doi:10.1016/j.jmb.2004.06.030
J. Mol. Biol. (2004) 341, 635–649
In Vivo Selection of Combinatorial Libraries and
Designed Affinity Maturation of Polydactyl Zinc Finger
Transcription Factors for ICAM-1 Provides New
Insights into Gene Regulation
Laurent Magnenat, Pilar Blancafort and Carlos F. Barbas III*
The Skaggs Institute for
Chemical Biology and the
Department of Molecular
Biology, The Scripps Research
Institute, 10550 North Torrey
Pines Road, La Jolla, CA 92037
USA
Zinc finger DNA-binding domains can be combined to create new proteins of desired DNA-binding specificity. By shuffling our repertoire of
modified zinc finger domains to create randomly generated polydactyl
zinc finger proteins with transcriptional regulatory domains, we developed large combinatorial libraries of zinc finger transcription factors
(TFZFs). Millions of TFZFs can then be simultaneously screened in mammalian cells. Here, we successfully isolated specific TFZFs that significantly
positively and negatively modulate the transcription of the ICAM-1 gene
in primary and cancer cells, which are relevant to ICAM-1 biology and
tumor development. We show that TFZFs can work in a general and in a
cell-type specific manner depending on the regulatory domain and the
zinc finger protein. We show that a TFZF that interacts directly with the
ICAM-1 promoter at an overlapping NF-kB binding enhancer can overcome or synergistically cooperate with NF-kB induction of ICAM-1. For
this TFZF, rational design was used to optimize the binding of the zinc finger protein to its DNA element and the resulting TFZF demonstrated a
direct correlation between increased affinity and efficiency of target gene
regulation. Thus, combining library and affinity maturation approaches
generated superior TFZFs that may find further applications in therapeutic
research and in ICAM-1 biology, and also provided novel mechanistic
insights into the biology of transcription factors. Transcription factor
libraries provide genome-wide approaches that can be applied towards
the development of TFZFs specific for virtually any gene or desired phenotype and may lead to the discovery of new genetic functions and
pathways.
q 2004 Elsevier Ltd. All rights reserved.
*Corresponding author
Keywords: polydactyl zinc finger; designer transcription factor; ICAM-1
regulation; in vivo library selection; affinity maturation
Introduction
The modularity of the zinc finger (ZF) domains
Abbreviations used: ZF, zinc finger; TFZFs, ZF
transcription factors; ZFPs, ZF proteins; IRES, internal
ribosome entry site; GFP, green fluorescent protein;
KRAB, Krüppel-associated box; SID, mSin3 interaction
domain; HUVEC, human umbilical vein endothelial
cells; ChIP, chromatin immunoprecipitation; MFI, mean
fluorescence intensities; EMSA, electrophoretic mobilityshift assays; EGF, epidermal growth factor; ICAM-1,
intercellular cell adhesion molecule.
E-mail address of the corresponding author:
carlos@scripps.edu
allows for the development of ZF transcription factors (TFZFs) that control the expression of genes of
biological and therapeutic interest. Prototypical ZF
domains bind 3 bp of DNA sequences through the
formation of specific contacts primarily within the
major groove of the DNA. By using selective strategies, our laboratory and others have successfully
changed the sequence specificity of ZF domains in
a directed fashion and have generated polydactyl
zinc-finger proteins for targeting unique sites
within complex genomes.1 – 5 When fused to
transcription activation or repression domains,
designed ZF proteins (ZFPs) become regulators
of the transcriptional activity of target genes in
cultured cells and in living plants and animals.5 – 11
0022-2836/$ - see front matter q 2004 Elsevier Ltd. All rights reserved.
636
Directed artificial gene regulation with rationally
designed ZFPs can be limited by a lack of information concerning the target gene, including
chromatin structure, the presence of endogenous
transcription factors and DNA accessibility. As an
alternative to the design and testing of singular
ZFPs, we recombined our set of predefined ZF
domains to construct random libraries of threeand six-ZF proteins. When attached to the desired
effector domain, the large libraries of new polydactyl-ZF DNA binding proteins become genomewide tools that can be screened in vivo (in this context referring to events occurring in living cells but
potentially in whole organisms) by selection in
mammalian cells for the discovery of novel functional transcription factors. We recently reported
preliminary studies concerning the construction
and screening of TFZF libraries for the selection of
VE-cadherin gene regulators.12
The transcriptional regulation of the gene encoding for the intercellular cell adhesion molecule 1
(ICAM-1, CD54) is dynamic and is implicated in
biology and in a variety of diseases. Disorders
associated with ICAM-1 deregulation include,
malignancies, inflammatory disorders, atherosclerosis, ischemia, neurological disorders and
organ transplantation. ICAM-1 is expressed at a
basal level in many cell types, including leukocytes
and endothelial cells, and binds to the b2 integrins
present on the cell surface of leukocytes.13 This
interaction promotes adhesion and signaling for
transendothelial migration of leukocytes and
for T-cell co-activation during inflammatory
and immune responses.14 Significantly, ICAM-1
transcription is spatiotemporally regulated in
endothelial and cancer cells during tumor angiogenesis, metastasis and progression.
Given the diverse roles of ICAM-1 in biology,
directed regulation of ICAM-1 expression with
novel TFZFs might be an important tool in vivo for
the development of anti-inflammatory and anticancer therapies. Here, we have chosen ICAM-1
regulation as a model system and report a new
approach to the discovery and optimization of
transcriptional regulators. This study led to the
development of a set of ICAM-1 regulators
(CD54-TFZFs), that are able to significantly up-regulate or completely suppress ICAM-1 expression in
primary cells and a variety of cell lines of special
interest for ICAM-1 biology. Moreover, we demonstrate that one of the selected CD54-TFZFs interacts
directly with the ICAM-1 promoter at a site,
normally known to confer responsiveness to natural inducers via NF-kB signaling pathway. The
other CD54-TFZFs may regulate via unknown DNA
elements, genes and genetic pathways involved in
ICAM-1 expression. In order to understand the
generality and the particulars of this approach, the
activities of the TFZFs were evaluated in different
contexts by testing different cell-types, by comparing regulatory domains, and by modifying zinc finger characteristics, including DNA-binding affinity
and specificity. Thus, in this study, we detail a
Novel ICAM-1 Zinc Finger Transcription Factors
powerful strategy for generating new transcription
factors for the potent activation and repression of
endogenous genes. Additionally, this study brings
valuable new mechanistic insight into the biology
of TFZFs. These insights will be important for others
interested in either engineered or natural transcription factors.
Results
Selection of TFZFs libraries in mammalian cells
for ICAM-1 regulators
Whereas three-ZF proteins recognize a 9 bp site
with affinities in the nanomolar range, proteins
containing six-ZF domains typically bind to 18 bp
sequences with better affinities.2,5,15 Therefore, in
order to target sites that are in principle unique
within the human genome, we developed and
used a library of artificial transcription factors containing shuffled 6ZF modules with the canonical
TGEKP linker as a connector between the ZF
domains wherein the resulting DNA-binding
protein is fused to the VP16-derived transactivation domain VP6412 (Figure 1(a)). Most of the
domains included in the library were selected in
vitro by phage display and optimized by sitedirected mutagenesis for specific binding to each
of the possible GNN subsites.3,16 Additional ANN
and TNN domains4,17 were also utilized to prepare
a six-finger library with a diversity of 8.42 £ 107
proteins, approximately 2000 times as many transcription factors as there are genes in the human
genome.
The 6ZF-VP64 library was introduced into A431
cells by retroviral infection using the pMX-IRESGFP retroviral vector, which expresses a single
bicistronic message for the translation of an effector gene and, from an internal ribosome entry site
(IRES), the green fluorescent protein (GFP).7,18
Since both proteins are coded on the same RNA,
GFP expression is used as an indicator of infection
efficiency and ZF expression. For screening the
pMX-6ZF-VP64 library for ICAM-1 regulators,
infected cells showing ICAM-1 immunofluorescence up-regulation were isolated using flow cytometry. After cell sorting, the recovered cells were
harvested to prepare genomic DNA. The retrovirally integrated ZF pools were PCR amplified
and recloned into pMX-VP64-IRES-GFP vector for
further rounds of selection (Figure 1(b)). During
the three rounds of selection, the ICAM-1 cell-surface expression gradually increased and plateaued.
A total of 35 individual CD54-TFZFs clones were
then screened. Among the most potent transactivators of ICAM-1, four 6-ZF proteins, CD54-3, CD5413, CD54-30 and CD54-31, were independently
cloned multiple times and reproducibly up-regulated endogenous ICAM-1 expression in A431
cells (Figure 1(c)). The expression level of the
endogenous ICAM-1 protein, which is already
present at moderate levels on normal A431 cells,
Novel ICAM-1 Zinc Finger Transcription Factors
637
Figure 1. TFZF library, selection strategy and isolation of TFZF clones. (a) Shuffling of ZF domains generates a large
library of 6-ZFPs. Fusion of the library to transcriptional effector domains allows for the screening of TFZFs regulating
virtually any gene. (b) ICAM-1 up-regulators were isolated by FACS sorting of GFPþ and ICAM-1þ cell population
from the gated region R1 and by recovery from integrated retroviral 6ZF DNA pools and subsequent rounds of selection. The relative increase in ICAM-1 mean fluorescence intensity (MFI) during three rounds of selection is compared
to mock infected cells (pcDNA) or cells infected with the unselected library (6ZFlib). (c) ICAM-1 cytometric profiles
of cells infected with the cloned retroviral CD54-transactivators (bold line) were compared to cells infected with the
unselected pMX-6ZF-VP64 library (thin line). The basal autofluorescence of A431 cells is represented by a broken line.
was increased two to four times as compared to
cells infected with the unselected TFZFs library
(Table 1). Interestingly, up-regulation was not correlated with the level of GFP expression (data not
shown), suggesting that CD54-TFZFs clones may
have different intrinsic efficiencies as regulators.
The polydactyl ZF domains were sequenced and
their predicted target sites were deduced from our
database of ZF domains and corresponding 3 bp
subsites. The CD54-6ZF proteins, expressed and
purified as maltose-binding protein (MBP-CD54)
fusions, bound their predicted 18 bp target
sequences with affinities ranging from 1 nM to
10 nM as determined in vitro by EMSA (Table 1).
Effector domain swapping in CD54-TFZFs for
ICAM-1 repression
In order to direct repression of ICAM-1, CD546ZF proteins were subcloned into the pMX retroviral vector as fusions to the transcriptional repressors Krüppel-associated box (KRAB) and the
mSin3 interaction domain (SID).6,19,20 The resulting
pMX-CD54-KRAB and pMX-CD54-SID constructs
were tested for repression of endogenous ICAM-1
expression in A431 carcinoma, C8161 melanoma
and Kaposi’s sarcoma SLK cells (Figure 2). The
KRAB domain only functioned as an ICAM-1
repressor in fusions with ZFPs CD54-13 and
CD54-31. Potent ICAM-1 knock-down was
achieved with CD54-31-KRAB in three cell lines
(9 – 1% ICAM-1 remaining), whereas 50 – 13%
ICAM-1 still remained with the same ZFP in combination with the SID domain. On the other hand,
the SID domain worked as repressor with all four
CD54-6ZF at variable degrees depending on the
cell line. For instance, CD54-30 was able to significantly repress ICAM-1 down to 15% of the native
level, but only with SID and only in C8161 cells.
Also, CD54-3 worked exclusively with SID. However, CD54-13 paired equally with both repressors.
The CD54-TFZFs transcriptional levels measured
by semi-quantitative RT-PCR varied slightly
depending on the nature of zinc finger and
638
DCRDLAR
RSDDLVR
QAGHLAS
CD54-31Opt
The fold activation of the endogenous ICAM-1 gene in A431 cells and dissociation constants are shown.
a
Zinc finger helices are positioned in the anti-parallel orientation (C-terminal-F6 to F1-N-terminal) relatively to the DNA target sequence. Amino acid position 21 to þ 6 of each DNA recognition helix is shown.
b
Predicted target DNA sequences are presented in the 50 to 30 orientation.
c
Fold increase ICAM-1 expression in A431 cells. Mean fluorescence intensity (MFI) corrected for background auto-fluorescence and standardized to the ICAM-1 intensity of A431 cells infected
with the unselected 6ZF-VP64 library ( ¼ 1). Average of relative MFI from 2 to 11 independent experiments (statistical significance of observed differences was determined using the Student’s t-test,
p , 0:06).
d
Dissociation constant ðKd Þ determined by gel shift assay. Data represent the average of two to four independent EMSA experiments. Zinc finger domains used for CD54-31 optimization (CD5431Opt) are presented in bold. Nucleotide positions that do not match the predicted target sequence are in lowercase letters.
9.1 £
7.1
5.1
8.5
1
3.1
0.16
3.6 £
2.2 £
4.0 £
3.6 £
GCC-30
GTA-30
GCA-30
GCC-30
GCC-30
GCC-30
GTT GAC
GCC GGG
GTA GGA
AAA GCG
gAA GCG
GAA GCG
–
–
–
–
–
–
50 -AAA GTT AAA
50 -GAC GGT AAA
50 -GAA GTT GTA
50 -TGA TGA GTT
50 -Tcc gGA GcT
50 -Tcc GGA GCT
DCRDLAR
QSSSLVR
QSGDLRR
DCRDLAR
QRANLRA
DPGNLVR
QSSNLVR
QAGHLAS
CD54-3
CD54-13
CD54-30
CD54-31
TSGSLVR
QRANLRA
TSGSLVR
TSGHLVR
QRANLRA
DCRDLAR
TSGSLVR
QSSSLVR
QSSSLVR
QAGHLAS
TSGSLVR
QRANLRA
Natural ICAM-1 promoter site (pro-220):
QRAHLER
TSGELVR
QSSNLVR
DPGNLVR
RSDKLVR
QRAHLER
RSDDLVR
Half-site 1
N-term F1
F2
F3
F4
F5
C-term F6
6ZF
Zinc finger helicesa
Table 1. DNA interacting helices of 6ZF clones activating ICAM-1 and their predicted 18 bp target sites
Target sitesb
Half-site 2
Fold (MFI)c
Kd (nM)d
Novel ICAM-1 Zinc Finger Transcription Factors
repression domains and correlated well with
the fluorescence levels of the bicistronically
co-expressed GFP marker in A431 cells (Figure
2(a)). However, the variations of both TFZFs
mRNA and GFP marker did not correlate with the
efficiency of ICAM-1 repression by the different
CD54-TFZFs and could not account for the striking
differences observed above (Figure 2(b) –(d)).
Thus, the choice of repressor domain is important
depending on the ZF protein and the cell line to
repress ICAM-1. We chose CD54-31-KRAB as a
general repressor for further studies.
Artificial TFZFs function efficiently in
primary cells
Artificial TFZFs perform well in cancer and transformed cells, but their activity in primary human
cells remains to be fully explored. Therefore,
we assessed CD54-31 TFZFs in primary cells and
telomerase (hTERT)-immortalized primary cells,
shown to maintain a normal karyotype and
phenotype.21,22 The effect of CD54-31 fusion proteins was considerable in human umbilical vein
endothelial cells (HUVEC), where the ICAM-1
level was reduced by 83% in combination with
KRAB and increased 63-fold with VP64 (Figure 3).
In immortalized human primary mammary epithelial cell line (hTERT-HME1) and in human
hTERT-fibroblasts, where ICAM-1 was present at
much higher extent, regulation was also effective
(Figure 3). These results demonstrate the wide
potential of the TFZFs in normal cells.
ICAM-1 promoter scanning and the search for
the CD54-31-binding site
TFZFs act by binding to DNA and recruiting
different factors of the transcription machinery. In
order to determine if the selected CD54-TFZFs
bound directly to the ICAM-1 promoter or may
have acted indirectly through the activation of
other genes involved in the expression of ICAM-1,
CD54-6ZF coding DNAs were subcloned into the
pcDNA-VP64 transient expression vector for activation studies with a luciferase reporter construct
driven by a 1.6 kb fragment of the ICAM-1 promoter (pGL3-ICAM-1). Two of the CD54-TFZFs,
CD54-3 and CD54-31, were able to significantly
up-regulate luciferase activity in transient transfection assays in A431 cells (Figure 4(a)). Comparison
of the promoter sequences and the predicted 18 bp
target sequence of CD54-31 ZFP suggested several
potential target sites. In order to determine the site
the TFZF functions through, purified MBP-CD54-31
ZF fusion protein was tested in an ELISA assay
for binding to the candidate target DNAs. Whereas
the CD54-31 protein bound preferentially to its predicted DNA sequence (cd54-31) at concentrations
in the nanomolar range, it also bound effectively
to an 18 bp promoter site, ICAM-1 pro-220 (Figure
4(b)). This 18 bp sequence shared 13 nt with
the predicted sequence (Table 1) and was found
639
Novel ICAM-1 Zinc Finger Transcription Factors
Figure 3. ICAM-1 regulation in primary cells mediated
by CD54-TFZFs. HUVEC cells, telomerase immortalized
primary mammary epithelial cells hTERT-hME1 and
hTERT-fibroblast retrovirally expressing CD54-31 ZFP
fused to either VP64 (red line) or KRAB (blue line) domains
were compared to the normal ICAM-1 expression of cells
infected with the unselected pMX-6ZF library (green line)
and the basal autofluorescence (broken line).
220 nt upstream of the initiation of translation of
ICAM-1.23 In agreement with this result, the affinity of the MBP-CD54-31 protein for the predicted
cd54-31 oligonucleotide was 1 nM and 3.1 nM for
the ICAM-1 pro-220 DNA (Table 1). Significantly,
this promoter sequence overlaps an enhancer
element, shown to bind to NF-kB and convey
responsiveness to tumor necrosis factor-a (TNFa),
interleukin 1b (IL-1b) and other factors.13
Natural target site validation and CD5431 optimization
When the ICAM-1 pro-220 site was replaced by
the target site of an unrelated 6-ZFP E2C in the
luciferase construct (pGL3-E2C/ICAM-1), CD5431-VP64 was no longer able to efficiently up-regulate luciferase activity. However, the mutant promoter was nonetheless inducible by the E2C-VP64
control protein, showing that the substitute e2c
sequence was not detrimental to the reporter construct, further indicating that ICAM-1 pro-220 is
the target site of CD54-31 (Figure 4(c)). Because
the 18 bp target site found in the ICAM-1 promoter
was not optimal for the CD54-31 protein, the 6ZF
protein was optimized for binding the natural
promoter site using rational design and protein
grafting of specific DNA-binding a-helices6
(Table 1). For instance, the ZF domains selected in
Figure 2. ICAM-1 repression and relative CD54-TFZFs
expression in different cancer cell lines. (a) Correlation
of mean fluorescence intensities (MFI) of the coexpressed GFP marker (black bars, average of three independent experiments) with TFZF – KRAB mRNA levels
measured by RT-PCR (white bars) in A431 cells. Values
are represented relatively to the CD54-3-KRAB levels
( ¼ 1). Decrease in ICAM-1 MFI in (b) A431 carcinoma,
(c) C8161 melanoma and (d) Kaposi sarcoma SLK cells
infected with CD54-3, CD54-13, CD54-30 and CD54-31
6-ZFPs fused to either KRAB (black bars) or SID (white
bars) repression domains. Mean fluorescence intensities
(MFI) are represented as percent ICAM-1 remaining and
standardized to normal ICAM-1 levels in cells infected
with the unselected pMX-6ZF library (6ZFlib ¼ 100%
ICAM-1 remaining) and to basal autofluorescence
( ¼ 0% ICAM-1) as determined by FACS. In parenthesis
are indicated the relative bicistronically expressed GFP
levels (MFI) corrected for background autofluorescence
for each CD54-TFZFs.
640
Novel ICAM-1 Zinc Finger Transcription Factors
Figure 4. ICAM-1 promoter regulation, natural target site determination and CD54-31 optimization. (a) ICAM-1
promoter activation in transient luciferase reporter assays. Normalized luciferase activity of pGL3-ICAM-1 reporter
induced by CD54-transactivators was compared relatively to pcDNA vector control ( ¼ 1). (b) Promoter scanning
ELISA. The DNA-binding intensity of purified MBP-CD54-31 protein for potential natural ICAM-1 promoter sites is
presented relatively to the predicted target site cd54-31 ( ¼ 1). (c) Mutant reporter assay. The pGL3-E2C/ICAM-1
mutant reporter construct was tested for lack of transactivation with the CD54-31-VP64 construct. The E2C-VP64 positive control was known to activate the e2c target site in the ErbB2 gene.6 (d) Effect of CD54-31 optimization on the
transactivation of the ICAM-1 promoter. Normalized luciferase activity of pGL3-ICAM-1 reporter induced by CD5431 and CD54-31Opt transactivators is presented as described in (a). (e) DNase I footprints of purified CD54-31 (left
gel) and CD54-31Opt (right gel) MBP fusion proteins on the ICAM-1 promoter. DNA incubated with tenfold dilutions
of ZFP (1000 –1 nM) was run in parallel to DNA with DNase I only (DNase I), DNA alone (Control), and chemical
sequencing of the 200 bp DNA probe (G þ A ladder). The ICAM-1 pro-220 target site (capital letters), site of DNase I
protection (dotted line) and the overlapping NF-kB enhancer (broken line) are presented.
CD54-31 for finger positions F3-AAA (QRANLRA),
F4-GTT (TSGSLVR) and F5-TGA (QAGHLAS)
were, respectively, shown to bind in vitro to the triplets GNN, GCT and GGA with reduced affinity.3,16
Suitably, those triplets matched the ICAM-1
pro-220 sequence (50 -TCC GGA GCT GAA GCG
GCC-30 ). Therefore, three out of six ZF domains
were replaced in a newly designed and optimized
Novel ICAM-1 Zinc Finger Transcription Factors
641
Figure 5. In vivo binding of CD54-TFZFs to the ICAM-1 promoter and effects on the NF-kB signaling pathway
inducing ICAM-1 in response to proinflammatory cytokines. (a) ChIP assay with normal (A431) and TFZFs expressing
(31-KRAB and 31Opt-KRAB) cells. Formaldehyde cross-linked chromatin was immunoprecipitated with a TFZF-specific
antibody (ZF) or without antibody (No) and analyzed by PCR using primers specific to the ICAM-1 promoter region
surrounding the NF-kB enhancer. Total input chromatin (In) was used as positive control. (b)– (e) Inhibition and
synergistic up-regulation of NF-kB mediated cytokine induction of ICAM-1 with CD54-TFZFs. FACS analysis of
ICAM-1 in HUVEC (b) and (c) and A431 cells (d) and (e) retrovirally infected with CD54-31-KRAB (red line) and
CD54-31Opt-KRAB (orange line) TFZFs and subsequently treated with TNF-a (b) or Il-1b (c) – (e). Normal ICAM-1
expression in control cells without (green broken line) and with cytokine treatment (green plain line), and basal autofluorescence (black broken line) are presented. (e) ICAM-1 mean fluorescence intensities of untreated (black bars) and
IL-1b (white bars) treated A431 cells infected with CD54-31 and CD54-31Opt zinc finger proteins alone (31- and
31Opt-) or either coupled to KRAB (31-KRAB and 31Opt-KRAB) or VP64 (31-VP64 and 31Opt-VP64) domains.
protein (CD54-31Opt) with domains determined to
have better specificity in vitro, namely F3-GAA
(QSSNLVR), F4-GCT (TSGELVR) and F5-GGA
(QRAHLER) (presented in bold in Table 1). Since
the finger positions F1-GCC and F2-GCG were
optimal and because no domains were available
for a F6-TCC subsite, the originally selected DNA
binding a-helices were retained at these positions
in the protein. The affinity of purified MBP-CD5431Opt fusion protein for the ICAM-1 pro-220 site
was determined to be 0.16 nM, approximately 20
times better than the dissociation constant of
CD54-31 for the same DNA (Table 1). As a result,
CD54-31Opt-VP64 induced twice the level of
luciferase activity as the original CD54-31-VP64
construct in transiently transfected A431 cells
(Figure 4(d)). Finally, DNase I footprinting
demonstrated actual binding of both proteins,
CD54-31 and CD54-31Opt, to the same ICAM-1
pro-220 promoter sequence in vitro (Figure 4(e)).
Together, these results convincingly demonstrate
that ICAM-1 pro-220 is the endogenous target
site of CD54-31 and that it can be further
induced by designed affinity maturation of CD5431Opt.
In vivo binding, competition and synergistic
cooperation of CD54-TFZFs with endogenous
factors at the ICAM-1 promoter
A chromatin immunoprecipitation (ChIP)
experiment was performed to establish the interaction of the selected and optimized CD54-TFZFs
with the ICAM-1 promoter site in vivo (Figure
5(a)). A chromatin fragment comprising the
ICAM-1 pro-220 site was efficiently cross-linked
by the CD54-31-KRAB and -31Opt-KRAB proteins
expressed in A431 cells and co-immunoprecipitated with a polyclonal antibody raised specifically
against the framework of our designer zinc finger
domains. This framework consist of a consensus
peptide sequence derived from 131 zinc finger
domains of different origins,24 that was shown to
provide better stability and affinity properties to
Sp1C, a hybrid zinc finger protein in which only
the residues from the three DNA recognition
helices of the natural Sp1 transcription factor
were grafted into the consensus framework.25
While the same antibody specifically reacted with
retrovirally expressed CD54-TFZFs in crude cell
extracts on Western blot (data not shown), no
642
Novel ICAM-1 Zinc Finger Transcription Factors
Figure 6. Specificity of the CD54-TFZFs. (a) Multitarget specificity in vitro assay. Purified MBP CD54-6ZF fusion proteins CD54-3, CD54-13, CD-54-30 and CD54-31 were tested in a DNA binding ELISA against each corresponding target
DNA oligonucleotides, cd54-3, cd54-13, cd54-30 and cd54-31. The CD54-31Opt protein and the ICAM-1 pro-220 promoter target site were also included. The DNA-binding intensity (A405 nm) is presented for each protein relatively to
the respective target site ( ¼ 1) as determined from duplicate experiments, for which dilution series of 60 nM protein
were used. (b)– (h) Multitarget specificity of CD54-TFZFs in cancer and primary cells. Retrovirally expressed CD54TFZFs, CD54-31Opt (orange line) and CD54-31 (red line) VP64 activators in (b) A431 and (c) HUVEC cells, (d) KRAB
repressors in A431 cells, and (f) CD54-3-VP64, (g) CD54-13-VP64 and (h) CD54-30-VP64 (all red line) in A431 cells,
were compared by FACS for the regulation of endogenous ICAM-1 and evaluated for regulation of other cell surface
markers Epidermal growth factor (EGF), 3-FAL selectin ligand (CD15, FUT4), integrin-a6 (CD49f, ITGA6), leukocyte
function-associated antigen (CD58, LFA-3), Apo1-FAS antigen (CD95, TNFRSF6), integrin-b4 (CD104, ITGB4) and vascular endothelial VE-cadherin (CD144, CDH5). Control cells are labeled as in Figure 3. (e) Semi-quantitative RT-PCR
analysis of A431 cells retrovirally infected with CD54-TFZF KRAB repressors for specificity of ICAM-1 and ITGA6 regulation at the transcriptional level in relation to TFZFs and GAPDH mRNAs levels. Controls experiments include RNAs
from mock infected A431 cells (M), cells infected with the unselected pMX-6ZF library (6ZFlib) and from CD54-31OptKRAB cells in the absence of reverse transcriptase (2).
643
Novel ICAM-1 Zinc Finger Transcription Factors
immunoprecipitated DNA fragment was detected
by PCR in normal A431 cells, nor without the
addition of the antibody. This demonstrates that
the immunoprecipitation complex was only
formed with our artificial TFZFs but not with
endogenous zinc finger containing factors.
While CD54-31-KRAB was able to efficiently
repress the constitutive expression of ICAM-1
(Figure 2) and bound to a sequence that was overlapping with the binding element of NF-kB, it was
interesting to determine the effect on the induced
over-expression of ICAM-1 in response to cytokines via NF-kB signaling (Figure 5(b) – (e)).
Mutation of this site completely abolished ICAM-1
promoter activation by TNF-a and IL-1b,26 the
major inducers of ICAM-1 expression in most cell
types which act through NF-kB signaling pathway.
Amongst all the ICAM-1 inducing factors assayed,
TNF-a was a better inducer than Il-1b by increasing expression more than 20-fold over the weak
constitutive ICAM-1 expression observed in
HUVEC cells (Figure 5(b) and (c)) and IL-1b was
the most potent inducer in A431 cells (Figure 5(d)
and (e)). CD54-31 and -31Opt TFZFs coupled to
VP64 activators up-regulated ICAM-1 through a
range far exceeding the induction provided by
natural factors (Figure 5(e)). Most importantly,
when coupled to KRAB repressors they were still
able to completely repress the inducible and constitutive ICAM-1 expression in the presence of
cytokines, indicating their ability to either compete
or overcome NF-kB regulation in both cell lines.
Similarly, total inhibition of ICAM-1 induction
was also seen in A431 and HUVEC with other factors such as lipopolysaccharide, phorbol 12-myristate-13-acetate, and interferon gamma, which
signals through other transcription factors that
bind to ICAM-1 regulatory elements different
from the NF-kB enhancer (data not shown). Interestingly, the same zinc finger proteins coupled to
the VP64 activation domain, in combination
with Il-1b together increased ICAM-1 expression
greater than the additive increase provided by the
factors individually. Zinc finger proteins lacking
the effector domain had no measurable effect
(Figure 5(e)).
Specificity of CD54-TFZFs and activity in
biologically relevant cell lines
As the specificity and modularity of individual
ZF domains assembled in a polydactyl ZF protein
can vary, it was important to evaluate the specificity of the CD54-6ZF proteins. Initially, purified
fusion proteins were assessed in a multi-target
specificity ELISA for binding to each of the predicted
or natural target sites. As shown in Figure 6(a), each
protein bound specifically to its corresponding
DNA hairpin oligonucleotide. Note that the CD5431Opt protein was redirected towards binding to
the ICAM-1 pro-220 target DNA with a fivefold preference over the originally predicted cd54-31 target
site. Next, the retrovirally expressed CD54-VP64
TFZFs were tested for their ability to alter ICAM-1
expression. Study of cytometric profiles of ICAM-1
and seven other cell surface markers in A431 and
HUVEC cells infected with CD54-31-VP64 and
CD54-31Opt-VP64 revealed that both TFZFs reproducibly and preferentially up-regulated ICAM-1 compared with the other markers (Figure 6(b) and (c)).
Further, CD54-31Opt was more than twice as potent
as CD54-31 in both cell lines (Table 2), thus reaching
one order of magnitude increase in ICAM-1 fluorescence signal in A431 and two in HUVEC. Optimization of CD54-31 also corrected the non-targeted
regulation found with CD54-31 for one marker
(CD144) in A431 and another in HUVEC (CD95),
while slightly affecting others (CD104 and CD58).
The specificity of the KRAB effector variants was
also tested in A431 cells (Figure 6(d)), where both
selected and optimized CD54-31 TFZFs repressed
completely ICAM-1 but did not affect the other
genes. A semi-quantitative RT-PCR was performed
(Figure 6(e)) and only the ICAM-1 mRNA was
Table 2. Fold activation and repression ICAM-1 expression with CD54-31 and CD54-31Opt TFZFs in cell lines
CD54-31
Type
Cancer
Primary
Cell line
Colon
Colon
Colon
Breast
Breast
Breast
Melanoma
Epidermoid
Endothelium
Lim1215
HT29
SW1222
T47D
SKBR3
MD-AMB-435S
C8161
A431
HUVEC
a
b
Relative ICAM-1 fluorescence
VP64
85
62
1
33
4
627
31
93
6
1.9
7.3
137.0
13.0
59.0 (14.5)
3.1
22.1 (0.27)
3.6 (1)
63.0 (2.83)
CD54-31Opt
c
b
KRAB
VP64
(Fold opt.d)
KRABc
3 (2.8)
5 (1.4)
23 (8)
3.4
10.1
142.0
9.6
58.0
4.5
59.0
9.1
135.0
1.8 £
1.4 £
1.0 £
0.7 £
1.5 £
1.0 £
2.7 £
2.5 £
2.1 £
10
6
13
a
Normal ICAM-1 mean fluorescence intensity is corrected for background and standardized to the autofluorescence ( ¼ 1).
Average value from two to seven experiments.
b
Fold increase in ICAM-1 expression. Mean fluorescence intensity with VP64 transactivators relatively to cells infected with
unselected 6ZF-VP64-library ( ¼ 1). In parentheses is presented the standard deviation of two independent experiments.
c
Percentage remaining ICAM-1 mean fluorescence intensity with the KRAB repressors relatively to cells infected with unselected
6ZF KRAB-library ( ¼ 100%) and autofluorescence ( ¼ 0%). In parentheses is presented the standard deviation of two independent
experiments.
d
Additional fold VP64 activation obtained with CD54-31Opt as compared to CD54-31.
644
Novel ICAM-1 Zinc Finger Transcription Factors
Figure 7. ICAM-1 regulation in different cell types. Activators CD54-31-VP64 (red line) and CD54-31Opt-VP64
(orange line) and repressors CD54-31-KRAB (light blue line) and CD54-31Opt-KRAB (dark blue line) were retrovirally
expressed in cells derived from colon cancer (Lim1215, HT29 and SW1222), breast cancer (T47D, SKBR3 and MD-AMB435S), melanoma (C8161), carcinoma (A431) and primary human umbilical vein endothelial cells (HUVEC). Control
cells are labeled as in Figure 3.
repressed with the CD54-31 and 31-Opt KRAB TFZFs,
whereas ITGA6 and GAPDH retained normal levels
of mRNA. Of the remaining CD54-TFZF regulators,
CD54-3 presented the best overall specificity towards
ICAM-1, while CD54-13 and CD54-30 presented
varying degrees of specificity (Figure 6(f)–(h)). Overall, CD54-TFZFs that showed regulation of the ICAM1 promoter in transient assays (namely, CD54-3, -31
and -31Opt) were specific to ICAM-1 and showed no
or little change in expression for the other markers.
Since ICAM-1 transcriptional regulation plays
major roles in biology and cancer development in
endothelial and cancers such as breast, colon and
melanoma, and the major transcription-regulatory
response element is the kB enhancer that overlaps
with the ICAM-1 pro-220 target site, the ability to
effectively interfere with the ICAM-1 expression in
this context is of considerable interest. Therefore,
the activity of CD54-31 and its optimized version
was examined for positive and negative regulation
of ICAM-1 in relevant cell lines (Figure 7 and
Table 2). Both regulators acted equally as repressors with the KRAB domain in A431 carcinoma
and in C8161 melanoma and HUVEC cells, where
ICAM-1 is moderately expressed as determined
by FACS. Levels of mean fluorescence intensity
were reproducibly knocked-down over 90%.
When used as VP64-based transactivators,
CD54-31Opt was twofold more potent in ICAM-1
activation in these cell lines, resulting in 135 and
59 times more ICAM-1 fluorescence in HUVEC
and melanoma, respectively. In breast and colon
cancer, ICAM-1 may function as a suppressor of
tumor progression by promoting necessary interactions with the immune system. ICAM-1 levels
in related cell lines were increased by the TFZFs
several fold in colon (Lim1215 and HT29) and
breast (MD-AMB-435S and T47D) cell lines already
showing significant expression, and were increased
by up to two orders of magnitude in breast
(SKBR3) and colon (SW1222) cell lines, lines that
normally show little ICAM-1 expression by FACS.
Discussion
In the light of the results presented here, we
achieved successful selection of specific regulators
of ICAM-1 from a library of nearly 108 6ZF
transcription factors. First, this strategy overcomes
the problems of chromatin structure and accessibility of the target gene and has a great potential for
discovery of genes and genetic pathways. Second,
considering TFZF biology itself, we found that the
Novel ICAM-1 Zinc Finger Transcription Factors
selected C54-TFZFs up- and down-regulated ICAM1 in a broad range of cells, including established
cancer cell lines, and that TFZFs can also function
in primary cells, which is of particular relevance
for the biological activity of TFZFs in living organisms. Accordingly, TFZFs can work in a general
manner, as seen in the CD54-31 protein, but also
in a more complicated fashion. For instance, by
doing comparative studies of the SID and KRAB
repression domains, we demonstrate for the first
time the generation of cell-type specific TFZFs and
TFZFs specific to a regulatory domain. CD54-31KRAB most consistently provided full repression,
whereas
CD54-30-SID
efficiently
repressed
ICAM-1 in C8161 melanoma cells only, and CD543 paired exclusively with the SID domain in the
three cell lines tested. Modest differences in the
level of expression of the TFZFs seemed not to be
indicative of effectiveness, e.g. the best expressing
CD54-3-KRAB TFZFs in the three cell lines
was inactive, whereas CD54-3-SID which was
expressed at , 50% the level of the KRAB factor
repressed well. This may be due to the fact that
TFZFs are typically expressed in excess compared
to their typical targets which are found in two
copies per nuclei. Consequently, engineered TFZFs
that activate the target endogenous gene, may not
necessarily work as a repressor, i.e. there is more
at work than simple chromatin accessibility, a
requirement that has been heretofore stressed in
the literature as perhaps the only requirement for
regulation. Accordingly, the distribution of regions
accessible to DNase I in the promoter of the
VEGF-A locus appeared to be specific to the cell
type and was essential in the design of TFZFs targeting VEGF-A.9,27 The differences in activity may
also reflect variations in the availability of interacting accessory factors among cell types and their
localization on the chromosomes. Accordingly,
complex regulation mechanisms involving abundant signaling pathways and transcription factors
in ICAM-1 expression have been revealed to be
cell-type specific.28 Thus, the unpredictable combination of parameters including zinc finger, DNAbinding element, regulatory domain and cell type,
show us a mechanism of action of the TFZFs that
is more complex than anticipated. These results better reflect the in vivo situation where chromatin structure is spatio-temporally dynamic and there is a
differential expression of factors of the transcriptional machinery. These results together with our
further study of ICAM-1 promoter regulation
suggest that targeting the proximal promoter of the
target gene allows for general activation and repression, whereas targeting sites further away or through
indirect regulation of intermediate genes could
provide more unique types of regulation.
Two of the CD54-TFZFs, CD54-3 and -31, interacted directly with the transcription of the 1.6 kb
ICAM-1 promoter and showed better gene specificity than the others studied. This correlation
leads us to speculate that CD-54-13 and CD54-30,
which did not interact with the reporter construct
645
and appeared to be less gene-specific, may
bind less specifically and farther away from the
ICAM-1 promoter, or may act by indirect means,
e.g. by regulating other genes involved in ICAM-1
expression. It is conceivable that targeting an
upstream regulator of ICAM-1 may also result in
changes of other genes involved in the same pathway or function, such as the adhesion molecules
tested in the gene specificity experiment. Also,
selected CD54-TFZFs could regulate genes involved
in described ICAM-1 post-transcriptional regulation mechanisms controlling RNA stability, cellsurface localization or proteolytic cleavage from
the cell surface,13 or other protein internalization
and specific proteolysis mechanisms. CD54-13
could be such a candidate since it represses
ICAM-1 cell surface expression consequently
(Figure 2), does not seem to affect ICAM-1 mRNA
and is not the most gene-specific CD54-TFZF
(Figure 6). Blast searches did not reveal near
perfect matches in either 100 kb of the ICAM-1
loci, or in the human genome.
Previous studies have shown that 6ZFPs provide
for higher affinity and specificity compared to
3ZFPs,29 and provided endogenous regulation of
the ErbB2 gene, but not the ErbB3 gene, which
shared 15 out of 18 identical base-pairs in their
respective target sites.7 The selected ZFs bound
their predicted target sequences with better than
10 nM affinity. Quite unexpectedly, we found that
a 6ZF protein like CD54-31 can regulate a target
site through imperfect recognition of 18 bp target
sites if they are bound with sufficient affinity, a
situation that may reflect the mechanism of general
natural transcription factors. However, optimization of the TFZF DNA binding domains using
rational design provided a 20-fold increase in
affinity to its DNA element. This translated into
substantial increases in transient reporter activation and endogenous ICAM-1 regulation. This is
the first direct evidence that affinity modulation of a
TFZF can be used to modulate the extent of endogenous gene regulation once threshold-binding
affinity has been obtained. Further, increased affinity
is accompanied by increased specificity. We can
speculate that, despite the uniqueness factor of the
target site, this may be a mechanism that differentiates general transcription factors from specific ones.
Finally, regarding the ICAM-1 target gene
biology, the target site cd54-31 was identified on a
NF-kB binding element of the ICAM-1 promoter
and was verified by promoter site deletion and
DNase 1 footprinting in vitro and chromatin
immunoprecipitation in vivo. Whereas the nearby
Sp1 site is responsible for constitutive ICAM-1 promoter activity, the overlapping NF-kB enhancer
confers responsiveness to TNF-a, IL-1b and other
signaling molecules.30 Transcription factor complexes of NF-kB, RelA and c-Rel dimers were
shown to bind to this element.13 The CD54 regulators
described here likely impact the effect of natural factors. Our results indicate that the CD54-31 and
CD54-31Opt TFZFs do not compete for the NF-kB
646
enhancer but, when coupled to the N-terminal
KRAB repression domain, rather overcome the
ability of NF-kB, and other transcription factors, to
up-regulate ICAM-1 in response to a variety of
inducers, including IL-1b and TNF-a. On the other
hand, when coupled to the C-terminal VP64 activation domain, they cooperate synergistically with
NF-kB. Note that steric hindrance cannot be the
reason for ICAM-1 repression while the TFZF sits 30
of the NF-kB site in an antiparallel orientation relatively to the DNA strand it binds and the C-terminal
VP64 activation domain does not seem to interfere
with NF-kB dimer binding. Thus, we present evidence that part of the NF-kB enhancer element in
the ICAM-1 promoter can also work as repressor
element given a specific repressor binds to it and
that artificial transcription factors can overcome or
synergistically cooperate with endogenous factors.
Interestingly, 6-ZFP alone (lacking effector domains)
were not able to block either constitutive or induced
ICAM-1 expression, despite their subnanomolar
binding constants. We also provide a study of the
directed gene regulation in a wide panel of cell
types, including primary cells and cancer cells that
are relevant to its function in the development of
tumors. Both CD54-31 and its optimized version,
CD54-31Opt, were able to up-regulate endogenous
levels of ICAM-1 on the cell surface of colon (up to
142-fold in colon SW1222 cells) and breast (up to 59fold in breast SKBR3 cells) cancer cell lines. During
cancer progression, low ICAM-1 correlates with a
reduced disease-free state and poor prognosis in
colorectal carcinoma patients.31 Accordingly,
expression of ICAM-1 in invasive breast cancer
reflects low growth potential and good prognosis.32
Reciprocally, ICAM-1 expression in melanoma cells
and primary vascular endothelial cells could be
completely suppressed and significantly overexpressed (59-fold and 135-fold, respectively). This
is of particular importance, since ICAM-1 is specifically down-regulated in tumor endothelial cells
during tumor angio-genesis33 and de novo expression of ICAM-1 in melanoma correlates with
increased risk of metastasis.34 Therefore, our
library and affinity maturation approach provide
powerful TFZFs to study ICAM-1 regulation and
function by imposed regulation.
The main advantage of our strategy is the large
complexity of available proteins that can be tested
simultaneously, taking into consideration the
chromatin accessibility of the targeted endogenous
gene and the availability of the transcription
machinery in living cells, therefore, overcoming
the need for an initial mapping of chromatin-free
zones in a target gene of interest.9 Alternatively,
our approach could be used to isolate TFZFs associated with specific complex phenotypes beyond
the surface expression phenotype selected here.
A limitation of this approach is the difficulty in
assigning genomic binding sites due to binding of
related DNA sequences; however, this has yet to
pose significant obstacles. It should be possible to
favorably exploit degenerate binding through the
Novel ICAM-1 Zinc Finger Transcription Factors
use of less specific DNA binding domains. This
would allow for more extensive coverage of the genome assuming the corresponding DNA-binding
activity was sufficient to impose transcriptional control. This should be feasible, since natural transcription factors bind families of DNA sequences in
much the same way. Such approaches might further
facilitate the discovery of new genes and pathways
involved in the expression of the target gene or
phenotype of interest. For instance, the CD54-13 and
CD-30 TFZF selected in this study may lead to the discovery of indirect target sites that regulate new genes
or genetic pathways involved in the expression of
ICAM-1 at the transcriptional and post-transcriptional levels. TFZFs can also be seen as “compact”
cDNA libraries when they are expressed with activation domains or their RNAi-like complement
when expressed with repression domains. Thus, a
wide variety of applications can be envisioned for
the transcription factor libraries described here.
Materials and Methods
Human cell lines
A431 epidermoid carcinoma cells and breast cancer
cell lines T47D, SKBR3 and MDA-MB-435s were
obtained from the American Type Culture Collection
(Manassas, VA). Colon cancer cell lines Lim1215,
SW1222, HT29 were obtained from the cell bank of the
Ludwig Institute for Cancer Research (New York).
Kaposi’s sarcoma cell line SLK was provided by R.
Pasqualini (University of Texas M.D. Anderson Cancer
Center, Houston) with permission from S. LevintonKriss (Tel-Aviv). Melanoma C8161 cells were a generous
gift from R. A. Reisfeld (The Scripps Research Institute,
La Jolla). Human umbilical vein endothelial cells
(HUVEC)
were
purchased
from
BioWhittaker
(Walkersville, MD) and maintained as indicated by the
manufacturer.
Immortalized
primary
mammary
epithelial cell line hTERT-HME1 was purchased from
CLONTECH (Palo Alto, CA) and hTERT-fibroblasts
(J.W. Shay, The University of Texas Southwestern
Medical Center, Dallas, TX) were a kind gift of Steven
I. Reed (The Scripps Research Institute, La Jolla). The
293-GagPol packaging cell line was obtained from I.
Verma (Salk Institute, San Diego). Cells were cultured in
RPMI 1640 medium (SLK, SKBR3, Lim1215, SW1222,
HT29) or Dulbecco’s modified Eagle’s medium (DMEM)
(A431, T47D, MDA-MB-435s, C8161, hTERT fibroblast
and 293-gagpol) supplemented with 10% (w/v) FCS
and antibiotics.
Construction of retroviral and transient expression
TFZFs plasmids
The pMX-6ZF-VP64 library of 8.4 £ 107 members was
constructed as described.12 Retroviral and transient
expression of TFZFs, constructs were prepared by subcloning 6ZF-coding DNA into pcDNA- and pMX-effector vectors as described.6,7 For the construction of the
CD54-31 and -31Opt proteins lacking an effector domain
fusion, the C-terminal AscI-PacI VP64 domain of the
pMX-CD54-31 and -31Opt VP64 vectors was removed
by replacing the AscI site for a second PacI site using
647
Novel ICAM-1 Zinc Finger Transcription Factors
site-directed mutagenesis, followed by PacI digestion
and self-religation. The resulting construct conserves the
C-terminal hemagglutinin peptide tag in frame with
the 6-ZFP for detection of the full-length protein by
Western blot (data not shown).
Retroviral gene delivery and FACS
Screening of the pMX-6ZF-VP64 library for ICAM-1
regulators was performed as described for the isolation
of TFZFs specific to the VE-cadherin gene12 with modifications. For optimal infection efficiency, we used a
293-GagPol packaging cell line lacking envelope coding
retroviral genes and VSV envelope pseudo-typing by
co-transfection of the plasmid pMD.G (obtained from
I. Verma, Salk Institute, San Diego). Retroviral infections
were set up as following: 3.5 £ 106 293-GagPol cells
were plated on a 10 cm polylysine-coated dish and cotransfected with 1.25 mg of pMD.G plasmid encoding
the vesicular stomatitis virus-G envelope protein,35 thus
conferring a broader host-cell range, a higher viral
stability and high titers (.106 cfu/ml),36 and 3.75 mg of
ZF-effector pMX retroviral vector using Lipofectamin
PLUS transfection reagents as recommended by the
manufacturer (Invitrogen). As mock infection control,
the same infection conditions were used with the
pcDNA3.1 vector (Invitrogen). Two days later, the viral
supernatant was cleared by centrifugation and applied
for eight hours with 8 mg/ml polybrene to 1 – 5 £ 105
host cells per 10 cm plate and infection was repeated
overnight. The complexity of the 6ZF library (8.4 £ 107)
necessitates a large number of infected cells, which was
achieved by up-scaling the infection protocol. Infected
cells were harvested for analysis 72 hours after infection.
FACS analysis and cell sorting were done using Becton
Dickinson flow cytometers. The anti-ICAM-1 primary
antibody was purchased from BD PharMingen (clone
HA58). Antibodies used for staining epidermal growth
factor (EGF), 3-FAL selectin ligand (CD15, FUT4), integrin-a6 (CD49f, ITGA6), leukocyte function-associated
antigen (CD58, LFA-3), Apo1-FAS antigen (CD95,
TNFRSF6), integrin-b4 (CD104, ITGB4) and vascular
endothelial VE-cadherin (CD144, CDH5) were used as
reported.12 Comparative analysis of ICAM-1 flow cytometric data, mean fluorescence intensities (MFI), was
determined using CELLQuest software (Becton
Dickinson) as described.33 For each sample analyzed,
the MFI was corrected for background by subtraction of
the autofluorescence of cells incubated with the secondary antibody only and normalized to the value of cells
infected with the unselected 6ZFlibrary and presented
as fold activation or percentage remaining ICAM-1. For
ICAM-1 induction studies, 20 ng/ml IL-1b (Sigma) and
10 ng/ml TNF-a (Sigma) were added to the cells 12
hours before the FACS analysis.
CCATCTACAGCTT-30 ) and ICAMRT-1265-r reverse
primers (50 -CAATCCCTCTCGTCCAGTCG-30 ); integrin
alpha 6 (ITGA6) at 54 8C for 25 cycles using ITGA6-f1
(50 -AACTTGGACACTCGGGAGGACAAC-30 )
and
ITGA6-b1 (50 -GGGGTCAGCATCGTTATCAAACTC-30 )
primers; glyceraldehyde-3-phosphate dehydrogenase
(GAPDH) at 52 8C for 20 cycles with GAPDH forward
and reverse primers,9 and TFZF-KRAB at 52 8C for 30
cycles with pMX-SKD-f (50 -ATCCGCCACCATGGATG30 ) and NLS-seq-b (50 -CTGGGCCACTTTGCGTTTC-30 )
primers, which amplify the 327 bp KRAB domain of the
TFZF. The levels of mRNA expression were determined
by quantifying the amount of PCR product on 1.5%
agarose gel electrophoresis with ImageQuant software
(Molecular Dynamics), background was subtracted and
the signal normalized to GAPDH. The 303 bp ICAM-1
and 243 bp ITGA6 PCR products were confirmed by
sequencing.
Generation of CD54-31Opt protein
The six-finger CD54-31Opt was assembled from two
three-finger proteins constructed by PCR grafting of the
appropriate DNA recognition helices into the framework
of the three-finger protein Sp1C as described.6 The corresponding ZF domains and DNA subsites are shown in
Table 1.
Characterization of 6ZF proteins in vitro
The selected and designed 6ZF protein coding regions
were subcloned into Escherichia coli expression vector
pMal-C2 (NEB) by SfII digestion. The purification of the
resulting MBP fusion proteins and their DNA-binding
properties were determined by multitarget specificity
ELISA and electrophoretic mobility-shift assays (EMSA)
and were carried out in duplicates essentially as
described.3 For this purpose, biotinylated dsDNA oligonucleotides containing the 18 bp target sequences were
synthesized (MWG-Biotech) and are presented in Table
1. DNA binding for ELISA is determined from serial
dilutions in duplicate experiments, from which dilution
series of 60 nM protein were used. The ICAM-1 promoter target oligonucleotides used by ELISA were named
after their position relative to the start of translation.
The DNA strand and the number of matching residues
relative to the predicted 18 bp cd54-31 sequence are presented in parentheses. ICAM-1 pro-1511 (þ, 11) 50 -AGC
TCAGTGGAACCCGCC-30 pro-940 (2 , 10) 50 -TGAGGA
GTTCTGAATTCC-30 pro-1000 (þ, 11) 50 -GCAGGGGTT
CGAGCGCCC-30 pro-793; (þ , 10) 50 -GGGAGCTGTAAA
GACGCC-30 pro-757 (þ , 9) 50 -AGGGGAAGCGAGGA
GGCC-30 pro-501 (þ, 11) 50 -CACTCGATTAAAGGGCC30 pro-220 (þ , 13) 50 -TCCGGAGCTGAAGCGGCC-30
pro-59 (- 2 , 11) 50 -TGATCCTTTATAGCGCTA-30 .
Semi-quantitative RT-PCR
DNase I footprinting
Total RNAs were isolated from A431 cells using TRI
reagent (Molecular Research Center), 1st strand cDNA
was made using the Superscript reverse transcriptase
(Invitrogen) as recommended by the manufacturer.
Semi-quantitative RT-PCR was performed as described12
with the following modifications. In order to compare
PCR products in the linear range of the PCR reaction,
2 ml of cDNA was used as template and ICAM-1 was
amplified at a 54 8C annealing temperature for 25 cycles
using the ICAMRT-963-f forward (50 -GCAGACAGTGA
Binding of ZFPs to ICAM-1 DNA sequences was
examined by DNase I footprinting by a following the
method of Trauger & Dervan.37 The 200 bp DNA probe
containing the ICAM-1 promoter site pro-220 was generated by PCR from the ICAM-1 promoter reporter construct as following. A forward DNA oligonucleotide
primer 50 -GTCATCGCCCTGCCACCG-30 was labeled
with polynucleotide kinase (Roche) and [g-32P]ATP
(8000 Ci/mmol) (Perkin Elmer Life Sciences) as per
manufacturer’s instruction. The radioactively labeled
648
primer was purified and used together with a reverse
primer 50 -TTTATAGCGCTAGCCACCTGGGG-30 for
PCR amplification with High Fidelity PCR Master mix
(Roche) and 100 ng pGL3-ICAM-1 template plasmid for
35 amplification cycles (30 seconds at 94 8C, 30 seconds,
at 55 8C, 30 seconds at 72 8C). The resulting DNA probe
was recovered from a non-denaturing polyacrylamide
gel and 17 – 24 ng was used in each DNase I protection
assay. Binding equilibrations of ZFPs at 1 mM, 100 nM,
10 nM and 1 nM concentrations with DNA was done in
400 ml binding buffer (10 mM Tris – HCl (pH 7.0), 10 mM
KCl, 10 mM MgCl2, 5 mM CaCl2, 10 mM ZnCl2 and
5 mM DTT) overnight at 4 8C. DNase I (RNase-free,
Roche) was added to 3 mU/ml and digestion was
stopped after seven minutes at room temperature. DNA
was ethanol precipitated and electrophoretically separated on a 6% (w/v) polyacrylamide/8 M urea denaturing gel along with guanine-specific chemical cleavage
products of the same DNA probe as marker.38
Transient reporter assay, ICAM-1 reporter plasmid
and deletion mutant thereof
A 1.3 kb upstream promoter fragment of the ICAM-1
gene was shown to drive luciferase expression and mediates responsiveness to ICAM-1 inducers.23 To generate
the pGL3-ICAM-1 reporter construct, a 1.6 kb PCR fragment of the ICAM-1 promoter (nucleotide positions
2 1592 to 2 15 from the start of translation) was PCR
amplified from HUVEC genomic DNA using the forward 50 -GAGGAGGAGGAGGAGGGTACCTGAGAAA
AGAACGGCACCATTG-30 and reverse 50 -GAGGAGGA
GGAGGAGACGCGTTGCAACTCTGAGTAGCAGAGG
AGC-30 primers and was cloned into the pGL3-basic
luciferase vector (Promega). The deletion mutant pGL3E2C/ICAM-1 was created by PCR mutagenesis. Transient reporter luciferase assays in A431 cells were done
as reported.6 Luciferase activity was normalized to
b-galactosidase activity using the Galacto-light Plus
assay (Tropix). Data represent the average of three
experiments.
ZF antibody generation and ChIP assay
Since at most seven out of 28 amino acid residues
differ from one domain to another, a minimum of 75%
conserved residues is provided overall between any
designer zinc finger domain. Therefore, the polyclonal
serum that was generated by rabbit immunization with
a mixture of three purified 3ZF proteins encoding the
consensus framework of Sp1C, was expected to recognize all Sp1C variants. ChIP assays were performed as
described with the following modifications.39 Nuclei
from a minimum of 107 cells per immunoprecipitation
were cross-linked, isolated and sonicated to produce
500 bp chromatin fragments. TFZF : chromatin complexes
were captured by centrifugation using 10 ml of ZF antibody or no antibody and lyophilized StaphA cells (Calbiochem), washed extensively and analyzed by PCR at a
60 8C annealing temperature for 40 cycles using the
ICAMpro-1381-f1 GTACTTAATAAACCGATTAAGCG
forward and ICAMpro-b5 TGCAACTCTGAGTAGCAG
AGGAGC reverse primers producing a 343 bp PCR fragment containing the ICAM-1 pro-220/NF-kB target site.
PCR reactions were performed with High fidelity Taq
DNA polymerase (Roche) and 2 ml of IP sample or 0.2%
of the total input chromatin that was not immunoprecipitated from the “no antibody” sample.
Novel ICAM-1 Zinc Finger Transcription Factors
Acknowledgements
We thank Dave Valente for technical support.
This study was supported in part by the National
Institutes of Health grant CA086258 (to C.F.B.).
L.M was the recipient of postdoctoral fellowships
from the Swiss National Science Foundation.
References
1. Greisman, H. A. & Pabo, C. O. (1997). A general
strategy for selecting high-affinity zinc finger proteins for diverse DNA target sites. Science, 275,
657– 661.
2. Liu, Q., Segal, D. J., Ghiara, J. B. & Barbas, C. F., III
(1997). Design of polydactyl zinc-finger proteins for
unique addressing within complex genomes. Proc.
Natl Acad. Sci. USA, 94, 5525– 5530.
3. Segal, D. J., Dreier, B., Beerli, R. R. & Barbas, C. F., III
(1999). Toward controlling gene expression at will:
selection and design of zinc finger domains recognizing each of the 50 -GNN-30 DNA target sequences.
Proc. Natl Acad. Sci. USA, 96, 2758– 2763.
4. Dreier, B., Beerli, R. R., Segal, D. J., Flippin, J. D. &
Barbas, C. F., III (2001). Development of zinc finger
domains for recognition of the 50 -ANN-30 family of
DNA sequences and their use in the construction
of artificial transcription factors. J. Biol. Chem. 276,
29466– 29478.
5. Beerli, R. R. & Barbas, C. F., III (2002). Engineering
polydactyl zinc-finger transcription factors. Nature
Biotechnol. 20, 135– 141.
6. Beerli, R. R., Segal, D. J., Dreier, B. & Barbas, C. F., III
(1998). Toward controlling gene expression at will:
specific regulation of the erbB-2/HER-2 promoter
by using polydactyl zinc finger proteins constructed
from modular building blocks. Proc. Natl Acad. Sci.
USA, 95, 14628– 14633.
7. Beerli, R. R., Dreier, B. & Barbas, C. F., III (2000). Positive and negative regulation of endogenous genes
by designed transcription factors. Proc. Natl Acad.
Sci. USA, 97, 1495– 1500.
8. Zhang, L., Spratt, S. K., Liu, Q., Johnstone, B., Qi, H.,
Raschke, E. E. et al. (2000). Synthetic zinc finger
transcription factor action at an endogenous chromosomal site. Activation of the human erythropoietin
gene. J. Biol. Chem. 275, 33850– 33860.
9. Liu, P. Q., Rebar, E. J., Zhang, L., Liu, Q., Jamieson,
A. C., Liang, Y. et al. (2001). Regulation of an
endogenous locus using a panel of designed zinc
finger proteins targeted to accessible chromatin
regions. Activation of vascular endothelial growth
factor A. J. Biol. Chem. 276, 11323– 11334.
10. Rebar, E. J., Huang, Y., Hickey, R., Nath, A. K., Meoli,
D., Nath, S. et al. (2002). Induction of angiogenesis
in a mouse model using engineered transcription
factors. Nature Med. 8, 1427– 1432.
11. Guan, X., Stege, J., Kim, M., Dahmani, Z., Fan, N.,
Heifetz, P. et al. (2002). Heritable endogenous gene
regulation in plants with designed polydactyl zinc
finger transcription factors. Proc. Natl Acad. Sci.
USA, 99, 13296– 13301.
12. Blancafort, P., Magnenat, L. & Barbas, C. F. (2003).
Scanning the human genome with combinatorial
transcription factor libraries. Nature Biotechnol. 21,
269– 274.
649
Novel ICAM-1 Zinc Finger Transcription Factors
13. van de Stolpe, A. & van der Saag, P. T. (1996). Intercellular adhesion molecule-1. J. Mol. Med. 74, 13 – 33.
14. Sligh, J. E., Jr, Ballantyne, C. M., Rich, S. S., Hawkins,
H. K., Smith, C. W., Bradley, A. & Beaudet, A. L.
(1993). Inflammatory and immune responses are
impaired in mice deficient in intercellular adhesion
molecule 1. Proc. Natl Acad. Sci. USA, 90, 8529– 8533.
15. Kim, J. S. & Pabo, C. O. (1998). Getting a handhold
on DNA: design of poly-zinc finger proteins with
femtomolar dissociation constants. Proc. Natl Acad.
Sci. USA, 95, 2812– 2817.
16. Dreier, B., Segal, D. J. & Barbas, C. F., III (2000).
Insights into the molecular recognition of the
50 -GNN-30 family of DNA sequences by zinc finger
domains. J. Mol. Biol. 303, 489– 502.
17. Segal, D. J. (2002). The use of zinc finger peptides to
study the role of specific factor binding sites in the
chromatin environment. Methods, 26, 76 –83.
18. Liu, X., Constantinescu, S. N., Sun, Y., Bogan, J. S.,
Hirsch, D., Weinberg, R. A. & Lodish, H. F. (2000).
Generation of mammalian cells stably expressing
multiple genes at predetermined levels. Anal. Biochem. 280, 20 – 28.
19. Margolin, J. F., Friedman, J. R., Meyer, W. K., Vissing,
H., Thiesen, H. J. & Rauscher, F. J., III (1994).
Kruppel-associated boxes are potent transcriptional
repression domains. Proc. Natl Acad. Sci. USA, 91,
4509–4513.
20. Ayer, D. E., Laherty, C. D., Lawrence, Q. A.,
Armstrong, A. P. & Eisenman, R. N. (1996). Mad
proteins contain a dominant transcription repression
domain. Mol. Cell. Biol. 16, 5772– 5781.
21. Jiang, X. R., Jimenez, G., Chang, E., Frolkis, M.,
Kusler, B., Sage, M. et al. (1999). Telomerase
expression in human somatic cells does not induce
changes associated with a transformed phenotype.
Nature Genet. 21, 111 – 114.
22. Morales, C. P., Holt, S. E., Ouellette, M., Kaur, K. J.,
Yan, Y., Wilson, K. S. et al. (1999). Absence of cancerassociated changes in human fibroblasts immortalized with telomerase. Nature Genet. 21, 115 – 118.
23. Voraberger, G., Schafer, R. & Stratowa, C. (1991).
Cloning of the human gene for intercellular adhesion
molecule 1 and analysis of its 50 -regulatory region.
Induction by cytokines and phorbol ester. J. Immunol.
147, 2777– 2786.
24. Krizek, B. A., Amann, B. T., Kilfoil, V. J., Merkle, D. L.
& Berg, J. M. (1991). A consensus zinc finger
peptide—design, high-affinity metal-binding, a
Ph-dependent structure, and a His to Cys sequence
variant. J. Am. Chem. Soc. 113, 4518– 4523.
25. Shi, Y. & Berg, J. M. (1995). A direct comparison of
the properties of natural and designed zinc-finger
proteins. Chem. Biol. 2, 83 – 89.
26. Ledebur, H. C. & Parks, T. P. (1995). Transcriptional
regulation of the intercellular adhesion molecule-1
gene by inflammatory cytokines in human endothelial cells. Essential roles of a variant NF-kappa B
site and p65 homodimers. J. Biol. Chem. 270, 933– 943.
27. Liang, Y., Li, X. Y., Rebar, E. J., Li, P., Zhou, Y., Chen,
B. et al. (2002). Activation of vascular endothelial
growth factor A transcription in tumorigenic
28.
29.
30.
31.
32.
33.
34.
35.
36.
37.
38.
39.
glioblastoma cell lines by an enhancer with cell
type-specific DNase I accessibility. J. Biol. Chem. 277,
20087 – 20094.
Roebuck, K. A. & Finnegan, A. (1999). Regulation
of intercellular adhesion molecule-1 (CD54) gene
expression. J. Leukoc. Biol. 66, 876– 888.
Segal, D. J., Beerli, R. R., Blancafort, P., Dreier, B.,
Effertz, K., Huber, A. et al. (2003). Evaluation of a
modular strategy for the construction of novel
polydactyl zinc finger DNA-binding proteins. Biochemistry, 42, 2137– 2148.
van de Stolpe, A., Caldenhoven, E., Stade, B. G.,
Koenderman, L., Raaijmakers, J. A., Johnson, J. P. &
van der Saag, P. T. (1994). 12-O-tetradecanoylphorbol-13-acetate- and tumor necrosis factor alphamediated induction of intercellular adhesion molecule-1 is inhibited by dexamethasone. Functional
analysis of the human intercellular adhesion molecular-1 promoter. J. Biol. Chem. 269, 6185– 6192.
Mulder, W. M., Stern, P. L., Stukart, M. J., de Windt,
E., Butzelaar, R. M., Meijer, S. et al. (1997). Low
intercellular adhesion molecule 1 and high 5T4
expression on tumor cells correlate with reduced disease-free survival in colorectal carcinoma patients.
Clin. Cancer Res. 3, 1923– 1930.
Ogawa, Y., Hirakawa, K., Nakata, B., Fujihara, T.,
Sawada, T., Kato, Y. et al. (1998). Expression of intercellular adhesion molecule-1 in invasive breast
cancer reflects low growth potential, negative lymph
node involvement, and good prognosis. Clin. Cancer
Res. 4, 31 – 36.
Griffioen, A. W., Damen, C. A., Mayo, K. H.,
Barendsz-Janson, A. F., Martinotti, S., Blijham, G. H.
& Groenewegen, G. (1999). Angiogenesis inhibitors
overcome tumor induced endothelial cell anergy.
Int. J. Cancer, 80, 315– 319.
Johnson, J. P., Stade, B. G., Holzmann, B., Schwable,
W. & Riethmuller, G. (1989). De novo expression of
intercellular-adhesion molecule 1 in melanoma correlates with increased risk of metastasis. Proc. Natl
Acad. Sci. USA, 86, 641– 644.
Naldini, L., Blomer, U., Gallay, P., Ory, D., Mulligan,
R., Gage, F. H. et al. (1996). In vivo gene delivery and
stable transduction of nondividing cells by a lentiviral vector. Science, 272, 263– 267.
Burns, J. C., Friedmann, T., Driever, W., Burrascano,
M. & Yee, J. K. (1993). Vesicular stomatitis virus G
glycoprotein pseudotyped retroviral vectors: concentration to very high titer and efficient gene transfer
into mammalian and nonmammalian cells. Proc.
Natl Acad. Sci. USA, 90, 8033– 8037.
Trauger, J. W. & Dervan, P. B. (2001). Footprinting
methods for analysis of pyrroleimidazole polyamide/DNA complexes. Methods Enzymol. 340,
450 –466.
Maxam, A. M. & Gilbert, W. (1980). Sequencing endlabeled DNA with base-specific chemical cleavages.
Methods Enzymol. 65, 499– 560.
Boyd, K. E., Wells, J., Gutman, J., Bartley, S. M. &
Farnham, P. J. (1998). c-Myc target gene specificity is
determined by a post-DNA binding mechanism.
Proc. Natl Acad. Sci. USA, 95, 13887– 13892.
Edited by M. Yaniv
(Received 19 January 2004; received in revised form 11 June 2004; accepted 14 June 2004)