TRANSDIFFERENCIATION PDF

Nom original: TRANSDIFFERENCIATION.pdf

Ce fichier au format PDF 1.4 a été envoyé sur Fichier-PDF le 25/08/2014 à 16:03, depuis l'adresse IP 212.195.x.x. La présente page de téléchargement du fichier a été vue 856 fois.
Taille du document: 1.2 Mo (5 pages).
Confidentialité: fichier public

Aperçu du document

Sequential histone-modifying activities determine the robustness of
transdifferentiation
Steven Zuryn et al.
Science 345, 826 (2014);
DOI: 10.1126/science.1255885

This copy is for your personal, non-commercial use only.

If you wish to distribute this article to others, you can order high-quality copies for your
colleagues, clients, or customers by clicking here.

The following resources related to this article are available online at
www.sciencemag.org (this information is current as of August 14, 2014 ):
Updated information and services, including high-resolution figures, can be found in the online
version of this article at:
http://www.sciencemag.org/content/345/6198/826.full.html
Supporting Online Material can be found at:
http://www.sciencemag.org/content/suppl/2014/08/13/345.6198.826.DC1.html
This article cites 42 articles, 15 of which can be accessed free:
http://www.sciencemag.org/content/345/6198/826.full.html#ref-list-1
This article appears in the following subject collections:
Development
http://www.sciencemag.org/cgi/collection/development

Science (print ISSN 0036-8075; online ISSN 1095-9203) is published weekly, except the last week in December, by the
American Association for the Advancement of Science, 1200 New York Avenue NW, Washington, DC 20005. Copyright
2014 by the American Association for the Advancement of Science; all rights reserved. The title Science is a
registered trademark of AAAS.

Downloaded from www.sciencemag.org on August 15, 2014

Permission to republish or repurpose articles or portions of articles can be obtained by
following the guidelines here.

R ES E A RC H | R E PO R TS

TRANSDIFFERENTIATION

Sequential histone-modifying
activities determine the robustness
of transdifferentiation
Steven Zuryn,1 Arnaud Ahier,1 Manuela Portoso,2 Esther Redhouse White,1*
Marie-Charlotte Morin,1 Raphaël Margueron,2 Sophie Jarriault1†
Natural interconversions between distinct somatic cell types have been reported in
species as diverse as jellyfish and mice. The efficiency and reproducibility of some
reprogramming events represent unexploited avenues in which to probe mechanisms that
ensure robust cell conversion. We report that a conserved H3K27me3/me2 demethylase,
JMJD-3.1, and the H3K4 methyltransferase Set1 complex cooperate to ensure invariant
transdifferentiation (Td) of postmitotic Caenorhabditis elegans hindgut cells into motor
neurons. At single-cell resolution, robust conversion requires stepwise histone-modifying
activities, functionally partitioned into discrete phases of Td through nuclear degradation of
JMJD-3.1 and phase-specific interactions with transcription factors that have conserved roles
in cell plasticity and terminal fate selection. Our results draw parallels between epigenetic
mechanisms underlying robust Td in nature and efficient cell reprogramming in vitro.

T

issue and organ regeneration occurs in
multiple animal species and can originate
from transdifferentiation (Td) of surrounding cell populations (1, 2). Understanding
how individual cells interconvert efficiently
and precisely to contribute to or result in the
robust regeneration of whole tissues is a major
goal of regenerative medicine. To characterize
these mechanisms in a physiological setting, we
investigated a traceable and predictable Td event
in Caenorhabditis elegans. The Y cell is a postmitotic hindgut cell that constitutes the half section of the third ring of the rectal tube. This cell
exhibits a remarkable behavior by which it disengages from the tube and changes into a motor
neuron called PDA during normal development
(Fig. 1A) (3). We found that conversion between
the two specialized cell types occurred with virtually invariant precision in each animal (100%,
n = 2209 animals). To identify those factors that
endowed such precision, we chemically mutagenized worms, screened for low-penetrance
mutants in which Td was no longer invariant
(Td–), and isolated several mutant alleles (fp11,
fp13, fp15, and fp25) (Fig. 1B and fig. S1, A and
B). Using deep-sequencing–based mapping (4),
we cloned and confirmed that the causal genetic lesions of all mutants affected jmjd-3.1
(fig. S1, C and D).
jmjd-3.1 encodes an ortholog of human Jmjd3,
which displays specificity for demethylating triand dimethylated histone H3 lysine 27 (H3K27)
via a highly conserved Jumonji C terminus (JmjC)

domain (5, 6). In each jmjd-3.1 mutant, the JmjC
domain is either mutated, truncated, or absent,
suggesting that H3K27me3/me2 demethylase activity mediates its Td role (Fig. 1C). Generating
jmjd-3.1(fp15) strains harboring Cherry-tagged
JMJD-3.1 forms, we found that nuclear JMJD-3.1
acted cell-autonomously [as well as nonredun-

dantly (fig. S2)] and that disrupting its ability to
demethylate lysine residues or recognize histone
H3 tails abolished its activity during Td (Fig. 1C
and figs. S3 and S4). Moreover, expressed and
purified JMJD-3.1 mutant proteins mimicking
fp11, fp15, and fp25 had completely lost their
capacity to demethylate H3K27me3 (fig. S5).
This correlated with the Td– phenotypes and suggested that each allele encoded enzymatically
null products. Last, deficiencies in the worm Polycomb Repressive Complex 1 component SPAT-3
(Ring1B) (7), which acts to fortify gene repression
upon recruitment to trimethylated H3K27 residues
(8), partially restored Td in jmjd-3.1(fp15) mutants (fig. S6). These data indicated that jmjd-3.1
acts antagonistically to H3K27me3-mediated transcriptional repression during Td by demethylating
H3K27me3.
Alongside H3K27, other histone tail residues
can exist in several states of methylation, each
of which can be modulated by specific enzymes.
We performed RNA interference (RNAi) screens
to determine whether other histone methyltransferases or demethylases, and hence other specific histone modifications, performed a role
similar to jmjd-3.1 in ensuring precise Td (fig. S7
and table S1). Only RNAi of the mammalian
SET1A/SET1B ortholog, set-2, disrupted invariant
Td. SET-2 is the catalytic subunit of the Set1 complex, a highly conserved multi-subunit ensemble that methylates H3K4 (9). Null deficiencies
in set-2, as well as other Set1 complex subunits
(wdr-5.1/WDR5, ash-2/ASH2L, rbbp-5/RbBP5,

1

Department of Development and Stem Cells, Institut de
Génétique et de Biologie Moléculaire et Cellulaire, CNRS
UMR 7104/INSERM U964, Université de Strasbourg, 67404
Illkirch CU Strasbourg, France. 2Institut Curie, INSERM U934,
CNRS UMR3215, 26, Rue d’Ulm, 75005 Paris, France.
*Present address: Department of Cell and Developmental Biology,
University College London, Gower Street, London WC1E 6BT, UK.
†Corresponding author. E-mail: sophie@igbmc.fr

826

15 AUGUST 2014 • VOL 345 ISSUE 6198

Fig. 1. jmjd-3.1 and the Set1 complex determine invariant Td. (A) Schematic of Y-to-PDA Td. (B) jmjd-3.1
mutations cause defects in Td. (C) Schematic of the JMJD-3.1 protein. NLS, nuclear localization signal;
JmjC, Jumonji catalytic domain; ZnB, zinc-binding, H3 tail recognition domain. Amino acids highlighted
in gray boxes were mutated in order to disrupt either lysine demethylation (blue) or H3 tail recognition
(red). (D) Effect of deficiencies in Set1 complex subunits on Td.

sciencemag.org SCIENCE

RE S EAR CH | R E P O R T S

dpy-30/DPY30, and cfp-1/CFP1), resulted in a
low-penetrance Td– phenotype, which again is
not attributable to hypomorphism or redundancy (Fig. 1D and fig. S8). Set1 deficiencies could
be phenocopied by overexpressing several functionally antagonistic enzymes (H3K4me3/me2
demethylases) in the Y cell in a demethylaseactivity–dependent manner (fig. S9). The same
effect was observed by overexpressing a K4 unmethylable replication-independent H3.3K4A
in Y (fig. S9). Further implicating Set1 H3K4 methyltransferase activity in Td, only a catalytically
active set-2 transgene expressed in Y could rescue the defect in set-2 mutants (discussed below).
Thus, multiple specific histone-modifying activities
acting on distinct lysines of histone H3 mediate
invariant cell conversion in a natural setting.
How do specific histone methylation states
elicit these effects? To answer this question, we
investigated the progression of Td at the singlecell level in live mutant animals. Td– cells in Set1
mutants, and in transgenic animals overexpressing Set1 antagonists or H3K4A in Y, could be
broken into two distinct classes: Approximately
half displayed a characteristic, hindgut fried-egg–
shaped nuclear morphology, whereas the rest
displayed speckled nuclear morphology (Fig. 2A
and fig. S10A). In contrast, Td– cells in jmjd-3.1
mutants had exclusively speckled nuclei. These
initial observations hinted that specific K4 and

K27 methylation states mediated different transitions through successive stages of cell conversion, which became blocked in the absence of
their respective modifiers. We have previously
shown that Y-to-PDA proceeds through discrete
phases: dedifferentiation and then redifferentiation into the new cell type (10), as observed in
vertebrate examples of Td, such as newt lens regeneration (1). We thus tested whether jmjd-3.1
and Set1 mediated distinct Td phases. Indeed,
those Td– cells with hindgut morphology in Set1
mutants persistently expressed hindgut molecular markers and remained attached to the rectal
tube via intercellular adherens junctions (fig. S10,
B and C), indicating a failure in dedifferentiation.
In Td– cells with speckled morphology in either
jmjd-3.1 or Set1 backgrounds, hindgut markers
were correctly extinguished, and the cells disengaged from the rectal tube, but they did not
properly activate a range of PDA neural markers
(fig. S10B), indicating a failure to properly redifferentiate. In addition, timely rescue of wdr-5.1
and jmjd-3.1 mutant strains by heat shock–
inducible rescue constructs demonstrated that
the temporal requirements of each factor correlated precisely with their successive cellular
roles (Fig. 2B). Therefore, Set1 acts dichotomously, mediating both dedifferentiation and redifferentiation, whereas jmjd-3.1 mediates only
redifferentiation. This suggests that stepwise

Fig. 2. Set1 and jmjd-3.1 mediate different Td phases. (A) Fluorescent micrographs of cell fate
markers in Td– cells in live animals. Asterisk indicates signal from adjacent cell. (B) Timed heat shock
pulses show that wdr-5.1 and jmjd-3.1 have distinct temporal requirements matching their successive
cellular roles. Ø, no heat-shock. (C) Schematic showing requirements for histone-modifying activities.
(D) Photoconversion of Dendra2 fused to JMJD-3.1 indicates that photoconverted protein is degraded
during dedifferentiation and resynthesized during redifferentiation. (E) Modulation of JMJD-3.1 nuclear
localization indicates that degradation is nuclear-dependent and that removal from the nucleus is sufficient for eliminating its activity. **P < 0.01, *P < 0.05; n > 100 animals.

SCIENCE sciencemag.org

modifications to distinct lysine residues on H3
mediate precise transition through distinct cellular phases of Td (Fig. 2C).
The model above implies a mechanism for efficient partitioning (temporally and functionally)
of K4- and K27-modifying activities within a postmitotic cell. To determine the mechanisms at
play, we first addressed how each distinct requirement was established. Consistent with Set1’s
action at different steps of the process, functional, green fluorescent protein (GFP)–tagged
Set1 subunits were nuclear-localized in Y/PDA
throughout Td (fig. S11A). Contrastingly, Cherrytagged JMJD-3.1 and a jmjd-3.1 fosmid-based
GFP construct were dynamically localized during
Td (fig. S11, A to C). Both JMJD-3.1 reporters were
detectable in the nucleus at all stages during,
before, and after Td, except for a precise window
when Y dedifferentiated. This pattern matched
jmjd-3.1’s temporal requirements and exclusive
redifferentiation role, suggesting that regulation
of JMJD-3.1 levels may provide a mechanism to
ensure stepwise K4 and K27 modification. Through
the use of a photoconvertible JMJD-3.1::Dendra2
fusion protein, as well as artificial modulation of
JMJD-3.1’s nuclear localization, we determined
that JMJD-3.1 levels were dynamically regulated
through nuclear-dependent degradation during
dedifferentiation (Fig. 2, D and E, and figs. S12).
Moreover, precise temporal regulation of JMJD3.1 levels, and hence stepwise histone modification, was critical for precise Y-to-PDA conversion
because forced ectopic JMJD-3.1 expression during Y dedifferentiation induced a Td– phenotype
and resulted in cells displaying a mixture of abnormal identities (fig. S13). Catalytic inactivation
of JMJD-3.1 largely mitigated these effects (fig.
S13D). We conclude that regulation of nuclear
JMJD-3.1 protein levels ensures partitioning of
K27 demethylase activity into the redifferentiation phase of Td and that such control is critical
for invariant Td.
We next turned our attention to the mechanism (or mechanisms) that partitioned Set1’s
K4 methylase activity into temporally separable,
seemingly opposite cellular functions (dedifferentiation and redifferentiation). We reasoned
that each distinct role might be segregated functionally through discrete temporal associations
with coregulators. Using coimmunoprecipitation, we found that the core Set1 component
WDR-5.1 associates with several subunits of the
Nanog and Oct4 Deacetylase (NODE)–like complex (Fig. 3A) (11). We have previously found that
the NODE-like complex is required for Y dedifferentiation (12) and is composed of factors that
have conserved roles in the maintenance and induction of cell plasticity in a variety of species (13).
Using Set1;NODE double mutants carrying catalytically active or inactive set-2 rescue constructs,
we found that the two complexes functionally
cooperated—in a Set1-H3K4 methyltransferase–
dependent manner—to specifically mediate Y
dedifferentiation (Fig. 3B). This suggested that
the NODE-like complex, which can bind DNA
targets directly (11), provides dedifferentiation
functional specificity to Set1.
15 AUGUST 2014 • VOL 345 ISSUE 6198

827

R ES E A RC H | R E PO R TS

Fig. 3. Phase-specific interactions specify Set1 and JMJD-3.1’s Td roles. (A) WDR-5.1 immunoprecipitates members of the C. elegans NODE complex, CEH-6(OCT), SOX-2(SOX2), SEM-4(SALL),
and EGL-27(MTA). (B) Set1 functionally cooperates with the NODE-like complex to mediate hindgut
dedifferentiation. m, mutant; wt, wild type. (C and D) JMJD-3.1 immunoprecipitates WDR-5.1 and
UNC-3 via an N-terminal domain. (E) unc-3 cooperates with jmjd-3.1 and Set1 to mediate PDA redifferentiation. Red and blue P values correspond to dedifferentiation and redifferentiation, respectively. Error bars indicate SEM. ND, not determined; ns, not significant. Statistical significance was
determined by using two-way analysis of variance.

Fig. 4. jmjd-3.1 and Set1 cooperate to ensure robust Td against stress. (A) jmjd-3.1 and Set1
components exhibit synergistic genetic interactions in a H3K27 demethylase and H3K4 methyltransferase activity–dependent manner during redifferentiation. m, null mutant; wt, wild type. (B to
E) Stressful environmental conditions enhance Td defects in jmjd-3.1 and Set1 null mutants. Error bars
indicate SEM. ND, not determined; ns, not significant. n > 100 animals; ***P < 0.001, **P < 0.01, *P <
0.05; n > 100 animals.

828

15 AUGUST 2014 • VOL 345 ISSUE 6198

Next, we uncovered an alternative functional
interaction network between WDR-5.1, JMJD-3.1,
and the phylogenetically conserved COE (Collier,
Olf, EBF)–type transcription factor (TF) UNC-3
that appeared to specify Set1’s redifferentiation
role. UNC-3 transcriptionally regulates batteries
of genes that specify terminal features of motor
neurons in C. elegans as well as chordates (14). We
showed that UNC-3, which is required for PDA
redifferentiation (10), interacts with a functionally important N-terminal region of JMJD-3.1.
JMJD-3.1 in turn associates with WDR-5.1 (Fig. 3,
C and D, and fig. S14), providing a mechanistic
rationale for the precise regulation of JMJD-3.1
protein levels that we unraveled, one that may
serve to prevent interaction with Set1 until redifferentiation, for which both activities are
needed. Moreover, double unc-3;jmjd-3.1 and
unc-3;wdr-5.1 mutants suggested that these three
factors functionally cooperate specifically during
the redifferentiation phase of Td and not during
dedifferentiation (Fig. 3E). Thus, UNC-3 appears
to act as an important regulatory hub for the redifferentiation phase of Td onto which chromatin modifiers may log their activities.
To gain deeper insight into SET1 and JMJD-3.1’s
mode of action, we analyzed double Set1;jmjd-3.1
mutant strains carrying active or inactive set-2
or jmjd-3.1 transgenes, or jmjd-3.1(fp15) mutants
overexpressing H3.3K4A (Fig. 4A). Our results
suggested that K4 methylation and K27 demethylation (i) acted in parallel, rather than in a mutually
dependent manner, and (ii) ultimately converged
on a common downstream target (or targets)
to specifically promote redifferentiation. One
possibility—suggested by their physical association
as well as data presented earlier indicating that
Set1 and JMJD-3.1 cellular localizations and activities intersected precisely during redifferentiation—
is that both activities act in a bifunctional K4/K27–
modifying complex. Such complexes have been
shown in other species (6, 15–17) and have been
predicted to resolve bivalent domains: nucleosomes
containing both H3K4me3 and H3K27me3 signatures (18). Evidence suggests that such domains act
to poise developmental loci for timely and efficient
activation, representing an attractive mechanism
for ensuring precise PDA redifferentiation (19).
However, bivalency has only been reported thus far
in vertebrates (20). Here, we demonstrate the cooccurrence of H3K4me3 and H3K27me3 marks on
mononucleosomes purified from worms, pointing
to the existence of bivalent domains in C. elegans
(fig. S15) and opening the possibility that targets of
Set1 and JMJD-3.1 may be bivalently marked.
Our results suggest a model in which Set1 and
jmjd-3.1’s role is to ensure precise Td, whereas
key TFs, which exhibit complete penetrance in
null mutants (3, 10, 12), act as drivers. We found
that histone modifications additionally are critically needed when the worm is exposed to adverse conditions. In absence of jmjd-3.1 or Set1,
Y-to-PDA Td was hypersensitive to a variety of
stressful treatments, whereas Td remained unaffected in wild-type animals (Fig. 4, B to E). This
suggests that both Set1-dependant H3K4 methylation and jmjd-3.1–dependent H3K27me3/me2
sciencemag.org SCIENCE

RE S EAR CH | R E P O R T S

demethylation provide protection against variations that may be encountered in the worm’s
natural environment. These roles in ensuring the
robustness of cell conversion in stressful circumstances may underlie invariant Td under normal
conditions. Last, our in vivo results suggest that
Set1 and jmjd-3.1 perform highly conserved roles
across phyla and cell types to reinforce TF-driven
changes in identity. Indeed, modulation of H3K4
and H3K27 methylation state has been shown
to regulate somatic reprogramming (21–23).
Specifically, the core Set1 subunit, Wdr5, was recently shown to interact with and potentiate
Oct4’s roles in the initial phases of mammalian
induced pluripotent stem (iPS) cell induction (22).
Furthermore, Jmjd3 was found to specifically
inhibit the initial stages of iPS reprogramming
(23), drawing a distinct parallel to its disruption of the initial stages of Td we observed.
RE FE RENCES AND N OT ES

1. C. Jopling, S. Boue, J. C. Izpisua Belmonte, Nat. Rev. Mol. Cell
Biol. 12, 79–89 (2011).
2. E. M. Tanaka, P. W. Reddien, Dev. Cell 21, 172–185 (2011).
3. S. Jarriault, Y. Schwab, I. Greenwald, Proc. Natl. Acad. Sci. U.S.A.
105, 3790–3795 (2008).
4. S. Zuryn, S. Le Gras, K. Jamet, S. Jarriault, Genetics 186,
427–430 (2010).
5. K. Agger et al., Nature 449, 731–734 (2007).
6. F. De Santa et al., Cell 130, 1083–1094 (2007).
7. O. Karakuzu, D. P. Wang, S. Cameron, Development 136,
943–953 (2009).
8. N. J. Francis, R. E. Kingston, C. L. Woodcock, Science 306,
1574–1577 (2004).
9. A. Shilatifard, Annu. Rev. Biochem. 81, 65–95 (2012).
10. J. P. Richard et al., Development 138, 1483–1492 (2011).
11. J. Liang et al., Nat. Cell Biol. 10, 731–739 (2008).
12. K. Kagias, A. Ahier, N. Fischer, S. Jarriault, Proc. Natl. Acad. Sci.
U.S.A. 109, 6596–6601 (2012).
13. I. Chambers, S. R. Tomlinson, Development 136, 2311–2322
(2009).
14. P. Kratsios, A. Stolfi, M. Levine, O. Hobert, Nat. Neurosci. 15,
205–214 (2012).
15. I. Issaeva et al., Mol. Cell. Biol. 27, 1889–1903 (2007).
16. Y. W. Cho et al., J. Biol. Chem. 282, 20395–20406 (2007).
17. S. A. Miller, A. C. Huang, M. M. Miazgowicz, M. M. Brassil,
A. S. Weinmann, Genes Dev. 22, 2980–2993 (2008).
18. B. E. Bernstein et al., Cell 125, 315–326 (2006).
19. P. Voigt, W. W. Tee, D. Reinberg, Genes Dev. 27, 1318–1338
(2013).
20. P. Voigt et al., Cell 151, 181–193 (2012).
21. A. A. Mansour et al., Nature 488, 409–413 (2012).
22. Y. S. Ang et al., Cell 145, 183–197 (2011).
23. W. Zhao et al., Cell 152, 1037–1050 (2013).
ACKN OW LEDG MEN TS

We are grateful to S. Gasser, R. Klose, M. Labouesse, B. Prud’homme,
and R. Schneider for their comments on the manuscript. Work
in R.M.’s laboratory is funded by a European Research Council–Stg
(REPODDID) grant, an ATIP-Avenir grant, and a Fondation pour
la Recherche Medicale (FRM) grant. Work in S.J.’s laboratory
is funded by grants from FRM, the Fondation ARC pour la
recherche sur le cancer, the Association Française contre les
Myopathies (AFM), the Fondation Schlumberger pour
l’Enseignement et la Recherche (FSER), and the European
Molecular Biology Organization Young Investigator program (EMBO
YIP). S.Z. is a FRM and ARC postdoctoral fellow, and S.J. is a
Centre National de la Recherche Scientifique research director.
SUPPLEMENTARY MATERIALS

www.sciencemag.org/content/345/6198/826/suppl/DC1
Materials and Methods
Figs. S1 to S16
Table S1
References (24–42)
12 May 2014; accepted 17 July 2014
10.1126/science.1255885

SCIENCE sciencemag.org

SYNTHETIC BIOLOGY

Programmable on-chip DNA
compartments as artificial cells
Eyal Karzbrun,1* Alexandra M. Tayar,1* Vincent Noireaux,2 Roy H. Bar-Ziv1†
The assembly of artificial cells capable of executing synthetic DNA programs has been
an important goal for basic research and biotechnology. We assembled two-dimensional
DNA compartments fabricated in silicon as artificial cells capable of metabolism,
programmable protein synthesis, and communication. Metabolism is maintained by
continuous diffusion of nutrients and products through a thin capillary, connecting protein
synthesis in the DNA compartment with the environment. We programmed protein
expression cycles, autoregulated protein levels, and a signaling expression gradient,
equivalent to a morphogen, in an array of interconnected compartments at the scale of
an embryo. Gene expression in the DNA compartment reveals a rich, dynamic system that
is controlled by geometry, offering a means for studying biological networks outside a
living cell.

I

n the past decade, cell-free gene expression
reactions have been used to design synthetic biological systems, including droplets for
molecular evolution (1), a vesicle bioreactor
toward an artificial cell (2), regulatory (3–6)
and morphogenetic-like genetic circuits (7), as
well as assembly of protein complexes (8–10).
Thus far, the lack of efficient protein turnover
has prevented the emergence of expression dynamic patterns, such as oscillations, in a continuous expression bioreactor or vesicle (2, 11).
Microfluidic chips containing switching valves
and addressable fluidic chambers (12, 13) have
succeeded in implementing steady-state and
dynamic protein synthesis reactions (14). However, micrometer-scale positional information
encoded in diffusive concentration gradients of
proteins and mRNA, as in a morphogenetic scenario (15), is washed away in flow-driven expression compartments. Here, we present a solid-state
biochip approach for the assembly of an artificial
cell, which enables protein turnover, materials
exchange with the environment, a capacity to encode and express genes at high surface density
within a controlled geometry, and the ability to
maintain micrometer-scale positional information in diffusive molecular gradients.
Dense phases of end-attached, linear doublestranded DNA templates (DNA brushes) were
assembled by chemical photolithography (16, 17)
on the surface of circular compartments carved
in silicon with radius R = 50 mm and depth h = 1
to 3 mm (Fig. 1, A and B, and figs. S1 to S3). The
DNA compartments were connected to a 30-mmdeep flow channel through thin capillaries of
width W = 20 mm and length L = 50 to 300 mm.
The device was sealed, and Escherichia coli
cell extract (18, 19) was continuously flown in
the main channel. Reaction components were
transported by diffusion into the DNA compart1

Department of Materials and Interfaces, Weizmann Institute
of Science, Rehovot 76100, Israel. 2Department of Physics,
University of Minnesota, Minneapolis, MN 55455, USA.

*These authors contributed equally to this work. †Corresponding
author. E-mail: roy.bar-ziv@weizmann.ac.il

ment because of the high resistance to flow
through the capillary (fig. S4). Proteins were
synthesized in the DNA compartment and
diffused out to the flow channel through the
capillary. A linear protein concentration gradient formed along the capillary, decaying from
a maximal value in the DNA compartment, where
its distribution was homogenous, down to zero at
the channel junction (Fig. 1C). The protein linear
gradient persisted throughout the duration of
expression, which reached a steady-state concentration (fig. S5). The dilution through the capillary
leads to an emergent effective protein lifetime,
2
t ¼ pR
WD L, that is obtained by solving the diffusion equation in the compartment geometry,
where D is the protein diffusion constant (supplementary materials).
The effective protein lifetime enabled us to
observe gene expression dynamics, including
steady states (Fig. 1D) and oscillations (Fig. 1E).
Figure 1D shows expression of green fluorescent
protein (GFP) through a positive feedback gene
construct, which is self-activated by the AraC protein dimer in the presence of arabinose (20) (fig.
S6). The kinetics was characterized by a sharp
onset after a 2-hour delay and reached a steadystate level for over 8 hours. To implement an oscillatory gene expression dynamics, we used an
activator-repressor network with sigma factor
s28 for activation and the lambda phage cI repressor (fig. S6 and tables S1 to S3). The network exhibited emergent oscillations for many
hours with a period of ~2.5 hours (Fig. 1E). Furthermore, the high concentration of regulatory
proteins near the DNA brush enabled direct imaging of transcription regulation. We imaged
repression of transcription by using a negativefeedback construct that codes for a fusion of a
Cro repressor dimer and GFP, expressed under
a Cro-regulated promoter (fig. S6). The synthesized fusion protein bound the repressor site adjacent to the promoter, thereby localized the
GFP signal to the DNA brush (Fig. 1F) and led
to a reduced self-regulated expression levels in
steady state.
15 AUGUST 2014 • VOL 345 ISSUE 6198

829