Novel screening methods - The key to cloning commercially successful biocatalysts

Article abstract of DOI:10.1016/S0968-0896(99)00146-7

Providing sufficient biocatalyst to support the demands of multi tonne product supply can be problematical. Here we describe how screening for and cloning a γ-lactamase overcame biocatalyst supply issues, and greatly improved the actual biocatalytic process. The isolation of an expressing γ- lactamase clone from a gene library necessitated a combination of classical molecular biology techniques together with innovative screening methods to identify a functional clone. Once isolated the enzyme was characterised with regard to its process performance and proved to be active at 500 g L^-1 substrate. Further development of the recombinant fermentation and downstream processing has resulted in the ability to produce sufficient biocatalyst from one 500 l fermentation to resolve 5 metric tonnes of (±)-lactam, whilst simplifying the process chemistry greatly.

Full text of DOI:10.1016/S0968-0896(99)00146-7

Bioorganic & Medicinal Chemistry 7 (1999) 2163±2168
Novel Screening MethodsÐThe Key to Cloning Commercially
Successful Biocatalysts
Stephen J. C. Taylor,* Rob C. Brown, Phil A. Keene and Ian N. Taylor
Chirotech Technology Ltd., Unit 283, Cambridge Science Park, Milton Road, Cambridge, CB4 0WE, UK
Received 27 January 1999
AbstractÐProviding sucient biocatalyst to support the demands of multi tonne product supply can be problematical. Here we
describe how screening for and cloning a g-lactamase overcame biocatalyst supply issues, and greatly improved the actual bioca-
talytic process. The isolation of an expressing g-lactamase clone from a gene library necessitated a combination of classical mole-
cular biology techniques together with innovative screening methods to identify a functional clone. Once isolated the enzyme was
�
1
characterised with regard to its process performance and proved to be active at 500 g L substrate. Further development of the
recombinant fermentation and downstream processing has resulted in the ability to produce sucient biocatalyst from one 500 l
fermentation to resolve 5 metric tonnes of (‹)-lactam, whilst simplifying the process chemistry greatly. # 1999 Elsevier Science
Ltd. All rights reserved.
Introduction
We reported a biocatalytic resolution of 2-aza-
3
bicyclo[2.2.1]hept-5-en-3-one several years ago. By
screening several soil and sewage samples, using N-ace-
tyl-l-phenylalanine as a sole source of carbon, many
microbes with the ability to perform amide bond
hydrolysis were isolated. Subsequent rescreening of
these for hydrolysis of the g-lactam revealed a strain of
Pseudomonas cepaecia which was highly selective for the
hydrolysis of the (+)-lactam, leaving the desired (� )
enantiomer. The whole-cells gave an enantiomeric ratio
(E value) of 94 (Scheme 1).
When a commercially available enzyme cannot be found
for a biotransformation then the researcher has to ®nd
one from the environment, and usually this will be from
a microbial source. As a biotransformation based pro-
cess matures, demand for the product increases, and
investment is required to ®nd the best possible biocata-
lyst which is highly ecient, selective and cheap. This
will most commonly emerge from a focussed pro-
gramme of screening for the microbe, enzyme isolation
and characterisation, then cloning of the enzyme for
over-expression. In this paper we describe such a pro-
gramme for the production of optically pure 2-aza-
Several attempts were made to isolate and purify the
lactamase, but it proved too unstable for the manipula-
tions required for isolation and puri®cation. As a result,
the biotransformation process was developed using the
whole-cell biocatalyst. Whilst this process allowed the
development of a tonne scale (� )-lactam process, as
required volumes grew further, it became more dicult
to satisfy demand. The process had several drawbacks.
The use of whole cells, and the subsequent lysis of these
during the biotransformation (and lysis from frozen
storage) complicated the isolation of the lactam. Direct
solvent extraction was impossible and a complex carbon
adsorption and elution cycle was needed. Whilst the
volume eciency of the biotransformation was toler-
able, large volumes were encountered during the work-
up. Also, the cells proved to be somewhat sensitive to
the quality of racemic lactam used, which varied
depending on the source. Furthermore, the quantity of
cells required (a kg of cells yielded a kg of (� )-lactam)
bicyclo[2.2.1]hept-5-en-3-one [(� )-g-lactam] using
a
lactamase; highlighting this as an example of a novel
method used to identify clones, and to show the bene®ts
of a cloned enzyme on the overall process.
2-Azabicyclo[2.2.1]hept-5-en-3-one is now well estab-
lished as a very versatile synthon for the production of
1
carbocyclic nucleosides (Scheme 1). Such compounds,
where the ribose oxygen of the nucleoside has been
replaced by a methylene, have become very valuable as
chemotherapeutic agents in the ®ght against viral infec-
2
tions such as HIV or herpes.
* Corresponding author. Tel.: +44(0)-1223-420430; e-mail: stevetaylor
@chirotech.com
0968-0896/99/$ - see front matter # 1999 Elsevier Science Ltd. All rights reserved.
PII: S0968-0896(99)00146-7

2
164
S. J. C. Taylor et al. / Bioorg. Med. Chem. 7 (1999) 2163±2168
�
1
� 1
ꢀ
Scheme 1. Carbocyclic nucleosides from 2-azabicyclo[2.2.1]hept-5-en-3-one. (i) 50 g L Pseudomonas cepaecia, 100 g L racemic lactam, 25 C 24 h.
contributed to a signi®cant proportion of the manu-
facturing cost. Thus, there was a considerable driving
force to invest in the research needed to ®nd a better
biocatalyst and process. A further screen for lactamases
was performed, again using N-acetyl-l-phenylalanine as
the sole source of carbon, and from the several positives
isolated, a temperature stability study was performed on
the lactamase in crude cell lysates. This indicated one
to be considerably more stable than the P. cepaecia
successful due to the degeneracy of the sequence. This
approach may also identify non-expressing forms of the
lactamase such as truncated versions of the gene, and a
direct screen for activity is more desirable, and in this
case neccessary.
Conventional biocatalyst screening methods such as
emulsi®ed substrate overlay (where enzyme activity can
be visualised directly in the form of a clearing zone
around the biocatalyst) were not appropriate for the
4
strain, and indeed retained half of its activity after
incubation at 60 C for 4 h. The strain was later identi®ed
ꢀ
as Comamonas acidovorans.
� 1
lactam substrate due to its high solubility (>500 g L in
water). Furthermore, other staining methods such as
ninhydrin or glutaraldehyde did not work in an agar
overlay with a soluble substrate because di€usion
quickly diluted any visible e€ects. A di€erent approach
was needed, and so a very simple screen was developed,
combining the classical techniques of replica plating and
ninhydrin staining. A Whatman ®lter-paper was dipped
in a methanolic solution of (+)-lactam, then allowed to
dry. This impregnated the paper with (+)-lactam sub-
strate which was then placed directly onto the agar plate
where it picked up moisture and most of the cells in
each colony. The physical transfer of cells gave sucient
biomass to allow a biotransformation to occur, but the
aqueous content of the paper was limited and prevented
excessive di€usion of product away from the cells. The
paper was incubated for several hours, dried and stained
with ninhydrin. A positive clone was clearly visualised
as a colony with a brown halo on the light background
(Fig. 2). The paper colony pattern was then matched
with the regrown cells on the agar plate, and the desired
colony isolated.
Results and Discussion
The g-lactamase was puri®ed to homogeneity by a
sequence of ammonium sulfate precipitation, HIC
butyl Sepharose), anion exchange chromatography (Q
Sepharose) and gel ®ltration (Sephacryl S200), giving a
single band by SDS-PAGE. This indicated a molecular
weight of 53±55 KDa. The N-terminal sequence was
then determined, as shown in Figure 1.
(
A random gene library of the Comamonas acidovorans
was prepared by standard techniques. Genomic DNA
was partially digested by Sau3A I, then DNA fragments
from 2±6 kb isolated after size separation by electro-
phoresis. Fragments were ligated into the dephos-
phorylated BamH1 site of pUC19, and the vector
transformed into E. coli DH5a.
The detection of an expressing clone was attempted by
the use of a radioactive oligonucleotide probe based on
the N-terminal amino acid sequence. The probes used
are shown in Figure 1. However, this approach was not
DNA sequence analysis of the insert showed the frag-
ment to incorporate an open reading frame (ORF) of
1.6 kb, shown in Figure 3. This was driven by the
Figure 1. Derived N-terminal amino acid sequence of puri®ed Comamonas acidovorans g-lactamase.

S. J. C. Taylor et al. / Bioorg. Med. Chem. 7 (1999) 2163±2168
2165
was used. Regulation of enzyme expression was not
particularly well controlled and the fermentation did
not require induction. A 500 L, 4 day fermentation yiel-
�
1
ded approximately 100 g L (wet weight) of cell paste
�
1
and 3000 units g cell paste, sucient to resolve about
tonnes of racemic substrate.
5
Having optimised the fermentation, we focused on
recovering the lactamase in a form suitable for use in
the biotransformation. The goal was a simple process
that gave suciently pure lactamase such that the pro-
tein content did not cause emulsions during isolation of
the lactam. After lysis of the cells with a combination of
lysozyme and Triton-X100, the bulk of nucleic acids
and cell debris were removed by polyethyleneimine pre-
cipitation and centrifugation. The lactamase was pre-
cipitated from the ®ltrate by standard ammonium
sulfate precipitation, before recovery by centrifugation.
The pellet was ®nally redissolved in Tris±HCl bu€er,
sterile ®ltered ready for use. This was achieved in an
overall yield of about 70%.
With a readily available supply of semi-puri®ed recom-
binant lactamase available, a much improved process
was quickly developed. The biotransformation was
optimised with respect to substrate concentration,
enzyme loading, pH, bu€er and temperature. This
Figure 2. Ninhydrin stain of ®lter paper colony transfer. Genomic
library of Comamonas acidovorans in E. coli screened for (+)-g-lacta-
mase acitivity by incubation of adsorbed colonies on (+)-lactam
impregnated ®lter discs. Brown halo apparent around single, central
colony indicating the conversion of lactam to amino acid product.
� 1
highlighted 500 g L substrate, 100 mM Tris±HCl at pH
ꢀ
has been much improved (Scheme 2). Volume eciency
7.5, 25 C to be optimal. The new (� )-lactam process then
�
1
is much higher where the enzyme tolerates 500 g L input
of lactam, a 5-fold increase. Since a clean concentrated
enzyme is used, the isolation was greatly simpli®ed where
direct dichloromethane solvent extraction of the lactam
can be achieved, avoiding the carbon adsorption step, and
associated problems of low volume eciency, handling
and safety issues. The ®nal step is a recrystallisation of the
product from the solvent concentrate.
upstream lac promoter of pUC19, and translates to a
protein of 575 residues (61 kDa). The deduced amino
acid sequence of the translated ORF showed >65%
homology to the acetamidase from Mycobacterium
8
smegmatis and Methylophilus methylotrophus. These
enzymes have been shown to hydrolyse short chain fatty
8
acylamides.
From the cloned gene sequence the N-terminal sequence
was identi®ed. The lactamase was found to have an
extra 31 amino acids upstream of the N-terminus, which
arose by a fortuitous ligation of a short piece of DNA
during the cloning. However, experiments showed that
enzyme expression was good, and subsequent manip-
ulation of the gene to remove these gave no added ben-
e®t. Final modi®cation of the recombinant plasmid was
the insertion of the cer element responsible for multi-
meric resolution and stable inheritance of the wild type
Conclusion
Cloning an enzyme has the obvious bene®t of reducing
the cost of the biocatalyst through over-expression.
However, what can be equally important is the dramatic
impact that use of a cloned enzyme can have on the
overall design of a process, where product recovery in
particular becomes much easier. By carefully planning
the screening strategy, a cloning process can be stream-
lined, as we have shown for the lactamase. The impreg-
nated ®lter paper assay should be applicable for any
soluble substrate/product where a selective staining
method exists.
6
E. coli plasmid ColE I. The cer element was transferred
6
from construct pKS492. The ®nal construct and
expression vector were designated pPET1 (Fig. 4). The
vector was transformed into E. coli MC1061 the desig-
nated recombinant lactamase host. The level of recom-
binant protein was signi®cant to be visualised on
Coomassie SDS-PAGE under induced growth.
Experimental
De®nition of units
The fermentation of the recombinant E. coli was based
on a complex media/feed with glycerol as the carbon
source and peptone/casein as a source of amino acids,
peptides and nitrogen. Initial growth was in a batch
mode after which a continuous feed containing glycerol
For our purpose 1 unit is de®ned as the amount of bio-
catalyst required to produce 1 g of product per hour.
�
1
This may be expressed in units mL for fermentation
�
1
broth or units g for puri®ed protein.

2
166
S. J. C. Taylor et al. / Bioorg. Med. Chem. 7 (1999) 2163±2168
Figure 3. Complete nucleotide sequence of the (+)-g-lactamase gene and deduced amino acid sequence from Comamonas acidovorans strain CMC
093. The 1.9 kb Sau3A I fragment ligated in-frame with the BamH I restriction sites of pUC19 backbone.
4

S. J. C. Taylor et al. / Bioorg. Med. Chem. 7 (1999) 2163±2168
2167
Plasmids, bacterial strains and culture conditions
5
Plasmid pUC19 was used for the library construction,
sequencing and cloning. The cloning vector was
improved for stable expression by incorporation of the
6
cer element. C. acidovorans strain CMC4093, used a
source of genomic DNA, was grown on BA medium
(
50 mM KH PO , 10 mM K PO , 4 mM MgSO .
2 4 2 4 4
7
trace elements solution (25 mM CaCl . 2H O, 25 mM
.5 mM (NH ) SO , 1% (w/v) yeast extract, 0.1% (v/v)
4 2 4
2
2
ZnO, 5 mM CuCl , 10 mM MnCl , 20 mM FeCl
2
2
2
1
0 mM CoCl , 20 mM Na MoO ., 5 mM H BO , 3.3 M
2 2 4 3 3
HCl [pH 7.0 with NaOH])). E. coli DH5a was used for
library construction and preliminary expression of
genomic fragments ligated into pUC19. E. coli DH5a
ꢀ
harbouring recombinant plasmids was grown at 37 C
on TSB medium (Oxoid Ltd.) with ampicillin and iso-
propyl b-d-thiogalactopyranoside (IPTG). Recombi-
nant E. coli strain was inoculated into a 1 L ba‚ed
shake ¯ask containing 100 ml TSB medium (Oxoid Ltd.)
�
1
supplemented with ampicillin (100 mg mL ). The ¯ask
and inoculum were incubated for 16 h at 37 C, shaking
ꢀ
at 300 rpm in an orbital shaker (25 mm throw). The seed
culture was inoculated (1% v/v) into a 2.8 L laboratory
bioreactor vessel containing 1.5 L TSB medium. The
Figure 4. Schematic diagram of plasmid pPET1. E. coli plasmid
pPET1 was derived from pUC19, which harbours a 1.9 kb Sau3A I
genomic fragment from Comamonas acidovorans CMC1493, ligated
into the BamHI restriction site. The cer stability element of the wild
type plasmid ColE 1 was inserted at the 3 to the lactamase fragment
via a BamHI (partial) and NdeI restriction.
ꢀ
temperature was maintained at 25 C, pH 7, and dis-
solved O tension at >50%. Growth was monitored at
0
2
5
After 24 h growth, cells were harvested by centrifuga-
20nm optical density against a TSB medium blank.
ꢀ
20 C until required.
tion (5000 g at 4 C for 10 min) Cells were stored at
ꢀ
Materials
�
All restriction enzymes and DNA modi®cation enzymes
were purchased from NBL Gene Sciences Ltd.; Agarose
was obtained from IBI Kodak Ltd.; qualitative ®lter
papers Type 2, 125 mm 1 (Whatman Ltd.); all other
chemical reagents were acquired from Fisher Scienti®c
UK.
Puri®cation of lactamase from C. acidovorans
A 10% (w/v) cell suspension in 1 mM NaHCO , 10 mM
3
�
1
EDTA, 0.1% (w/v) triton X-100, 1 mg mL lysozyme
was incubated overnight at 25 C. Debris was removed
ꢀ
by the addition of 0.5% w/v PEI and centrifugation at
ꢀ
1
0,000 g, 10 min, 4 C.
Biotransformation assay
A 1 h single time point assay is routinely used. A dilu-
tion of biocatalyst is incubated in 0.1 M Tris±HCl, pH
Hydrophobic interaction chromatography
7
.5 (900 ml) in a 20 mL glass scintillation vial. The reac-
After precipitation of lactamase from the cell lysate by
addition of solid (NH ) SO to 50% saturation, the pellet
was resuspended in 25 mM KH PO /K HPO +1.5 M
2 4 2 4
�
1
tion is started by addition of 100 mL 100 gL (‹) lac-
tam. The reaction is then incubated for 1 h at 25 C in an
4
2
4
ꢀ
orbital shaker of 2.5 cm throw at 250 rpm. The reaction
is stopped by taking a 100 ml aliquot and dilution in to
(NH ) SO pH 7.0. This was then applied to a butyl
4 2 4
sepharose (Pharmacia) column (XK16Â16 cm), total
volume 32 mL, pre equilibrated in the same bu€er. Elu-
tion was with a decreasing salt gradient over 20 column
900 ml HPLC mobile phase. The extent of the reaction
is then monitored using HPLC. Column: Kromasil
KR100 C8, mobile phase 50:50 (v/v) MeOH: 10mM
�
1
volumes at 2 mL min . Lactamase eluted at around
0.5 M (NH ) SO . Active fractions were pooled and
dialysed against 100 mM Tris±HCl pH 8.0.
�
1
K HPO pH 7.0. Flow 1 mL min . Detection by UV
2
4
4 2
4
monitor at 225 nm.
�
1
� 1
Scheme 2. Bioresolution of 2-azabicyclo[2.2.1]hept-5-en-3-one. (i) 5 mL cloned lactamase, 500 gL racemate, E>400.

2
168
S. J. C. Taylor et al. / Bioorg. Med. Chem. 7 (1999) 2163±2168
Anion exchange chromatography
chloroform extraction and subsequent ethanol pre-
cipitation. Puri®ed genomic fragments were ligated into
the dephosphorylated BamHI site of pUC19. The liga-
tion mix was then used to transform competent E. coli
The lactamase was applied to a Q sepharose (Pharma-
cia) column (XK10Â8 cm), 6 mL bed volume, eluting
with a 0.1 M NaCl increasing linear gradient in 100 mM
Tris pH 8, ¯ow 2 mL min . Lactamase eluted at 0.25 M
NaCl.
5
DH5a. Transformants were screened on TSB agar
�
1
� 1
plates containing ampicillin (100 mg ml ) and 1 mM
IPTG. The screening procedure involved adsorption of
�
1
recombinant colonies onto (+)-lactam (2 mg mL dis-
solved in methanol) impregnated ®lter paper discs (Type
Gel ®ltration
2
were incubated at 37 C for 4 h, dried and developed
, Whatman). The replicate colonies and ®lter papers
ꢀ
with 2% (w/v) ninhydrin in acetone at 80 C for 5 min.
Active fractions were concentrated using Amicon 30 K
nmwl ultrafree microcentrifuge concentrators. Protein
was applied to a Sephacryl S200HR (XK16Â60 cm),
ꢀ
The presence of the acid product was identi®ed as a
brown halo surrounding the recombinant colony.
�
1
1
00 mM NaCl 10 mM Tris (pH 8.0), ¯ow 1 mL min .
Protein was pure by Coomassie stained SDS-PAGE,
approximate mw 53±55 kDa. Electroblotting of the
lactamase from the SDS-PAGE gel onto a PVDF
membrane (Biorad) allowed the N-terminal sequence to
be determined. The N-terminal sequence is shown in
Figure 1.
A single colony displaying the characteristic brown halo
of activity was isolated (Fig. 2). Plasmid DNA was
prepared from the lactamase expressing clone, and
restriction digest analysis showed the presence of a 1.9
kb Sau3A I restriction fragment.
Subsequent DNA sequencing. was performed by Dye
Terminator Cycle Sequencing using AmpliTaq poly-
merase FS (ABI) with the ABI 377 DNA Sequencer
(ABI). Overlapping sequence reads were aligned into a
single open reading frame using the DNAstar Inc.
Lasergene software.
Cloning and sequencing
Total genomic DNA was prepared from C. acidovorans
by an initial lysozyme treatment. 2 g of cell pellet was
mixed with 10 mL of lysis bu€er (10 mg mL lysozyme
in 50 mM glucose, 20 mM EDTA, 25 mM Tris±HCl pH
�
1
ꢀ
8
.0), cell suspension was incubated at 37 C for 30 min,
�
1
1
incubation was elevated to 50 C for 30 min. Cell lysis
mL of proteinase K (100 mg mL ) was added and the
ꢀ
was completed by the addition of 5 mL 10% (w/v) SDS
References
ꢀ
and gentle inversion at 50 C for a further 30 min. Solid
cesium chloride was dissolved in the cell lysate at 1 g
1. Daluge, S.; Vince, R. J. J. Org. Chem. 1978, 43, 2311.
2. Marquez, V. E.; Lim, M. I. Med. Res. Rev. 1986, 6, 1.
3. Taylor, S. J. C. et al. Tetrahedron: Asymmetry 1993, 4 (6),
1117.
4. Brabban, A. B.; Littlechild, J.; Wisdom, R. J. Industrial
Microbiol. 1996, 16, 8±14.
�
1
mL . Finally ethidium bromide was added to the
1
�
suspension at a concentration of 200 mg mL
and
2
loaded into a 13 mL ULTRACRIMP (Sorvall) ultra-
centrifuge tube. The genomic DNA was resolved by
ultracentrifugation (140,000 g, 20 C, 72 h), extracted
ꢀ
5
1
6
7
. Yanish-Peron, C.; Vieira, J.; Messing, J. Gene 1985, 33, 103±
19.
. Summers, D. K.; Sherratt, D. J. EMBO J. 1988, 7, 851±858.
. Sambrook, J.; Fritsch, E. F.; Maniatis, T. Molecular Clon-
with cesium chloride saturated butanol and dialysed in
TE bu€er pH 8.0. The DNA (250 mg) was then partially
digested with Sau3A I (20U) restriction endonuclease
for 1±10 min and electrophoresed in a 0.8% agarose
ing: A Laboratory Manual. Cold Spring Harbor Laboratory
Press, New York, 1989.
8. Draper, P. J. Gen. Microbiol. 1969, 46, 111±123.
7
TAE gel. Fragments ranging from 2±6 kb were excised
from the gel, electroeluted and puri®ed by phenol: