramanan et al. / Process Biochemistry 45 (2010) 196–202 201

Table 6Comparison of IFN-a2b release and selective product release for different cell concentration ratio and volume of sample.

Sample Concentration

of cells in hypertonic

solution (g/mL)

Concentration of

cells in hypotonic

solution (g/mL)

Volume of

sample (mL)

IFN-a2b release

(10�6 mg/g)

Selective product

release (10�6 mg/mg)

1 0.05 0.05 1 3520.3 (52.7) 280.1 (6.0)

2 0.05 0.2 1 3111.1 (95.4) 304.1 (8.6)

3 0.05 0.2 10 3179.2 (95.5) 344.6 (10.6)

4 0.05 0.2 100 3700.9 (61.8) 302.3 (12.2)

All experiments were performed in triplicates at 18% sucrose concentration. The value in the brackets represents the standard error between the replicates.

R.N. Ramanan et al. / Process Biochemistry 45 (2010) 196–202202

were similar up to 4 times of recycling. This result shows that theamount of hypertonic solution sent for waste treatment could alsobe reduced by reusing the solution at least for 4 times. It is alsoimportant to note that the reduction in process yield was not yetobserved.

4. Conclusions

The osmotic shock procedure and process variables wereoptimized for the enhancement of the release of periplasmic IFN-a2b from the disrupted E. coli cells. The proposed procedure andoptimized process variables were as follows: (i) treatment of thecells at a concentration of 0.05 g/mL of hypertonic solution for10 min; (ii) separation of cells by centrifugation from thehypertonic solution and then re-suspended in cold water (4 8C)for 10 min at a cell concentration of 0.2 g/mL; and (iii) separation ofcell by centrifugation to obtain the supernatant containingproteins. The optimized hypertonic solution contained 18% sucrosewith a buffer concentration of 100 mM of Tris, 2.5 mM of EDTA atpH 8.0 with a temperature of 25 8C. The optimized process reducedthe volume of extraction to fourfold without much loss in the yieldwhich would avoid the step of concentrating the proteins in thesubsequent purification process.

Acknowledgements

This work was funded by the Ministry of Science, Technologyand Innovation, Malaysia under the SR IRPA research grant (ProjectNumber: 03-02-04 SR2010 SR0008/05). Ramanan and Joo Shun arerecipients of graduate research fellowship from Universiti PutraMalaysia. The assistance provided by Goh Chin Chai in performinga part of the osmotic shock experiment was also acknowledged.

Appendix A. Supplementary data

Supplementary data associated with this article can be found, in

the online version, at doi:10.1016/j.procbio.2009.09.007.

References

[1] Bargiggia S, Thorburn D, Anderloni A, Ardizzone S, Giorgi A, Porro GB, et al. Isinterferon-alpha therapy safe and effective for patients with chronic hepatitisC and inflammatory bowel disease? A case–control study. Aliment PharmacolTher 2005;22:209–15.

[2] Billiau A. Interferons: the pathways of discovery. II. Immunomodulatory, invivo and applied aspects. J Clin Virol 2007;39:241–65.

[3] Lanford RE, Guerra B, Lee H, Averett DR, Pfeiffer B, Chavez D, et al. Antiviraleffect and virus-host interactions in response to alpha interferon, gammainterferon, poly(I)–poly(C), tumor necrosis factor alpha, and ribavirin inhepatitis C virus subgenomic replicons. J Virol 2003;77:1092–104.

[4] Mazzoran L, Zorat F, Chemello L, Croce LS, Rigato I, Cavalleto L, et al. Humanleucocyte interferon-alpha in the treatment of chronic hepatitis C. Dig Liver Dis2001;33:347–52.

[5] Rang A, Gunther S, Will H. Effect of interferon alpha on hepatitis B virusreplication and gene expression in transiently transfected human hepatomacells. J Hepatol 1999;31:791–9.

[6] Schumann W, Ferreira LCS. Production of recombinant proteins in Escherichiacoli. Genet Mol Biol 2004;27:442–53.

[7] Hannig G, Makrides SC. Strategies for optimizing heterologous protein expres-sion in Escherichia coli. Trends Biotechnol 1998;16:54–60.

[8] Balasundaram B, Harrison STL. Influence of the extent of disruption ofBakers’ yeast on protein adsorption in expanded beds. J Biotechnol 2008;133:360–9.

[9] Ramanan RN, Tey BT, Ling TC, Ariff AB. Classification of pressure range based onthe characterization of Escherichia coli cell disruption in high pressure homo-genizer. Am J Biochem Biotechnol 2009;5:21–9.

[10] Balasundaram B, Pandit AB. Selective release of invertase by hydrodynamiccavitation. Biochem Eng J 2001;8:251–6.

[11] Balasundaram B, Harrison STL. Study of physical and biological factorsinvolved in the disruption of E. coli by hydrodynamic cavitation. BiotechnolProg 2006;22:907–13.

[12] Bakir U, Hamamci H. The effect of freeze-thawing on the release of intracel-lular proteins from Escherichia coli by means of a bead mill. World J MicrobiolBiotechnol 1997;13:475–7.

[13] Anraku Y, Heppel LA. On the nature of the changes induced in Escherichia coliby osmotic shock. J Biol Chem 1967;242:2561–9.

[14] Neu HC, Heppel LA. The release of enzymes from Escherichia coli by osmoticshock and during the formation of spheroplasts. J Biol Chem 1965;240:3685–92.

[15] Neu HC, Heppel LA. The release of ribonuclease into the medium whenEscherichia coli cells are converted to spheroplasts. J Biol Chem1964;239:3893–900.

[16] Nossal NG, Heppel LA. The release of enzymes by osmotic shock from Escher-ichia coli in exponential phase. J Biol Chem 1966;241:3055–62.

[17] Bayer ME. Response of cell walls of Escherichia coli to a sudden reduction ofenvironmental osmotic pressure. J Bacteriol 1967;93:1104–12.

[18] Chen YC, Chen LA, Chen SJ, Chang MC, Chen TL. A modified osmotic shock forperiplasmic release of a recombinant creatinase from Escherichia coli. BiochemEng J 2004;19:211–5.

[19] Rastgar Jazii F, Karkhane AA, Yakhchali B, Fatemi SS, Deezagi A. A simplifiedpurification procedure for recombinant human granulocyte macrophage-col-ony stimulating factor from periplasmic space of Escherichia coli. J ChromatogrB 2007;856:214–21.

[20] Rathore AS, Bilbrey RE, Steinmeyer DE. Optimization of an osmotic shockprocedure for isolation of a protein product expressed in E. coli. BiotechnolProg 2003;19:1541–6.

[21] Guo Q-R, Wei D-Z, Tong W-Y. Partial purification of human parathyroidhormone 1–84 as a thioredoxin fusion form in recombinant Escherichia coliby thermoosmotic shock. Protein Expr Purif 2006;49:32–8.

[22] Rodriguez M, Guereca L, Valle F, Quintero R, LO’pez-Munguia A. Penicillinacylase extraction by osmotic shock. Process Biochem 1992;27:217–23.

[23] Middelberg APJ. Process-scale disruption of microorganisms. Biotechnol Adv1995;13:491–551.

[24] Bradford MM. A rapid and sensitive method for the quantitation of microgramquantities of protein utilizing the principle of protein dye binding. AnalBiochem 1976;72:248–54.