Development of a genetic selection for catalytic antibodies

Article abstract of DOI:10.1016/S0960-894X(02)00198-1

The design and evaluation of a new genetic selection system for evolving catalytic antibodies with aldolase activity are described. Through a series of model selections, we have identified selection conditions where expression of a catalytically active antibody confers a growth advantage to Escherichia coli. In addition, we provide evidence that the growth advantage is a direct result of catalytic activity.

Full text of DOI:10.1016/S0960-894X(02)00198-1

Bioorganic & Medicinal Chemistry Letters 12 (2002) 1691–1694
Development of a Genetic Selection for Catalytic Antibodies
Jeff Gildersleeve, Jeff Janes, Helle Ulrich, Priscilla Yang,
Carlos Barbas and Peter G. Schultz*
Department of Chemistry, The Scripps Research Institute, La Jolla, CA 92037, USA
Received 21 January 2002; accepted 14 February 2002
Abstract—The design and evaluation of a new genetic selection system for evolving catalytic antibodies with aldolase activity are
described. Through a series of model selections, we have identified selection conditions where expression of a catalytically active
antibody confers a growth advantage to Escherichia coli. In addition, we provide evidence that the growth advantage is a direct
result of catalytic activity. # 2002 Published by Elsevier Science Ltd.
Genetic selections have proven extremely valuable for
evolving enzymes with improved catalytic activity,
improved stability, and altered specificity.1 Conse-
quently, there has been considerable interest in devel-
oping genetic selections for enhancing the catalytic
activity of antibodies.2 For a genetic selection to work,
one must identify a set of conditions whereby expression
of an active enzyme/antibody confers a growth advan-
tage. The high catalytic activity of enzymes makes it
relatively easy to identify such conditions whereas the
catalytic activity of antibodies is typically lower than
that of enzymes, making it more difficult to develop
schemes in which a catalytically active antibody affords
a growth advantage. Moreover, other factors that affect
the growth rates of bacteria such as differences in toxi-
city and expression levels of antibodies can further
complicate the development of genetic selections. Here
we report a general selection strategy in which expres-
sion of a catalytically active antibody confers a growth
advantage to Escherichia coli. In addition, we provide
evidence that the observed growth advantage is a direct
result of catalytic activity.
use an E. coli strain (HDU78) that cannot biosynthesize
chorismate due to a defective chorismate synthase gene,
aroC.3 Chorismate is a key intermediate in the shiki-
mate pathway and is required for the biosynthesis of the
essential metabolites tyrosine, phenylalanine, trypto-
phan, para-aminobenzoic acid (PABA), para-hydroxy-
benzoic acid (PHB), and isochorismate.4 In principle,
any one of these nutrients could be used as a basis for
complementation as long as the remaining nutrients are
supplemented in the media making HDU78 a highly
flexible host. Importantly, we have shown that HDU78
is capable of expressing Fabs in active form. Yields of
Fab are in the range of 100 to 200 mg/L and are com-
parable with good E. coli expression strains such as
Top10 cells.5
The next step is the choice of selection marker and
reaction system. Complementation based on PABA
should be quite general since a range of chemical reac-
tions (e.g., hydrolytic, elimination, redox reactions, etc.)
can be coupled to the release of this nutrient. In addi-
tion, complementation based on PABA should be sen-
sitive enough to translate even modest catalytic activity
into a measurable growth advantage since very low
concentrations of PABA (1–10 nM) are required to sus-
tain growth of E. coli. To test the utility of PABA in
antibody-catalyzed complementation reactions, we
investigated the aldolase antibody 38C2developed by
the Lerner and Barbas groups.6 The antibody has broad
substrate specificity and, in particular, has been shown
to catalyze the activation of prodrugs such as 1 via a
retro-aldol retro-Michael reaction sequence (see Fig.
1).7,8 We anticipated that this system could be easily
modified for the release of PABA.
A first consideration for the development of a genetic
selection is the choice of the host auxotroph and com-
plementation strategy. A good selection host should be
easy to manipulate, capable of producing antibodies in
active form, and suitable for selections with a range of
substrates. Based on these requirements, we decided to
*Corresponding author. Fax: +1-858-784-9339; e-mail: schultz@
scripps.edu
0960-894X/02/$ - see front matter # 2002 Published by Elsevier Science Ltd.
PII: S0960-894X(02)00198-1

1692
J. Gildersleeve et al. / Bioorg. Med. Chem. Lett. 12 (2002) 1691–1694
Table 1. Results from the model selections
Entry
Substrate
(mM)
Temp.
(ꢁC)
Arabinosea
(%)
pHb
Enrichment
for 38C2
1
2
3
4
5
6
7
8
2 (20)
2 (10)
2 (5)
2 (1)
2 (5)
2 (5)
5 (10)
5 (10)
37
37
37
37
30
37
37
37
0.05
0.05
0.05
0.05
7.0
7.0
7.0
7.0
7.0
7.0
7.0
7.0
9
11
21
None
None
None
None
None
0.05
Figure 1. Retro-aldol retro-Michael activation of prodrug 1 and sub-
strate 2.
None
0.05
None
aSelections with 0.05% and 0.005% arabinose produced similar
results.
bSelections were also conducted at pH 7.5 and 8.0, but pH 7.0 was
found to be optimal.
containing colonies to 38KM containing colonies for
each of the cultures was then compared to the ratio for
the starting mix.
Figure 2. Synthesis of substrate 2. Reaction conditions: (a) ethyl iso-
cyanatobenzene, toluene, 105 ꢁC (99%); (b) NaOEt, EtOH/H2O (2:1),
55 ꢁC (75%); (c) OsO4, NMO, CH2Cl2, 1 h then Pb(OAc)4 (63%).
The results from the model selection are summarized in
Table 1. Using the best growth conditions (see Table 1,
entry 3), we obtained a 21-fold enrichment of bacteria
expressing 38C2demonstrating that expression of an
active antibody can confer a growth advantage using
our selection system. An enrichment of 21 after growth
over 30 generations indicates that bacteria expressing
38C2have an average growth rate that is ꢂ10% faster
than bacteria expressing 38KM. The growth advantage
is noteworthy given the modest activity of 38C2for
substrate 2 and the concentration of 2 in the selection. It
is important to note that a growth advantage of 10% or
more should be sufficient for selections on antibody
libraries as long as multiple rounds of selection are
conducted. For example, a 10% growth advantage over
150 generations would produce an enrichment of ꢂ106.
A genetic selection involving 38C2-catalyzed release of
PABA requires that substrate 2 is not converted to
PABA by any endogenous enzymes and the rate of the
uncatalyzed reaction in culture media is sufficiently low
to provide a selectable window. Substrate 2 was syn-
thesized as shown in Figure 2and then evaluated by
comparing the growth rates of HDU78 in the presence
of various concentrations of either PABA or 2.9
Approximately 1 nM PABA is required to detect growth
while 10 nM PABA is required to produce normal
growth of HDU78. For comparison, 1 mM 2 is required
to detect growth and >20 mM 2 is required for normal
growth. These results demonstrate that the uncatalyzed
rate of conversion of 2 to PABA is sufficiently low to
permit selections using this substrate. Finally, com-
pound 2 was found to be an acceptable substrate for
38C2with similar kinetic parameters as other retro-
aldol-retro-Michael substrates (kcat=0.0069 minꢀ1 at
37 ꢁC, Km=150 mM).10,11
Although enrichment for bacteria expressing an active
antibody over bacteria expressing an inactive antibody
was observed, this result alone did not rule out the pos-
sibility that the difference in growth rates was a result of
subtle differences in toxicity, expression levels, etc.,
between the antibodies. However, no enrichment could
be detected in the absence of Fab expression (i.e., no
induction with arabinose – see Table 1, entry 6), and no
enrichment could be detected when cells were grown in
the presence of 5 (an isostere of 2 that is not a substrate
for 38C2, see entry 7). These additional experiments
provide evidence that the growth advantage is a direct
result of catalytic activity.
To determine if expression of an active antibody confers
a growth advantage and identify the optimal growth
conditions, a series of model selections were carried out
with bacteria expressing 38C2. Bacteria expressing
38KM, a catalytically inactive mutant of 38C2, were
used as a control.12,13 By comparing 38C2with a closely
related mutant, any differences in growth rates due to
differences in toxicity, expression levels, etc. should be
minimized. Bacteria expressing 38C2and bacteria
expressing 38KM were mixed and small aliquots of this
stock mix were used to inoculate selection cultures. The
cultures were then grown under various conditions (i.e.,
1–20 mM substrate 2, 0–0.05% arabinose, pH 7.0–8.0,
30–37 ꢁC) to an OD600 of 1.0, diluted, and then re-
grown over approximately 30 generations.14 The starting
mix and cultures were plated out on LB/amp/glucose
plates and colony PCR reactions were performed with
primers specific for either 38C2or 38KM to determine
the identities of individual colonies. The ratio of 38C2
In summary, we have developed a new genetic selection
system for catalytic antibodies based on complementa-
tion of an E. coli auxotroph lacking chorismate synthase
activity. Through a series of model selections, we have
successfully identified selection conditions where
expression of a catalytically active antibody confers a
growth advantage. In addition, we provide evidence
that the growth advantage is a direct result of catalytic
activity. It is important to note that the selection system
is sensitive enough to translate even modest catalytic
activity into a measurable growth advantage. By gen-
erating large combinatorial libraries of antibodies and

J. Gildersleeve et al. / Bioorg. Med. Chem. Lett. 12 (2002) 1691–1694
1693
conducting iterative rounds of selection and DNA
shuffling, we hope to identify highly active aldolase
antibodies.
3H), 1.25 (s, 3H); 13C NMR (CDCl3, 62.5 MHz) d 166.2,
153.2, 142.1, 130.8, 125.0, 117.5, 115.4, 77.2, 71.4, 62.2, 60.8,
50.0, 40.5, 27.1, 24.9, 14.3; MALDI FTMS calcd for
C18H25NO5Na (M+Na+) 358.1625, found 358.1619. Com-
pound 5: Rf 0.30 (20% acetone/chloroform +0.5% acetic
1
acid); H NMR (CDCl3, 250 MHz) d 8.08 (d, J=8.8 Hz, 2H),
Acknowledgements
7.52(d, J=8.8 Hz, 2H), 7.10 (br s, 1H), 4.97 (br s, 1H), 4.78
(br s, 1H), 4.35–4.45 (m, 2H), 2.27 (d, J=13.3 Hz, 1H), 2.22
(d, J=13.3 Hz, 1H), 1.87–1.99 (m, 2H), 1.85 (s, 3H), 1.28 (s,
3H); 13C NMR (acetone-d6, 62.5 MHz) d 167.9, 154.9, 145.3,
144.7, 132.2, 125.7, 118.8, 115.3, 72.2, 63.3, 51.7, 42.1, 29.5,
28.1; MALDI FTMS calcd for C16H21NO5Na (M+Na+)
330.1312, found 330.1315.
This research was supported by the National Institutes
of Health (GM56528). J.G. gratefully acknowledges a
postdoctoral fellowship from the NIH (Grant No. 5
F32AI10419). J.J. gratefully acknowledges a pre-doc-
toral fellowship from the Howard Hughes Medical
Institute. The authors also wish to thank Professor
R.A. Lerner for a generous gift of 38C2IgG. This is
manuscript number 14412-CH of The Scripps Research
Institute.
10. Initial reaction rates for the retro-aldol-retro-Michael
reaction of substrate 2 to PABA were measured using a
reverse-phase HPLC assay. Reactions were incubated at 37 ꢁC
and reaction progress was monitored by periodic injection of a
60 mL sample from the reaction mixture onto a reverse-phase
C-18 HPLC column (Rainin protein and peptide C18) with
UV detection at 275 nm. Products were eluted using the fol-
lowing conditions: buffer A=water+0.1% acetic acid, buffer
B=acetonitrile+0.1% acetic acid; HPLC gradient=1.0–7 min
95:5 A:B; 2. ramp to 50:50 A:B over 3 min; 3. ramp to 20:80
A:B over 10 min; 4. ramp to 0:100 A:B over 2min; 5. 0:100
A:B for 1 min; 6. ramp to 95:5 A:B over 2min; 7. 95:5 A:B for
5 min. Substrate 2 and PABA had retention times of 14.3 min
and 7.7 min, respectively.
References and Notes
1. For a recent review, see: Kasr, P.; Hilvert, D. Pure Appl.
Chem. 1996, 68, 2017.
2. (a) For previous work on the development of genetic selec-
tions for catalytic antibodies, see: Tang, Y.; Hicks, J. B.; Hil-
vert, D. Proc. Natl. Acad. Sci. U.S.A. 1991, 88, 8784. (b)
Lesley, S. A.; Patten, P. A.; Schultz, P. G. Proc. Natl. Acad.
Sci. U.S.A. 1993, 90, 1160. (c) Smiley, S. J.; Benkovic Proc.
Natl. Acad. Sci. U.S.A. 1994, 91, 8319.
11. With a kcat=0.0069 minꢀ1, 100 mg/L 38C2Fab (2nM)
could produce ꢂ20 nM PABA in a 24 h period.
12. The expression vector for the selections, pFH4, was gen-
erated from plasmid p4xH5 as follows. First, the alkaline
phosphatase promotor (PhoA) was removed by EcoRI/SpeI
digestion and replaced by an insert containing the araC gene,
the arabinose promotor (PBAD) and its regulatory sequences
(O1, O2, I1, I2, CAP-binding site) amplified from vector
pBAD/lamB (Maxygen) using primers PLY100 (50-AGGTC-
3. E. coli strain HDU78 [hdsF mcr13 araD13 9 Á(araABC-
leu) 7679 D LacX74 galUgalK rpsL thi- aroC purF::Tn10 ]
was generated from strain MC1061 [hdsF mcr13 araD139
D(araABC-leu)7679 D LacX74 galUgalK rpsL thi- ] through a
series of P1 transductions. Briefly, a P1 lysate was prepared
from strain NK6073 (obtained from B. Bachmann, E. coli
Genetic Stock Centre) which contains a tetracycline resistance
GAATTCGCATAATGTGCCTGTCAAATGGA-30)
and
marker (TetR) in the form of
a
transposable element
PLY101 (50-AGGTCACTAGTTGATGTATATCTCCTTCT-
TAAAGT-30). The BstEII restriction site located between the
O2 region and PBAD was mutated to introduce a SmaI site,
leaving a unique BstEII site in the JH region of the heavy chain
for cloning of VH genes. A ClaI site was similarly introduced
using primer S033 (50-CGCTTCAGCCAATACTTTAT-
CGATTCCCGCCATTCAGAGA-30) between the araC and
the O2 region to facilitate the diversification of the promotor
region along with the Fab fragment genes in expression
libraries. Finally, the stII leader peptide upstream of Vk was
replaced by the pelB leader sequence using primers S044 (50-
CTAGTATGAAATACCTATTGCCTACGGCAGCCGCTG-
GATTGTTATTACTCGCGGCCCAGCCGGCCATGGCCG-
AGCT-30) and S045 (50-CGGCCATGGCCGGCTGGGCC-
GCGAGTAATAACAATCCAGCGGCTGCCGTAGGCAA-
TAGGTATTTCATA-30). An optimized ribosomal binding
site was also introduced in this step.
(purF::Tn10) close to the wild type aroC locus. This lysate was
used to introduce the TetR marker into AB2849 (B. Bach-
mann), an E. coli strain that carries a defective chorismate
synthase gene (aroC ). A resulting strain possessing both the
TetR marker and the aroC auxotrophy (HDU73) was used to
transduce the aroC TetR combination into strain MC1061 to
produce HDU78.
4. Walsh, C. T.; Lui, J.; Rusnak, R.; Sakaitani, M. Chem.
Rev. 1990, 90, 1105.
5. Ulrich, H. D.; Patten, P. A.; Yang, P. L.; Romesberg, F. E.;
Schultz, P. G. Proc. Natl. Acad. Sci. U.S.A. 1995, 92, 11907.
6. For a lead reference see: Wagner, J.; Lerner, R. A.; Barbas,
C. F., III Science 1995, 270, 1797.
7. Shabat, D.; Rader, C.; List, B.; Lerner, R. A.; Barbas, C. F.,
III Proc. Natl. Acad. Sci. U.S.A. 1999, 96, 6925.
8. While 38C2catalyzes both steps of prodrug activation, the
retro-aldol reaction is the rate-determining step.
13. 38KM was produced by mutating the codon for the active
site lysine (AAA, heavy chain residue 93) to ATG (coding for
methionine) using a QuikChange kit (Stratagene) with primers
JG1 (50-CTGAAGACACTGGAATTTATTACTGTATGA-
9. Experimental data for compounds 2, 4, and 5 are listed
below. Compound 2: Rf 0.30 (25% acetone/chloroform
+0.5% acetic acid); 1H NMR (CDCl3+5% CD3OD,
250 MHz) d 8.31 (br s, 1H), 7.99 (d, J=8.8 Hz, 2H), 7.50
(d, J=8.8 Hz, 2H), 4.32 (t, J=6.9 Hz, 2H), 2.70 (m, H), 2.20
(s, 3H), 1.85–2.01 (m, 2H), 1.30 (s, 3H); 13C NMR (CDCl3,
62.5 MHz) d 210.2, 167.8, 154.2, 145.2, 132.1, 125.8, 118.8,
71.3, 62.9, 54.5, 41.6, 32.5, 28.2; MALDI FTMS calcd for
C15H19NO6Na (M+Na+) 332.1105, found 332.1098. Com-
pound 4: Rf 0.30 (30% ethylacetate/hexane); 1H NMR
(CDCl3, 250 MHz) d 7.99 (d, J=8.8 Hz, 2H), 7.48 (d, J=8.8Hz,
2H), 7.25 (br s, 1H), 4.96 (br s, 1H), 4.77 (br s, 1H), 4.30–4.50
(m, 4H), 2.26 (d, J=13.5 Hz, 1H), 2.20 (d, J=13.5 Hz, 1H),
2.1 (br s, 1H), 1.86–1.95 (m, 2H), 1.84 (s, 3H), 1.38 (t, 7.0 Hz,
0
CCTATTTTTACTCATTTTCTTACTGGG-30) and JG2(5 -
CCCAGTAAGAAAATGAGTAAAAATAGGTCATACAG-
TAATAAATTCCAGTGTCTTCAG-30).
14. Model selections were conducted as follows. A stock mix-
ture of 38C2and 38KM containing cells was produced by
suspending a single colony of HDU78 containing the plasmid
pFH4.38C2and a single colony of HDU78 containing the
plasmid pFH4.38KM in 1 mL of supplemented minimal media
(M9 salts containing 0.5% glycerol, 0.8 mM ala, 5.2mM arg,
0.4 mM asp, 0.4 mM asn, 0.1 mM cys, 0.6 mM glu, 0.6 mM
gln, 0.8 mM gly, 0.2mM his, 0.4 mM ile, 0.8 mM leu, 0.4 mM

1694
J. Gildersleeve et al. / Bioorg. Med. Chem. Lett. 12 (2002) 1691–1694
lys, 0.2mM met, 0.4 mM phe, 0.4 mM pro, 10 mM ser, 0.4 mM
thr, 0.1 mM trp, 0.2mM tyr, 0.6 mM val, 0.01 mM thiamine,
0.01 mM pantothenate, 0.01 mM para-hydroxybenzoic acid,
0.01 mM dihydroxybenzoic acid, 0.4 mM adenine, 1 mM
MgSO4, 0.1 mM CaCl2, and 0.01 mM FeSO4). This stock mix
was then diluted by a factor of 50 and 50 mL of the diluted
stock mix was used to inoculate each 5 mL selection culture
(e.g., supplemented minimal media containing 0.05% arabi-
nose, 70 ug/mL ampicillin, pH 7.0, and 5 mM substrate 2).
Selection cultures were grown in a shaker (250 rpm, 30 ꢁC or
37 ꢁC) to an OD600=1.0, diluted into fresh selection media
and then re-grown to an OD600=1.0 etc. over approximately
30 doublings. The starting stock mix and the final selection
cultures were plated on LB/Amp/Glucose plates and single
colonies were analyzed by two different colony PCR reactions:
one containing primers jj24 (50-AATTTTGTTTAACTTTAA-
GAAGGAG-30) and jg5 (50-GTAAGAAAATGAGTAAAA-
ATAGGTTT-30, specific for 38C2) and another containing
primers jj24 and jg6 (50-GTAAGAAAATGAGTAAAAA-
TAGGTCA-30, specific for 38KM).