Crystal structure of a domain-swapped photoactivatable sfGFP variant provides evidence for GFP folding pathway

Article abstract of DOI:10.1111/febs.14797

Photoactivatable fluorescent proteins (PA-FPs) are a powerful non-invasive tool in high-resolution live-cell imaging. They can be converted from an inactive to an active form by light, enabling the spatial and temporal trafficking of proteins and cell dynamics. PA-FPs have been previously generated by mutating selected residues in the chromophore or in its close proximity. A new strategy to generate PA-FPs is the genetic incorporation of unnatural amino acids (UAAs) containing photocaged groups using unique suppressor tRNA/aminoacyl-tRNA synthetase pairs. We set out to develop a photoactivatable GFP variant suitable for time-resolved structural studies. Here, we report the crystal structure of superfolder GFP (sfGFP) containing the UAA ortho-nitrobenzyl-tyrosine (ONBY) at position 66 and its spectroscopic characterization. Surprisingly, the crystal structure (to 2.7?? resolution) reveals a dimeric domain-swapped arrangement of sfGFP66ONBY with residues 1–142 of one molecule associating with residues 148–234 from another molecule. This unusual domain-swapped structure supports a previously postulated GFP folding pathway that proceeds via an equilibrium intermediate.

Full text of DOI:10.1111/febs.14797

Received Date : 07-Sep-2018
Revised Date : 07-Feb-2019
Accepted Date : 27-Feb-2019
Crystal structure of a domain-swapped photoactivatable sfGFP variant provides evidence for
GFP folding pathway
Stephanie Kesgin-Schaefer1,2, Johannes Heidemann3, Anke Puchert1,4, Knut Koelbel3, Briony Yorke1,
Nils Huse1,4, Arwen R. Pearson1,4, Charlotte Uetrecht3,5, and Henning Tidow1,2*
1 The Hamburg Centre for Ultrafast Imaging and Department of Chemistry, Luruper Chaussee 149, D-
22761 Hamburg, Germany
2
Institute for Biochemistry and Molecular Biology, University of Hamburg, Martin-Luther-King-
Platz 6, D-20146 Hamburg, Germany
3
Heinrich Pette Institute, Leibniz Institute for Experimental Virology, Martinistrasse 52, D-20251
Hamburg, Germany
4
Institute for Nanostructure and Solid State Physics, Department of Physics and Center for Free-
Electron Laser Science, University of Hamburg, Luruper Chaussee 149, D-22761 Hamburg, Germany
5 European XFEL GmbH, Holzkoppel 4, D-22869 Schenefeld, Germany
* corresponding author
This article has been accepted for publication and undergone full peer review but has not
been through the copyediting, typesetting, pagination and proofreading process, which may
lead to differences between this version and the Version of Record. Please cite this article as
doi: 10.1111/febs.14797
This article is protected by copyright. All rights reserved.

Running title: Crystal structure of a domain-swapped GFP variant
Article type : Original Article
Corresponding author:
Henning Tidow
Department of Chemistry, Institute for Biochemistry and Molecular Biology, University of Hamburg,
Martin-Luther-King-Platz 6, D-20146 Hamburg, Germany
Tel: +49 40428388984
e-mail: tidow@chemie.uni-hamburg.de
Abstract
Photoactivatable fluorescent proteins (PA-FPs) are a powerful non-invasive tool in high-resolution
live-cell imaging. They can be converted from an inactive to an active form by light, enabling the
spatial and temporal trafficking of proteins and cell dynamics. PA-FPs have been previously
generated by mutating selected residues in the chromophore or in its close proximity. A new strategy
to generate PA-FPs is the genetic incorporation of unnatural amino acids containing photocaged
groups using unique suppressor tRNA/aminoacyl-tRNA synthetase pairs.
We set out to develop a photoactivatable GFP variant suitable for time-resolved structural studies.
Here, we report the crystal structure of superfolder GFP (sfGFP) containing the unnatural amino acid
ortho-nitrobenzyl-tyrosine (ONBY) at position 66 and its spectroscopic characterization. Surprisingly,
the crystal structure (to 2.7 Å resolution) reveals a dimeric domain-swapped arrangement of
sfGFP66ONBY with the residues 1–142 of one molecule associating with residues 148–234 from
another molecule. This unusual domain-swapped structure supports a previously postulated GFP
folding pathway that proceeds via an equilibrium intermediate.
This article is protected by copyright. All rights reserved.

Keywords: GFP, domain-swap, unnatural amino acid, ortho-nitrobenzyl-tyrosine, crystal
structure
Introduction
Photoactivatable fluorescent proteins (PA-FPs) are molecular probes which can be controlled by light.
Their spectral characteristics can be controlled and modified by irradiating them with light of a
specific wavelength, intensity, and duration, leading to a fluorescent state [1]. Inactivated PA-FPs
are non-fluorescent, whereas activated PA-FPs yield bright signals in contrast to a dark background.
These properties allow the spatial and temporal labeling of specific subcellular structures and thus
enable the visualization of protein, organelle, and cell dynamics. This makes PA-FPs a powerful non-
invasive tool for high-resolution live-cell imaging [2] [3]. Various PA-FPs have been engineered,
especially proteins of the green fluorescent protein (GFP) family, with altered properties such as
shifted fluorescence spectra, enhanced fluorescence and higher quantum yields [4] [5] [6] [7] [2] [8].
GFP, discovered in the jellyfish Aequorea victoria, exhibits bright green fluorescence when exposed
to light in the blue to ultraviolet range [9]. The eleven β-strands form a cylindrical β-barrel with the α-
helix harboring the chromophore spanning straight through the barrel centre [10] [11]. Due to its
ability to auto-catalytically form its highly visible, efficiently emitting internal chromophore [12] [13]
[14], GFP has become a well-established and powerful tool for many applications in cell and
molecular biology [15].
Time-resolved (TR) structural biology is key to understand the mechanisms of biological processes
[16]. In order to perform time-resolved experiments – either time-resolved X-ray diffraction (TR-
XRD) or time-resolved small-angle X-ray scattering (TR-SAXS) – an ensemble of molecules first
needs to be populated/trapped in a defined state. Conformational changes can then be triggered by
means of rapid mixing or light irradiation [16] [17] [18] [19]. The use of light in principle allows
observations in the ps-regime if the decaging reaction is fast enough.

The incorporation of unnatural amino acids into the chromophore of GFP has previously been
accomplished by introducing boronate, azido, keto, and nitro substituents [20], phenyl azide [21], and
azobenzene [22], thus establishing photosensitive GFP variants. In order to develop and apply a GFP-
based system that can be used for simultaneous time-resolved structural and spectroscopic studies, we
further investigated a photoactivatable GFP variant, originally designed by Schultz and coworkers
[23], that contained the photocaged tyrosine analogue ortho-nitrobenzyl-tyrosine (ONBY) [24] [25].
Photocaged unnatural amino acids (UAAs) are synthetically modified amino acids which can be
controlled by light, usually by photolytic conversion from an inactive to an active form [26]. The most
common photocaging groups are ortho-nitrobenzyl-groups and their derivatives, which can be readily
cleaved upon irradiation with 355 nm light [27]. The incorporation of photocaged ONBY in E. coli is
achieved by genetically encoding it using an amber stop codon (TAG) in combination with an
orthogonal suppressor tRNA/ aminoacyl-tRNA synthetase pair [24]. Incorporation of ONBY at
residue 66, a key residue of the chromophore, leads to a non-fluorescent (“dark-state”) GFP molecule.
Upon irradiation with light (355 nm), the photocaging ortho-nitrobenzyl-group dissociates and the
fluorescent state is re-established. However, the structure of this photocaged GFP variant, the
decaging mechanism, as well as its suitability for time-resolved structural studies are still not
understood.
In this study, we determined the crystal structure of the dark-state sfGFP66ONBY variant. The
structure reveals an unusual and unexpected dimeric domain-swapped arrangement, which in
combination with spectroscopic and native mass spectrometry (NMS) experiments provides
experimental evidence for a GFP folding pathway.
Results and Discussion
Spectroscopic and hydrodynamic properties of sfGFP66ONBY
In order to investigate the suitability of GFP variants for time-resolved structural and spectroscopic
studies, we expressed and purified a superfolder GFP (sfGFP) variant, which had the genetically
This article is protected by copyright. All rights reserved.

encoded unnatural amino acid ONBY incorporated at position 66, replacing the central Tyr residue of
the chromophore. The protein predominantly adopts a monomeric state as judged from size-exclusion
chromatography (Figure 1A). It also migrates as monomer on an SDS-PAGE gel when samples were
fully denatured (Figure 1B).
The wt-GFP chromophore is formed by the autocatalytic cyclization of the Ser/Thr 65-Tyr 66-Gly 67
tripeptide [10] [11], which is associated to an intramolecular hydrogen-bonding network linking the
two ends of the chromophore in the protein pocket. Upon excitation with light of wavelength 355 nm,
an excited-state proton transfer (ESPT) is induced which enables structural transitions and involves
multiple proton transfers through the hydrogen bonding network resulting in fluorescence [28] [29].
In contrast to wt-GFP, the sfGFP66ONBY variant protein is non-fluorescent. The bulky ortho-
nitrobenzyl-group most likely blocks the proton-wire, resulting in a non-functional chromophore [30].
Moreover, it has been reported, that ortho-nitrobenzyl-groups quench excited fluorophores through a
photo-induced electron transfer from the S1 excited chromophore to the ortho-nitrobenzyl-group upon
irradiation [31] [32].
The decaging characteristics of sfGFP66ONBY were investigated by absorption spectroscopy. Upon
photolysis with 355 nm pulsed light, the UV/Vis absorption spectrum exhibits a loss of the absorption
peak at 390 nm for the caged protein, while a new absorption peak at 488 nm appears for the
photolyzed protein (Figure 2A, 2B). Photolyzed sfGFP66ONBY shows an absorption spectrum
similar to wild-type sfGFP. Additionally, fluorescence spectra of photolyzed sfGFP66ONBY and
wild-type sfGFP using 355 nm excitation are similar with fluorescence maxima at 509 nm and 517
nm, respectively (Figure 2C). A pronounced spectral shoulder at 541 nm is also present in both
fluorescence spectra. The similarity of absorption and fluorescence spectra of photolyzed
sfGFP66ONBY and wild-type sfGFP confirm previous measurements and dissociation of the ONBY
group from the mutant [20].

Crystal structure of dark-state sfGFP66ONBY shows (dimeric) domain-swapped arrangement
To visualize the chemical environment of the caged chromophore we crystallized dark-state
(photocaged) sfGFP66ONBY and determined its structure to 2.7 Å resolution (Table I). To our great
surprise, the sfGFP66ONBY structure revealed a domain-swapped dimeric arrangement with residues
1- 143 (domain 1) of the first molecule associating with the residues 148 – 235 (domain 2) of the
second molecule (Figure 3A). Clear electron density identified the connecting loops ranging from
residues Tyr143 to His148 (Figure 3C), while no electron density was present in the position
corresponding to β-strand β7 in the search model (as previously observed by Groff et al. [23]). The
electron density extends away from the search model, leading to the second half-subunit where the
protein backbone is also shifted due to the steric interference of the bulky ortho-nitrobenzyl-group
with the normal tight packing of the β7-strand against the chromophore. The peptide side chains of the
β7-strand are reoriented and the bulky side chain of Phe145 is rotated by 180° to the outside of the
barrel. These backbone rearrangements lead to the exchange of half of the structure of one subunit
with the other half of the other subunit, as the backbone folds back on itself between residues 143 and
148 to form intertwined domain-swapped dimers (Figure 3A).
The domain-swapped dimer structure of sfGFP66ONBY shows a significantly different dimer
interface compared to the wild-type sfGFP structure, which also crystallizes as a dimer (Figure 4A).
In the wt-sfGFP structure the dimer interface is formed by β-strands β7 and β10. In the domain-
swapped sfGFP66ONBY structure the β10 strands are far apart and the only connection between the
two barrels are the loops formed by residues 143-148 (Figure 4A). While the loops (res. 143-148)
connecting the swapped domains are clearly distinct from the wild-type sfGFP structure, the overall
structures of the individual barrel moieties are very similar with an all-atom RMSD of 1.27 Å (0.35 Å
for Cα atoms). Both structures have an 11-strand β-barrel as well as a mature cofactor derived from
cyclisation of the Thr 65 – Tyr 66/ONBY 66 – Gly 67 tripeptide, that differs only in the presence of
the photocaged tyrosine (ONBY) in the sfGFP66ONBY structure, which is clearly visible in the
electron density (Figure 3B).
This article is protected by copyright. All rights reserved.

The phenomenon of domain-swapped oligomers has been described in several proteins [33] [34], and
in 3DSwap, the curated knowledgebase of proteins involved in 3D domain swapping, nearly 300
depositions can be found [35]. Many domain-swapped dimers are only observed in crystals due to
their low affinity resulting from relatively small differences in free energy of monomers and dimers
[36] [37]. Hence, clear evidence for their functional role in vivo is lacking. However, three possible in
vivo functions have been suggested: regulation of protein function, a mechanism for protein
misfolding and aggregation, and structural diversification during evolution [33]. The sfGFP66ONBY
structure presented here constitutes a bona fide domain-swapped structure [38], an arrangement where
the dimer adopts a domain swapped conformation and the monomer adopts a closed conformation
[35], as previously observed for diphtheria toxin or RNaseA [39] [40].
Implications for the folding pathway of GFP
The surprising observation of a domain-swapped GFP structure prompted us to speculate about its
assembly mechanism. In general, the native and the domain-swapped state are separated by a large
kinetic barrier [41] [42], while the mechanism by which a monomeric protein transforms into a
domain-swapped dimer is highly dependent on the folding mechanism of the particular protein [33].
Due to the huge thermodynamic stability of folded GFP [43] [44] [45] it is very unlikely that the
domain-swapped structure forms via unfolding of properly folded monomeric GFP, for example
during crystallization. Therefore we investigated whether we could infer anything about folding
intermediates from this domain-swapped structure.
The folding of GFP and its variants has been excessively investigated experimentally and
theoretically [46] [47] [48] [49] [50] [51] [52] [43]. Thirumalai and coworkers used molecular
simulations to map the folding landscape of GFP and, consistently together with previous studies,
suggest that GFP folds via equilibrium or kinetic intermediates along one of four possible pathways
(EQL, KIN1, KIN2, KIN3) with the EQL pathway characterized by an equilibrium intermediate [53].
The sfGFP protein chain is built by eleven β-strands forming a cylindrical barrel, with one α-helix

threading straight through the β-barrel. The β-strands can be divided into four folding units: the N-
terminal β-strands (strand 1 – 3), the chromophore-containing helix at the center of the β-strand
barrel, the three β-strands in the center (strands 4 – 6), which form local contacts, and the five C-
terminal β-strands (strands 7 – 11) (Figure 5). In the EQL folding pathway, the N-terminal β-sheets β1
– β6 and the central helix fold and join to build half of the β-barrel structure. The C-terminal β-sheets
β7 – β11 are unstructured and flexible and are not part of the β-barrel structure. Hence, the C-terminal
β-sheets do not interact with the ordered N-terminal strands β1 – β6 in this equilibrium intermediate
state. Finally, the C-terminal β-sheets β7 – β11 fold and join with the rest of the structure folding into
the β-barrel structure. This EQL folding pathway agrees well with experimental HDX/NMR data that
fit a three-state model and indicate higher flexibility for β-strands β7 – β10 as well as stable
intermediate states along the folding pathway of GFP. These folding intermediates retain considerable
secondary and tertiary structure [47].
The domain-swapped sfGFP66ONBY structure presented here contains two domains, β1 – β6 (res. 1-
143) and β7 – β11 (res. 148-235), which makes it tempting to speculate that the domain swapped
structure is formed from an intermediate resembling the EQL intermediate postulated by Reddy et al.
[53]. We identified Phe145 as a key residue preventing folding of sfGFP66ONBY into a monomeric
structure due to its steric clashes with the bulky ortho-nitrobenzyl moiety of the ONBY chromophore
if it were to adopt the conformation observed in wild-type and sfGFP. These steric clashes would
prevent the folding and joining of β7 – β11 to the structured β1 – β6 and inhibits folding into the correct
monomeric form. Hence, it is not surprising that sfGFP66ONBY is able to access an alternate domain-
swapped folded structure (Figure 4, Figure 5C) under certain conditions.
Therefore, we expressed and purified the sfGFP66ONBY-F145A point mutant and compared its
oligomeric status to wild-type sfGFP and sfGFP66ONBY. All investigated sfGFP variants migrate as
monomers using denaturing SDS-PAGE analysis and predominantly elute as monomers from size-
exclusion chromatography (SEC) [54] (Figure 1). In order to further verify the oligomeric state of the
sfGFP variants, we also employed native mass spectrometry (NMS) [55]. For all sfGFP variants (wt-
sfGFP / sfGFP66ONBY before and after decaging / sfGFP66ONBY-F145A), monomeric and dimeric
This article is protected by copyright. All rights reserved.

species were present at low µM concentration with the dimeric fraction representing the minor
population (Figure 6). For wt-sfGFP we assigned the monomeric and dimeric species to the properly
folded monomer and dimer structures as observed in the crystal structure (pdb: 2B3Q [54]). For
sfGFP66ONBY, we assigned the dimer mass to the domain swapped dimer, while the monomer mass
could correspond to the partly unfolded intermediate (Figure 5C). Indeed, we have observed that
sfGFP66ONBY protein samples are less stable compared sfGFP and prone to aggregation within
days. This indicates that partly unfolded barrel structures exist for sfGFP66ONBY in solution,
probably because of its inability to fold into a proper barrel structure due to steric clashes between
strand β7 and the ONBY moiety. We conclude that the samples used for spectroscopy and
crystallography contained a mixture of monomeric and dimeric domain-swapped sfGFP66ONBY and
that only decaging of the monomeric sfGFP66ONBY (folding intermediate) population is able to
restore the wt-sfGFP phenotype and fluorescence, while the crystallization process may have forced
the monomeric folding intermediate into a domain-swapped dimer structure. The sfGFP66ONBY-
F145A variant showed the lowest dimer population in NMS analysis (Figure 6) suggesting that this
mutant indeed allows proper folding of sfGFP66ONBY and thus might be a suitable variant for future
time-resolved spectroscopic and structural studies.
In summary, the crystal structure of sfGFP66ONBY containing a genetically incorporated unnatural
amino acid as part of the chromophore revealed a novel and unexpected domain-swapped
arrangement. It is likely that the domain-swapped form of sfGFP66ONBY derives from the same
folding intermediate as that postulated for the EQL folding pathway.
Materials and methods
Ortho-nitrobenzyl-tyrosine (ONBY) synthesis
L-Tyrosine (2.0 g, 11.0 mmol) was dissolved in 2 M NaOH aq. (10 ml), and CuSO4·5H2O (1.9 g, 7.28
mmol), dissolved in a minimal amount of water, was added slowly at room temperature (RT). The
solution was heated to 60 °C and stirred for 20 min and then allowed to cool to RT before adjusting to

pH 7 using 1 M HCl. The light-blue solid was filtered and washed three times with 25 ml water,
before it was suspended in 75% aqueous dimethylformamide (60 ml). K2CO3 (1.5 g, 11.04 mmol) and
ortho-nitrobenzyl bromide (1.8 g, 8.49 mmol) were added and the reaction was allowed to proceed for
72 h at RT while kept in the dark. The solid was filtered, washed with 75% aqueous
dimethylformamide (40 ml x 2), water (40 ml x 2), 75% aqueous acetone (40 ml), and ice-cold
acetone (10 ml), and then suspended in 1 M HCl (100 ml) to stir for 2 h at RT. The white solid was
filtered and stirred once more with fresh 1 M HCl (100 ml) for another 30 min. The solid was finally
filtered, washed with water (40 ml x 2) and ice-cold acetone (10 ml), and dried to give ortho-
nitrobenzyl-tyrosine as an off-white solid (1.85 g, 68%).
Cloning, protein expression and purification
To produce photoactivatable ortho-nitrobenzyl-photocaged sfGFP (sfGFP66ONBY), the stabilized
GFP variant ‘superfolder GFP’ [54] was cloned into a pET28a vector with an N-terminal fusion
consisting of a hexahistidine-tag and a TEV protease cleavage site. For incorporation of ONBY at
residue position 66, tyrosine 66 was mutated to the amber codon, TAG, via site-directed mutagenesis,
to generate pET-sfGFP66TAG-His6. This plasmid was cotransformed into E. coli BL21 (DE3) cells
together with the plasmid pEVOL-ONBY, containing the orthogonal aminoacyl tRNA synthetase
(aaRS) and an amber suppressor tRNA, both derived from the Methanococcus jannaschi tyrosyl
tRNA/aaRS pair [56] [24].
Cells were grown at 37 °C to an OD600= 0.9 in Terrific Broth and shifted to 20 °C. ONBY was added
to a final concentration of 1 mM, and protein production was induced by the addition of 1 mM IPTG
with a delay of 60 min. After 20 h, the cells were harvested by centrifugation at 4000 g, lysed by
sonication and GFP66ONBY was purified using standard His6-tag purification protocols followed by
TEV protease digestion and a second Ni-affinity chromatography step to separate the His6-tag. The
protein was kept in 40 mM Tris, pH 8.0. Protein identity and purity were assessed using mass
spectrometry and SDS–PAGE.
This article is protected by copyright. All rights reserved.

Size-exclusion chromatography and SDS-PAGE analysis
Size-exclusion chromatography (SEC) was used to separate the monomer and dimer species from
aggregates. For the SEC a Superdex75 (GE Healthcare, Freiburg, Germany) column connected to a
ÄKTApure (GE Healthcare, Freiburg, Germany) was used at 4 °C and equilibrated with 40 mM Tris,
pH 8.0. Loaded protein concentrations were 5 - 20 mg/ml. The protein was detected by measuring the
UV absorbance at 280 nm. The resulting fractions of the purified protein were analyzed by SDS-
PAGE. Prior to being applied to the SDS-PAGE gel, protein samples where mixed with an
appropriate amount of 5x sample buffer (50 mM Tris-HCl pH 6.8, 10% (w/v) SDS, 50% (v/v)
glycerol, 125 mM DTT, 0.1% (w/v) bromophenol blue). Samples were directly loaded onto the gel or
heated for 5 min to 96 °C.
Crystallization and data collection
Initial crystallization conditions were screened for using commercially available matrices from
Hampton Research and Molecular Dimensions using the vapour-diffusion technique in sitting drops.
Crystallization of sfGFP66ONBY was accomplished by the sitting-drop vapour-diffusion method.
Sitting drops containing 2 μl of protein solution at 12 mg/ml and 2 μl of buffer (0.1 M Na-HEPES, pH
7.5, 1.1 M tri-Na-citrate) were equilibrated at 20 °C against 1 ml of the same buffer in the reservoir.
Small rodlike crystals formed after 8 weeks. Crystals were mounted and flash-cooled in liquid
nitrogen. Diffraction data were collected at 100 K at the MASSIF1 beamline at ESRF, Grenoble,
France, at a fixed energy of 12.835 keV (0.966 Å). Full datasets with an oscillation angle of 0.1° per
image were collected to a resolution limit of 2.7 Å. All datasets were processed with XDS [57] and
merged with POINTLESS [58]. GFP66ONBY crystals belonged to space group P41212 with unit-cell
dimensions a = 155.3, b = 155.3, c = 162.0, α = β = γ = 90°. A summary of the data statistics is given
in Table I.

Structure determination, refinement, and analysis
The structures were solved by molecular replacement with PHASER [59] using pdb:1ema [10] as
search model. Building of the photocaged chromophore was achieved using eLBOW [60]. Subsequent
rounds of manual building using Coot [61] and refinement using phenix.refine [62] allowed complete
model building in space group P 41212. The final model yielded crystallographic R factors of
0.20/0.24 (Rwork/Rfree) with 97% of residues falling within the Ramachandran favoured region and no
outliers in disallowed regions, and a MolProbity [63] clashscore of 4.34. Molecular graphics images
were produced using the UCSF Chimera package [64]. The structural data have been deposited in the
Protein Data Bank with accession code pdb: 6H01.
Spectroscopy
UV/Vis absorption spectra of sfGFP66ONBY during pulsed 355-nm irradiation were measured using
a QE Pro spectrometer and a DH-2000 light source (both from Ocean Optics). A Q-switched diode-
pumped Nd:YAG laser (Standa-Q1-TH), operating at 1 kHz and producing 35 mW average power of
355 nm light, was used to initiate photo-dissociation of the ONBY group. The laser was continuously
pulsed with pulse durations of 750 ps and a beam size of 600 microns FWHM. 1 ml protein solution
(30 µM protein in 40 mM Tris, pH 8.0) was used in a rectangular quartz cuvette (4 mm x 10 mm
lateral dimensions) without stirring with 355 nm excitation along the short cuvette axis and
absorption/fluorescence measurements in orthogonal direction. The integration time of each
absorption spectrum was 22 ms. Fluorescence spectra of photolyzed sfGFP66ONBY were recorded
with an integration time of 100 ms using an Ocean Optics QE Pro spectrometer via a quartz lens and a
UV grade optical fiber (400 microns diameter). Samples were photolyzed for 30 min before
fluorescence spectra were recorded. The Standa-Q1-TH laser was used in the operating mode
described above for both, sfGFP66ONBY photolysis and fluorescence spectroscopy.
This article is protected by copyright. All rights reserved.

Native Mass Spectrometry
Proteins were exchanged into 50 mM ammonium acetate pH 8.0 using centrifugal filter units
(Vivaspin 500, 10k MWCO, Sartorius). Samples were filled into ESI capillaries that were prepared as
follows: Borosilicate capillaries (1.2 mm OD, 0.68 mm ID, with filament, World Precision
Instruments) were produced with a micropipette puller (P-1000, Sutter Instruments), equipped with a
squared box filament (2.5 x 2.5 mm, Sutter Instruments). Capillaries were gold-coated using a sputter
coater (Q150R, Quorum Technologies, 40 mA, 200 s, tooling factor 2.3, end bleed vacuum 8 x 10-2
mbar). Samples were analyzed in positive ion mode on a QToF2 (Waters and MS Vision) modified
for high mass experiments [65]. Data were acquired with 7 mbar source pressure and 1.5 x 10-2 mbar
argon as collision gas, 1.3 kV capillary voltage and 100 V sample cone voltage. MS measurements
were performed with 2 µM protein concentration and spectra were acquired with 30 V collision
energy.
Acknowledgements
We thank K. Veith for technical assistance and members of the Tidow lab for helpful discussions. We
are grateful to Matthew Bowler for support at the MASSIF1 beamline (ESRF) and the Radiation
Damage BAG (mx1932) for beamtime, Henning Mootz for the ONBY synthesis protocol, and John
Joseph Zaitsev-Doyle for help with the ONBY synthesis. We acknowledge access to the Sample
Preparation and Characterization (SPC) Facility of EMBL, Hamburg. The pET-His6-GFP-TEV-LIC
cloning vector (1GFP) was a gift from Scott Gradia (Addgene plasmid # 29663). We thank Peter
Schultz (TSRI) for the pEVOL plasmid. The Heinrich Pette Institute, Leibniz Institute for
Experimental Virology is supported by the Free and Hanseatic City of Hamburg and the Federal
Ministry of Health. J.H. and C.U. acknowledge funding by BMBF VISAVIX 05K16BH1 and the
Leibniz Association through SAW-2014-HPI-4 grants. C.U. and K.K. acknowledge funding through
EU Horizon 2020 ERC StG-2017 759661. B.Y acknowledges funding by the Wellcome Trust (grant
code: 110296/Z/15/Z). This work has been supported by the Cluster of Excellence 'The Hamburg

Centre for Ultrafast Imaging' of the Deutsche Forschungsgemeinschaft (DFG) - EXC 1074 - project
ID 194651731.
Author contributions
Conceptualization, SKS and HT; Investigation, SKS, JH, AP, BY, and KK; Analysis and
Interpretation, SKS and HT; Manuscript Writing, SKS and HT; Funding Acquisition, HT;
Supervision, NH, ARP, CU and HT.
Footnotes
The abbreviations used are: aaRS, aminoacyl-tRNA synthetase; EQL, equilibrium; GFP, green
fluorescent protein; MS, mass spectrometry; NMS, native mass spectrometry; ONBY, ortho-
nitrobenzyl-tyrosine; RMSD, root mean square deviation; SEC, size-exclusion chromatography;
sfGFP, superfolder GFP; UAA, unnatural amino acid.
Competing Financial Interest statement
We declare no competing financial interests.
References
1. Lippincott-Schwartz, J. & Patterson, G. H. (2009) Photoactivatable fluorescent proteins
for diffraction-limited and super-resolution imaging, Trends Cell Biol. 19, 555-65.
2. Lukyanov, K. A., Chudakov, D. M., Lukyanov, S. & Verkhusha, V. V. (2005) Innovation:
Photoactivatable fluorescent proteins, Nat Rev Mol Cell Biol. 6, 885-91.
3. Fernandez-Suarez, M. & Ting, A. Y. (2008) Fluorescent probes for super-resolution
imaging in living cells, Nat Rev Mol Cell Biol. 9, 929-43.
4. Patterson, G. H. & Lippincott-Schwartz, J. (2002) A photoactivatable GFP for selective
photolabeling of proteins and cells, Science. 297, 1873-7.
5. Ando, R., Mizuno, H. & Miyawaki, A. (2004) Regulated fast nucleocytoplasmic shuttling
observed by reversible protein highlighting, Science. 306, 1370-3.
This article is protected by copyright. All rights reserved.

6. Wiedenmann, J., Ivanchenko, S., Oswald, F., Schmitt, F., Rocker, C., Salih, A., Spindler,
K. D. & Nienhaus, G. U. (2004) EosFP, a fluorescent marker protein with UV-inducible
green-to-red fluorescence conversion, Proc Natl Acad Sci U S A. 101, 15905-10.
7. Tsutsui, H., Karasawa, S., Shimizu, H., Nukina, N. & Miyawaki, A. (2005) Semi-rational
engineering of a coral fluorescent protein into an efficient highlighter, EMBO Rep. 6, 233-8.
8. Ando, R., Hama, H., Yamamoto-Hino, M., Mizuno, H. & Miyawaki, A. (2002) An optical
marker based on the UV-induced green-to-red photoconversion of a fluorescent protein, Proc
Natl Acad Sci U S A. 99, 12651-6.
9. Shimomura, O., Johnson, F. H. & Saiga, Y. (1962) Extraction, purification and properties
of aequorin, a bioluminescent protein from the luminous hydromedusan, Aequorea, J Cell
Comp Physiol. 59, 223-39.
10. Ormo, M., Cubitt, A. B., Kallio, K., Gross, L. A., Tsien, R. Y. & Remington, S. J. (1996)
Crystal structure of the Aequorea victoria green fluorescent protein, Science. 273, 1392-5.
11. Yang, F., Moss, L. G. & Phillips, G. N., Jr. (1996) The molecular structure of green
fluorescent protein, Nat Biotechnol. 14, 1246-51.
12. Reid, B. G. & Flynn, G. C. (1997) Chromophore formation in green fluorescent protein,
Biochemistry. 36, 6786-91.
13. Cubitt, A. B., Heim, R., Adams, S. R., Boyd, A. E., Gross, L. A. & Tsien, R. Y. (1995)
Understanding, improving and using green fluorescent proteins, Trends Biochem Sci. 20,
448-55.
14. Heim, R., Prasher, D. C. & Tsien, R. Y. (1994) Wavelength mutations and
posttranslational autoxidation of green fluorescent protein, Proc Natl Acad Sci U S A. 91,
12501-4.
15. Tsien, R. Y. (1998) The green fluorescent protein, Annu Rev Biochem. 67, 509-44.
16. Levantino, M., Yorke, B. A., Monteiro, D. C., Cammarata, M. & Pearson, A. R. (2015)
Using synchrotrons and XFELs for time-resolved X-ray crystallography and solution
scattering experiments on biomolecules, Curr Opin Struct Biol. 35, 41-8.
17. Kubelka, J. (2009) Time-resolved methods in biophysics. 9. Laser temperature-jump
methods for investigating biomolecular dynamics, Photochem Photobiol Sci. 8, 499-512.
18. Lee, H. M., Larson, D. R. & Lawrence, D. S. (2009) Illuminating the chemistry of life:
design, synthesis, and applications of "caged" and related photoresponsive compounds, ACS
Chem Biol. 4, 409-27.
19. Trincao, J., Hamilton, M. L., Christensen, J. & Pearson, A. R. (2013) Dynamic structural
science: recent developments in time-resolved spectroscopy and X-ray crystallography,
Biochem Soc Trans. 41, 1260-4.
20. Wang, F., Niu, W., Guo, J. & Schultz, P. G. (2012) Unnatural amino acid mutagenesis of
fluorescent proteins, Angew Chem Int Ed Engl. 51, 10132-5.
21. Reddington, S. C., Rizkallah, P. J., Watson, P. D., Pearson, R., Tippmann, E. M. &
Jones, D. D. (2013) Different photochemical events of a genetically encoded phenyl azide
define and modulate GFP fluorescence, Angew Chem Int Ed Engl. 52, 5974-7.
22. Padilla, M. S. & Young, D. D. (2015) Photosensitive GFP mutants containing an
azobenzene unnatural amino acid, Bioorg Med Chem Lett. 25, 470-3.
23. Groff, D., Wang, F., Jockusch, S., Turro, N. J. & Schultz, P. G. (2010) A new strategy to
photoactivate green fluorescent protein, Angew Chem Int Ed Engl. 49, 7677-9.
24. Deiters, A., Groff, D., Ryu, Y., Xie, J. & Schultz, P. G. (2006) A genetically encoded
photocaged tyrosine, Angew Chem Int Ed Engl. 45, 2728-31.
25. Krafft, G. A., Sutton, W. R. & Cummings, R. T. (1988) Photoactivable fluorophores. 3.
Synthesis and photoactivation of fluorogenic difunctionalized fluoresceins, J Am Chem Soc.
110, 301-303.

26. Adams, S. R. & Tsien, R. Y. (1993) Controlling cell chemistry with caged compounds,
Annu Rev Physiol. 55, 755-84.
27. Givens, R. S., Weber, J. F., Jung, A. H. & Park, C. H. (1998) New photoprotecting
groups: desyl and p-hydroxyphenacyl phosphate and carboxylate esters, Methods Enzymol.
291, 1-29.
28. Brejc, K., Sixma, T. K., Kitts, P. A., Kain, S. R., Tsien, R. Y., Ormo, M. & Remington,
S. J. (1997) Structural basis for dual excitation and photoisomerization of the Aequorea
victoria green fluorescent protein, Proc Natl Acad Sci U S A. 94, 2306-11.
29. Chattoraj, M., King, B. A., Bublitz, G. U. & Boxer, S. G. (1996) Ultra-fast excited state
dynamics in green fluorescent protein: multiple states and proton transfer, Proc Natl Acad Sci
U S A. 93, 8362-7.
30. Fang, C., Frontiera, R. R., Tran, R. & Mathies, R. A. (2009) Mapping GFP structure
evolution during proton transfer with femtosecond Raman spectroscopy, Nature. 462, 200-4.
31. Ueno, T., Urano, Y., Setsukinai, K., Takakusa, H., Kojima, H., Kikuchi, K., Ohkubo, K.,
Fukuzumi, S. & Nagano, T. (2004) Rational principles for modulating fluorescence
properties of fluorescein, J Am Chem Soc. 126, 14079-85.
32. Kobayashi, T., Urano, Y., Kamiya, M., Ueno, T., Kojima, H. & Nagano, T. (2007)
Highly activatable and rapidly releasable caged fluorescein derivatives, J Am Chem Soc. 129,
6696-7.
33. Rousseau, F., Schymkowitz, J. W. & Itzhaki, L. S. (2003) The unfolding story of three-
dimensional domain swapping, Structure. 11, 243-51.
34. Liu, Y. & Eisenberg, D. (2002) 3D domain swapping: as domains continue to swap,
Protein Sci. 11, 1285-99.
35. Shameer, K., Shingate, P. N., Manjunath, S. C., Karthika, M., Pugalenthi, G. &
Sowdhamini, R. (2011) 3DSwap: curated knowledgebase of proteins involved in 3D domain
swapping, Database (Oxford). 2011, bar042.
36. Carey, J., Lindman, S., Bauer, M. & Linse, S. (2007) Protein reconstitution and three-
dimensional domain swapping: benefits and constraints of covalency, Protein Sci. 16, 2317-
33.
37. Lee, S. J., Cobb, M. H. & Goldsmith, E. J. (2009) Crystal structure of domain-swapped
STE20 OSR1 kinase domain, Protein Sci. 18, 304-13.
38. Bennett, M. J., Schlunegger, M. P. & Eisenberg, D. (1995) 3D domain swapping: a
mechanism for oligomer assembly, Protein Sci. 4, 2455-68.
39. Bennett, M. J., Choe, S. & Eisenberg, D. (1994) Domain swapping: entangling alliances
between proteins, Proc Natl Acad Sci U S A. 91, 3127-31.
40. Liu, Y., Gotte, G., Libonati, M. & Eisenberg, D. (2002) Structures of the two 3D
domain-swapped RNase A trimers, Protein Sci. 11, 371-80.
41. Rousseau, F., Schymkowitz, J. W., Wilkinson, H. R. & Itzhaki, L. S. (2001) Three-
dimensional domain swapping in p13suc1 occurs in the unfolded state and is controlled by
conserved proline residues, Proc Natl Acad Sci U S A. 98, 5596-601.
42. Schlunegger, M. P., Bennett, M. J. & Eisenberg, D. (1997) Oligomer formation by 3D
domain swapping: a model for protein assembly and misassembly, Adv Protein Chem. 50,
61-122.
43. Jackson, S. E., Craggs, T. D. & Huang, J. R. (2006) Understanding the folding of GFP
using biophysical techniques, Expert Rev Proteomics. 3, 545-59.
44. Penna, T. C., Ishii, M., Junior, A. P. & Cholewa, O. (2004) Thermal stability of
recombinant green fluorescent protein (GFPuv) at various pH values, Appl Biochem
Biotechnol. 113-116, 469-83.
This article is protected by copyright. All rights reserved.

45. Pitman, D. J., Banerjee, S., Macari, S. J., Castaldi, C. A., Crone, D. E. & Bystroff, C.
(2015) Exploring the folding pathway of green fluorescent protein through disulfide
engineering, Protein Sci. 24, 341-53.
46. Hsu, S. T., Blaser, G. & Jackson, S. E. (2009) The folding, stability and conformational
dynamics of beta-barrel fluorescent proteins, Chem Soc Rev. 38, 2951-65.
47. Huang, J. R., Craggs, T. D., Christodoulou, J. & Jackson, S. E. (2007) Stable
intermediate states and high energy barriers in the unfolding of GFP, J Mol Biol. 370, 356-71.
48. Andrews, B. T., Schoenfish, A. R., Roy, M., Waldo, G. & Jennings, P. A. (2007) The
rough energy landscape of superfolder GFP is linked to the chromophore, J Mol Biol. 373,
476-90.
49. Andrews, B. T., Gosavi, S., Finke, J. M., Onuchic, J. N. & Jennings, P. A. (2008) The
dual-basin landscape in GFP folding, Proc Natl Acad Sci U S A. 105, 12283-8.
50. Orte, A., Craggs, T. D., White, S. S., Jackson, S. E. & Klenerman, D. (2008) Evidence of
an intermediate and parallel pathways in protein unfolding from single-molecule
fluorescence, J Am Chem Soc. 130, 7898-907.
51. Enoki, S., Maki, K., Inobe, T., Takahashi, K., Kamagata, K., Oroguchi, T., Nakatani, H.,
Tomoyori, K. & Kuwajima, K. (2006) The equilibrium unfolding intermediate observed at
pH 4 and its relationship with the kinetic folding intermediates in green fluorescent protein, J
Mol Biol. 361, 969-82.
52. Dietz, H. & Rief, M. (2004) Exploring the energy landscape of GFP by single-molecule
mechanical experiments, Proc Natl Acad Sci U S A. 101, 16192-7.
53. Reddy, G., Liu, Z. & Thirumalai, D. (2012) Denaturant-dependent folding of GFP, Proc
Natl Acad Sci U S A. 109, 17832-8.
54. Pedelacq, J. D., Cabantous, S., Tran, T., Terwilliger, T. C. & Waldo, G. S. (2006)
Engineering and characterization of a superfolder green fluorescent protein, Nat Biotechnol.
24, 79-88.
55. Chandler, S. A. & Benesch, J. L. (2018) Mass spectrometry beyond the native state, Curr
Opin Chem Biol. 42, 130-137.
56. Young, T. S., Ahmad, I., Yin, J. A. & Schultz, P. G. (2010) An enhanced system for
unnatural amino acid mutagenesis in E. coli, J Mol Biol. 395, 361-74.
57. Kabsch, W. (2010) Xds, Acta Crystallogr D Biol Crystallogr. 66, 125-32.
58. Evans, P. R. (2011) An introduction to data reduction: space-group determination,
scaling and intensity statistics, Acta Crystallogr D Biol Crystallogr. 67, 282-92.
59. McCoy, A. J., Grosse-Kunstleve, R. W., Adams, P. D., Winn, M. D., Storoni, L. C. &
Read, R. J. (2007) Phaser crystallographic software, J Appl Crystallogr. 40, 658-674.
60. Moriarty, N. W., Grosse-Kunstleve, R. W. & Adams, P. D. (2009) electronic Ligand
Builder and Optimization Workbench (eLBOW): a tool for ligand coordinate and restraint
generation, Acta Crystallogr D Biol Crystallogr. 65, 1074-80.
61. Emsley, P. & Cowtan, K. (2004) Coot: model-building tools for molecular graphics, Acta
Crystallogr D Biol Crystallogr. 60, 2126-32.
62. Afonine, P. V., Grosse-Kunstleve, R. W., Echols, N., Headd, J. J., Moriarty, N. W.,
Mustyakimov, M., Terwilliger, T. C., Urzhumtsev, A., Zwart, P. H. & Adams, P. D. (2012)
Towards automated crystallographic structure refinement with phenix.refine, Acta
Crystallogr D Biol Crystallogr. 68, 352-67.
63. Chen, V. B., Arendall, W. B., 3rd, Headd, J. J., Keedy, D. A., Immormino, R. M.,
Kapral, G. J., Murray, L. W., Richardson, J. S. & Richardson, D. C. (2010) MolProbity: all-
atom structure validation for macromolecular crystallography, Acta Crystallogr D Biol
Crystallogr. 66, 12-21.

64. Pettersen, E. F., Goddard, T. D., Huang, C. C., Couch, G. S., Greenblatt, D. M., Meng, E.
C. & Ferrin, T. E. (2004) UCSF Chimera--a visualization system for exploratory research and
analysis, J Comput Chem. 25, 1605-12.
65. van den Heuvel, R. H., van Duijn, E., Mazon, H., Synowsky, S. A., Lorenzen, K.,
Versluis, C., Brouns, S. J., Langridge, D., van der Oost, J., Hoyes, J. & Heck, A. J. (2006)
Improving the performance of a quadrupole time-of-flight instrument for macromolecular
mass spectrometry, Anal Chem. 78, 7473-83.
This article is protected by copyright. All rights reserved.

Tables
Table I: Data collection and refinement statistics
sfGFP66ONBY (pdb:6H01)
Data collection
Space group
P 41 21 2
Cell dimensions
a, b, c (Å)
155.3, 155.3, 162.0
90, 90, 90
)
Resolution (Å)
90.93 - 2.70 (3.34 – 2.70)
0.055 (0.609)
9.7
Rmerge
I
I
CC(1/2)
0.997 (0.546)
99.90 (99.87)
2.0 (2.0)
Completeness (%)
Multiplicity
Refinement
Resolution (Å)
No. reflections
Rwork / Rfree
49.32 - 2.70 (2.78-2.70)
54873 (5387)
0.201/ 0.239 (0.35/0.37)
No. atoms
Protein
7273
128
Ligand
Water
133
Average B-factors (Å2)
Protein
47.14
44.28
48.20
Chromophore
Water
R.m.s. deviations
Bond lengths (Å)
Bond angles ()
0.009
1.37
*Values in parentheses are for the highest-resolution shell.

This article is protected by copyright. All rights reserved.

Figure 1: SDS-PAGE and size-exclusion chromatography (SEC) analysis of sfGFP variants. A)
wt-sfGFP (green), sfGFP66ONBY (blue), sfGFP66ONBY-F145A (red) and decaged sfGFP66ONBY
(black) predominantly elute as monomers from SEC. B) Without heat treatment wt-sfGFP,
sfGFP66ONBY, sfGFP66ONBY-F145A and decaged sfGFP66ONBY migrate as dimers (54 kDa) in
the SDS-PAGE gel (5 – 8). In contrast, after heat treatment (1 – 4) proteins migrate as monomer
(27 kDa) (M = molecular weight marker, 1 + 5 = wt-sfGFP, 2 + 6 = sfGFP66ONBY, 3 + 7 =
sfGFP66ONBY-F145A, 4 + 8 = decaged sfGFP66ONBY).

Figure 2: Spectroscopic characterization of sfGFP66ONBY. A) Evolution of UV/Vis absorption
spectra of sfGFP66ONBY during 355-nm pulsed irradiation. Samples were irradiated for 0 min
(black), 1 min (purple), 4 min (red), 9 min (orange), 30 min (blue). B) UV/Vis absorption spectra of
sfGFP66ONBY before (black) and after photolysis (blue) and wt-sfGFP (green). Spectrum of sfGFP
(green) was rescaled by a factor of 1.8 for better comparison. C) Fluorescence spectra of photolyzed
sfGFP66ONBY (blue) and wt-sfGFP (green) after pulsed excitation at 355 nm. For B) and C) samples
were irradiated with 355-nm pulses at 1 kHz and 35 mW average power for 30 min. D)
Photochemical reaction of the decaging of the non-fluorescent sfGFP66ONBY by UV photolysis.
This article is protected by copyright. All rights reserved.

Figure 3: Crystal structure of the domain-swapped sfGFP66ONBY. A) Structure of domain-
swapped sfGFP66ONBY dimer with each chain colored in blue and green, respectively (pdb: 6H01).
B) sfGFP66ONBY was purified and crystallized to yield crystals that diffracted to 2.7 Å. The 2Fo-Fc
map contoured at 1 r.m.s.d. clearly shows the presence of the ortho-nitrobenzyl group and the
cyclized backbone forming the chromophore. C) Omit map of the open interface and hinge loop
region of the domain-swapped dimer contoured at 1 r.m.s.d showing the continuous electron density
along the loop connecting the two domains of each chain. This open interface exists only in the
domain-swapped dimer, but not in the wild-type monomer or dimer.

Figure 4: Structural comparison of sfGFP66ONBY domain-swapped structure (cyan, (pdb:
6H01)) with wild-type sfGFP dimer structure (green, pdb:2b3q). A) One monomer of the dimer is
superimposed. The different orientation of the second monomer in the domain-swapped structure and
the corresponding lack of dimer interface are clearly visible. The second dimer is shifted by 19 Å. B)
Chromophores are shown as sticks. Tyr66 of the chromophore is superimposed. Comparison of the
orientation of Phe145 in the sfGFP66ONBY structure (cyan) with Phe145 in the wt-sfGFP structure
(green) indicates that a clear steric clash between Phe145 in its wt-sfGFP conformation and the ortho-
nitrobenzyl-group of ONBY at position 66 would occur, thus causing the structural rearrangement.
This article is protected by copyright. All rights reserved.

Figure 5: Postulated folding mechanism of wt-GFP and its implications for the formation of
domain-swapped sfGFP66ONBY (pdb: 6H01). A) Topological depiction of wt-GFP. The N-
terminal β-strands are represented in blue, the helix containing the chromophore in the centre of the β-
strand barrel is in green, the central three β-strands are in grey, and the C-terminal β-strands are in
red. B) The postulated folding of wt-GFP in the EQL pathway occurs in a two-state manner via
equilibrium intermediates. The EQL intermediate comprises two domains: a structured N-terminus (β1
– β6) and structured C-terminus (β7 – β11 in red) connected by a loop. The EQL intermediate is able to
fold into the correct monomeric species. C) In the sfGFP66ONBY protein, steric clashes of Phe145
with the ortho-nitrobenzyl group interferes with folding and joining of C-terminal β7 – β11 (red) to the
structured β1 – β6 at the N-terminus and prevent folding of the correct monomer. Hence, the
intermediate of the EQL pathway can only fold into the unusual domain-swapped sfGFP66ONBY
dimer.

Figure 6: Native mass spectrometry analysis of sfGFP variants. A) Native mass spectra indicate
the presence of monomeric and dimeric species for wt-sfGFP (green), sfGFP66ONBY (blue), decaged
sfGFP66ONBY (black), and sfGFP66ONBY-F145A (red). Main peaks are annotated with their
charge states; positions of equal charge states are depicted in grey. B) Masses from GFP monomers
and dimers were determined from at least three MS or MS/MS measurements and listed with standard
deviations and the average full width half maximum (FWHM) value as a measure of protein
heterogeneity and desolvation (all in Da).
This article is protected by copyright. All rights reserved.