Zhuet al.,Science 375 , eabg9765 (2022) 21 January 2022 3 of 11
mCitrine- mCitrine+
WT R39A R2AR39AR2AR39AR67A
Mutations
1
10
20
30
40
Fold Activation
ErbB2ZF
ZF-AD
ZF-GCN4-AD
C TMP and AP1903 modulate autoregulation
0.0
0.2
0.4
0.6
0.8
1.0
mCit+ fraction, norm. to Dox group
FKBP-BCRZFR39A-VP48-DHFR self-activation
AP1903
TMP (nM)
0
10000 0
+
1
+
10
+
100
+
10000
+
1000
+
A Engineered TFs exhibit dimer-dependent activity
D Competitive dimerization allows inhibition of positive autoregulation
ZF-AD
AD
Citrine
ZFbs-ZFbs
AD AD
X
TMP
Dox
mCitrine-PEST
ZFbs-ZFbs
FKBP-ZF-VP48-DHFRIRES
AP1903
AD AD AD
AD AD
B FKBP12F36V allows inducible dimerization
ZF-GCN4-AD
Citrine
ZFbs-ZFbs
AD AD
AP1903
FKBP-ZF-AD
Citrine
ZFbs-ZFbs
AD
AD AD
6x
2x
2x
2x
AD
Dox+ Dox+ (for norm.)
Dox+ AP1903+, TMP 10 μM
Dox+ AP1903+, TMP 1 nM
Dox+ no AP1903, TMP 10 μM
0 0.1 1 10 100
AP1903 (nM)
1
5
10
15
Fold Activation
DHFR degron
Zinc finger DNA-binding domains FKBP12F36V homodimerization domain
ADTranscriptional activation domain
Monomeric or dimeric GCN4 domain
AP1903 dimerizer TMP stabilizing drug
AD AD
Monoclonal
stable line 1
AD
inactive
complex
active
complex
AD AD
perturbation
ZFbs-ZFbs
IRESmCitrine-PEST
6x
ZF-GCN4-VP48-DHFR
AD
AD AD
Monoclonal
stable line 2
inactive
complex
active
complex
ZFbs-ZFbs
IRESmCitrine-PEST
6x
FKBP-ZF-VP48-DHFR
AD AD
Perturbation
None AD AD
AD
0.0
0.2
0.4
0.6
0.8
1.0
mCitrine+ fraction
Monoclonal stable line 1
None AD
AD
Perturbation
AD
0.0
0.2
0.4
0.6
0.8
1.0
mCitrine+ fraction
Monoclonal stable line 2
perturbation
AD AD
AD
or or
AD AD
AD
or or
Fig. 2. Engineered transcription factors enable homodimer-dependent auto-
regulation and heterodimerization-based inhibition.(A) ZF transcription factors
enable homodimer-dependent activation. Left: Design of test constructs in which
ErbB2ZF ( 31 ) (red circle), fused to VP48 (AD) and in some cases GCN4 (blue squiggle)
domains, binds to target sites (red pads) to activate Citrine expression. Activators were
expressed from a constitutive CAG promoter ( 52 ). Right: R-to-A mutations in ZF-
modulated reporter activation by ZF-GCN4-AD and ZF-AD. The R2AR39A variant was
selected because of high ZF-GCN4-AD activation and minimal ZF-AD activation. Fold
activation is defined in fig. S4A. WT, wild-type variant. (B) Left: Design of test constructs in
which FKBP12F36V (FKBP, light cyan partial box), fused to BCRZFR39A (red circle) and
VP48 (AD), binds to target sites (red pads) to activate Citrine expression in the presence of
AP1903 (cyan circle). Right: This design allows dose-dependent control of activation by
AP1903. (C) Transcription factor self-activation can be controlled by TMP and AP1903.
Left: Design of the controllable self-activation circuit. IRES, internal ribosome entry site;
PEST, constitutive degradation tag ( 53 ). Center: Stable polyclonal cells showed bimodal
mCitrine distribution upon circuit activation. An empirical threshold at mCitrine = 10^4
separates the distribution into two fractions, and the normalized mCitrine+fraction was
used to quantify the self-activation strength ( 25 ). Right: Colored arrows indicate data
from the center panel. AP1903+samples had 100 nM AP1903. (D) Self-activation was
inhibited by proteins with a different ZF and matching dimerization domains. Two
monoclonal stable lines could spontaneously self-activate in media containing AP1903
and TMP (fig. S5B). Each perturbation construct is introduced by stable integration ( 25 ).
The integrated construct in the“None”group did not express any perturbation protein.
Red circle denotes 42ZFR2AR39AR67A; green circle denotes BCRZFR39A. In all
panels, each dot represents one biological replicate, and each red line or bar indicates the
mean of replicates. Lists of constructs and cell lines are in tables S2 and S3.
RESEARCH | RESEARCH ARTICLE