Localized light-induced protein dimerization in living cells using a photocaged dimerizer

Abstract

Regulated protein localization is critical for many cellular processes. Several techniques have been developed for experimental control over protein localization, including chemically induced and light-induced dimerization, which both provide temporal control. Light-induced dimerization offers the distinct advantage of spatial precision within subcellular length scales. A number of elegant systems have been reported that utilize natural light-sensitive proteins to induce dimerization via direct protein-protein binding interactions, but the application of these systems at cellular locations beyond the plasma membrane has been limited. Here we present a new technique to rapidly and reversibly control protein localization in living cells with subcellular spatial resolution using a cell-permeable, photoactivatable chemical inducer of dimerization. We demonstrate light-induced recruitment of a cytosolic protein to individual centromeres, kinetochores, mitochondria and centrosomes in human cells, indicating that our system is widely applicable to many cellular locations.

ARTICLE

Received 28 Aug 2014 | Accepted 6 Oct 2014 | Published 17 Nov 2014

DOI: 10.1038/ncomms6475

Localized light-induced protein dimerization in living cells using a photocaged dimerizer

Edward R. Ballister1, Chanat Aonbangkhen2, Alyssa M. Mayo1, Michael A. Lampson1 & David M. Chenoweth2

Regulated protein localization is critical for many cellular processes. Several techniques have been developed for experimental control over protein localization, including chemically induced and light-induced dimerization, which both provide temporal control. Light-induced dimerization offers the distinct advantage of spatial precision within subcellular length scales. A number of elegant systems have been reported that utilize natural light-sensitive proteins to induce dimerization via direct proteinprotein binding interactions, but the application of these systems at cellular locations beyond the plasma membrane has been limited. Here we present a new technique to rapidly and reversibly control protein localization in living cells with subcellular spatial resolution using a cell-permeable, photoactivatable chemical inducer of dimerization. We demonstrate light-induced recruitment of a cytosolic protein to individual centromeres, kinetochores, mitochondria and centrosomes in human cells, indicating that our system is widely applicable to many cellular locations.

1 Department of Biology, School of Arts and Sciences, University of Pennsylvania, Philadelphia, Pennsylvania 19104, USA. 2 Department of Chemistry, School of Arts and Sciences, University of Pennsylvania, Philadelphia, Pennsylvania 19104, USA. Correspondence and requests for materials should be addressed to M.A.L. (email: mailto:[email protected]

Web End [email protected] ) or to D.M.C. (email: mailto:[email protected]

Web End [email protected] ).

NATURE COMMUNICATIONS | 5:5475 | DOI: 10.1038/ncomms6475 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 1

& 2014 Macmillan Publishers Limited. All rights reserved.

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms6475

Vital cellular processes, such as signalling, migration, growth and division, require the participation of tens to hundreds of different proteins whose localization and

interactions are dynamically regulated in space and time. Experimentally controlling protein localization is a powerful approach for investigating these complex systems. Light-induced protein dimerization can provide both temporal and spatial precision for such manipulations1.

Similar to the more mature technique of chemically induced dimerization24, light-induced dimerization exploits a pair of specialized protein domains that can be driven into a high-afnity binding state by an experimental stimulus, such as the addition of a small molecule or illumination with a specic wavelength of light. These dimerization domains are genetically fused to two proteins of interest, which allows for experimental control over their interaction. The state-of-the-art for light-induced protein dimerization techniques utilize either photosensitive proteins58 or a photocaged chemical dimerizer, rapamycin9,10, and each technique offers a different set of strengths and weaknesses.

Inducing dimerization with subcellular spatial control requires that the photoactivated component remain persistently localized at the site of activation. The reported forms of photocaged rapamycin release diffusible, free rapamycin upon uncaging and therefore have limited potential for spatially controlled dimerization3. In principle, a photosensitive protein can be stably targeted to any location in the cell by genetic fusion to an appropriate anchoring domain. In practice, the photosensitive protein systems all function at the plasma membrane, and some have been used to regulate transcription57, but application of most of these systems to a variety of intracellular locations has not been reported. Only one system5 has been tested with a diverse set of anchoring domains, and in that study approximately half of the fusion proteins tested localized properly11.

Here we report a light-induced dimerization system that is broadly applicable at a variety of locations in human cells. This system utilizes a modular dimerizer consisting of a photocaged ligand joined by a exible linker to a different, non-caged ligand. The two orthogonal ligands allow specic, high-afnity binding to their cognate receptors and, therefore, heterodimerization of proteins of interest fused to the receptors. We demonstrate light-induced recruitment of a cytosolic reporter protein to centromeres, kinetochores, centrosomes and mitochondria in living human cells using this novel cell-permeable photocaged chemical dimerizer.

ResultsMolecular design of the photocaged dimerizer. To function successfully, a photocaged chemical dimerizer should enter living cells, stably bind the receptor for the non-caged ligand and bind the receptor for the photocaged ligand only after uncaging (Fig. 1). To ensure that the dimerizer remains stably bound to the receptor for the uncaged ligand, we employed a selective, covalent labelling system. This approach allows saturation of the rst receptor with the caged dimerizer followed by washout of any free dimerizer, which would otherwise nonproductively compete for binding the second receptor after uncaging. For our non-caged covalent ligand/receptor pair, we chose the Halotag system, which consists of a bacterial alkyldehalogenase enzyme mutated to irreversibly form a covalent bond with chloroalkyl-containing substrates12. This system offers rapid covalent protein labelling that is bioorthogonal in human cells, and it has been successfully utilized for protein heterodimerization13,14. For clarity, we will refer to the Halotag protein as Haloenzyme and the chlorohexane group as Haloligand.

For the photocaged ligand, we chose trimethoprim (TMP), a dihydrofolate reductase (DHFR) inhibitor, which is selective for

bacterial versus mammalian forms of DHFR15. The combination of TMP and Escherichia coli DHFR (eDHFR) has been well validated as a bioorthogonal ligand/receptor pair in eukaryotic cells1620. TMP is small and structurally simple but also amenable to synthetic derivatization. Linkers can be added to the para position of the trimethoxyphenyl ring of TMP without interfering with DHFR binding16. By contrast, the 2,4-diaminopyrimidine ring of TMP is buried in the DHFR substrate pocket21, so we predicted that adding a bulky photocage to this group would effectively block DHFR binding. Neither photocaged TMP nor a photocaged modular chemical inducer of dimerization have been reported. Haloenzyme and eDHFR are globular proteins of 294 and 158 amino acids, respectively, comparable to GFP (239

NO2

NH2

HN O

N

N

OMe

N H

O

MeO

OMe

O

O

O Cl

NH OMe

O

O

NVOC photocage 405 nm cleavage

Trimethoprim (TMP) binds eDHFR

Halotag ligand

(Haloligand) covalently labels Halotag protein

Linker

cTMP-Htag (1)

NVOC

Trimethoprim

Haloligand

E. coli DHFR (eDHFR)

Halotag protein (haloenzyme)

Enters cell

Dimerization OFF

Dimerization ON

Covalent halotag labeling reaction

Reacts with Haloenzyme

eDHFR binding blocked

385405 nm light

Allows thorough

washout

Prevents release

of dimerizer

after uncaging

385405 nm light

Compatible with

fluorescent imaging Not cytotoxic

eDHFR, Haloenzyme can be fused to a variety of proteinsof interest

Uncaging

eDHFR binds uncaged TMP

Free, excess dimerizer removed by washout

Figure 1 | Design of photocaged dimerizer cTMP-Htag. (a) Chemical structure and schematic diagrams of cTMP-Htag 1 and its two receptors: E. coli DHFR (eDHFR) and the Halotag protein (Haloenzyme). (b) Schematic of light-induced protein dimerization in living cells via cTMP-Htag. The cell-permeable photocaged dimerizer enters cells and irreversibly reacts with the Haloenzyme. Any unreacted dimerizer is removed by washout. The photocage prevents binding with eDHFR before illumination, and is removed by irradiation with B385405 nm light, allowing eDHFR to bind the uncaged TMP group, thus dimerizing eDHFR and Haloenzyme.

2 NATURE COMMUNICATIONS | 5:5475 | DOI: 10.1038/ncomms6475 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

& 2014 Macmillan Publishers Limited. All rights reserved.

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms6475 ARTICLE

amino acids). Like GFP, both have been successfully fused to the amino (N) and carboxy (C) termini of a variety of proteins12,18,19,22.

The diaminopyrimidine ring of TMP was caged as the 6-nitroveratryl carbamate (NVOC)23. Exocyclic amines analogous to those of TMP have been photocaged using carbamate linkages24, which rapidly decarboxylate upon photocleavage, releasing the native amine23,25. Combining photocaged TMP with the Haloligand yielded cTMP-Htag (1, Fig. 1a). We also synthesized and tested the non-photocaged analogue TMP-Htag (2, Fig. 2). For synthetic scheme and characterization details, see Supplementary Figs 115 and Supplementary Methods. NVOC can be released with light at 365420 nm, which is not acutely cytotoxic23,26. In addition, the absorption spectrum of NVOC is sufciently blueshifted to protect it from undesired cleavage at wavelengths commonly used for imaging (for example, 488 nm) (Fig. 2).

Light-induced protein dimerization in living cells. We rst tested the ability of cTMP-Htag to recruit a freely diffusible protein to centromeres. Centromeres are specialized chromatin domains that host the dynamically regulated localization of 4100 proteins essential for cell division, making them highly attractive targets for light-inducible dimerization27. Vital centromere functions that require regulated protein localization include maintenance of sister chromatid cohesion and kinetochore assembly. Light-induced dimerization at centromeres has not been reported using any of the published systems. We constructed a HeLa cell line expressing E. coli DHFR (eDHFR) fused to mCherry (mCherryeDHFR) and the human centromere protein CENPB fused to GFP and Haloenzyme (CENPBGFPHalo). CENPB specically binds to a short sequence in the repetitive DNA present at all typical human centromeres (except the Y chromosome)28,29. As expected, CENPBGFP-Halo localizes to centromeres, and mCherryeDHFR localizes

diffusely in untreated cells. cTMP-Htag enters living cells, and incubation with 20 mM cTMP-Htag for 30 min results in full occupancy of the available Haloenzyme sites in the cell (Fig. 3). Cells treated with cTMP-Htag and exposed to a short pulse of 387 (6) nm light exhibited dramatic recruitment of mCherry eDHFR to centromeres with a t1/2 of B15 s (Fig. 4a,b). No effect

was observed in control experiments testing extended exposure to 488 nm light or the absence of cTMP-Htag (Supplementary Fig. 16). Next, we used a targeted 405 nm laser to achieve rapid and spatially dened recruitment to an individual centromere (Fig. 4c).

Photochemical uncaging is irreversible, but TMP-eDHFR binding is reversible, and there is precedent for using free TMP to compete with a TMP-based chemical dimerizer19. Therefore we predicted that addition of free TMP would compete with uncaged TMP-Htag for eDHFR binding, leading to reversal of induced dimerization. Following light-induced recruitment, incubation with excess TMP released mCherryeDHFR from centromeres within 6 min, demonstrating a full OFFONOFF cycle of experimental control (Fig. 5a,b). Furthermore, washout of excess TMP led to relocalization of mCherryeDHFR to centromeres (Fig. 5c).

To test the general applicability of our system, we constructed HaloenzymeGFP fusion proteins anchored to kinetochores, centrosomes and mitochondria. Kinetochores are complexes of B100 proteins which assemble at centromeres and serve as attachment points for spindle microtubules and regulatory hubs for mitotic checkpoint signalling27. To target Haloenzyme to kinetochores, we fused it to the kinetochore protein Nuf2, which localizes to the outer kinetochore, proximal to the microtubule binding site30. Centrosomes are protein-based organelles that are the primary microtubule organizing centers of animal cells. Centrosome replication and positioning are tightly regulated and linked to polarized cell division and cancer3133. We used the centrosome targeting domain from AKAP9 (ref. 34) to localize Haloenzyme to centrosomes. Centromeres, kinetochores and

cTMP-Htag (1)

NO2

NH2

N

N

OMe

N H

MeO

OMe

O

O

NH OMe

O

H

O N O

O

O Cl

TMP-Htag (2)

NH2

TMP-Htag (2)

cTMP-Htag (1)

Light sources used to release photocage

N

OMe

N H

O H

N O

O

1.0

0

O

O Cl

N

H2N OMe

(normalized intensity)

Light sources used for live-cell fluorescent imaging

Absorbance

387/11 nm bandpass

GFP 488 nm

mCherry 594 nm

Spot laser 405 nm

NVOC photocage absorbance

250 350 450 550 650Wavelength (nm)

Figure 2 | Ultraviolet (UV)visible absorbance spectra of dimerizers. Chemical structures and normalized UVvisible spectra (260600 nm) of cTMP-Htag (1, in 90% phosphate-buffered saline (PBS) pH 7.4/10% dimethyl sulphoxide) and the non-caged analogue TMP-Htag (2, in 75% PBS pH 7.4/25% DMSO). Light sources used for uncaging (387/11, 405 nm) or for imaging (488, 594 nm) are indicated.

NATURE COMMUNICATIONS | 5:5475 | DOI: 10.1038/ncomms6475 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 3

& 2014 Macmillan Publishers Limited. All rights reserved.

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms6475

cTMP-Htag concentration

1.2

1

Halo-OreGrn

Low brightness display

High brightness display

1 M 5 M 10 M 20 M

Halo-OreGrn

intensity (a.u.)

0.8

0.6

0.4

0.2

0 0 5 10 15 20

[cTMP-Htag] (M)

cTMP-Htag incubation duration

1.2

Low brightness display

High brightness display

Untreated 15 min 30 min 60 min

1

Halo-OreGrn

Halo-OreGrn

intensity (a.u.)

0.8

0.6

0.4

0.2

0 0 20 40 60

Time (min)

Figure 3 | cTMP-Htag enters living cells. The extent of cTMP-Htag reaction with CENPB-Haloenzyme as a function of cTMP-Htag concentration or treatment time was measured using a dye-blocking assay. Cells expressing CENPB-Haloenzyme (without GFP) were incubated with cTMP-Htag, washed, treated with 100 nM Halotag-Oregon Green (Halo-OreGrn) for 20 min, then washed again before imaging. Decreased Halo-OreGrn indicates cTMP-Htag occupancy of CENPB-Haloenzyme sites. Each Halo-OreGrn image is displayed at two brightness levels to aid visualization. Within each row, all the images are displayed using identical brightness levels. (a,b) Cells were treated with 1, 5, 10 or 20 mM cTMP-Htag for 1 h, then treated with Halo-OreGrn as described above. Treatment with 10 mM cTMP-Htag for 1 h is sufcient to block B90% of Halo-OreGrn binding. (c,d) Cells were treated with 20 mM cTMP

Htag for 15, 30 or 60 min, or left untreated as control, then treated with Halo-OreGrn as described above. Treatment with 20 mM cTMP-Htag for 30 min is sufcient to block B90% of Halo-OreGrn binding. Images (a,c) are maximum-intensity projections of representative cells from each condition. Average

Halo-OreGrn intensity at centromeres was quantied for each condition (b,d). Error bars represent s.d. (nZ15 elds for each data point, multiple cells per eld). a.u., arbitrary unit.

centrosomes are not membrane-bound organelles. As a representative membrane-bound organelle, we chose mitochondria, which are involved in oxidative metabolism and apoptotic signalling, among many other critical cellular pathways. We targeted Haloenzyme to the cytosolic face of the mitochondrial outer membrane using the C-terminal domain of the Listeria monocytogenes ActA protein35.

All three of these HaloenzymeGFP-anchor constructs localized properly, and we successfully induced recruitment of mCherryeDHFR to individual structures in all cases using short pulses (10100 ms) of 405 nm laser illumination (Fig. 6). Whole-cell activation of mitochondrial-targeted cTMP-Htag resulted in a dramatic depletion of free mCherryeDHFR in the cytoplasm (Supplementary Fig. 17). This result suggests that this system could be used for protein inactivation in a manner similar to rapamycin-induced re-routing36, with the added potential for spatial control (on the level of individual cells). Together, these results show that cTMP-Htag functions robustly at a membrane-bound organelle (mitochondria) and at three diverse nonmembranous structures, including a DNA binding protein (CENPB) and two proteins which are localized entirely through proteinprotein interactions (Nuf2 and AKAP9). The ability to selectively recruit a protein to an individual kinetochore within a pair of sister kinetochores separated by B1 mm (Fig. 6c)

highlights the power of this system for spatially precise manipulation of protein localization.

DiscussionWe present a new strategy for controlling protein localization with light using a modular photocaged dimerizer covalently

pre-localized to an anchoring protein. This technique should be readily and widely applicable at a variety of cellular locations to address a range of biological questions. We demonstrate localized light-induced protein recruitment to three structures (that is, centrosomes, centromeres and kinetochores) that have not been reported with any other published system, as well as mitochondria, a representative membrane-bound organelle that has been previously targeted8. Possible applications for cTMP-Htag include recruitment of microtubule associated proteins (such as motors) to one of the two centrosomes in a mitotic spindle37, chromatin remodelling factors to individual centromeres or mitotic checkpoint signalling proteins to individual kinetochores38,39. There is precedent for similar experiments using chemically induced dimerization or constitutive tethering to operator arrays, but these lacked either spatial or temporal control. The ability to manipulate protein localization at individual cellular structures using cTMP-Htag will enable comparisons between experimentally perturbed and unperturbed structures within single cells, or the introduction of imbalances between naturally paired structures such as kinetochores and spindle poles.

Each of the published systems for light-inducible dimerization has a particular set of strengths and weaknesses, and none has emerged as optimal for all the applications. One fundamental difference between cTMP-Htag and the natural light-sensitive protein systems is that photochemical uncaging is irreversible, whereas most light-sensitive dimerization proteins will relax to a non-dimerizing state in the dark. Irreversibility may be an advantage or disadvantage, depending on experimental context, and cTMP-Htag dimerization can be reversed by addition of free TMP, albeit without spatial control. Identifying the best

4 NATURE COMMUNICATIONS | 5:5475 | DOI: 10.1038/ncomms6475 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

& 2014 Macmillan Publishers Limited. All rights reserved.

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms6475 ARTICLE

UV

Pre

Post 10 s 20 s 40 s

2 s UV

1

GFP mCherry

Centromere-recruited

mCherry-eDHFR (a.u.)

0.8

0.6

0.4

0.2

0 10

0 10 20 30 40 50 60

Time (s)

405 nm laser

Pre Post 20 s

Pre Post 10 s 20 s 30 s 40 s 80 s

GFP mCherry

Figure 4 | Light-induced dimerization at centromeres. Cells expressing CENPBGFP-Haloenzyme and mCherryeDHFR were treated with 20 mM cTMPHtag for 1 h, then washed before imaging. (a) Cell-wide recruitment of mCherryeDHFR to centromere-localized CENPBGFPHaloenzyme in response to a 2-s pulse of 387(5.5) nm light. (b) Average mCherryeDHFR centromere intensity at time points before and after uncaging, error bars represent s.d. (n 10 cells). (c) A single centromere (indicated by arrowhead in inset) was irradiated with a 405 nm laser to induce mCherryeDHFR recruitment. Insets

show boxed regions in GFP (top row) and mCherry (middle row) and colour-merge (bottom row) from indicated time points. Scale bars, 5 or 1 mm in insets. a.u., arbitrary unit.

dimerization system for a given experiment will likely require

empirical comparison.

One strength of this system is its modular design, in which each function of the molecule is devolved to a specialized, independent chemical component. Modularity also allows for future development of dimerizers with different properties. Alternative photocages could be used to allow uncaging at longer wavelengths, or the protein-binding ligands could be replaced to allow the use of a different receptor. For example, a completely orthogonal dimerizer could be designed by replacing the Halotag ligand with a SNAP-tag ligand13, TMP with dexamethasone18 and NVOC with a redshifted coumarin-based photocage23, which would allow for two different proteins of interest to be selectively recruited using different wavelengths of light. An expanding toolkit for light-induced protein dimerization will enable an exciting new era of experimental cell biology, in which researchers can optically control protein localization and interactions.

Methods

Dimerizer synthesis and characterization. 1 and 2 were synthesized via a convergent scheme (Supplementary Fig. 1). Amine-functionalized TMP and carboxylic acid-functionalized Haloligand intermediates were prepared following literature precedent4042. The functionalized TMP intermediate was photocaged in a single step using commercially available NVOC (Sigma-Aldrich)43. In the nal, convergent step, the Haloligand module was coupled to either photocaged TMP or non-caged TMP via standard amide coupling conditions. Synthesis, purication and characterization of all products are detailed in Supplementary Figs 215 and Supplementary Methods. Ultravioletvisible absorption spectrophotometry inFig. 2 was performed on a JASCO V-650 spectrophotometer with a PAC-743R multichannel Peltier using quartz cells with a 1 cm cell path length.

Cell culture and transfection. All experiments were performed with HeLa cells44 (obtained from E.V. Makayev, Nanyang Technological University, Singapore) either stably or transiently transfected with a plasmid for bicistronic expression of mCherryeDHFR and indicated Haloenzyme constructs (see below). CENPB and mitochondria experiments were performed with stable cell lines; centrosome and kinetochore experiments were performed by transient transfection. Stable cell lines were created by recombinase-mediated cassette exchangeusing the HILO recombinase-mediated cassette exchange system44 (obtained from E.V. Makeyev, Nanyang Technological University, Singapore). Briey: a monoclonal acceptor cell line with LoxP and Lox2272 recombination sites at a single chromosomal locus was cotransfected with a donor plasmid containing a transgenic cassette anked by LoxP and Lox2272 sites and a second plasmid expressing Cre recombinase, followed by selection for a marker in the donor cassette. Cells were cultured in growth medium (Dulbeccos Modied Eagles medium with 10% FBS and penicillinstreptomycin) at 37 C in a humidied atmosphere with 5% CO2. Cells at B60% conuency in a single well of a6-well plate were transfected with 1 mg of donor plasmid 10 ng of Cre plasmid

pEM784 (see below for plasmid details) using 3 ml of Fugene 6 (Promega). Two days after transfection, 1 mg ml 1 puromycin was added to the growth medium for the selection of stable cell lines. Transient transfections were performed as above using 3 ml of Fugene 6 and 1 mg of plasmid DNA, 40 h before experiment.

Dimerizer treatment. cTMP-Htag (1) was dissolved in dimethyl sulphoxide at 10 mM and stored in amber plastic microcentrifuge tubes at 80 C, then diluted

in medium to a nal working concentration of 20 mM (except as indicated in Fig. 3). Care was taken to minimize incidental exposure of cTMP-Htag or treated cells to light before the experiment. We found that working quickly in low levels of normal room lighting did not cause any detectable premature uncaging. The low levels of white light necessary for differential interference contrast microscopy also did not cause any detectable cTMP-Htag uncaging. TMP-Htag (2) and Halo-OreGrn were dissolved in dimethyl sulphoxide at 10 mM and 100 mM, respectively, and stored at 80 C.

NATURE COMMUNICATIONS | 5:5475 | DOI: 10.1038/ncomms6475 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 5

& 2014 Macmillan Publishers Limited. All rights reserved.

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms6475

UV TMP

Pre-UV 25 s 8 min 14 min

GFP mCherry

Dimerization: OFF ON OFF

UV

2 s

+TMP

1

10 0 10 20 30 40 50 60

* *

1

0.8

0.8

Centromere-recruited

mCherry (a.u.)

Individual fields

1 2 3 4 5 6 7 8 9 10 11

average

Centromere-recruited

mCherry (a.u.)

0.6

0.6

0.4

0.4

0.2

0.2

*

0

*

0

100

0 100 200 300 400 500 600

Time (s) after UV exposure Time (s) after TMP addition

+TMPHtag +TMP

Washout TMP

0:00 h:min 2:12 2:20 2:40 3:00

GFP mCherry

Dimerization: OFF OFF

ON ON

Figure 5 | Dimerization reversal by addition of free TMP. Cells were expressing CENPBGFP-Haloenzyme and mCherryeDHFR. (a,b) Cells were treated with 20 mM cTMP-Htag for 1 h, then washed before imaging. Eleven elds of interphase cells were sequentially activated with a 2 s pulse of ultraviolet (UV)

light, then imaged every 10 s for 60 s. Eight minutes after the 11th eld was photoactivated, all the elds were imaged again, then TMP was added to a nal concentration of 100 mM. Each eld was imaged every minute for 10 min. Image acquisition and stage movement took B4.5 s for each eld. Selected images from the 11th eld (a). mCherry intensity at centromeres was quantied (b). For the photoactivation phase of the experiment, centromere intensity levels were normalized for each eld, then averaged over all the elds for each time point. This average, as well as the normalized values for the 11th eld are plotted. For the TMP-addition phase of the experiment, measurements for each eld at each time point are plotted. mCherry intensity was normalized for each eld, then averaged across all the elds for each 1-min image acquisition cycle. Averages are plotted as a function of the average image acquisition time during each cycle (for instance, B24 s after the addition of TMP for the rst cycle). Measurements of eld 11 corresponding to the images displayed in a are marked with an asterisk (*). Error bars represent s.d. (c) Cells were treated with 1 mM TMP-Htag for 30 min, then washed and imaged B2 h later. TMP was added to a nal concentration of 100 mM, resulting in the loss of dimerization within 8 min. TMP was removed by washout, resulting in the recovery of dimerization within 20 min. Scale bars, 5 mm.

Plasmids and primers. pEM784, expressing nuclear-localized Cre recombinase and the donor cassette plasmid pEM705 were obtained from E.V. Makeyev44. All the dimerization protein expression plasmids used for this study were derived from pEM705, which contains a CAG promoter and a viral Internal Ribosome Entry

Sequence (IRES) for constitutive bicistronic cDNA expression in mammalian cells. For this study mCherryeDHFR was cloned downstream of the IRES, and the Haloenzyme constructs were cloned upstream of the IRES. For centromere experiments in Figs 4 and 5, and Supplementary Fig. 16, the full sequence of

6 NATURE COMMUNICATIONS | 5:5475 | DOI: 10.1038/ncomms6475 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

& 2014 Macmillan Publishers Limited. All rights reserved.

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms6475 ARTICLE

Pre Post 50 s

Pre 10 s 20 s

Post 50 s 80 s

Pre 10 s 20 s

Post 40 s 80 s

Pre 80 s 160 s

Post 220 s 260 s

Mitochondria

GFP mCherry

Pre Post

Pre Post

40 s

160 s

Centrosomes

GFP mCherry

Kinetochores

GFP mCherry

Figure 6 | Dimerization at individual mitochondria, centrosomes and kinetochores. Cells expressing mCherryeDHFR and HaloenzymeGFP-anchor domain fusion proteins specic for (a) mitochondria (ActA), (b) centrosomes (AKAP9) or (c) kinetochores (Nuf2) were incubated with 20 mM cTMP-Htag for 1 h, then washed before imaging. Cells were imaged before and after targeted laser illumination, as indicated. Individual structures in these cells (indicated by arrowheads in insets) were targeted with a 10100 ms pulse from a 405 nm laser immediately before the post image. Insets show boxed regions in GFP (top row) and mCherry (middle row) and colour-merge (bottom row) from indicated time points. GFP is locally photobleached by the 405 nm uncaging pulse. Gaussian smoothing with a radius of 1 pixel was applied to mCherry images in c. Scale bars, 5 or 1 mm in insets.

human CENPB was fused to the N terminus of GFP-Haloenzyme to produce CENPBGFPHaloenzyme. For centromere experiments in Fig. 3, the DNA binding domain (DBD) comprising the rst 167 amino acids of human CENPB was fused to the N terminus of Haloenzyme to produce CENPBDBDHaloenzyme. For mitochondrial targeting in Fig. 6a and Supplementary Fig. 17, the mitochondrial outer membrane targeting domain comprising the C-terminal 47 amino acids of the Listeria monocytogenes ActA gene was fused to the C terminus of HaloenzymeGFP. For centrosome targeting in Fig. 6b, the PACT (Pericentrin and AKAP Centrosome Targeting) domain of AKAP9 (residues 36423807) was fused to the C terminus of HaloenzymeGFP. For kinetochore targeting in Fig. 6c, the full sequence of the human outer kinetochore protein Nuf2 was fused to the C terminus of HaloenzymeGFP. All primers used for cloning were obtained from Integrated DNA Technologies.

The following primers were used: mCherry forward 50-acgatgataatatggtgagcaagggcgaggagg-30 and reverse 50-tctagactccaggccggccttatacagctcg-30; eDHFR forward 50-ggcctggagtctagaatcagtctgattgcggc-30 and reverse

50-cgcctcgagcgcttaattaaagatcctcttctgagatgagtttttgttcc-30; CENPBDBD forward 50-ccggcgcgccaccggtcaccatgggccccaagagg-30 and reverse 50-ttctgcaccacgcgtcccgccactgccctccgag-30; C-terminal Haloenzyme forward 50-acgcgtggtgcagaaatcggtactggctttccattcg-30 and reverse 50-taggctgcagaccggctagccggaaatctcgagcgtcgac-30; full-length CENPB-GFP forward 50-ccggcgcgccaccggtcaccatgggccccaagaggc-30 and reverse 50-tttctgcaccacgcgtttcaccacccttgtacagctcgtc-30; N-terminal Haloenzyme forward 50-atcaccctgcagtacggcgcgccaccatggcagaaatcggtactggctttcc-30 and reverse 50-atatcaagcttatcgtcatccggaaatctcgagcgtcgac-30; GFP forward 50-gctcgagatttccggttacgtagccaccatggtgagcaagggc-30 and reverse 50-cttatcgtcatccggattcaccacccttgtacagctcgtc-30; ActA forward 50-gggtggtgaatccggtgatgcatcgggaagtggaagtggatctcgagc-30 and reverse 50-cttatcgtcatccggagatcagttatctagatccggtggatcc-30; AKAP9 forward 50-gggtggtgaatccggtgatgcatcgtatcgattgctagccaacattgaagcc-30 and reverse 50-cttatcgtcatccggatgcaccttgattcagtccaaagccatctc-30; Nuf2 forward 50-gggtggtgaatccggtgatgcatcgcgagctcaagcttcgaattcaggc-30 and reverse 50-cttatcgtcatccggacattttgaacatcttcctcttcagttcagc-30.

NATURE COMMUNICATIONS | 5:5475 | DOI: 10.1038/ncomms6475 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 7

& 2014 Macmillan Publishers Limited. All rights reserved.

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms6475

Image acquisition and photoactivation. For live imaging, cells were plated on 22 22 mm glass coverslips (no. 1.5; Fisher Scientic) coated with poly-D-lysine

(Sigma-Aldrich) or 4-chamber CELLview #1.5 glass coverslip 35 mm dishes (Greiner Bio-One). Coverslips were mounted in magnetic chambers (Chamlide CM-S22-1, LCI). During imaging, cells were maintained in L-15 medium without phenol red (Invitrogen) supplemented with 10% FBS and penicillin/streptomycin. Temperature was maintained at B35 C using an environmental chamber (Incubator BL; PeCon GmbH).

All images were acquired with a spinning disk confocal microscope (DM4000; Leica) with a 100 1.4 NA objective, an XY Piezo-Z stage (Applied Scientic

Instrumentation), a spinning disk (Yokogawa), an electron multiplier charge-coupled device camera (ImageEM; Hamamatsu Photonics), and a laser merge module equipped with 488- and 593-nm lasers (LMM5; Spectral Applied Research) controlled by MetaMorph software (Molecular Devices). Images in Figs 46 and Supplementary Figs 16 and 17 are single confocal sections, images in Fig. 3 are maximum-intensity projections of ve confocal sections, 1 mm spacing.

For whole-cell ultraviolet exposure experiments in Figs 4a and 5a, and Supplementary Figs 16a and 17, light from a mercury arc lamp (Osram HXP R 120W/45c Vis) was ltered through a 387/11 nm bandpass lter (Semrock part #FF01-387/11 as a component in a 4,6-diamidino-2-phenylindole lter cube) and focused through the objective. Targeted laser experiments in Figs 4c and 6 used an iLas2 illuminator system (Roper Scientic), equipped with a 405 nm laser (CrystaLaser LC model # DL405-050-O) with a measured output of 27 mW after bre coupling, controlled using the iLas2 software module within Metamorph. Individual centromeres, mitochondria, centrosomes and kinetochores were photoactivated by dening rectangular regions of interest (26 mm2, depending on the individual structure) and rasterized 10 times with the laser at 15% power. Other standard commercial imaging systems, such as a scanning confocal microscope, could also be used for these experiments.

Image processing. All image processing and analysis was performed using ImageJ45. Image quantication in Figs 35 was performed on maximum-intensity Z-projections (ve confocal sections, 1 mm spacing). Image quantication in

Supplementary Fig. 17 was performed on single confocal sections.

For mCherryeDHFR recruitment quantication in Figs 4 and 5, centromere regions were dened by thresholding the GFP channel using the MaxEntropy algorithm. Whole-cell regions were dened by thresholding the mCherry channel using the Huang algorithm. Average mCherry intensity values within the centromere and whole-cell regions were measured. Whole-cell average mCherry intensity was subtracted from centromere mCherry intensity for each time point. These values were normalized over the entire time course for each eld, then normalized centromeric intensities were averaged over all the elds at eachtime point.

For Halo-OreGrn staining quantication in Fig. 3, centromere regions were dened in the green uorescence channel using the MaxEntropy algorithm. Average background intensity was measured in several areas between cells and subtracted from the average centromeric green uorescence values for each eld. These background-subtracted values were then averaged over all the elds for each condition.

References

1. Gautier, A. et al. How to control proteins with light in living systems. Nat. Chem. Biol. 10, 533541 (2014).

2. Fegan, A., White, B., Carlson, J. C. T. & Wagner, C. R. Chemically controlled protein assembly: techniques and applications. Chem. Rev. 110, 33153336 (2010).

3. Putyrski, M. & Schultz, C. Protein translocation as a tool: the current rapamycin story. FEBS Lett. 586, 20972105 (2012).

4. DeRose, R., Miyamoto, T. & Inoue, T. Manipulating signaling at will: chemically-inducible dimerization (CID) techniques resolve problems in cell biology. Pugers Arch. 465, 409417 (2013).

5. Levskaya, A., Weiner, O. D., Lim, W. A. & Voigt, C. A. Spatiotemporal control of cell signalling using a light-switchable protein interaction. Nature 461, 9971001 (2009).

6. Yazawa, M., Sadaghiani, A. M., Hsueh, B. & Dolmetsch, R. E. Induction of protein-protein interactions in live cells using light. Nat. Biotechnol. 27, 941945 (2009).

7. Kennedy, M. J. et al. Rapid blue-light-mediated induction of protein interactions in living cells. Nat. Methods 7, 973975 (2010).

8. Strickland, D. et al. TULIPs: tunable, light-controlled interacting protein tags for cell biology. Nat. Methods 9, 379384 (2012).

9. Karginov, A. V. et al. Light regulation of protein dimerization and kinase activity in living cells using photocaged rapamycin and engineered FKBP.J. Am. Chem. Soc. 133, 420423 (2010).10. Umeda, N., Ueno, T., Pohlmeyer, C., Nagano, T. & Inoue, T. A photocleavable rapamycin conjugate for spatiotemporal control of small GTPase activity.J. Am. Chem. Soc. 133, 1214 (2011).

11. Yang, X., Jost, A. P.-T., Weiner, O. D. & Tang, C. A light-inducible organelle-targeting system for dynamically activating and inactivating signaling in budding yeast. Mol. Biol. Cell 24, 24192430 (2013).

12. Los, G. V. et al. HaloTag: a novel protein labeling technology for cell imaging and protein analysis. ACS Chem. Biol. 3, 373382 (2008).

13. Erhart, D. et al. Chemical development of intracellular protein heterodimerizers. Chem. Biol. 20, 549557 (2013).

14. Zimmermann, M. et al. Cell-permeant and photocleavable chemical inducer of dimerization. Angew. Chem. Int. Ed. 53, 47174720, 2014).

15. Baccanari, D. P., Daluge, S. & King, R. W. Inhibition of dihydrofolate reductase: effect of NADPH on the selectivity and afnity of diaminobenzylpyrimidines. Biochemistry 21, 50685075 (1982).

16. Miller, L. W., Cai, Y., Sheetz, M. P. & Cornish, V. W. In vivo protein labeling with trimethoprim conjugates: a exible chemical tag. Nat. Methods 2, 255257 (2005).

17. Calloway, N. T. et al. Optimized uorescent trimethoprim derivatives for in vivo protein labeling. Chembiochem 8, 767774 (2007).

18. Gallagher, S. S., Miller, L. W. & Cornish, V. W. An orthogonal dexamethasonetrimethoprim yeast three-hybrid system. Anal. Biochem. 363, 160162 (2007).

19. Czlapinski, J. L. et al. Conditional glycosylation in eukaryotic cells using a biocompatible chemical inducer of dimerization. J. Am. Chem. Soc. 130, 1318613187 (2008).

20. Chen, Z., Jing, C., Gallagher, S. S., Sheetz, M. P. & Cornish, V. W. Second-generation covalent TMP-tag for live cell imaging. J. Am. Chem. Soc. 134, 1369213699 (2012).

21. Heaslet, H. et al. Structural comparison of chromosomal and exogenous dihydrofolate reductase from Staphylococcus aureus in complex with the potent inhibitor trimethoprim. Proteins 76, 706717 (2009).

22. Saul, J. et al. Development of a full-length human protein production pipeline. Protein Sci. 23, 11231135 (2014).

23. Kln, P. et al. Photoremovable protecting groups in chemistry and biology: reaction mechanisms and efcacy. Chem. Rev. 113, 119191 (2013).

24. Furuta, T., Watanabe, T., Tanabe, S., Sakyo, J. & Matsuba, C. Phototriggers for nucleobases with improved photochemical properties. Org. Lett. 9, 47174720 (2007).

25. Cummings, R. T. & Krafft, G. A. Photoactivable uorophores. 1. Synthesis and photoactivation ofo-nitrobenzyl-quenched uorescent carbamates. Tetrahedron Lett. 29, 6568 (1988).

26. Bochet, C. G. Orthogonal photolysis of protecting groups. Angew. Chem. Int. Ed. 40, 20712073 (2001).

27. Westhorpe, F. G. & Straight, A. F. Functions of the centromere and kinetochore in chromosome segregation. Curr. Opin. Cell Biol. 25, 334340 (2013).

28. Earnshaw, W. C. & Rotheld, N. Identication of a family of human centromere proteins using autoimmune sera from patients with scleroderma. Chromosoma 91, 313321 (1985).

29. Masumoto, H., Masukata, H., Muro, Y., Nozaki, N. & Okazaki, T. A human centromere antigen (CENP-B) interacts with a short specic sequence in alphoid DNA, a human centromeric satellite. J. Cell Biol. 109, 19631973 (1989).

30. Joglekar, A. P., Bloom, K. & Salmon, E. D. In vivo protein architecture of the eukaryotic kinetochore with nanometer scale accuracy. Curr. Biol. 19, 694699 (2009).

31. Elric, J. & Etienne-Manneville, S. Centrosome positioning in polarized cells: Common themes and variations. Exp. Cell Res. 328, 240248 (2014).

32. Godinho, S. A. & Pellman, D. Causes and consequences of centrosome abnormalities in cancer. Philos. Trans. R. Soc. Lond. B. Biol. Sci. 369, 20130467 (2014).

33. Frat-Karalar, E. N. & Stearns, T. The centriole duplication cycle. Philos. Trans.R. Soc. Lond. B. Biol. Sci 369, 20130460 (2014).34. Gillingham, A. K. & Munro, S. The PACT domain, a conserved centrosomal targeting motif in the coiled-coil proteins AKAP450 and pericentrin. EMBO Rep. 1, 524529 (2000).

35. Bear, J. E. et al. Negative regulation of broblast motility by Ena/VASP proteins. Cell 101, 717728 (2000).

36. Robinson, M. S., Sahlender, D. A. & Foster, S. D. Rapid inactivation of proteins by rapamycin-induced rerouting to mitochondria. Dev. Cell 18, 324331 (2010).

37. Kapitein, L. C. et al. Probing intracellular motor protein activity using an inducible cargo trafcking assay. Biophys. J. 99, 21432152 (2010).

38. Ballister, E. R., Riegman, M. & Lampson, M. A. Recruitment of Mad1 to metaphase kinetochores is sufcient to reactivate the mitotic checkpoint. J. Cell Biol. 204, 901908 (2014).

39. Kuijt, T. E. F., Omerzu, M., Saurin, A. T. & Kops, G. J. P. L. Conditional targeting of MAD1 to kinetochores is sufcient to reactivate the spindle assembly checkpoint in metaphase. Chromosoma 123, 471480 (2014).

40. Jing, C. & Cornish, V. W. A uorogenic TMP-tag for high signal-to-background intracellular live cell imaging. ACS Chem. Biol. 8, 17041712 (2013).

41. Singh, V., Wang, S. & Kool, E. T. Genetically encoded multispectral labeling of proteins with polyuorophores on a DNA backbone. J. Am. Chem. Soc. 135, 61846191 (2013).

8 NATURE COMMUNICATIONS | 5:5475 | DOI: 10.1038/ncomms6475 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

& 2014 Macmillan Publishers Limited. All rights reserved.

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms6475 ARTICLE

42. Passemard, S. et al. Convenient synthesis of heterobifunctional poly(ethylene glycol) suitable for the functionalization of iron oxide nanoparticles for biomedical applications. Bioorg. Med. Chem. Lett. 23, 50065010 (2013).

43. Wysocki, L. M. et al. Facile and general synthesis of photoactivatable xanthene dyes. Angew. Chem. Int. Ed. 50, 1120611209 (2011).

44. Khandelia, P., Yap, K. & Makeyev, E. V. Streamlined platform for short hairpin RNA interference and transgenesis in cultured mammalian cells. Proc. Natl Acad. Sci. USA 108, 1279912804 (2011).

45. Schneider, C. A., Rasband, W. S. & Eliceiri, K. W. NIH Image to ImageJ: 25 years of image analysis. Nat. Methods 9, 671675 (2012).

Acknowledgements

We thank Dr George Furst and Dr Jun Gu for NMR assistance and insightful discussions. We thank Dr Rakesh Kohli for High Resolution Mass Spectrometry (HRMS) assistance. C.A. thanks the Royal Thai Government for fellowship funding through the Development and Promotion of Science and Technology (DPST) Project.

Author contributions

E.R.B., M.A.L. and D.M.C. designed the experiments. E.R.B. initiated dimerizer synthesis and conducted the biological experiments. C.A. conducted synthesis and chemical characterization. A.M.M. contributed to cloning. E.R.B., M.A.L. and D.M.C. analysed the data and wrote the manuscript.

Additional information

Supplementary Information accompanies this paper at http://www.nature.com/naturecommunications

Web End =http://www.nature.com/ http://www.nature.com/naturecommunications

Web End =naturecommunications

Competing nancial interests: The authors declare no competing nancial interests.

Reprints and permission information is available online at http://npg.nature.com/reprintsandpermissions/

Web End =http://npg.nature.com/ http://npg.nature.com/reprintsandpermissions/

Web End =reprintsandpermissions/

How to cite this article: Ballister, E. R. et al. Localized light-induced protein dimerization in living cells using a photocaged dimerizer. Nat. Commun. 5:5475 doi: 10.1038/ ncomms6475 (2014).

NATURE COMMUNICATIONS | 5:5475 | DOI: 10.1038/ncomms6475 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 9

& 2014 Macmillan Publishers Limited. All rights reserved.