- Email: [email protected]
Regulation and expression of human Fabs under the control of the Escherichia coli arabinose promoter, PBAD
Immunotechnology
3 (1997) 217-226
Regulation and expression of human Fabs under the control of the Escherichia coli arabinose promoter, P,,, Michelle A. Clark a, Fiona R. Hammond ‘, Arthur Papaioannou Nicholas J. Hawkins b, Robyn L. Ward ‘G*
‘,
’ Department of Medical Oncology, St Vincent’s Hospital, Darlinghurst, NS W, 2010, Australia b School of Pathology, University of NS W, Sydney, NS W, 2052, Australia c CRC for Biopharmaceutical Research, Darlinghurst, NSW, 2010, Australia
Received
13 February
1997; received
in revised form 23 June 1997; accepted
26 June
1997
Abstract Background: The L-arabinose operon from E. coli contains an inducible promoter P BADwhich has been extensively studied for the control of gene expression. PsAD has a number of potential advantages over PI,,., and has been used successfully to promote high level expression of recombinant proteins. Objectives: The aim of this study was to investigate PsAD as an alternative system to P,, for the bacterial expression of recombinant Fabs. Study design: The promoter PBaD from the E. coli arabinose operon araBAD and the gene encoding the regulator of this promoter, were cloned into the phagemid expression vector MCOl. Expression of .human recombinant tetanus toxoid (TT) and c-erbB2 Fabs under the control of P,, was compared at two induction temperatures with the same Fabs produced under the control of P,a,. Results: Expression of TT and c-erbB2 Fabs under the control of PsAD was comparable to was localised to the soluble Fab expression from P,,.. However, highly expressed TT Fab under the control of P,, periplasmic fraction whereas under the control of PI,,, there was greater leakage of Fab into the culture supernatant. could be more tightly repressed than from P,a,. Conclusion: P,, is a useful In addition, Fab expression from P,, and cheaply inducible alternative to the more commonly used P,fl, for the rapid expression of soluble recombinant human antibody fragments. 0 1997 Elsevier Science B.V.
Keywords: Arabinose;
Promoter;
Fabs
1. Introduction promoter region of the araBAD Abbreviations: P,,, operon; P,a,, promoter region of the lac operon; IPTG, isopropylthiogalactoside; TT, tetanus toxoid.
* Corresponding author. Tel.: + 61 2 92958412;fax: + 61 2 92958451. 1380-2933/97/$17.00
0 1997 Elsevier
PIISl380-2933(97)00016-X
Science B.V. All rights
reserved.
The source
utilisation of L-arabinose as an energy by E. coli requires the metabolic operon, araBAD, which is both positively and negatively
218
M.A.
Clark et ul. /Immunotechnology
regulated by the protein product of the araC gene [1,2]. Activation of the operon requires both the removal of a repressor and the addition of an activator [3]. It has been shown that the arabinose promoter P,,, is tightly regulated in the presence of the anti-inducers glucose and fucose, while in the presence of the inducer arabinose it permits high-level expression of cloned genes [1,4,5]. There are many examples of bacterial expression under the control of the T7 RNA polymerase promoter [6- 131. In most cases, although high level expression is achieved, the recombinant protein is present in an insoluble form sequestered in inclusion bodies. Derivatives of P,,. are also amongst the strongest and most commonly used bacterial promoters for the inducible overexpression of soluble.recombinant proteins [14]. Recent reports which have looked at alternative promoters for tightly regulated and high level expression of soluble product have used aromatic chemicals [I 51 and low temperature [16] as inducers. To date, most phagemid vectors have used P,,, for the expression of soluble human recombinant antibodies [17-201. However, Better et al. (1988) [21] reported tlje expression of a functional chimeric antibody fragment in E. coli under the control of the araB promoter from Salmonella typhimurium.
Expression of Fab fragments in vitro is variable and appears to be closely dependent on the particular Fab [19,22]. Expression levels of recombinant antibody fragments in small scale cultures are often as low as 0.1 fig/ml secreted into the supernatant and 0;4 pug/ml in the periplasmic fraction [23-271. It would be advantageous to develop alternative expression systems particularly where they may offer tight regulation, rapid and inexpensive induction, and potentially higher levels of expression than are currently obtained with P,,.. For these reasons, the expression of two human Fabs, tetanus toxoid (TT) and c-erbB2, was assessed under the control of P,,,. P,,, was cloned into the phagemid expression vector MC01 and Fab expression levels were analysed by capture ELISA. The amount of Fab secreted into the soluble periplasmic fraction by P,,, was comparable with that obtained under the control
3 (1997) 217-226
of P/,,.. Constructs containing PSAD were present in this fraction at a similar copy number to constructs containing P,,. TT Fab was present in the supernatant at a higher level when driven by P,, than by PRAUindicating a greater degree of leakiness. Furthermore, repression of Fab expression could be more tightly regulated from PRAn than from P,u,. Thus, the araBAD promoter offers a tightly regulated, alternative system for the expression of soluble recombinant Fabs.
2. Materials and methods 2.1. Cloning of’PBAD and construction of’nra-TT Vector MC01 containing TT Fab under the control of P,Lc,,(lac-TT, 5176 bp) [28] was digested with excess restriction enzyme AJI III (New England BioLabs, MA). The 5’overhang was filled in using 10 U Klenow fragment (Promega, WI) and 5 ~1 of 2 mM dNTP mix in a final volume of 107 ~11for 15 min at 30°C, then heated to 75°C for 10 min to inactivate the enzyme. The DNA was purified using the Qiaquick PCR Purification Kit (Qiagen, Germany), eluted with 40 ,ul of water then digested again with EcoRI to remove P,,,, (325 bp). Vector araXT (3579 bp; gift of The R.W. Johnson Pharmaceutical Research Institute, San Diego, CA) contains the E. coli promoter (PBAD) of the L-arabinose operon, as well as the region encoding the positive and negative regulator araG, and the araC protein binding site. This construct was digested with Sal1 (Promega), 5’ overhangs were end-filled then the DNA was purified as for lac-TT, and digested with EcoRI to remove PBAD and the regulatory sequence (1 I83 bp). Double-digested lac-TT and araXT were electrophoresed on a 0.8% low melting temperature gel (Nu-Sieve, FMC BioProducts, ME) in TAE buffer, and the 4851 bp AflIIIjSalI lac-TT and 1183 bp SalI/EcoRI P,,,IaruC fragments were recovered from the gel using the Qiaquick Gel Extraction Kit (Qiagen). Fragments were quantified by absorbance at 260 nm and ligated overnight at 15°C in the presence of 2000 U of T4 ligase (New England BioLabs). Clones containing P BADand uruC elements (ara-TT) were verified by
M.A. Clark et al. /Immunotechnology 3 (1997) 217-226
restriction enzyme digestion and sequence analysis. The c-&B2 Fab 3F2, isolated by panning a human lymph node library against purified antigen [29], was also subcloned into both the MC01 vector containing P,, (lac-erbB2) and the new P BAD containing construct (ara-erbB2). 2.2. Expression of soluble fabs and determination of j34actamase activity All constructs were electroporated into the E. coli strains HB2151 [K12, ara, A(Zac-pro), thi, F’ proA+B+, lac IqAAM15] and MC1061 [F-, araD139, A(ara-Zeu)7696, gaZE15, galK16, mcrA, A(lac)X74, rpsL (SW) hsdR2, (t-cm;), mcrBl]. HB2151 is wild-type with regards metabolism of arabinose whilst MC1061 has a deletion in the arabinose gene which prevents it from metabolising the inducer arabinose supplied for Fab expression in PBAD containing constructs. Control constructs (Zac-control and ara-control) were also electroporated into HB2151 and MC1061. These constructs contained no Fab but instead had irrelevant, non-coding stuffer fragments cloned into the light chain and heavy chain restriction enzyme sites (1300 bp and 300 bp, respectively). Duplicate cultures of all constructs were grown overnight at 37°C in 2YT containing 2% D( + )glucose and 50 pg/ml carbenicillin (2YT/glc/carbSO). D( + )fucose (0.2%, Sigma, MI) was also included in media for all constructs under control of PBAD [3]. Overnight cultures were inoculated l/l00 into fresh 10 ml 2YT/glc/carbSO, f fucose as appropriate, and grown at 37°C until the cells were in late log phase as determined by OD at 595 nm. Cells were pelleted by centrifugation (1700 x g, 10 min) then resuspended in 10 ml 2YT/carbSO containing either 1 mM isopropylthiogalactoside (IPTG) for induction of expression in constructs containing PI,,, or 0.2% L( + ) arabinose (Sigma) [1,2] for induction of expression in constructs containing P,,,. Control inductions were also performed with cells which were maintained under repressed conditions (either 2% glucose, or 2% glucose + 0.2% fucose, as appropriate) in the absence of inducer. Cultures were induced at both 20°C and 30°C and 2.1 ml samples were removed at 0, 3, 7 and 2 1
219
h. OD at 595 nm was determined to construct growth curves and the supernatant was recovered from the remaining 2 ml by centrifugation as above, followed by the preparation of soluble and insoluble periplasmic fractions (total volume of 500 and 535 ~1, respectively) [22]. All samples were stored at 4°C prior to assay. /J -1actamase activity [30] in culture supernatant, soluble periplasmic and insoluble periplasmic fractions was determined by measuring the change in absorbance at 390 nm using the substrate Nitrocefin (gift of Glaxo-Wellcome, Victoria, Australia). p -1actamase in the supernatant provided an estimate of the cell leakiness whilst enzyme activity in the periplasmic fractions was used as an estimate of the plasmid copy number [22,31]. 2.3. Quantljication of soluble Fabs by capture ELISA The concentration of Fab in each cell fraction was assayed by a capture ELISA which detected only correctly assembled Fab fragments comprised of a heavy and light chain. Multiwell plates (Maxisorp, Denmark) were coated Nunc, overnight at 4°C with 1 ,ugg/ml sheep anti-human ICchain-specific antibody (The Binding Site, Birmingham, UK) in carbonate buffer pH 9.6, washed with PBS then blocked with 2% milk powder in PBS (2% MPBS) for 2 h at 37°C. Fab samples were added in 0.2% MPBS and incubated for 2 h at room temperature. Wells were washed three times with PBS/O.l% Tween 20 (PBST) then three times with PBS. Fab heavy chain was detected with biotinylated anti-myc monoclonal 9ElO (10 pg/ml) [32] in 0.2% MPBS and the plate incubated for a further 1 h at room temperature. Wells were washed as above and streptavidin-alkaline phosphatase conjugate (1 fig/ml; Jackson Immunoresearch Laboratories, USA) in 0.2% MPBS added for 30 min at RT. Wells were washed as above then twice in carbonate buffer prior to the addition of p-nitrophenyl phosphate substrate (1 mg/ml in carbonate buffer; Pharmacia, Uppsala, Sweden) and determination of the absorbance at 410 nm. A standard curve was constructed with TT Fab which had been precipitated with ammonium sulfate (50%) from the
220
M.A. Clark et al. /Immunotechnology 3 (1997) 217-226
soluble periplasmic fraction, purified under denaturing conditions on Ni-NTA Agarose (Qiagen), and renatured in situ [33]. TT Fab was checked for purity by SDS-PAGE and quantified using the BCA protein assay (Pierce, Rockford, IL). Standard curve and test samples were assayed in duplicate, and sample absorbances plotted to the linear region of the curve using AssayZap software (Biosoft, Version 2.5; Cambridge, MO). The presence of Fab heavy and light chains in .a11cell fractions was verified by immunoblotting as previously described [28] after normalising to cell number ( lo9 cells/ml).
3. Results 3.1. /I-lactamase activity and plasmid copy number Growth curves for the Fab-containing constructs, lac-TT, ara-TT, lac-erb B2 and ara-erb B2, and the control constructs lac-control and aracontrol, propagated in HB2151, were similar both before and after induction indicating that expression from either of the promoters did not have an effect on the viability of this host strain (data not shown). Strain MC1061 containing lac-TT, however, showed relatively poor growth. Whilst the reasons for this are unclear, this was observed on a number of occasions in both liquid and solid culture. The copy number of all constructs was estimated from the activity of /?-lactamase which is constitutively expressed in the periplasmic fraction of each plasmid [31]. At 20°C /3-lactamase activity in strains HB21.51 and MC1061 was relatively high in the soluble periplasmic fraction (2-6 pmol substrate destroyed/min per ml enzyme), with all constructs showing a similar level over the time course (Fig. 1). This indicates that the copy number of constructs under the control of p,,, and PBADis similar. Generally, there was higher /?-lactamase activity at 21 h than at other times examined with the exception of the 0 h time point for lac-TT in both host strains. Very low or negligible levels of enzyme activity ( < 0.25 pmol substrate destroyed/min per ml enzyme) were
found in the supernatant (Fig. 1) and insoluble periplasmic fractions (not shown). When inductions were performed at 30°C (Fig. 2) all constructs had a similar copy number at each time point although the distribution of plactamase activity was different from that observed at 20°C. The soluble periplasmic fraction showed high enzyme activity which peaked at 3-7 h (2-6 pmol substrate destroyed/min per ml enzyme) and decreased markedly by 21 h; in particular, lac-TT showed a lo-fold lower copy number at this time. The p-lactamase activity in the supernatant of all constructs progressively increased after induction, with a peak at 21 h (up to 0.96 pmol substrate destroyed/min per ml enzyme). Both host strains demonstrated this trend in /?lactamase activity at 30°C. When constructs were maintained under repressed conditions in host strains HB2151 and MC1061, the distribution of p-lactamase activity was similar for both Fab-containing and control vectors (data not shown). As for induced cultures, cultures maintained in the presence of anti-inducer showed high enzyme activity in the soluble periplasmic fraction (2-4 pmol substrate destroyed/min per ml p-lactamase) at both 20°C and 30°C and at all time points; in the supernatant there was little or no enzyme detected ( < 0.5 pmol substrate destroyed/min per ml /?-lactamase) with highest levels being measured for lacTT and ara-TT at 21 h at 20°C and at 7 h at 30°C. Very low levels of enzyme activity were measured in the insoluble periplasmic fractions. 3.2. Expression of soluble Fab At 20°C, soluble periplasmic Fab levels were similar for both lac-TT and ara-TT at each time point in host strain HB2151 (Fig. 3). Relative to plasmid copy number, however, more Fab appeared to be expressed at 3 and 7 h than at 21 h. In the supernatant, there was notably more Fab measured in lac-TT than ara-TT at each time point suggesting a higher degree of leakiness into this fraction. In contrast to HB2 151, expression of lac-TT in strain MC1061 was considerably lower in spite of a similar plasmid copy number in both the periplasmic and supematant fractions. The
M.A. Clark et al. /Immunotechnology
3 (1997) 217-226
182151
1
lac-TT
lac-erbB2
lac-control
ara-TT
ara-erbB2
ara-control
lac-TT
laeerbB2
lac-control
ara-TT
ara-erbB2
ara-control
PP
’ sl
I
PP
’ sl
I
I
PP
Fig. 1. j?-lactamase activity in the soluble periplasmic (PP) and supernatant (SN) fractions of HB2151 and MC1061 at 20°C. Times after induction are shown [0 (unshaded), 3 (light shaded), 7, (dark shaded) and 21 (filled) h]. Values are normalised to lo9 cells/ml.
reasons for this are unclear but, as noted above, difficulties were encountered in obtaining good growth of lac-TT in MC1061. There was no difference in production of c-e&B2 Fab under the control of either promoter at 20°C in either host strain as expression levels remained low compared with TT Fab. As was the case for induction at 20°C induction at 30°C in HB2151 yielded generally similar levels and cellular distribution of Fab from lac-TT and ara-TTY (Fig. 4). A notable difference from Fab expression at 20°C was the relative decrease in the level of Fab in the soluble periplasmic fraction at 21 h from lac-TT and ara-TT. This is consistent with the observed p-lactamase activity. Further, there also appeared to be an increase in Fab from lac-TT secreted into the culture supernatant at all time points compared with expres-
sion at 20°C; such high levels of Fab in the supernatant were not measured consistently for ara-TT in HB2151 at the higher temperature. Also in contrast to HB2151, in MC1061 there was an increase in Fab from ara-TT in the culture supernatant at all time points. Expression of Fab from lac-TT was again low in MC 1061. When cultures were maintained under repressed conditions, non-induced expression of Fab from lac-TT was comparatively higher than ara-TT in HB2151 after 3, 7 and 21 h (results not shown). At 20°C and at 30°C peak levels of Fab from lac-TT were measured at 7 and 21 h ( N 20 pg/ ml); these data are consistent with the apparently greater promoter strength of P,,. Under repressed conditions, expression of Fab from ara-TT remained at background levels during the time course.
M.A.
Clark et al. / Immunotechnology
3 (1997) 217-226
7.5 HE2151
lac-erbB2
ara-erbB2
ara-TT
lac-control
ara-control
2.5
I
PP
I
PP
ara-erbB2
ara-TT
lac-control
lac-erbB2
lac-TT
I
PP
’ 94
I
PP
’ SJ
ara-control
I
PP
Fig. 2. ,94actamase activity in the soluble periplasmic (PP) and supernatant (SN) fractions of HB2151 and MC1061 at 30°C. Times after induction are shown [0 (unshaded), 3 (light shaded), 7, (dark shaded) and 21 (filled) h]. Values are normalised to 10’cells/ml.
4. Discussion
The production of large amounts of recombinant antibody fragments has often proved difficult [23,24,34]. There is a need, therefore, to investigate alternative expression systems which may help to alleviate problems of low expression. The ,8-lactamase activity in the periplasmic fraction of induced P,a,. and P,, constructs at 20°C, and in two E. coli host strains, indicated that the copy number of plasmids containing these promoters was not significantly different. The consistent profile of plasmid copy number apparent at 20°C was however, not observed at 30°C. At the higher temperature, there was more variability in p-lactamase activity, most notably a decrease at 21 h in most samples.
With regard to antibody production, the major issue is not the amount of Fab per plasmid but rather the total amount produced and its cellular localisation. Fab collected in the soluble periplasmic fraction is effectively concentrated and hence more readily purified. In the soluble periplasmic fraction, TT is expressed at high levels and in this sense may be atypical of the behaviour of most Fabs encountered. Fab c-erbB2 is perhaps more typical in that it is expressed at a lower level than TT Fab. Whilst PBAD did not increase the yield of c-erbB2 Fab in the periplasmic space, irrespective of the time of induction, it was no less efficient than P,,.. This result highlights the difficulties in obtaining a high yield of some Fabs in a bacterial expression system, a difficulty which may ultimately depend on the primary protein sequence [22].
223
M.A. Clark et al. / Immunotechnology 3 (1997) 217-226
40
ara-erbB2
ara-control
ara-erbB2
ara-control
30
Fig. 3. Fab expressed in the soluble periplasmic (PP) and supernatant (SN) fractions of HB 215 1 and MC 1061 at 20°C. Times after induction are shown [0 (unshaded), 3 (light shaded), 7, (dark shaded) and 21 (filled) h]. Values are normalised to lo9 cells/ml.
While Fabs are often purified from the periplasmic fraction the bacterial supernatant is also widely used [21,27,35] and it is therefore important to consider levels of recombinant protein in this fraction. Lac-TT yielded more Fab in the supernatant of strain HB2151 than ara-TT at 20 and 30°C suggesting a higher degree of leakiness with this construct. This may be related to a higher level of Fab expressed from lac-TT than ara-TT which, in view of their similar copy number, is consistent with the previously reported superior strength of P,, [14]. In terms of localisation of Fab, however, PsAD appears to be preferable in that comparably high levels of Fab are invariably present in the soluble periplasmic fraction. It is also apparent from this data that leakage of TT Fab into the supernatant is reduced by induction at 20°C and that this temperature is preferable to 30°C. In terms of the induction time,
3 h at 20°C appears to be sufficient to obtain adequate periplasmic levels of a highly expressing Fab such as TT although longer times do not seem to be detrimental. In contrast, at 3O”C, 3-7 h appears optimal with a notable decrease in Fab concentration at 21 h. Leakiness into the supernatant was reduced further by using HB2151 as the expression host strain rather than MC1061 and no difficulties were encountered propagating all constructs in HB2151. These results clearly demonstrate the influence of the host strain on production and cellular localisation of recombinant protein. MC1061, being defective in the arabinose metabolic pathway, was used to ensure the continued availability of the inducer arabinose by preventing its degradation. However, the level of arabinose used for induction in this study (0.2%) was considerably higher than the concentration
224
M.A. Clark et al. / Immunotechnology 3 (1997) 217-226
“” 1HB2151 ara-erbB2
40
ara-control
30
c
s
50
s g 6
40-
9
30-
MC1061
lac-TT
lac-erbB2
ara-TT
lac-control
ara-erbB2
ara-control
n
Fig. 4. Fab expressed in the soluble periplasmic (PP) and supematant (SN) fractions of HB2151 and MC 1061 at 30°C. Times after induction are shown [0 (unshaded), 3 (light shaded), 7, (dark shaded) and 21 (filled) h]. Values are normalised to lo9 cells/ml.
(0.05%) used by Perez-Perez and Gutierrez (1995) [36], and higher than the concentration (0.008%) reported by Johnson and Schlief (1995) [2] to be the minimum sufficient for induction of expression. It is possible that in the presence of very low levels of inducer, the choice of strain becomes more critical and modulation of expression under the control of PsAD may be more dependent on the aru allele of the host strain [1,2]. Under conditions of constant repression, the copy number of plasmids was similar and there were no signficant differences in the amount of /?-lactamase measured in the supernatant. However, differences in levels of non-induced Fab were notable with lac-TT being the only construct to yield measurable levels of periplasmic Fab. This is consistent with the greater strength of P,, and there are many reports of the production of
high levels of recombinant protein from promoters such as P,,even in the absence of inducer or under repressing conditions [1,14,36]. The araTT construct, whilst capable of producing as much Fab as lac-TT, could be more efficiently repressed which is clearly an advantage in controlling expression. The results presented here support the findings of Guzman et al. (1995) [l] that P,,,is very tightly regulated under repressed conditions by a combination of glucose and fucase although it should be noted that such differences may be construct and Fab. dependent. A further advantage of P,,, over Plo, is that the inducer arabinose is inexpensive compared with IPTG making it a preferable inducer in fermentation processes. Finally, it is important to note that this study was performed using a rich growth medium (2YI9. An alternative strategy to
M.A. Clark et al. / Immunotechnology 3 (1997) 217-226
increase the production of soluble Fab from PBAD would be to grow the cells in minimal medium. Guzman et al. (1995) [l] and Lee et al. (1987) [3] reported that the ratio of induction to repression in rich medium was 250 x whilst in minimal medium it was increased to - 1000 x .
5. Conclusion
In conclusion, P,, is a useful alternative promoter to P,, in the expression of recombinant antibody fragments. Efficient induction of TT Fab in the host strain HB2151 was obtained at 20°C after 3 h. TT Fab accumulated almost exclusively in the soluble periplasmic fraction which is desirable for subsequent purification. Compared with PI,,., production of TT Fab could be more strictly repressed under the control of P,,, and this may represent a significant advantage where production of a highly expressed Fab is toxic to the E. coli host. Whilst PBAD did not increase the yield of Fab c-e&B2 which is expressed at a low level, expression levels were comparable with those observed under the control of PI,,. Another potential advantage of P,,, is in fermentation systems since very low levels of arabinose are required for induction of high levels of expression.
Acknowledgements This work was supported by the National Health and Medical Research Council of Australia, and by the Cooperative Research Centre for Biopharmaceutical Research.
References [l] Guzman L-M, Belin D, Carson MJ, Beckwith J. J Bacteriol 1995;177:4121-30. [2] Johnson C, Schlief RF. J Bacterial 1995;177:3438-42. [3] Lee N, Francklyn C, Hamilton EP. Proc Nat1 Acad Sci USA 1987;84:8814-8. [4] Johnston S, Lee J-H, Ray DS. Gene 1985;34:137745. [5] Slos P, Speck D, Accart N, Kolbe HV, Schubnel D,
225
Bouchon B, Bischoff R, Kieny MP. Prot Exp Purif 1994;5:518-26. 161Nassau PM, Martin SL, Brown RE,, Weston A, Monsey D, McNeil MR, Duncan K. J Bacterial 1996;178:104752. D, Nayak SK, Batra JK. FEBS Lett [71 Rathore 1996;392:259-62. 181Mattanovich D, Weik R, Thim S, Kramer W, Bayer K, Katinger H. Annals NY Acad Sci 1996;782:182-90. [91 Bassuk JA, Baneyx F, Vernon RB, Funk SE, Sage EH. Arch Biochem Biophys 1996;325:8-19. DOI Lullien-Pellerin V, Gavalda S, Joudrier P, Gautier MF. Prot Exp Purif 1994;5:218-24. Pll Freije JM, Munoz M, Vinuela E, Lopez-Otin C. Gene 1993;123:259-62. E, lngendoh A, Hil[W Specht B, Oudenampsen-Kruger lenkamp F, Lezius AG, Spener F. J Biotechnol 1994;33:259969. u31 MacLaren DC, Clarke S. Prot Exp Purif 1995;6:99%108. [I41 Neubauer P, Hofmann K. FEMS Microbial Rev 1994;14:99-102. u51 Suarez A, Staendner LH, Rohde M, Piatti G, Timmis Guzman CA. Microbial KN, APP~ Environ 1997;63:12227. U61 Vasina JA, Baneyx F. Appl Environ Microbial 1996;62:144&7. P71 Marks JD, Hoogenboom HR, Bonnert TP, McCafferty J, Griffiths AD, Winter G. J Mol Biol 1991;222:581-97. [181 Persson MA. Int Rev Immunol 1993;10:153363. 1191Vaughan TJ, Williams AJ, Pritchard K, Osbourn JK, Pope AR, Eamshaw JC, McCafferty J, Hodits RA, Wilton J, Johnson KS. Nature Biotechnol 1996;14:30914. PO1 Tyutyulkova S, Gao Q-S, Thompson A, Rennard S, Paul S. Biochim Biophys Acta 1996;1316:217-23. Pll Better M, Chang CP, Robinson RR, Horwitz AH. Science 1988;240:1041-3. P21 Knappik A, Pluckthun A. Prot Eng 1995;8:81-9. ~231Barbas 111 CF, Kang AS, Lerner RA, Benkovic SJ. Proc Nat1 Acad Sci USA 1991;88:7978-82. 1241Abraham R, Buxbaum S, Link J, Smith R, Venti C, Darsley M. J Immunol Meth 1995;183:119-25. ~251Neri D, Natali PG, Petrul H, Soldani P, Nicotra MR, Vola R, Rivella A, Creighton AM, Neri P, Mariani M. J Invest Dermatol 1996;107:164-70, WI Duenas M, Ayala M, Vazquez J, Ohlin M, Soderlind E, Borrebaeck C, Gavilondo JV. Gene 1995;158:61-6. 1271Popov S, Hubbard JG, Ward ES. Mol Immunol 1996;33:493-502. WI Ward RL, Clark MA, Lees J, Hawkins NJ. J Immunol Meth 1996;189:73-82. v91 Clark M, Hawkins NJ, Papaioannou A, Fiddes RJ, Ward RL. Clin. Exp. Immunol. 1997;109:166-74. [301 O’Callaghan CH, Morris A, Kirby SM, Shingler AH. Antimicrob Ag Chemother 1972;1:283-8. [311 Cagnon C, Valverde V, Masson J-M. Prot Eng 1991;4:843-7.
226
M.A.
Clark et al. / Immunotechnology
[32] Evan GI, Lewis GK, Ramsay G, Bishop JM. Mol Cell
Biol 1985;5:3610&6. [33] McCafferty J, Fitzgerald KJ, Earnshaw J, Chiswell DJ, Link J, Smith R, Kenten J. Appl B&hem Biotechnol 1994;47:157-73.
3 (1997) 217-226
[34] Zhu ZP, Zapata G, Shalaby R, Snedecor B, Chen H, Carter P. Bio/Technology 1996;14:192-6. [35] Nissim A, Hoogenboom HR, Tomlinson IM, Flynn G, Midgley C, Lane D, Winter G. EMBO J 1994;13:692-8. [36] Perez-Perez J, Gutierrez J. Gene 1995;158:141~2.
3 (1997) 217-226
Regulation and expression of human Fabs under the control of the Escherichia coli arabinose promoter, P,,, Michelle A. Clark a, Fiona R. Hammond ‘, Arthur Papaioannou Nicholas J. Hawkins b, Robyn L. Ward ‘G*
‘,
’ Department of Medical Oncology, St Vincent’s Hospital, Darlinghurst, NS W, 2010, Australia b School of Pathology, University of NS W, Sydney, NS W, 2052, Australia c CRC for Biopharmaceutical Research, Darlinghurst, NSW, 2010, Australia
Received
13 February
1997; received
in revised form 23 June 1997; accepted
26 June
1997
Abstract Background: The L-arabinose operon from E. coli contains an inducible promoter P BADwhich has been extensively studied for the control of gene expression. PsAD has a number of potential advantages over PI,,., and has been used successfully to promote high level expression of recombinant proteins. Objectives: The aim of this study was to investigate PsAD as an alternative system to P,, for the bacterial expression of recombinant Fabs. Study design: The promoter PBaD from the E. coli arabinose operon araBAD and the gene encoding the regulator of this promoter, were cloned into the phagemid expression vector MCOl. Expression of .human recombinant tetanus toxoid (TT) and c-erbB2 Fabs under the control of P,, was compared at two induction temperatures with the same Fabs produced under the control of P,a,. Results: Expression of TT and c-erbB2 Fabs under the control of PsAD was comparable to was localised to the soluble Fab expression from P,,.. However, highly expressed TT Fab under the control of P,, periplasmic fraction whereas under the control of PI,,, there was greater leakage of Fab into the culture supernatant. could be more tightly repressed than from P,a,. Conclusion: P,, is a useful In addition, Fab expression from P,, and cheaply inducible alternative to the more commonly used P,fl, for the rapid expression of soluble recombinant human antibody fragments. 0 1997 Elsevier Science B.V.
Keywords: Arabinose;
Promoter;
Fabs
1. Introduction promoter region of the araBAD Abbreviations: P,,, operon; P,a,, promoter region of the lac operon; IPTG, isopropylthiogalactoside; TT, tetanus toxoid.
* Corresponding author. Tel.: + 61 2 92958412;fax: + 61 2 92958451. 1380-2933/97/$17.00
0 1997 Elsevier
PIISl380-2933(97)00016-X
Science B.V. All rights
reserved.
The source
utilisation of L-arabinose as an energy by E. coli requires the metabolic operon, araBAD, which is both positively and negatively
218
M.A.
Clark et ul. /Immunotechnology
regulated by the protein product of the araC gene [1,2]. Activation of the operon requires both the removal of a repressor and the addition of an activator [3]. It has been shown that the arabinose promoter P,,, is tightly regulated in the presence of the anti-inducers glucose and fucose, while in the presence of the inducer arabinose it permits high-level expression of cloned genes [1,4,5]. There are many examples of bacterial expression under the control of the T7 RNA polymerase promoter [6- 131. In most cases, although high level expression is achieved, the recombinant protein is present in an insoluble form sequestered in inclusion bodies. Derivatives of P,,. are also amongst the strongest and most commonly used bacterial promoters for the inducible overexpression of soluble.recombinant proteins [14]. Recent reports which have looked at alternative promoters for tightly regulated and high level expression of soluble product have used aromatic chemicals [I 51 and low temperature [16] as inducers. To date, most phagemid vectors have used P,,, for the expression of soluble human recombinant antibodies [17-201. However, Better et al. (1988) [21] reported tlje expression of a functional chimeric antibody fragment in E. coli under the control of the araB promoter from Salmonella typhimurium.
Expression of Fab fragments in vitro is variable and appears to be closely dependent on the particular Fab [19,22]. Expression levels of recombinant antibody fragments in small scale cultures are often as low as 0.1 fig/ml secreted into the supernatant and 0;4 pug/ml in the periplasmic fraction [23-271. It would be advantageous to develop alternative expression systems particularly where they may offer tight regulation, rapid and inexpensive induction, and potentially higher levels of expression than are currently obtained with P,,.. For these reasons, the expression of two human Fabs, tetanus toxoid (TT) and c-erbB2, was assessed under the control of P,,,. P,,, was cloned into the phagemid expression vector MC01 and Fab expression levels were analysed by capture ELISA. The amount of Fab secreted into the soluble periplasmic fraction by P,,, was comparable with that obtained under the control
3 (1997) 217-226
of P/,,.. Constructs containing PSAD were present in this fraction at a similar copy number to constructs containing P,,. TT Fab was present in the supernatant at a higher level when driven by P,, than by PRAUindicating a greater degree of leakiness. Furthermore, repression of Fab expression could be more tightly regulated from PRAn than from P,u,. Thus, the araBAD promoter offers a tightly regulated, alternative system for the expression of soluble recombinant Fabs.
2. Materials and methods 2.1. Cloning of’PBAD and construction of’nra-TT Vector MC01 containing TT Fab under the control of P,Lc,,(lac-TT, 5176 bp) [28] was digested with excess restriction enzyme AJI III (New England BioLabs, MA). The 5’overhang was filled in using 10 U Klenow fragment (Promega, WI) and 5 ~1 of 2 mM dNTP mix in a final volume of 107 ~11for 15 min at 30°C, then heated to 75°C for 10 min to inactivate the enzyme. The DNA was purified using the Qiaquick PCR Purification Kit (Qiagen, Germany), eluted with 40 ,ul of water then digested again with EcoRI to remove P,,,, (325 bp). Vector araXT (3579 bp; gift of The R.W. Johnson Pharmaceutical Research Institute, San Diego, CA) contains the E. coli promoter (PBAD) of the L-arabinose operon, as well as the region encoding the positive and negative regulator araG, and the araC protein binding site. This construct was digested with Sal1 (Promega), 5’ overhangs were end-filled then the DNA was purified as for lac-TT, and digested with EcoRI to remove PBAD and the regulatory sequence (1 I83 bp). Double-digested lac-TT and araXT were electrophoresed on a 0.8% low melting temperature gel (Nu-Sieve, FMC BioProducts, ME) in TAE buffer, and the 4851 bp AflIIIjSalI lac-TT and 1183 bp SalI/EcoRI P,,,IaruC fragments were recovered from the gel using the Qiaquick Gel Extraction Kit (Qiagen). Fragments were quantified by absorbance at 260 nm and ligated overnight at 15°C in the presence of 2000 U of T4 ligase (New England BioLabs). Clones containing P BADand uruC elements (ara-TT) were verified by
M.A. Clark et al. /Immunotechnology 3 (1997) 217-226
restriction enzyme digestion and sequence analysis. The c-&B2 Fab 3F2, isolated by panning a human lymph node library against purified antigen [29], was also subcloned into both the MC01 vector containing P,, (lac-erbB2) and the new P BAD containing construct (ara-erbB2). 2.2. Expression of soluble fabs and determination of j34actamase activity All constructs were electroporated into the E. coli strains HB2151 [K12, ara, A(Zac-pro), thi, F’ proA+B+, lac IqAAM15] and MC1061 [F-, araD139, A(ara-Zeu)7696, gaZE15, galK16, mcrA, A(lac)X74, rpsL (SW) hsdR2, (t-cm;), mcrBl]. HB2151 is wild-type with regards metabolism of arabinose whilst MC1061 has a deletion in the arabinose gene which prevents it from metabolising the inducer arabinose supplied for Fab expression in PBAD containing constructs. Control constructs (Zac-control and ara-control) were also electroporated into HB2151 and MC1061. These constructs contained no Fab but instead had irrelevant, non-coding stuffer fragments cloned into the light chain and heavy chain restriction enzyme sites (1300 bp and 300 bp, respectively). Duplicate cultures of all constructs were grown overnight at 37°C in 2YT containing 2% D( + )glucose and 50 pg/ml carbenicillin (2YT/glc/carbSO). D( + )fucose (0.2%, Sigma, MI) was also included in media for all constructs under control of PBAD [3]. Overnight cultures were inoculated l/l00 into fresh 10 ml 2YT/glc/carbSO, f fucose as appropriate, and grown at 37°C until the cells were in late log phase as determined by OD at 595 nm. Cells were pelleted by centrifugation (1700 x g, 10 min) then resuspended in 10 ml 2YT/carbSO containing either 1 mM isopropylthiogalactoside (IPTG) for induction of expression in constructs containing PI,,, or 0.2% L( + ) arabinose (Sigma) [1,2] for induction of expression in constructs containing P,,,. Control inductions were also performed with cells which were maintained under repressed conditions (either 2% glucose, or 2% glucose + 0.2% fucose, as appropriate) in the absence of inducer. Cultures were induced at both 20°C and 30°C and 2.1 ml samples were removed at 0, 3, 7 and 2 1
219
h. OD at 595 nm was determined to construct growth curves and the supernatant was recovered from the remaining 2 ml by centrifugation as above, followed by the preparation of soluble and insoluble periplasmic fractions (total volume of 500 and 535 ~1, respectively) [22]. All samples were stored at 4°C prior to assay. /J -1actamase activity [30] in culture supernatant, soluble periplasmic and insoluble periplasmic fractions was determined by measuring the change in absorbance at 390 nm using the substrate Nitrocefin (gift of Glaxo-Wellcome, Victoria, Australia). p -1actamase in the supernatant provided an estimate of the cell leakiness whilst enzyme activity in the periplasmic fractions was used as an estimate of the plasmid copy number [22,31]. 2.3. Quantljication of soluble Fabs by capture ELISA The concentration of Fab in each cell fraction was assayed by a capture ELISA which detected only correctly assembled Fab fragments comprised of a heavy and light chain. Multiwell plates (Maxisorp, Denmark) were coated Nunc, overnight at 4°C with 1 ,ugg/ml sheep anti-human ICchain-specific antibody (The Binding Site, Birmingham, UK) in carbonate buffer pH 9.6, washed with PBS then blocked with 2% milk powder in PBS (2% MPBS) for 2 h at 37°C. Fab samples were added in 0.2% MPBS and incubated for 2 h at room temperature. Wells were washed three times with PBS/O.l% Tween 20 (PBST) then three times with PBS. Fab heavy chain was detected with biotinylated anti-myc monoclonal 9ElO (10 pg/ml) [32] in 0.2% MPBS and the plate incubated for a further 1 h at room temperature. Wells were washed as above and streptavidin-alkaline phosphatase conjugate (1 fig/ml; Jackson Immunoresearch Laboratories, USA) in 0.2% MPBS added for 30 min at RT. Wells were washed as above then twice in carbonate buffer prior to the addition of p-nitrophenyl phosphate substrate (1 mg/ml in carbonate buffer; Pharmacia, Uppsala, Sweden) and determination of the absorbance at 410 nm. A standard curve was constructed with TT Fab which had been precipitated with ammonium sulfate (50%) from the
220
M.A. Clark et al. /Immunotechnology 3 (1997) 217-226
soluble periplasmic fraction, purified under denaturing conditions on Ni-NTA Agarose (Qiagen), and renatured in situ [33]. TT Fab was checked for purity by SDS-PAGE and quantified using the BCA protein assay (Pierce, Rockford, IL). Standard curve and test samples were assayed in duplicate, and sample absorbances plotted to the linear region of the curve using AssayZap software (Biosoft, Version 2.5; Cambridge, MO). The presence of Fab heavy and light chains in .a11cell fractions was verified by immunoblotting as previously described [28] after normalising to cell number ( lo9 cells/ml).
3. Results 3.1. /I-lactamase activity and plasmid copy number Growth curves for the Fab-containing constructs, lac-TT, ara-TT, lac-erb B2 and ara-erb B2, and the control constructs lac-control and aracontrol, propagated in HB2151, were similar both before and after induction indicating that expression from either of the promoters did not have an effect on the viability of this host strain (data not shown). Strain MC1061 containing lac-TT, however, showed relatively poor growth. Whilst the reasons for this are unclear, this was observed on a number of occasions in both liquid and solid culture. The copy number of all constructs was estimated from the activity of /?-lactamase which is constitutively expressed in the periplasmic fraction of each plasmid [31]. At 20°C /3-lactamase activity in strains HB21.51 and MC1061 was relatively high in the soluble periplasmic fraction (2-6 pmol substrate destroyed/min per ml enzyme), with all constructs showing a similar level over the time course (Fig. 1). This indicates that the copy number of constructs under the control of p,,, and PBADis similar. Generally, there was higher /?-lactamase activity at 21 h than at other times examined with the exception of the 0 h time point for lac-TT in both host strains. Very low or negligible levels of enzyme activity ( < 0.25 pmol substrate destroyed/min per ml enzyme) were
found in the supernatant (Fig. 1) and insoluble periplasmic fractions (not shown). When inductions were performed at 30°C (Fig. 2) all constructs had a similar copy number at each time point although the distribution of plactamase activity was different from that observed at 20°C. The soluble periplasmic fraction showed high enzyme activity which peaked at 3-7 h (2-6 pmol substrate destroyed/min per ml enzyme) and decreased markedly by 21 h; in particular, lac-TT showed a lo-fold lower copy number at this time. The p-lactamase activity in the supernatant of all constructs progressively increased after induction, with a peak at 21 h (up to 0.96 pmol substrate destroyed/min per ml enzyme). Both host strains demonstrated this trend in /?lactamase activity at 30°C. When constructs were maintained under repressed conditions in host strains HB2151 and MC1061, the distribution of p-lactamase activity was similar for both Fab-containing and control vectors (data not shown). As for induced cultures, cultures maintained in the presence of anti-inducer showed high enzyme activity in the soluble periplasmic fraction (2-4 pmol substrate destroyed/min per ml p-lactamase) at both 20°C and 30°C and at all time points; in the supernatant there was little or no enzyme detected ( < 0.5 pmol substrate destroyed/min per ml /?-lactamase) with highest levels being measured for lacTT and ara-TT at 21 h at 20°C and at 7 h at 30°C. Very low levels of enzyme activity were measured in the insoluble periplasmic fractions. 3.2. Expression of soluble Fab At 20°C, soluble periplasmic Fab levels were similar for both lac-TT and ara-TT at each time point in host strain HB2151 (Fig. 3). Relative to plasmid copy number, however, more Fab appeared to be expressed at 3 and 7 h than at 21 h. In the supernatant, there was notably more Fab measured in lac-TT than ara-TT at each time point suggesting a higher degree of leakiness into this fraction. In contrast to HB2 151, expression of lac-TT in strain MC1061 was considerably lower in spite of a similar plasmid copy number in both the periplasmic and supematant fractions. The
M.A. Clark et al. /Immunotechnology
3 (1997) 217-226
182151
1
lac-TT
lac-erbB2
lac-control
ara-TT
ara-erbB2
ara-control
lac-TT
laeerbB2
lac-control
ara-TT
ara-erbB2
ara-control
PP
’ sl
I
PP
’ sl
I
I
PP
Fig. 1. j?-lactamase activity in the soluble periplasmic (PP) and supernatant (SN) fractions of HB2151 and MC1061 at 20°C. Times after induction are shown [0 (unshaded), 3 (light shaded), 7, (dark shaded) and 21 (filled) h]. Values are normalised to lo9 cells/ml.
reasons for this are unclear but, as noted above, difficulties were encountered in obtaining good growth of lac-TT in MC1061. There was no difference in production of c-e&B2 Fab under the control of either promoter at 20°C in either host strain as expression levels remained low compared with TT Fab. As was the case for induction at 20°C induction at 30°C in HB2151 yielded generally similar levels and cellular distribution of Fab from lac-TT and ara-TTY (Fig. 4). A notable difference from Fab expression at 20°C was the relative decrease in the level of Fab in the soluble periplasmic fraction at 21 h from lac-TT and ara-TT. This is consistent with the observed p-lactamase activity. Further, there also appeared to be an increase in Fab from lac-TT secreted into the culture supernatant at all time points compared with expres-
sion at 20°C; such high levels of Fab in the supernatant were not measured consistently for ara-TT in HB2151 at the higher temperature. Also in contrast to HB2151, in MC1061 there was an increase in Fab from ara-TT in the culture supernatant at all time points. Expression of Fab from lac-TT was again low in MC 1061. When cultures were maintained under repressed conditions, non-induced expression of Fab from lac-TT was comparatively higher than ara-TT in HB2151 after 3, 7 and 21 h (results not shown). At 20°C and at 30°C peak levels of Fab from lac-TT were measured at 7 and 21 h ( N 20 pg/ ml); these data are consistent with the apparently greater promoter strength of P,,. Under repressed conditions, expression of Fab from ara-TT remained at background levels during the time course.
M.A.
Clark et al. / Immunotechnology
3 (1997) 217-226
7.5 HE2151
lac-erbB2
ara-erbB2
ara-TT
lac-control
ara-control
2.5
I
PP
I
PP
ara-erbB2
ara-TT
lac-control
lac-erbB2
lac-TT
I
PP
’ 94
I
PP
’ SJ
ara-control
I
PP
Fig. 2. ,94actamase activity in the soluble periplasmic (PP) and supernatant (SN) fractions of HB2151 and MC1061 at 30°C. Times after induction are shown [0 (unshaded), 3 (light shaded), 7, (dark shaded) and 21 (filled) h]. Values are normalised to 10’cells/ml.
4. Discussion
The production of large amounts of recombinant antibody fragments has often proved difficult [23,24,34]. There is a need, therefore, to investigate alternative expression systems which may help to alleviate problems of low expression. The ,8-lactamase activity in the periplasmic fraction of induced P,a,. and P,, constructs at 20°C, and in two E. coli host strains, indicated that the copy number of plasmids containing these promoters was not significantly different. The consistent profile of plasmid copy number apparent at 20°C was however, not observed at 30°C. At the higher temperature, there was more variability in p-lactamase activity, most notably a decrease at 21 h in most samples.
With regard to antibody production, the major issue is not the amount of Fab per plasmid but rather the total amount produced and its cellular localisation. Fab collected in the soluble periplasmic fraction is effectively concentrated and hence more readily purified. In the soluble periplasmic fraction, TT is expressed at high levels and in this sense may be atypical of the behaviour of most Fabs encountered. Fab c-erbB2 is perhaps more typical in that it is expressed at a lower level than TT Fab. Whilst PBAD did not increase the yield of c-erbB2 Fab in the periplasmic space, irrespective of the time of induction, it was no less efficient than P,,.. This result highlights the difficulties in obtaining a high yield of some Fabs in a bacterial expression system, a difficulty which may ultimately depend on the primary protein sequence [22].
223
M.A. Clark et al. / Immunotechnology 3 (1997) 217-226
40
ara-erbB2
ara-control
ara-erbB2
ara-control
30
Fig. 3. Fab expressed in the soluble periplasmic (PP) and supernatant (SN) fractions of HB 215 1 and MC 1061 at 20°C. Times after induction are shown [0 (unshaded), 3 (light shaded), 7, (dark shaded) and 21 (filled) h]. Values are normalised to lo9 cells/ml.
While Fabs are often purified from the periplasmic fraction the bacterial supernatant is also widely used [21,27,35] and it is therefore important to consider levels of recombinant protein in this fraction. Lac-TT yielded more Fab in the supernatant of strain HB2151 than ara-TT at 20 and 30°C suggesting a higher degree of leakiness with this construct. This may be related to a higher level of Fab expressed from lac-TT than ara-TT which, in view of their similar copy number, is consistent with the previously reported superior strength of P,, [14]. In terms of localisation of Fab, however, PsAD appears to be preferable in that comparably high levels of Fab are invariably present in the soluble periplasmic fraction. It is also apparent from this data that leakage of TT Fab into the supernatant is reduced by induction at 20°C and that this temperature is preferable to 30°C. In terms of the induction time,
3 h at 20°C appears to be sufficient to obtain adequate periplasmic levels of a highly expressing Fab such as TT although longer times do not seem to be detrimental. In contrast, at 3O”C, 3-7 h appears optimal with a notable decrease in Fab concentration at 21 h. Leakiness into the supernatant was reduced further by using HB2151 as the expression host strain rather than MC1061 and no difficulties were encountered propagating all constructs in HB2151. These results clearly demonstrate the influence of the host strain on production and cellular localisation of recombinant protein. MC1061, being defective in the arabinose metabolic pathway, was used to ensure the continued availability of the inducer arabinose by preventing its degradation. However, the level of arabinose used for induction in this study (0.2%) was considerably higher than the concentration
224
M.A. Clark et al. / Immunotechnology 3 (1997) 217-226
“” 1HB2151 ara-erbB2
40
ara-control
30
c
s
50
s g 6
40-
9
30-
MC1061
lac-TT
lac-erbB2
ara-TT
lac-control
ara-erbB2
ara-control
n
Fig. 4. Fab expressed in the soluble periplasmic (PP) and supematant (SN) fractions of HB2151 and MC 1061 at 30°C. Times after induction are shown [0 (unshaded), 3 (light shaded), 7, (dark shaded) and 21 (filled) h]. Values are normalised to lo9 cells/ml.
(0.05%) used by Perez-Perez and Gutierrez (1995) [36], and higher than the concentration (0.008%) reported by Johnson and Schlief (1995) [2] to be the minimum sufficient for induction of expression. It is possible that in the presence of very low levels of inducer, the choice of strain becomes more critical and modulation of expression under the control of PsAD may be more dependent on the aru allele of the host strain [1,2]. Under conditions of constant repression, the copy number of plasmids was similar and there were no signficant differences in the amount of /?-lactamase measured in the supernatant. However, differences in levels of non-induced Fab were notable with lac-TT being the only construct to yield measurable levels of periplasmic Fab. This is consistent with the greater strength of P,, and there are many reports of the production of
high levels of recombinant protein from promoters such as P,,even in the absence of inducer or under repressing conditions [1,14,36]. The araTT construct, whilst capable of producing as much Fab as lac-TT, could be more efficiently repressed which is clearly an advantage in controlling expression. The results presented here support the findings of Guzman et al. (1995) [l] that P,,,is very tightly regulated under repressed conditions by a combination of glucose and fucase although it should be noted that such differences may be construct and Fab. dependent. A further advantage of P,,, over Plo, is that the inducer arabinose is inexpensive compared with IPTG making it a preferable inducer in fermentation processes. Finally, it is important to note that this study was performed using a rich growth medium (2YI9. An alternative strategy to
M.A. Clark et al. / Immunotechnology 3 (1997) 217-226
increase the production of soluble Fab from PBAD would be to grow the cells in minimal medium. Guzman et al. (1995) [l] and Lee et al. (1987) [3] reported that the ratio of induction to repression in rich medium was 250 x whilst in minimal medium it was increased to - 1000 x .
5. Conclusion
In conclusion, P,, is a useful alternative promoter to P,, in the expression of recombinant antibody fragments. Efficient induction of TT Fab in the host strain HB2151 was obtained at 20°C after 3 h. TT Fab accumulated almost exclusively in the soluble periplasmic fraction which is desirable for subsequent purification. Compared with PI,,., production of TT Fab could be more strictly repressed under the control of P,,, and this may represent a significant advantage where production of a highly expressed Fab is toxic to the E. coli host. Whilst PBAD did not increase the yield of Fab c-e&B2 which is expressed at a low level, expression levels were comparable with those observed under the control of PI,,. Another potential advantage of P,,, is in fermentation systems since very low levels of arabinose are required for induction of high levels of expression.
Acknowledgements This work was supported by the National Health and Medical Research Council of Australia, and by the Cooperative Research Centre for Biopharmaceutical Research.
References [l] Guzman L-M, Belin D, Carson MJ, Beckwith J. J Bacteriol 1995;177:4121-30. [2] Johnson C, Schlief RF. J Bacterial 1995;177:3438-42. [3] Lee N, Francklyn C, Hamilton EP. Proc Nat1 Acad Sci USA 1987;84:8814-8. [4] Johnston S, Lee J-H, Ray DS. Gene 1985;34:137745. [5] Slos P, Speck D, Accart N, Kolbe HV, Schubnel D,
225
Bouchon B, Bischoff R, Kieny MP. Prot Exp Purif 1994;5:518-26. 161Nassau PM, Martin SL, Brown RE,, Weston A, Monsey D, McNeil MR, Duncan K. J Bacterial 1996;178:104752. D, Nayak SK, Batra JK. FEBS Lett [71 Rathore 1996;392:259-62. 181Mattanovich D, Weik R, Thim S, Kramer W, Bayer K, Katinger H. Annals NY Acad Sci 1996;782:182-90. [91 Bassuk JA, Baneyx F, Vernon RB, Funk SE, Sage EH. Arch Biochem Biophys 1996;325:8-19. DOI Lullien-Pellerin V, Gavalda S, Joudrier P, Gautier MF. Prot Exp Purif 1994;5:218-24. Pll Freije JM, Munoz M, Vinuela E, Lopez-Otin C. Gene 1993;123:259-62. E, lngendoh A, Hil[W Specht B, Oudenampsen-Kruger lenkamp F, Lezius AG, Spener F. J Biotechnol 1994;33:259969. u31 MacLaren DC, Clarke S. Prot Exp Purif 1995;6:99%108. [I41 Neubauer P, Hofmann K. FEMS Microbial Rev 1994;14:99-102. u51 Suarez A, Staendner LH, Rohde M, Piatti G, Timmis Guzman CA. Microbial KN, APP~ Environ 1997;63:12227. U61 Vasina JA, Baneyx F. Appl Environ Microbial 1996;62:144&7. P71 Marks JD, Hoogenboom HR, Bonnert TP, McCafferty J, Griffiths AD, Winter G. J Mol Biol 1991;222:581-97. [181 Persson MA. Int Rev Immunol 1993;10:153363. 1191Vaughan TJ, Williams AJ, Pritchard K, Osbourn JK, Pope AR, Eamshaw JC, McCafferty J, Hodits RA, Wilton J, Johnson KS. Nature Biotechnol 1996;14:30914. PO1 Tyutyulkova S, Gao Q-S, Thompson A, Rennard S, Paul S. Biochim Biophys Acta 1996;1316:217-23. Pll Better M, Chang CP, Robinson RR, Horwitz AH. Science 1988;240:1041-3. P21 Knappik A, Pluckthun A. Prot Eng 1995;8:81-9. ~231Barbas 111 CF, Kang AS, Lerner RA, Benkovic SJ. Proc Nat1 Acad Sci USA 1991;88:7978-82. 1241Abraham R, Buxbaum S, Link J, Smith R, Venti C, Darsley M. J Immunol Meth 1995;183:119-25. ~251Neri D, Natali PG, Petrul H, Soldani P, Nicotra MR, Vola R, Rivella A, Creighton AM, Neri P, Mariani M. J Invest Dermatol 1996;107:164-70, WI Duenas M, Ayala M, Vazquez J, Ohlin M, Soderlind E, Borrebaeck C, Gavilondo JV. Gene 1995;158:61-6. 1271Popov S, Hubbard JG, Ward ES. Mol Immunol 1996;33:493-502. WI Ward RL, Clark MA, Lees J, Hawkins NJ. J Immunol Meth 1996;189:73-82. v91 Clark M, Hawkins NJ, Papaioannou A, Fiddes RJ, Ward RL. Clin. Exp. Immunol. 1997;109:166-74. [301 O’Callaghan CH, Morris A, Kirby SM, Shingler AH. Antimicrob Ag Chemother 1972;1:283-8. [311 Cagnon C, Valverde V, Masson J-M. Prot Eng 1991;4:843-7.
226
M.A.
Clark et al. / Immunotechnology
[32] Evan GI, Lewis GK, Ramsay G, Bishop JM. Mol Cell
Biol 1985;5:3610&6. [33] McCafferty J, Fitzgerald KJ, Earnshaw J, Chiswell DJ, Link J, Smith R, Kenten J. Appl B&hem Biotechnol 1994;47:157-73.
3 (1997) 217-226
[34] Zhu ZP, Zapata G, Shalaby R, Snedecor B, Chen H, Carter P. Bio/Technology 1996;14:192-6. [35] Nissim A, Hoogenboom HR, Tomlinson IM, Flynn G, Midgley C, Lane D, Winter G. EMBO J 1994;13:692-8. [36] Perez-Perez J, Gutierrez J. Gene 1995;158:141~2.