Log in using OpenID

Methods in Enzymology
Volume 328
zca zons o
enes an
Part C
Protein-Protein Interactions and Genomics
Jeremy Thorner
Scott D. Emr
John N. Abelson
San Diego London
Boston New York Sydney ·rokyo 'f'oronto
[22] Selectively Infective Phage Technology
reprinted from:
K. M. Arndt, S. Jung, C. Krebber and A. Plückthun, Selectively infective phage technology.
Methods Enzymol. 328, 364-388 (2000).
In this chapter we describe a selection technology related to phage
display, the selectively infective phage technology (SIP). In both SIP and
phage display, libraries of proteins or peptides of interest are displayed on
the tip of a filamentous phage fused to the gene 3 protein (g3p). However,
in contrast to conventional phage display, in which separation of binders
with an immobilized ligand is required, in SIP technology the selection
step is carried out in solution and directly coupled to the infection event.
Therefore, no elution of binders is ever necessary, and incomplete elution
of the most tightly binding molecules is not a limitation.
This is achieved by taking advantage of the modular structure of g3p,
which occurs probably in five copies on the tip of the filamentous phage
and consists of the N-terminal domains Nl [68 amino acids (aa)], N2 (131
aa), and the C-terminal domain CT (150 aa), connected by the glycine-rich
linkers Gl (18 aa) and 02 (39 aa) (Fig. lA and Fig. 2A). 1•2 In a SIP phage
theN-terminal domains required for infection (Nl or Nl-N2) are replaced
by a protein or peptide of interest in all copies of g3p. This leads to a
noninfective phage that displays the protein or peptide to be selected as a
fusion to the C-terminal domain of g3p. The "adapter," consisting of either
Nl or Nl-N2, is fused to the second protein or peptide of interest or
chemically coupled to a nonpeptidic ligand (Fig. lB and C). No wild-type
g3p must be present on a SIP phage. Therefore, in its simplest form, the
Cf-fusion protein directly replaces the g3p in the phage genome. However,
we discuss below other strategies for carrying out SIP with phagemid/helper
phage systems. To simplify the nomenclature, we refer to the protein or
peptide library fused to the CT domain of g3p, and thus displayed on the .
phage, as "A" and denote the target fused to or chemically coupled to the
N-terminal part of g3p, the soluble adapter, as "B." Cognate interaction .
between these two molecules (A and B) restores the gene 3 protein in a :
noncovalent form and thus allows for phage infection. There are two vari- ·:
ants of this selection system, termed in vitro SIP and in vivo SIP.
In the in vitro SIP procedure only the fusion of library A to the C- ·:
I. Stengele, P. Bross, X. Garces, J. Giray, and I. Raschcd, J. Mol. Bioi. 212, 143 (1990).
J. Armstrong, R. N. Perham, and J. E. Walker, FEBS Lett. 135, 167 (1981).
- --
-- - - -- - - - - - - - - -- - - - -- - - - ·- - - - - - - -----·
fd phage with 3-5 copies of
g3p on the tip of the phage
"short"-SIP-phage: infection
is mediated by cognate
interaction of A and 8
adapter molecule
N 1-N2-B
"medium"-SIP-phage: infection
is mediated by cognate
interaction of A and B
adapter molecule
N1 -B
1. Overview of the arrangement of the gene 3 protein in various phages. (A) Fd
phages carry on their tip three to five copies of the gene 3 protein, which consists of three
domains, the two N-terminal domains Nl and N2, required for infection and binding to the
F pilus, respectively, and the C-terminal domain Cf, which anchors the N-terminal domains
in the phage particle. (B) SIP phage as used for in vivo and in vitro SIP. One protein of
interest is displayed as a fusion to Cf on . the phage ("short" phage), whereas Nl-N2 are
expressed as a fusion to or chemically coupled to the other protein/molecule of interest. In
in vitro SIP, the adapter molecule is expressed and purified separately and added later for
infection, whereas in in vivo SIP both proteins are expressed simultaneously on the phage
genome. (C) Alternative SIP phage used in in vitro SIP. In this construct, the "medium"
phage displays the protein of interest fused to N2- CT, whereas the adapte r consists only of
Nl and the second protein of interest.
terminal part of g3p is encoded by the phage or phagemid vector (Fig. 2C
and D). The adapter molecule with the N-terminal part of the gene 3
protein is expressed and purified separately and can then be chemically
coupled to a great variety of targets, including nonpeptidic ones. The sepa- _
rate expression of the adapter molecule allows for better control of the
infection experiments by enabling quality control of the adapter molecule,
titration of adapter, and measurement of background infectivity of purified
phage particles. For in vivo SIP, both parts of the gene 3 protein are
expressed simultaneously in vivo, and thus the genetic information of both
parts remains linked to the phage phenotype. This n1eans that, in principle,
a library-versus-library approach should be feasible, in which one library
is fused to the CT domain (library A) and the other library is fused to the
N-terminal (adapter) part of the gene 3 protein (library B) (Fig. 2E).
----------·- ----------term
- · - - - - - - - - - - -- ---- ------ ·- - - - --
geneS t - -
fd phage
'R ...,.<H.
. ...· .• •. .....
· ~~~~~·
. .-t,
~ "'
geneS t--
cam;R .. ,.;.
m-~ ·, . ~,.
. . .. .... ' · l
........... .. ..,.. ..; . ~j;,_,.,,
z ·llbrary A
geneS t--
geneS t----'pelS
lac P. o
(Sfil )/Bgll
Hind lll 9
N1 : : N2 . .•
" · library.
8'6'""" .
Fro. 2. (A) Organization of the fd phage genome. The genome of the wild-type phage is
subdivided by two intergcnic regions (IG) into a region of strongly expressed genes
2, 5, 7, 8, 9, and 10) and a region of weakly expressed genes (genes 1, 3, 4, and 6). A
terminator (term) separates the weakly expressed genes from the strongly expressed ones.
The organization of the gene segment of the fd phage showing gene 8, gene 3 promotor (g3P),
gene 3 terminator (term), gene 3 (dark gray) encoding the three domains Nl, N2, and Cf of
the wild-type gene 3 protein, and gene 6 is shown in more detail. p3sig indicates the signal
sequence for the gene 3 protein, necessary for transport into the peri plasm. (B) In the
derivative of the fd phage fCKC, which served as a starting construct for all other constructs,
a chloramphenicol resistance gene (CamR, light gray) is inserted downstream of gene 8 and
upstream of the natural terminator. The end of gene 8 is repeated, as it carries the promotor
for g3p. (C) Gene arrangement of the "short" phage (library A-Cf) and (D) "medium"
phage (library A-N2-Cf) as used for in vitro SIP. The fusion is expressed under the natural
gene 3 promotor (g3P), and library A can be exchanged by two Sfii sites with different cutting
sequences. (E) Gene arrangement of the two parts of the gene 3 protein used in in vivo SIP.
Libraries A and B mark the places where the proteins/p.eptides of interest are encoded, which
can be a single sequence or a library. In this case, both fusions are under the control of the
lac promotor/operator system (lac plo ), and a second terminator is inserted in front of gene
6. Both libraries can be cloned by unique restriction sites as indicated. Restriction enzymes
in parentheses indicate sites that are not present in all constructs.
.., .
- - -- --
- --
- -- - --
--···- -- - -
However, this system" which is more powerful hut also less controllable,
can be used only with proteins and peptides for both partners.
In the following section we provide several general protocols, which
apply for in vitro as well as in vivo SIP. In the second and third part, we
focus on in vitro and in vivo SIP, and compare in more detail the use of
both methods by means of a few examples. Two other names, DIRE (direct
rescue interaction) and SAP (selection and amplification of phage ), have
also been used for this technology.
General Procedures
Bacterial Strains
XL1-Blue (Stratagene, La Jolla, CA) (recAJ endA I gyrA96 thi-1 hsdR17
sup£44 relAJ lac [F' proAB laclqZ6.Ml5 TnlO (TetR)]) is the preferred
host for electroporation and for infection exp-e riments because of its recA I
genotype. For infection, it is possible to switch resistances by alternatingly
using the tetracycline- or the kanamycin-resistant strain XLl-Blue MRF'
Kan (Stratagene; ~(merA) 183 ~(mcrCB-hsdSMR-mrr) 173 endAJ sup£44
thi-1 recAJ gyrA96 relA1 lac [F' proAB lac/q Z~M15 Tn5 (KanR)]). This
ensures that during phage infection only phages but no cells are carried
over. An additional precaution is to filter the phage-containing supernatant
directly after phage production and prior to polyethylene glycol {PEG)
precipitation (see Phage Production, below). BL21 (DE3) (Novagen, Madison, WI) [F- ompT hsdS8 (r 8 - m 8 -)gal dcm, harboring DE3, a A prophage
carrying the T7 RNA polymerase gene] is used for the cytoplasmic production of Nl-N2 adapter molecules for in vitro SIP. BL21 (DE3)pLysS (Novagen) [same as BL21 (DE3), but harboring the plasmid pLysS CamR, which
encodes T7 lysozyme, a natural inhibitor of T7 RNA polymerase; resistant
to chloramphenicol at 40 ,u,g/ml] is used for the cytoplasmic production of
Nl adapter molecules for in vitro SIP.
Vector Constructs: Phage versus Phagemid System
In SIP, the infection is strictly dependent on the reconstitution of the
gene 3 protein. Therefore, when using a phagemid/helper phage system, it
is absolutely necessary that the helper phage does not deliver a.ny wildtype g3p to the phage particle. For this purpose, ag3 helper phages were
constructed. · These phages, while not carrying the gene, do need to carry
K. Gramatikoff, 0. Georgiev, and W. Schaffner, Nucleic Acids Res. 22, 5761 (1994).
M. Duenas and C. A. Borrebaeck, Bio!Technology 12, 999 ( 1994).
M. Duenas and C. A. Borrebaeck, FEMS Microbial. Lett. 125, 317 (1995).
F. K. Nelson, S. M. Friedman, and G. P. Smith, Virology 108, 338 ( 1981 ).
; [221
- -- ·- ···- ·- - · -- - - -----··-- ....·- -··------- -- ------·------- ·------- --··- · ..----·- ·------ - -
functional g3p on their coats, because the helper phage must be able to
infect the bacteria harboring the phagemid library for which it needs the
g3 protein. There are two ways to produce such a phage. First, the helper
phage can be produced in cells harboring an unpackageable g3p expression
plasmid, which does not contain an Fl origin. • Nonetheless, we still observed some packaging of the g3p expression plasmid, which increased the
background in subsequent infection experiments, as the population will
produce phages carrying wild-type g3p. This problem has more recently
been solved by engineering an Escherichia coli strain with a g3p expression
cassette integrated in the chromosome (S. Rondot and F. Breitling, personal
communication, 1999). An alternative way is to transform the phagemid
library into cells that already contain Llg3 helper phage. However, at least
in our experience, the helper phage Llg3 M13K07 showed genetic instabilit7
ies and lowered the viability and transformation frequency of bacteria.
A third approach, favored by the authors, is to directly use a phage
genome, in which the complete genetic information of the phage is encoded.
The circular genome of the filamentous phage consists of two main transcriptional units that are separated by a central terminator on one side and
the origin of replication on the other (Fig. 2A).8 The wild-type gene 3 is
modified accordingly for in vitro and in vivo SIP, and the resulting gene 3
cassettes are described further below. In addition, an antibiotic resistance
gene was inserted into the fd genome, immediately downstream of gene 8
and upstream of the original gene 3, to enable detection of infection events
by screening for resistance (Fig. 2B). Several antibiotic resistances were
tested, and resistance against chloramphenicol was chosen, because it gave,
along with ampicillin resistance, the highest level of infectivity, and it had
the advantage that it does not allow growth of plasmid-free cells due to
enzyme leakage as observed for ampicillin. This resulted in the phage vector
fCKC, which is the basis for all other constructs.
Vector Constructs: Different Gene 3 Fusions
Similar to phage display, a peptide or protein library of interest can be
fused to different parts of the gene 3 protein of the phage. In the "short"
fusion, the protein of interest is expressed as a fusion to the CT .domain,
while in the "medium'' fusion, the protein or peptide library of interest is
fused to the N terminus of the N2-CT fragment. Conversely, the adapter
can consist either of only the Nl "bait" fusion or an N1-N2 "bait" fusion
(Fig. lB and C). In one combination of phage and adapter, no N2 domain
at all is present, while in another N2 is present both on the adapter and
C. Krebber, S. Spada, D. Desplancq. and A. Pltickthun. FEBS Lett. 377, 227 (1995).
x E. Beck and B. Zink, Gene 16, 35 ( 19H1).
·--· --·-
-- - -------·--·----- - - - - - -- ------··---
the phage. All four combinations lead to infectivity as tested for in vitro
SIP, albeit at different levels and with a different concentration dependence
of the adapter. Only the "short" fusion has been tested for in vivo SIP so far.
Library Construction
The library will of course depend on the particular question to be
addressed with SIP. A number of standard libraries is available nowadays,
including peptide libraries, antibody libraries, or cDN A libraries, some of
which are even commercially available. In addition, several techniques have
been developed for generating genetic diversity using~ e.g., PCR (polymer14
10 13
ase chain reaction) techniques, - recombination sites. mutator strains or
15 16
degenerated codons, · and trinucleotides for randomization of synthetic
genes. These methods are numerous and well established, and therefore
are not detailed here. However, we discuss the issue of library complexity
and quality in connection with the in vivo SIP procedure.
Phage Production
Phage expression in XLl-Blue cells carrying the phage DNA is per18
formed in shake flasks in 2X YT medium supplemented with 1% (w/v)
glucose, chloramphenicol (Cam; 30 ~-tg/ml), and tetracycline (Tet; 5 JLgl
ml). Overnight cultures are grown at 37° (or 25° when there is reason to
believe that the in vivo folding yield of the ligands might be increased on
lowering the temperature). For in vitro SIP with the constructs shown, in
which the gene 3 fusion protein is expressed under the natural gene 3
promotor, phages are obtained directly from the supernatant of the overnight culture. For in vivo SIP, where both parts of gene 3 are under lac
promoter/operator (lac p/o) control, a main culture containing Cam (30
C. Krebber, S. Spada, D. Desplancq, A. Krebber, L. Ge, and A . PHickthun. J. Mol. Bioi.
268, 607 (1997).
R. C. Cadwell and G. F. Joyc.e, Genom.e Res. 3, S136 (1994).
A. Crameri, S. A. Raillard, E. Bermudez, and W. P. C. Stem.m.e r. Nature (London) 391,
288 (1998).
W. P. C. Stemmer, Nature (London) 370, 389 (1994). ·
H. Zhao, L. Giver, Z. Shao, J. A. Affholter, and F. H. Arnold. Nature Biotechnol. 16,
258 (1998).
N. Tsurushita, H. Fu, and C. Warren, Gene 172, 59 (1996).
S. Kamtekar, J. M. Schiffer, H . Xiong, J. M. Babik, and M. H. He.c ht. Science 262, 1680 (1993).
E. Wolf and P. S. Kim, Protein Sci. 8, 680 {1999).
B. Virnekas, L. Ge, A. Pltickthun, K. C. Schneider, G. Wellnhofer. and S. E. Moroney,
Nucleic Acids Res. 22, 5600 ( 1994 ).
J. Sam brook, E. F. Fritsch, and T. Maniatis, "Molecular Cloning: A Laboratory Manual,"
2nd Ed. Cold Spring Harbor Lahoratory Press, C.o ld Spring Harbor. New York. 19R9.
. [22]
'., Tet (5, and glucose (1 o/o, w/v) is inoculated from the overnight culture to an initial 00550 of 0.15 (typical dilution of 1:25-1: 30) and
grown at 37° until an 00 550 of 0.5 to 0.6 is reached. The cells are then
pelleted by centrifugation (3000g, 1.0 min, 4-10°) and resuspended in fresh
2X YT medium without glucose containing the appropriate antibiotics
and 0.5 mM isopropyl-13-o-thiogalactopyranoside (IPTG) to induce the
expression of the two fusion proteins, protein A-CT and N1-N2-protein
B, which results in phage packaging. Expression is continued for at least
6 hr at 37°, or in the case of less stable proteins at 24-26° overnight. In
both cases, phages are separated from the cells by centrifugation for 10-15
min at 5000g and 4°. The supernatant containing the phages can then be
optionally filtered ( pore size filters) to remove any remaining.cells.
The ·yield is usually in the range of 5 X 10 to 2 X 10 phages per milliliter
of culture supernatant.
Phage Purification
For the infection experiments, phage particles are enriched by two
precipitations with polyethylene glycol (PEG). The most convenient way
is to prepare a five-times concentrated stock solution with 16% (w/v) PEG
6000 and 3.3 M NaCI, passed through a pore size filter. For the
PEG precipitation, one part C?f this solution is added directly to four parts
of the supernatant obtained after phage production and cell precipitation,
and incubated on ice or at 4° for at least 1 hr or overnight. Phages are
pelleted by centrifugation at least at 5000g for 30 min at 4° and redissolved
in about 1 ml of TBS buffer [Tris-buffered saline, 50 mM Tris-HCl (pH
8.0), 150 mM NaCl]. If higher purity or concentrations are required, a
second PEG precipitation can be performed similar to the first one. For
this purpose, PEG-NaCl solution is added in the same ratio as for the first
PEG precipitation, and after 1 hr of incubation at 4° phages are pelleted
in a 2-ml tube by centrifuging at maximal speed for 20 min at 4°. Phages
are then dissolved in 50-200 JLl of TBS buffer. They can be stored at 4°
for several months, depending mainly on the stability of the fusion protein.
However, if phages of greater purity are required as, e.g., for a more
accurate quantification or electron microscopy, phages are best purified by
ultracentrifugation in a CsCl gradient. 19 Phages are first enriched 80-fold
by PEG precipitation, and then added to a solution of 1.6 g of CsCl in
TBS, filled to a final volume of 4 ml. The CsCl solution is transferred to a
13 X 38 mm polyallomer tube (Quick-Seal; Beckman, Fullerton, CA) and
centrifuged at 100,000 rpm (541,000g) for 4 hr in a TLN-100 rotor (Beckman) at 4°. After ultracentrifugation the phages become visible as a white .
G. P. Smith and J. K. Scott, Methods Enzymol. 217, 228 (1993).
- - - --
---····· ... --·-------- ··-.. ---- -·- -·--·---·- ---- ..
~---- -------
----· --· ···--·-~
.. ···--·--- ......
band through light scattering when shining light from the top to the bottom
·of the tube and by looking through the wall of the tube at a right angle to
the light beam. Phages are removed by puncturing the side of the tube
with a hypodermic needle and gently removing the banded phages. To
remove the remaining CsCl, phages are then pelleted by ultracentrifugation
(50,000 rpm/135,000g, 1 hr, 4°, TLA-100.3 rotor), redissolved in 3 ml of
TBS buffer, pelleted again (same conditions), and redissolved in 50 of
TBS buffer.
Determination of the Phage Concentration
Phage particles are quantified spectrophotometrically by measuring the
absorption between 200 and 350 nm. A broad peak between 260 and 280
nm is obtained due to the presence of both DNA and proteins, and the
absorption at 269 nm is used to estimate the phage concentration as
·. Phage particles _
number of base pairs of ssDNA
This method, however, gives only a crude estimate and upper limit of
phage concentration for phages that are prepared merely by PEG precipitation, because proteins from the culture supernatant can be coprecipitated
by PEG. Especially in cases of low phage production, coprecipitated proteins can influence the absorption significantly, as s.e.en by a red shift of the
absorption maximum. This can be avoided by using phages purified by
CsCI gradient ultracentrifugation. However, in our experience, "average"
phage titers (about 10 phages per milliliter of supernatant), prepared by
two subsequent PEG precipitations, have a "normal" absorption spectrum
with no red shift (comparable to purified phages), and usually give reliable
results in concentration determinations, when compared with phages purified by ultracentrifugation.
Another possibility for phage quantification is by using an enzymelinked immunosorbent assay (ELISA), by immobilizing the phage and
detecting with an antibody against gene 8 protein. ·However, this method
does not yield absolute phage concentrations, but relative numbers, and is
therefore best suited for comparative infection experiments. ln this case,
phages prepared by PEG precipitation are coated overnight at 4° on an
ELISA plate in a twofold dilution series. The ELISA plate is washed three
times with TBS and blocked with 4-5% (w/v) milk powder in TBS for 1
hr at room temperature. After washing twice with TBS, phages are detected
with a 1:5000 dilution of anti-M13-peroxidase ·conjugate (Pharmacia, Pis20
L. A. Day J. Mol. Rio/. 39, 265 (I 969) ..
cataway, NJ) in TBST [TBS with 0.05o/o (v/v) Tween 20]. The amount of
bound anti-Ml3-peroxidase is measured with peroxidase (POD) soluble
substrate (Boehringer Mannheim, Mann.heim, Germany) at 405 nm after
various incubation times. To obtain absolute concentrations, a control
phage dilution needs to be included with known concentration. Preferably,
the concentration of this calibration standard should have been determined
by UV absorption after purification by CsCI gradient ultracentrifugation.
In Vitro S·e lectively Infective Phage Procedures
In the in vitro SIP procedure, the adapter molecule is purified separately,
and later added to phage$ displaying the protein or peptide library of
interest, which is fused to the C-terminal part of the gene 3 protein. Thus,
nonpeptidic "baits" can also be used, which can be chemically coupled to
Nl or Nl-N2, respectively. In addition, the concentration of the adapte_r
molecule can b·e varied.
Design of Vectors Suitable for in Vitro Selectively Infective Phage
The phage vector used in these examples is based on fC.K C and encodes
the protein or library of interest with the gene 3 leader sequence as fusion
to either N2-CT ("medium" phage) or CT ("short'' phage) under the
control of th·e natural gene 3 promotor. The phage production is carried
out in XLl-Blue as .described in the general procedures.
Native, periplasmic expression of theN-terminal domains of gene 3 was
found to be difficult, as the native protein always led to complete lysis of
the culture and relatively low yields requiring a tedious concentration step.
On the other hand, the Nl domain alon·e, lacking the glycine-rich linker,
can be obtained by secretion. However, we favor cytoplasmic expression
of the gene 3 N-terminal domains under the control of the strong T7
promoter, using the vector pTFI74, as this appears to be more reliable
and simpler for downstream processing. The deletion of the signal sequence
and the 17 expression system leads to cytoplasmic inclusion bodies, which
are easily refolded. Interestingly, the toxicity effects of secreted proteins
vary significantly for the various constructs. The highest toxicity is observed
for the Nl do·m ain with a C-terminal glycine-rich linker, whereas th·e expression of the Nl-N2 complex caused significantly fewer problems. However,
Nl-N2 with a C-terminal glycine-rich linker showed again more problems ·
with expression (lower yield and higher toxicity) than Nl-N2 without the
P. Holliger and L. Riechmann, Structure 5, 265 (1997).
C. Freund, A. Ross, B. Guth, A. Pliickthun, and T. A. Holak, FEBS Lett. 320, 97 {1993).
Flo. 3. Amino acid sequence of (A) Nl-N2 and (B) NI as used for in vitro SIP. The
initiator methionine, which is needed for expression in the T7 system and that becomes cleaved
afterward, is not shown. ·The engineered cysteine is underlined and the hcxahistidinc (His6 )
tag is shown in boldface. (A) Nl-N2 domain: Nl (amino acids 1-68) followed hy G I (amino
acids 69-86), N2 (amino acids 87-217), and the Cys-Hisn tag (228 re.sidues; 24,854.9 Da, p/
4.58; e = 41,100 M- 1 em - 1). (B) Nl domain: N1 (amino acids 1-68) followed by the first 14
amino acids of Gl and the Cys-His6 tag for purification and coupling (91 residues; 9,658.37
Da; p/ 4.87; E = 21,210 M- 1 cm- 1).
long glycine-rich linker. To obtain cytoplasmic expression, the gene 3 signal
sequence (amino acids 1-18) is deleted, and a methionine is added as a
start codon, which becomes cleaved, as detected by mass spectrometry and
N-terminal sequencing. In addition, a C-terminal Ser-Gly-Cys-Pro-His6 tag
is introduced to allow chemical coupling (Cys) and easy purification (hexahistidine, His6) (Fig. 3A and B). We use the Nl-N2 domain with only three
additional residues before the engineered cysteine (Fig. 3A) and the Nl
domain alone followed by the first 14 amino acids from the linker Gl before
the cysteine (Fig. 3B) for in vitro SIP experiments. In addition, we could
produce in the same way the Nl-N2 complex followed by a glycine-rich
linker and the N2 domain alone followed by its glycine-rich linker. Similar
expression constructs have been used to produce Nl-N2 for X-ray crystal24
lography,23 and Nl-TolA for X-ray crystallography. As mentioned above,
Nllacking the glycine-rich linker and Cys-His6 tag could be purified from
culture supernatant for nuclear magnetic resonance (NMR). 21
J. Lubkowski, F. Hennecke, A. Pllickthun, and A. Wlodawer, Nature Struct. Bioi. 5, 140
J. Lubkowski, F. Hennecke, A. Pltickthun, and A. Wlodawer, Structure Fold. Des. 7, 711
- ·- -------
- - - ----- - - __.....---·--·----·--·--·----·-·-.......
··- ' - - -
- -------
Expression of the Nl-N2 Domain of the Gene 3 Protein
Plasmids carrying the gene for cytoplasmic expression of Nl-N2 (Fig.
3A) are transformed into BL21 (DE3), plated onto 2X YT agar plates,
containing Amp (200 ,ug/ml) and 1% (w/v) glucose, and grown overnight
at room temperature. A single colony of each host/plasmid is grown in 2 X
YT [containing Amp at 200 J.Lg/ml, 1% (w/v) glucose] for 8 hr. Subsequently,
a two-thirds volume of 50% (v/v) glycerol is added and the cells are
frozen as glycerol stocks at - 80°. For the expression culture, 50 JLl of
the glycerol stock is used to inoculate 50 ml of SB medium (2o/o (w/ v)
tryptone, 1% (w/v) yeast extract, 0.5% (w/v) NaCI], containing Amp
(200 ,ug/ml), 1% (w/v) glucose, 2% (w/v) glycero1,-·5o mM K 2HP04 , and
10 mM MgCl2 , and the ·culture is grown overnight at 37°. The overnight
culture is used to inoculate the main culture (2 liters of SB medium
with the same additives). The main culture is grown at 37° to an OD 550
of 0.9 to 1.2, and then IPTG is added to a final concentration of 1 mM,
and the cells are grown for another 3- 4. hr.
Expression of the Nl Domain of the Gene 3 Protein
Because of the high toxicity and thus instability of the expression culture
for the Nl domain alone (Fig. 3B), it is produced in BL21(DE3) harboring
the plasmid pLysS, which encodes TI lysozyme, a natural inhibitor of T7
RNA polymerase. This plasmid also confers resistance to chloramphenicol
(up to 40 ,u,g/ml). After. transformation, cells are plated onto 2X YT agar
plates, containing Amp at 200 J.Lg/ml, Cam at 30 JLglml, and 1% (w/v)
glucose-, and grown overnight at room temperature. For glycerol stocks
(see preceding section), 2X YT medium supplemented with Amp (200 J.Lgl
ml) and 1% (w/v) glucose is inoculated with a single colony and grown for
8 hr. For production of Nl alone a more elaborate protocol must be used
than in the case of N1-N2 production. A dilution of 50 J.Ll of glycerol stock
with 250 of SB medium is plated on three 100-mm SB-agar plates,
containing Amp (200 ,ug/ml), Cam (30 JLg/ml), 1% (w/v) glucose, and
5o/o (v/v) glycerol. After incubation overnight at 37°, the colonies (near
confluence) are scraped from the plates with 20 ml of SB medium, which
is then used to inoculate 2 liters of the main culture. The main culture [2
liters of SB medium containing Amp (200 J.Lg/ml), Cam (30 JLglml), 1%
(w/v) glucose, 2% (v/v) glycerol, 50 mM K 2HP04 , 10 mM MgCI2] is grown
at 37° to an 00 550 of 0.9 to 1.2, IPTG is added tp a final concentration of
1 mM, and the cells are grown for another 3-4 hr.
As the expression of the Nl domain with the single cysteine is not
F. W. Studier, J. Mol. Bioi. 219, 37 (1991).
- ----·- -·-..·· --··- --
I ~JIA(it:.
- ·--- --- - --- · ---~ ----- ·
_ __
··- - ··· ·-...- - -..... -.. . . . . .......
- -··--···-----
. . .... - ··· ----
totally reliable even in the described experimental setup, two or three main
cultures (2 liters each) are started separately and sn1all aliquots of each
are analyzed. In contrast, fusion proteins of N 1 with other proteins give
reliable inclusion bodies. Cells from 50 ml of culture are centrifuged and
resuspended in 1 ml of 25 mM Tris-HCl, pH 7.5. Four milliliters of 6 M
guanidine hydrochloride, 25 mM Tris-HCl, pH 7.4, is added to lyse the
cells. Because of the liberated nucleic acids the lysate becomes viscous and
must be sonicated. After centrifugation (SS34, 20,000 rpm/48,000g, 10 min,
4°) the supernatant is filtered through a 0.45-JLm pore size filter. The filtrate
is analyzed by coupled IMAC-AIEX as described in detail below. Small
aliquots of the peak fractions can be loaded onto a sodium dodecyl sulfate
(SDS)-polyacrylamide gel, as the sample is eluted from the AlEX in a
nonionic urea buffer.
Cell Rupture and Enrichment of the Inclusion Bodies
Cells are harvested by centrifugation (GS3, 5000 rpm/4000g, 10 min,
4°). The cell pellet of a 2-liter culture is resuspended in 25 ml of 50 mM
Tris-HCl, 1 mM MgCl 2 (pH 8.0), and 10 mg of RNase A, 10 mg of DNasc
I, and 25 mg of hen egg white lysozyme are added. The cell suspension is
ruptured in a French press (Aminco ). After centrifugation (SS34, 20,000
rpm/48,000g, 30 min, 4°) the pellet containing the inclusion bodies is resuspended in 25 ml of 20 mM Tris-HCI, 23o/o (w/v) sucrose, 0.5% (v/v) Triton
X-100, 1 mM EDTA, pH 8.0. The suspension is stirred by a magnetic stir
bar at 4° for 30 min, pelleted again, and the wash step is repeated. During
the washing procedure the inclusion body pellet should change its color
from brownish-yellow to white. To the final washed inclusion body pellet,
20 ml of 5.5 M guanidine hydrochloride, 25 mM Tris-HCl, pH 7.5, is added
and the pellet is solubilized by stirring at room temperature. Insoluble
material is removed by centrifugation (SS34, 20,000 rpm/48,000g, 30 min),
and the supernatant is filtered through a 0.45-,u,m pore size filter.
Protein Purification
Two routes can be chosen to obtain purified gene 3 N-terminal domains
from the inclusion bodies. The first procedure is the more classic way: ·
refolding followed by purification. For the second, the protein is first purified
under denaturing conditions. Both procedures are equivalent in yield. However, the second procedure may be advantageous especially when pro teases
are of concern and when a more defined and pure denatured sample is
necessary (e.g., to test refolding parameters for more complicated fusion
proteins). Using the second route, ·we were able to refold the protein by
dialysis and did not have to refold by dilution, thus saving an additional
- .. . --
. .
.. -·- .
. .
-- . -·· . . -- .
.. .
- -· ..
conccnt rat ion step. ()vcnlll, the second rncthod is faster and rnore straight for~ard. CJ1roinatography can be pcrforn1ed with any equiprncnt~ however,
the autorna1cd two-colurnn fonnat sin1plif1cs the procedure signiflcantly.Hl
Irnmobilizcd metal-ion affinity chromatography (IMAC) is usually able to
enrich the His-tagged adapter protein to a high extent, although some
contaminants are still present. Anion- or cation-exchange chromatography
is a good second step after IMAC, as the relatively pure IMAC-purified
protein can often be baseline separated. In both cases the chromatography,
performed with a BioCAD 60 system (PerSeptive Biosysten1s, Framingham,
MA) takes about 30 min to obtain pure protein.
Route 1: Refolding and Subsequent Purification of the Native Protein
Step A: Refolding by Dilution
Twenty milliliters of the solubilized inclusion bodies (in 5.5 M guanidine
hydrochloride, 25 mM Tris-HCI, pH 7.5) is reduced by addition of dithiothreitol (DTT) to 100 mM and EDTA to 10 mM final concentration. Most
conveniently, this is done overnight at 4°, but reduction can also be performed for 4 hr at room temperature. A.s the DTf interferes with the
formation of the disulfide bonds in the gene 3 N-terminal domains, the
OTT concentration is reduced to 0.1 mM by three dialysis steps for 2 hr
each at room temperature in 10 volumes (200 ml) of 5.5 M guanidine
hydrochloride, 25 mM Tris-HCl, 10 mM EDTA,, pH 6.0. The reduced,
denatured protein is then slowly added dropwise to 1 liter of refolding .
buffer [0.4 M L-arginine, 0.2 M Tris-HCI, 0.2 M guanidine hydrochloride
0.1 M (NH 4 ) 2S0 4 , 2 mM EDT A, pH 8.5] which contains 1 mM oxidized and
0.2 mM reduced glutathione. After overnight stirring at 10°, the refolding
mixture is concentrated in an RA2000 concentrator (Amicon, Danvers,
MA) to 120-150 ml, and further concentrated to a volume of 10-15 ml
with a Centriprep YMlO (Amicon). Precipitates (mainly made up of contaminating proteins) are removed by centrifugation. Finally, the protein is
dialyzed against 25 mM HEPES, 900 mM NaCl, pH 7 .5, and filtered through
a 0.45-,um pore size filter.
Step B: Coupled Immobilized Metal-/on Affinity-Anion Exchange
Chromatography under Native Conditions
To purify the refolded N-terminal domains of gene 3 protein, the protein
is first bound to an IMAC column, using the C-terminal His6 tag. After
A. Pliickthun, A. Krebber, C. Krebber, U. Horn, U. Kntipfer. R. Wenderoth, L. Nieba, K.
Proba, and D. Riesenberg, in "Antibody Engineering: A Practical Approach" (J. McCafferty
and H. R. H.oogenboom, eds.), p. 203. IRL Press, Oxford, 1996.
- .
the protein is el utcd direct Iy on to the an ion-exchange col un1n ( i\ I [~X)
under such conditions that it binds there. On the A I EX, the protein can
be purified to a high degree hy an Na( 'I gntdient. 'l'he whole procedure is
automated on a BioCAD 60 (PerSeptivc Biosystems): First, the 1.66-ml
HQ/M AlEX column (PerSeptive Biosystems) is equilibrated with 16 ml
of 50 mM Tris-HCl, pH 7.5. The switches are then set such that the column
is taken out of the flow and t~e 1.66-ml MC/M lMAC column (PerSeptive
Biosystems) is switched in-line, washed with 20 ml of water, loaded with
2 ml of 100 mM NiCh through one of the water-washed sample lines, and
subsequently washed with 20 ml of water. After Ni + charging the MC/M
column, it is equilibrated with 16 ml of 25 mM HEPES, 900 mM NaCl, pH
7.5. Five milliliters of sample is loaded onto the MC/M column. It is then
washed first with 12 ml of 25 mM HEPES, 150 mM NaCl, 1 mM imidazole,
pH 7.5, and then with 24 ml of 25 mM HEPES, .12 mM imidazole, pH 7.5.
After the wash steps, the HQ/M column is set in-line such that the flow
is directed from the MC/M to the HQ/M column. To transfer the protein from the MC/M onto the HQ/M column it is directly eluted from the
MC/M onto the HQ/M column by 12 ml of 120 mM imidazole, pH 7.5,
which results in strong binding of the protein to the HQ/M column. The
MC/M column is then set off-line and the HQ/M column is washed with
16 ml of 50 mM Tris-HCl, pH 7.5. A 20-ml gradient of 0 to 500 mM NaCI
in 50 mM Tris-HCl, pH 7.5, is employed to elute the N-terminal domain
protein. As part of the protein dimerizes in the refolding reaction due to
the free cysteine in the tag (to be used for chemical coupling), two major
gene 3 N-terminal protein domain peaks are observed. The monomer and
dimer peak fractions are combined and OTT is added to 2 mM and EDT A
to 10 mM final concentration, respectively. The protein solution is passed
through a 0.22-J.Lm pore size filter and stored at 4°. If necessary, the protein
solution can be concentrated with a Centriprep YMlO (Amicon).
Route 2: Purification and Subsequent Refolding
Step A: Coupled Immobilized Metal-Ion Affinity-Anion Exchange
Chromatography under Denaturing Conditions
The purification under denaturing conditions is basically the same as
under native conditions. However, because the purification starts with
freshly solubilized inclusion bodies, the reduction with OTT and the subsequent dialysis to remove it are not necessary. Oxidized protein is reduced
on the column and only the monomeric species must be separated from
contaminants. As the pure protein can be refolded at higher concentrations,
the protein is refolded by dialysis and not by dilution, which further simplifies downstream processing.
First, the 8-ml HQ/M AlEX column (PerSeptive Biosystems) is equilibrated with 60 n1l of 6 M urea, 50 mM Tris-HCI, pH 7.5. The column is
switched out of the flow anti the 4-nll MC/M lMAC colun1n (PcrScptivc
Biosystems) is set in-line, washed with 40 ml of water, loaded with 5 n1l of 100
1nM NiCb through one of the water-washed sample lines, and subsequently
washed with 80 ml of water. After charging the MC/M column it is equilibrated with 30 ml of 8 M urea, 25 mM HEPES, 1.5 M NaCl, pH 7.5. Five
milliliters of solubilized inclusion body sample is loaded onto the MC/M
column. It is then washed first with 20 ml of 8 M urea, 25 mM HEPES,
1.5 M NaCl, 4 mM imidazole, pH 7.5, and then with 60 ml of 8 M urea, 25
mM HEPES, 15 mM imidazole, pH 7.5. After the wash step the HQ/M
column is set in-line such that the flow would be directed from the MC/M
to the HQ/M colu1nn. To transfer the protein from the MC/M onto the
HQ/M column it is eluted from the MC/M onto the HQ/M column by 40
1rtl of 8 M urea, 100 mM imidazole, pH 7.5, which results in strong binding
of the protein to the HQ/M column. The MC/M column is then set offline and the HQ/M column is washed, and at the same time the protein is
reduced on the column with 20 ml of 8 M urea, 100 m.M Tris-HCl, 100 mM
DTT, pH 9.0, at a low flow rate of 2 ml/min (10 min). Subsequently, the
column is washed with 60 ml of 6 M urea, 25 mM Tris-HCl, pH 7.5. The
reduced, denatured protein is eluted by a 60-ml gradient from 0 to 250 mM
NaCl in 6 M urea, 50 mM Tris-HCl, pH 7.5.
Step B: Refolding. by Dialysis
The purified, reduced and denatured protein (10-20 ml) is immediately
dialyzed against 500 ml refolding buffer [0.4 M L-arginine, 0.2 M Tris-HCI,
0.2 M guanidine hydrochloride, 0.1 M (NH4 ) 2S04 , 2 mM EDT A, pH 8.5],
which contains 1 mM oxidized and 0.2 mM reduced glutathione, overnight
at 10°. The buffer is exchanged by two dialysis steps, each against 500 ml
of 50 mM Tris-HCI, 50 mM, NaCl, pH 7.5. DTf is added to 2 mM ·and
EDT A to 10 mM final concentration, respectively. The protein solution is
passed through a 0.22-p,m pore size filter and stored at 4°, and if necessary
concentrated on a Centriprep YMlO (Amicon). However, in case of direct
peptide or protein fusions to the C terminus of Nl or N1-N2, the Cys
residue in the C-terminal tag is replaced by the interaction partner B, and
the refolding and purification procedure might have to be varied accordingly.
Chemical Coupling of Nonpeptidic Ligands
Coupling is carried out via the free sulfhydryl group, introduced by
the Cys residue in the C-tcrminal tag (Fig. 3). Coupling chemistry and
·-----purification may vary depending on the bait. We will give an example for
coupling ftuorescein, which is already coupled to a 1-.~ys residue to give the
compound Flu Cad (5-[ (5-aminopcntyl)thiourcidyl )fluorescein). C:oupling
is achieved by using the hclcrobifunctional cross-linker N-succinirnidyl-6maleimidocaproate (Fiuka, Ronkonkoma, NY). The coupling of FluCad
to the cross-linker is carried out in ditncthylformamidc (DMF) for 1 hr at
30° in the dark, at a ratio of FluCad to cross-linker of 3: 2. Completion of
the reaction is controlled by thin-layer chromatography in 80% (v/v) ethyl
acetate_.lO% (v/v) methanol-10% (v/v) acetic acid. At the same time the
DTT in the Nl or Nl-N2 stock is removed by gel filtration on a short
Sephadex G-25 C<?lumn (Pharmacia) in 25 mM sodium phosphate, pH 6.8.
This leads to Nl or Nl-N2 molecules that contain a free sulfhydryl group
in the cysteine of the tag. The amount of protein is quantified by spectropho27
tometry, using the calculated extinction cocfficicnt.· Three n1olar equivalents of FluCad are then reacted with one molar equivalent of the free
sulfhydryl group of Nl or Nl-N2 in 25 mM sodium phosphate, pH 6.8, for
1 hr at 25° and then at 4° overnight. The pH of the reaction buffer is critical,
as at higher pH the maleimide would start to react with primary amines
in the Nl or Nl-N2 domain. The resulting adapter molecules are gel
· filtrated on a Sephadex G-25 column (Pharmacia) in 50 mM Tris-HCI, pH
7.5. Because fluorescein tends to noncovalently interact with the adapter
protein (Nl or Nl-N2), another purification step is needed. Therefore, the
fluorescein-coupled adapter molecules are separated from the uncoupled
Nl or Nl-N2 by anion-exchange chromatography on a perfusion chromatography HQ column NaCl gradient (0 to 600 mM NaCl in 50 mM TrisHCl, pH 7.5), using a BioCAD 60 system (PerSeptive Biosystems). Mass
spectrometry should be used to verify the success of the coupling reaction
and purification. In general, the coupling method and the heterobifunctional
linker can be used for the coupling of any primary amine compound that
does not contain a free sulfhydryl group. Using an analogous strategy every
compound activated by a maleimide should be linkable to the free cysteine
containing Nl or Nl-N2. The disulfide bonds of Nl or Nl-N2 were not
affected, neither by the DTI nor the maleimide under the conditions employed.
Phage Infection Experiment
In vitro SIP experiments are performed by incubating 1 ~-tl of SIPphage supernatant, concentrated 500 times by two PEG precipitations (as
described under General Procedures), with 350 nM adapter Nl-N2 bait
S. C. Gill and P. H. von Hippel, Anal. Biochen1. 182, 319 ( llJR9).
at 4°. For infection, 100 p,l of an exponentially growing XLl-Blue culture
is added to the mixture and incubated for 1 hr at 37° with shaking, and
plated subsequently on 2X YT-agar plates containing Cam (25-30 p,g/ml)
and Tet (15 p,g/ml). Addition of 50 mM MgC1 2 to the cells prior to and
during infection increases infectivity four- to sixfold.9 The number of phages
can be varied according to library size and expected infection rate. The
concentration of adapter molecule, especially for Nl-N2 bait, can influence
the infectivity, and thus, when maximal infectivity is needed, a titration of
adapter molecules might be advisable (see the next section).
Dependence of Infectivity on Concentration of Adapter Molecule
The concentration of adapter molecules influences the phage infectivity.
Because of the law of mass action, higher concentrations of adapter will shift
the equilibrium to the bound and thus infective state. However, differences
between Nl bait and Nl-N2 bait are apparent, presumably because of the
ab'ility of N2 to bind to the pili and thus block infection at high concentration. Both variants have previously been tested with fluorescein as bait,
fused to Nl or Nl-N2, respectively, and a fluorescein-binding single-chain
variable fragment (scFv), fused to N2-CT ("medium" phage) or CT
("short" phage), respectively (Fig. 4). The infectivity of 10 medium
u E
a> en
c co
.Q cn
...... a>
·- 0
concentration of adapter molecule [M]
F1o. 4. Dependence of infectivity on concentration of adapter molecules and g3p arrangement in in vitro SIP. Open circles represent the situation depicted in Fig. 1C with the "short"
phage and Nl-N2 bait as adapter molecule. In this case, the same infectivity was observed
when using "medium" phages instead (data not shown). Filled circles represent the infectivities
obtained with medium phage and Nl bait as adapter molecules (Fig. 10). (Adapted from
Krebber et al. 9)
., .
phages increases as a function of Nl-fiuorescein up to the highest concentration tested (lo-s M), at which point it levels off, yielding 1.2 X 106
colonies. In contrast, using the adapter Nl-N2-ftuorescein with short or
medium phages, a maximum of infectivity is reached at 3 x 10- 8 M, yielding
5 X 10 colonies from 10 input phages. In summary, using Nl bait as
adapter and medium phages, the maximal infectivity is about 20-fold higher
than with the Nl-N2 bait, but a higher adapter concentration is needed.
Consequently, the Nl-N2 bait is recommended for selection of high-affinity
systems, as the adapter concentrations can be kept low, whereas the Nl
bait is more suitable for lower affinities, as the adapter concentration can
be increased without inhibitory effects.
Examples from the Literature using in Vitro Selectively Infective
Phage Selection
In vitro SIP was used to select for novel, nonrepetitive linkers for
antibody scFv fragments. 2H In tbis case, medium phages and adapters with
Nl bait were used, and phage production was performed at 30 and 37°.
After one round of SIP selection, 22 of 22 clones gave positive signals in
ELISA, whereas before selection, only 1 of 23 clones was positive. Nine
clones were characterized further, and shown to be comparable to the
parental sequence, which had (Gly4 Ser) 3 as the linker. in terms of binding,
folding, expression, and solubility.
A selection from an antibody Fab library was carried out and compared
with selection results from conventional phage display. In this case, a
phagemid system was used, and the adapter, consisting of the first 98 amino .
acids of the mature gene 3 protein fused to the antigen, was secreted
from a separate E. coli culture, and the supernatant was used directly for
infection. The N2 domain was missing completely in this experimental
setup. Infection was carried out by adding 25% ( v /v) adapter-containing
supernatant to the supernatant from the phage-producing cells and incubating the mixture overnight at 4° before adding XLl-Blue recipient cells.
From this library, half the phages gave positive ELISA signals after three
rounds of SIP selection, while no antibodies against the antigen could be
obtained from phage display after three rounds of phage panning. Thus,
SIP is faster than conventional phage display, using the identical library.
In two further model systems, the properties of selection were investigated in more detail by using a small set of molecules with defined thermodynamic and kinetic parameters. The phage infectivity of various point mutants with slightly different K 0 values and activities was compared and a
F. Hennecke, C. Krebber, and A . PlUckthun, Protein Eng. II, 405 ( 1998).
----- - -- - - - -- -- -- -- - - - - - - - - - - - - - - PIIACiE DISPLAY AND ITS APPLICATIONS
clear correlation was observed. In these experiments, short phages and
Nl-N2 bait adapter molecules were used. In a competitive SIP experiment
with all mutants, only the tightest binders remained after three rounds of
selection. In another model selection with six different Fab fragments, the
ability to select for affinity or even kinetic constants was investigated. 30
In individual experiments using the phagemid system (mentioned above),
phages expressing higher affinity clones were enriched preferentially with
low concentrations of antigen. However, it was also possible to select clones
with lower affinity by increasing antigen concentration. In another experiment the incubation time was varied. Using short incubation times (30
min), clones with high association rate constants were preferred, whereas
after long incubation (16 hr), clones with the lowest dissociation rate constant were preferred.
Two further examples use techniques closely related to in vitro SIP. In
the first example, linkage of covalent catalysis to infectivity was demonstrated by fusing a catalytic antibody fragment to the CT domain and the
substrate was coupled to Nl-N2. Only covalent catalysis was able to pro31
duce infective phages. In the second example, selection for protease resistance was carried out by fu~ing the proteins of interest between Nl-N2
and CT and performing several rounds of in vitro proteolysis, infection,
32 33
and propagation. •
In Vivo Selectively Infective Phage Procedures
In the in vivo SIP procedure, all steps are carried out with crude E. coli
supernatant, without the need for purification of any compound or any in
vitro panning steps. In particular, the target for the library does not have
to be expressed and purified. Both interacting partners (protein A-CT and
Nl-N2-protein B) are encoded on the same phage vector, as described
below. Alternatively, a phagemid system3 .4 (see General Procedures) or a
combination of phage and phagemid vectors that are copackaged can be
used. Although two-vector systems provide for more convenient cloning,
G. Pedrazzi, F. Schwesinger, A. Honegger, C. Krebber, and A. Pltickthun, FEBS Lett. 415,
289 (1997).
M. Duenas, A . C. Malmborg, R. Casalvilla, M. Ohlin, and C. A. Borrebaeck, Mol. Immunol.
33, 279 (1996).
C. Gao, C. H. Lin, C. H. L. Lo, S. Mao, P. Wirsching, R. A. Lerner. and K. D. Janda, Proc.
Nat/. A cad. Sci. U.S.A. 94, 11777 ( 1997).
:u V_ Sieber, A. Pltickthun, and F. X. Schmid. Nature Biotechnol. 16, 955 ( 1998).
P. Kristensen and G. Winter, Fold. Des. 3, 321 (1998).
F. Rudert, C. Woltering, C. Frisch, C. Rottenberger, and L. L. Hag. FEBS Lett. 440, 135
- - --
- - - - - - - --- --- -
it is important to ensure the quality of both vectors and the complete
transformation or transfection of one library to the other, which is usually
verifiable only by sequencing cotransformants.
Cloning of Two Libraries in One Vector
If a library-versus-library experiment is to be carried out, two libraries
must be cloned (library A-CT and Nl - N2-libtary B). If the phage system is
used, both libraries must be cloned in one vector with the highest efficiency
possible. Two strategies are available. First, to avoid multiple transformations, both libraries can be combined with the interconnecting vector fragment in an assembly PCR, and then cloned in one step into the vector.
Thereby, only one transformation is required. Alternatively, both libraries
can be cloned one after the other, which then requires two ligations and
two transformations, however. We tested both possibilities by sequencing
individual clones obtained after cloning. Using the assembly PCR approach,
fewer clones analyzed .were generally correct, and the remainder were
mainly frameshift mutants with deletions of 1 base pair up to a stretch of
several base pairs. Also in the region between both libraries, deletions
were occasionally observed. However, using the second approach with two
subsequent ligations and transformations, the result was significantly better.
It appears that the assembly PCR, even though performed with a proofreading polymerase, introduces more errors, and because the overall transformation rates are comparable for both approaches, the sequential cloning appears to be the method of choice for library cloning in one-vector systems.
Design of Vectors Suitable for in Vivo Selectively Infective Phage to
A void Genetic Instabilities
In the case of in vivo SIP, where both parts of the gene 3 protein are
expressed in the same cell, whether on one or two vectors, great care
must be taken to prevent recombination. In initial constructs with the
arrangement Nl-N2-protein B and protein A-CT, homologous recombination between duplicated glycine-rich regions was observed, leading to
restoration of a wild type-like gene 3 protein at high frequency. Therefore,
in more recent vectors the order of the two fusion proteins has been
switched to protein A-CT being upstream of Nl-N2-protein B, and the
glycine-rich linkers between protein A-CT and N2- protein B were shortened such that there were no long identical sequence stretches present in
both fusion proteins. By this measure, recombination could be prevented
successfully in most cases.
However, the occurrence of recombination of course still depends on
the nature of the two interacting proteins or peptides A and B. In the case
-- ----------------------------.. -----~·t·_________· - -- that both parts have similar sequence stretches, either in constant parts or
in a randomized region, the risk of recombination increases significantly.
In an example of a library-versus-library experiment using two semirandomized libraries, recombinations were found only in clones in which an
identical stretch of 8 consecutive base pairs (bp) happened to exist in
libraries A and B. In these cases, the sequence stretch between those 8 bp,
which served as recombination sites, was doubled, which led to a wild typelike arrangement of gene 3. Such clones can therefore easily be recognized
by wild type-like infectivity.
However, such recombined clones can be eliminated in a simple and
fast way. Phage DNA of the entire library after the appropriate SIP round,
where recombined clones are found, is prepared, and the gene 3 cassette
is excised by restriction digest (in the vector described here, PinAl/ Hindlll;
Fig. 2E). The correctly sized band of the gene 3 cassette (1963 bp in our
example) can then be easily separated by gel electrophoresis from the larger
band created by the recombination (3379 bp ). The purified, correctly sized
band is cloned back into fresh vector and then used for further SIP selection.
With this approach, another round of SIP selection yields only clones with
correctly sized gene 3 cassette, as judged by analytical restriction digest, and
significant recombination usually does not occur for another two rounds. It
is advisable to perform an analytical restriction digest as a control for
recombination from the entire library or single clones obtained after selection.
Selective Infection Experiments
For infection, TetR or KanR XL1-Blue cells are grown in 2X YT medium,
supplemented with 1% (w/v) glucose, 50 mM MgC12 , and Tet (5 p,g/ml)
or Kan (50 p,g/ml), respectively, to an OD 550 of 0.5 to 0.8. It has been
shown previously that the addition of 50 mM MgC12 increases the infectivity
four- to sixfold. The number of phages used for infection should be on
the order of 10 to 10 phages, and is varied depending on the size of the
libraries and the expected infection rate. The appropriate amount of phages
is added to 0.5 ml of cells and shaken for 1 hr at 37°. Bacteria are plated
on 2X YT-agar containing 1% (w/v) glucose, Cam (30 p,g/ml), and Tet
(5 p,g/ml) or Kan (50 p,g/ml), depending on the host strain. For library
experiments 245 X 245 mm plates are usually most appropriate. The infectivity is determined by plating a series of 10-fold dilutions on small plates.
Plates are incubated overnight at 37 or 25° for more unstable proteins.
With this step, one round of SIP selection is finished. To proceed with the
library for further selection rounds, the cells are pooled from the plates
into 5-10 ml of rich medium, shaken for 5 min at 37° to allow for good
. ...
mixing and cell separation, and used for inoculation of a new overnight
culture. At the same time, glycerol stocks are prepared as a backup (stored
at -80°), by adding glycerol to a final concentration of 30% and making
aliquots of 500-1000 J.Ll. Individual clones can be analyzed by restriction
digest or sequencing of minipreparation DNA.
Occurence of Cysteine Mutations and Effects of Dithiothreitol
Libraries created by using synthetic oligonucleotides, even those of high
quality, almost always have a low percentage of frameshift mutants due to
imperfect oligonucleotide synthesis. Similarly, cysteines are introduced by
random mutagenesis techniques such as DNA shuffling or error-prone
PCR12.1 3,35 and will inevitably be present in eDNA libraries. Most of the
frameshift mutants will not be selected, because they lack the ability to
fold and bind their partner, but some might still show up in selection:
Provided a cysteine is introduced both at the end of the adapter and the
beginning of the cr domain, a covalently linked "wild type-like'' gene 3
protein is formed. In the following section, we highlight this problem and
point out general solutions.
In an example of a library-versus-library selection (library A-CT and
Nl-N2-library B) for noncovalently interacting pairs, strong selection for
a single cysteine in each library peptide was observed, even though the
designed sequences contained no cysteines at any place. The cysteines were
found to be caused by either point mutations (in library A) or by frameshifts
(in library B). The cysteine pairs caused a strong increase in infectivity of
tour orders of magnitude in one SIP round. The same phenomenon was
observed in a one-library approach (library A-CT and Nl-N2-peptide B),
in which also a point mutation in library A and a frameshift in peptide B
led to single cysteines (for details see Ref. 36). These covalent interactions
are mostly unspecific but nonetheless provide a strong selective advantage.
Consequently, the high infectivity of phages in the absence of DTI was
significantly reduced after DTI treatment. Control experiments showed
that incubation with DTI has only a minor effect on phage infectivity itself,
consistent with previous experiments, where it was reported that only phage
production but not phage infection and phage DNA replication is prevented
by 5 mM DTI. Furthermore, all four disulfide bridges in the native g3p
D. W. Leung, E. Chen, and D. V. Goeddel, Technique 1, 11 ( 1989).
S. Jung, K. M. Arndt, K. M. MUller, and A. Pltickthun. J. Jmmunol. Methods 231, 93 (1999).
M. Vaccaro, B. Boehler-Kohler. W. MUller, and I. Rasched, Biochinz. Biophys. Acta 923,
29 (1987).
are inaccessible to the alkylating agent vinylpyridine after treatment with
DTf (50°, 50 min, 100-fold molar excess of OTT over cysteines). 38
This stability of phages against reducing agents can be extremely useful
when working with complex libraries, where the occurrence of single cysteines cannot be ·excluded. In our experience, the purified wild-type phage
tolerates a 3- to 5-hr incubation at 37° with 5 mM DIT, pH 8.0, prior to
infection very well, with only a slight loss of infectivity. We usually dilute
the phages after DIT incubation and prior to infection to obtain a final
OTT concentration of abo.ut 10 ~-tM during infection, but the infection
worked equally well at a 100-fold higher OTT concentration (1 mM). However, the exact amount of OTT, the incubation time, and temperature
depend not only on the stability of the phage but also on those of the
proteins of interest. Nonetheless, OTT incubation can significantly reduce
but not fully eliminate the problem as some disulfide bridges might survive
the D1T treatment or reform during the infection process.
Detection of Translational Frameshifts by the Appearance of Polyphages
In a further SIP experiment using the same library-versus-library approach previously mentioned, two clones were selected with a -1 frameshift
in the library peptide before the CT domain. Extensive studies revealed
that neither the frameshift peptide, which would be produced instead of
the CT-library A fusion, nor the peptide from library B fused to Nl-N2,
could replace the function of CT. Most likely the observed -1 frameshift
is accompanied by a second, translational + 1 frameshift caused by two rare
arginine codons (AGO) that were present in both clones due to the -1
frameshift. This was further confirmed by the observation that these clones
formed polyphages, most likely due to insufficient capping of the nascent
phages caused by the limited amounts of CT domain, created by a translational + 1 frameshift (for details refer to Ref. 36).
It can therefore be concluded that the CT domain, and probably the
transmembrane helix, is not only needed for infection, but can definitely
not be replaced by "sticky pep tides," which would nonspecifically attach
Nl-N2 to the phage. Therefore, the SIP procedure must clearly bring
together Nl-N2 and. CT on the phage.
Examples from the Literature using in Vivo Selectively Infective ·
Phage Selection
In an example of in vivo SIP selection, a conserved region in the immunoglobulin variable domain was investigated by using a small synthetic
~ A. Krcmser and I. Rasched ,
Biochen1isrry 33, 13954 (1994 ).
Y R. A. Spanjaard and J. van Duin, Proc. Nat/. Acad. Sci. U.S.A. 85, 7967 (1988).
library. The library was fused to the CT domain, and the peptidic antigen
was fused to Nl-N2. After only three rounds of SIP, a strong enrichment
was found of clones similar to the most abundant sequences in the Kabat
database. In another example, the coiled-coil domain of c-J un with an
engineered free cysteine was fused to the CT domain of gene 3 protein
and used as bait to select against a human eDNA library fused to the soluble
N-terminal domains_3.4l Several clones were isolated that were predicted to
encode potential coiled-coil structures. However, in our hands, constructs
with free cysteines, even though they show higher infectivity, showed less
specificity in their interaction. In addition, in the one-library approach we
prefer to fuse the library to the CT domain rather than to Nl-N2. To keep
phenotype and genotype connected, it is advantageous to have the library
tightly connected to the phage, in case the ligand dissociate.s and exchanges
to a different phage during the infection in the library pool.
A further model experiment showed the feasibility of a two-vector
packaging of adapter and CT fusion, which would be useful for library34
versus-library selection. In this case, protein A-CT was displayed on
phage and encoded on the phage genome and the Nl-N2-bait was encoded
on a phagemid. After cotransformation both vectors were copackaged in
polyphages, which were screened on special filter plates.
SIP is a powerful strategy to select for protein-ligand interactions as
well as for other desired features such as protein folding and stability,
provided the binding function of the protein is limited to the native state.
The selection is carried out completely in solution without any need for
solid-phase panning steps. Furthermore, the enrichment cycle is fast and
thus less time consuming. Most importantly, the selection power is ex5
tremely high and enrichment factors of 10 to 106 have been observed, 42
while in phage display enrichment factors of 10 to 10 are normal. Therefore, in most applications SIP requires only one selection round to separate
binders from nonbinders, 28•29 whereas phage display usually needs three or
four rounds. In three or four SIP rounds, it is moreover possible even to
discriminate between, more subtle affinity, folding, and stability differences. 29•40
S. Spada, A. Honegger, and A. Pltickthun, 1. Mol. Bioi. 283, 395 ( 1998).
K. Gramatikoff, W. Schaffner, and 0. Georgiev, Bioi. Chen1. 376, 321 (1995).
M. Duenas, L. T. Chin, A. C. Malmborg, R. Casalvilla, M. Ohlin, and C. A. Borrebaeck,
lmn1unology 89, 1 (1996).
G. Winter, A. D. Griffith, R. E. Hawkins, and H. R. Hoogcnboom, Annu. Rev. /nununo/.
u, 433 (1994).
However, be aware that this extremely strong selection ability can sometimes lead to unwanted solutions, by covalently linking the N terminal to
the C-terminal part of g3p. This is o.nly a problem in the in vivo SIP
procedure, where both partners can potentially acquire mutations, because
in in vitro SIP the infection-mediating particle is produced separately. With
highly complex libraries, products of random mutagenesis, and eDNA libraries, a selection for disulfide bridges and enrichment of recombined
clones is possible. Genetic recombination is a rare event, easy to check and
to remedy by cutting out and recloning restriction fragments. The selection
for disulfide bonds from spurious cysteines can be most effectively conlibraries
trolled by working with high-quality
agents also help, but do not eliminate the problem. Currently, the work
with eDNA is still a challenge.
For a library-versus-library selection, only in vivo SIP is applicable, but
needs to be strictly controlled. It also remains to be determined what the
maximum size of the fusion protein can b·e, whether there are geometric
restrictions imposed by the infection process, and what the minimum affinity
for the interaction is. In an in vivo SIP library-versus-library approach, a
weak interaction might lead to the exchange of adapter molecules between
different clones and thus destroy the genotype-phenotype correlation. An ..
other limitation for complex libraries is the moderate infection efficiency.
The highest observed. was 1 infection per 10 SIP phages, but much lower
infectivities have also been reported.
Taking everything together, SIP is at this point most useful in the fast ·
screening of less complex libraries and in molecular improvement. We .
generally re~ommend the use of the in vitro SIP procedure over in vivo
SIP whenever possible, because it is more robust and easier to control. ...."
Furthermore, by choosing the right adapter and titrating its concentration;,;.
it allows for selection of medium as well as high-affinity interactions. Fur~.·:-·_
thermore, the in vivo SIP complements very well the other two techniques .-:~.
available to directly select two libraries simultaneous against each other,:~
the yeast two-hybrid system and the protein fragment complementatio~.lli
assay, which select in the nucleus of yeast or the cytosol of E. coli, respec~~
tively. In in vivo SIP proteins are folded in the oxidative environment o{;~
the periplasm, which is essential for selection from libraries containing1~-.
proteins that require the formation of disulfide bridges.
.· ~~
.. · ~
..., .f
•. .J"''
. •r-:\
. •'1'tz·.
.. Fi>o''
..-. ...:)!(.';
. -<..:f:·
· ~·t:' )
,.. .
. . .. .. '•j
' ·..· ·:"'
. ·t.
··· -.·~
S. Fields and 0. Song, Nature (London) 340, 245 (1989).
J. N. Pelletier, K. M. Arndt, A. Pliickthun, and S. W. Michnick, Nature Biotechnol. 11;;;
683 (1999).
. '·:i!r'·J.'l
Пожаловаться на содержимое документа