Cleavage of a carbon–fluorine bond by an engineered cysteine dioxygenase

Article abstract of DOI:10.1038/s41589-018-0085-5

Cysteine dioxygenase (CDO) plays an essential role in sulfur metabolism by regulating homeostatic levels of cysteine. Human CDO contains a post-translationally generated Cys93–Tyr157 cross-linked cofactor. Here, we investigated this Cys–Tyr cross-linking by incorporating unnatural tyrosines in place of Tyr157 via a genetic method. The catalytically active variants were obtained with a thioether bond between Cys93 and the halogen-substituted Tyr157, and we determined the crystal structures of both wild-type and engineered CDO variants in the purely uncross-linked form and with a mature cofactor. Along with mass spectrometry and ¹⁹F NMR, these data indicated that the enzyme could catalyze oxidative C–F or C–Cl bond cleavage, resulting in a substantial conformational change of both Cys93 and Tyr157 during cofactor assembly. These findings provide insights into the mechanism of Cys–Tyr cofactor biogenesis and may aid the development of bioinspired aromatic carbon–halogen bond activation.

Full text of DOI:10.1038/s41589-018-0085-5

Articles
https://doi.org/10.1038/s41589-018-0085-5
Cleavage of a carbon–fluorine bond by an
engineered cysteine dioxygenase
Jiasong Li1, Wendell P. Griffith1, Ian Davis1, Inchul Shin1, Jiangyun Wang2, Fahui Li2, Yifan Wang1,
Daniel J. Wherrittꢀ ꢀ1 and Aimin Liuꢀ ꢀ1*
Cysteine dioxygenase (CDO) plays an essential role in sulfur metabolism by regulating homeostatic levels of cysteine. Human
CDO contains a post-translationally generated Cys93–Tyr157 cross-linked cofactor. Here, we investigated this Cys–Tyr cross-
linking by incorporating unnatural tyrosines in place of Tyr157 via a genetic method. The catalytically active variants were
obtained with a thioether bond between Cys93 and the halogen-substituted Tyr157, and we determined the crystal structures
of both wild-type and engineered CDO variants in the purely uncross-linked form and with a mature cofactor. Along with mass
spectrometry and 19F NMR, these data indicated that the enzyme could catalyze oxidative C–F or C–Cl bond cleavage, resulting
in a substantial conformational change of both Cys93 and Tyr157 during cofactor assembly. These findings provide insights
into the mechanism of Cys–Tyr cofactor biogenesis and may aid the development of bioinspired aromatic carbon–halogen bond
activation.
ost-translational modification increases the number of pos-
When recombinant CDO is isolated, cross-linked and uncross-
sible molecular variations of proteins in living cells by several linked forms are both present. Separation of the two forms can be
orders of magnitude and hence is known as ‘the chemistry of achieved in the denatured state by SDS–PAGE; two distinct bands
P
proteome diversifications’1,2. Whereas reversible protein modifica- are observable, and the faster-migrating band is identified as the
tions play central roles in cellular regulation, unidirectional post- cross-linked form15. The cross-linking reaction takes place during
translational modifications generate novel cofactors to enhance catalysis as an autocatalytic reaction because of an uncoupled oxy-
or expand the catalytic repertoire of enzymes3–5. Irreversible post- gen activation at the nonheme iron center of the enzyme, where O2
translational modifications have become a fundamental challenge activation is not linked to the oxidation of the substrate cysteine but
for chemists in predicting the structures and functions of proteins. rather to its own residues. It takes an unclarified number, likely hun-
A protein-derived Cys–Tyr cofactor has recently been found in dreds to thousands, of turnovers to obtain a fully mature enzyme15.
mammalian cysteine dioxygenase (CDO, EC 1.13.11.20)6,7. Such Depending on the substrate concentrations (l-cysteine and O2), pH,
a cofactor is only known in a few proteins8–12. CDO is a nonheme and temperature, the time needed to reach the fully cross-linked
iron enzyme that catalyzes the conversion of l-cysteine to cysteine form in solution varies. Once the Cys–Tyr cofactor is generated, the
sulfinic acid (CSA; Supplementary Fig. 1). The product, CSA, is mature form of CDO performs the coupled dioxygenation reaction
ultimately catabolized to taurine and sulfate13. The Cys–Tyr cofac- more efficiently with an increased capability to metabolize high lev-
tor contains a thioether (C–S) bond between the side chains of a els of cysteine.
cysteine residue (Cys93; human CDO numbering) and a tyrosine
To date, understanding of the mechanism of Cys–Tyr cross-link
residue (Tyr157)14. The presence of such a Cys–Tyr cofactor boosts biogenesis in CDO has stagnated because of several technical chal-
the catalytic efficiency of CDO by one order of magnitude15 owing lenges in addition to the single-turnover nature discussed above.
to the concerted redox action of the metal ion and the protein- First, the uncross-linked form of active CDO, the starting material
derived cofactor.
for studying cofactor biogenesis, is difficult to isolate, though it is
CDO occupies a central position in biological sulfur metabolism; observable in denatured gels. Thus, little spectroscopic characteriza-
its enzymatic activity is crucial for maintaining proper cysteine lev- tion has been done, and no structural information is currently avail-
els for protein synthesis and for initiating cysteine catabolism. It has able for the uncross-linked form. Second, traditional site-directed
been established that the metabolites derived from cysteine and the mutagenesis approaches can support the catalytic importance of the
ratio of cysteine to sulfate and taurine exert a wide variety of physi- cofactor, but mutation of Cys93 or Tyr157 leads to disruption of the
ological effects in cells, including energy balance, fat metabolism16, formation of the Cys–Tyr cross-link and the inability of the protein
autoimmune and neurological conditions6,17, oxidative stress18, O2 variants to generate a mature enzyme active site, thus offering lim-
sensing and hypoxia19–21, and production of the signaling molecule ited information15,26–28
H2S22. Reduced activity of CDO results in elevated serum levels of
.
In the present work, we surmounted these challenges by employ-
neuroexcitatory cysteine, which has been associated with rheuma- ing a genetic method for site-specific incorporation of unnatural
toid arthritis23, breast cancer24, and several neurological disease amino acids29 into human CDO (hCDO) at position 157 of the
states, such as Alzheimer’s and Parkinson’s diseases25. In mouse protein sequence. The resulting protein variants were character-
models, knockout of CDO results in increased activity of the alter- ized by enzymatic activity and biophysical techniques to explore
native desulfhydration pathways, causing increased H2S production the mechanism of cofactor biogenesis in CDO. An oxidative car-
and cytotoxicity22.
bon–fluorine (C–F) or carbon–chlorine (C–Cl) bond cleavage was
1Department of Chemistry, University of Texas at San Antonio, San Antonio, TX, USA. 2Institute of Biophysics, Chinese Academy of Sciences, Beijing, P.R.
China. *e-mail: Feradical@utsa.edu
NAture ChemICAL BIoLoGY | www.nature.com/naturechemicalbiology
© 2018 Nature America Inc., part of Springer Nature. All rights reserved.

NATuRe CHeMiCAl BiOlOgy
Articles
COOH
COOH
a
b
COOH
NH2
COOH
NH2
NH2
NH2
Cl
Cys-Tyr
F
F
Cl
Cl
OH
OH
Tyr
OH
OH
Arg60
F2-Tyr (2)
Cl2-Tyr (3)
Cl-Tyr (1)
c
30
Fe
His88
CYS
U
C
Tyr58
20
His86
kDa
His140
Marker
1
2
3
4
Fig. 1 | Crystal structures of human CDo and incorporation of unnatural amino acids into the catalytic active site tyrosine. a, Overlay of the active
sites of the ligand-free (gray) and substrate (CYS)-bound complex (green) crystal structures of human CDO (PDB entries 6BGF and 6BGM). Arg60
presents two conformations in the ligand-free structure. The omit Fo − Fc electron densities for the water ligands (WAT) and the l-cysteine are shown in
Supplementary Fig. 2a,b. b, Native tyrosine and halogen-substituted unnatural tyrosine analogs used in this study; i.e., l-Tyr, 3-Cl-l-Tyr (Cl-Tyr, 1), 3,5-F2-l-
Tyr (F2-Tyr, 2), and 3,5-Cl2-l-Tyr (Cl2-Tyr, 3). c, SDS–PAGE shows two bands, with the slower moving band corresponding to the uncross-linked enzyme (U)
and the faster moving band to the mature CDO with a cross-linked cofactor (C). The lanes from left to right of the molecular weight marker are as follows:
(1) wild-type CDO; (2) Cl-Tyr157 CDO; (3) F2-Tyr157 CDO; and (4) Cl2-Tyr157 CDO. The full gel image is displayed in Supplementary Fig. 5b,c.
observed during cofactor formation in the dihalogen-substituted CDO in complex with l-cysteine (Fig. 1a) shows that the substrate
enzyme variants, indicating that the nonheme iron center gener- l-cysteine displaces two water molecules, with both the amino and
ates an oxidant that is strong enough to oxidize the adjacent second thiol groups coordinating to the ferrous ion, and the remaining
coordination sphere residues for cross-linking whether a stronger water ligand is only weakly associated to the Fe ion at a distance of
or weaker bond is present or the chemical properties of the tyrosine 2.6Å and is hydrogen bonded with the hydroxyl group of the Cys–
are altered. When the engineered protein with a monochlorine- Tyr cofactor. The substrate l-cysteine forms a salt bridge with Arg60
substituted tyrosine was exploited, the structural and MS analyses and is hydrogen bonded to Tyr58 and the Cys–Tyr cofactor in addi-
showed that steric hindrance is the next factor governing selectivity tion to its bidentate coordination to the Fe(ii) ion.
for generating a new C–S bond.
Genetic incorporation of unnatural amino acids. To investigate the
results
biogenesis mechanism in CDO, we genetically substituted Tyr157
Structures of ligand-free and substrate-bound human CDO. To with three unnatural amino acids: 3-chloro-l-tyrosine (Cl-Tyr, 1),
investigate the formation and key role of the Cys93–Tyr157 cross- 3,5-difluoro-l-tyrosine (F2-Tyr, 2), and 3,5-dichloro-l-tyrosine (Cl2-
link, we determined the crystal structures of mature hCDO both Tyr, 3) (Fig. 1b). The advantage of this approach is that it allows less
in the ligand-free status and in complex with l-cysteine (Fig. 1a) perturbation to the enzyme active site structure than site-directed
refined to 2.20 and 1.80Å resolution, respectively (Supplementary mutagenesis while specifically and chemically altering the amino acid
Table 1). Prior to this work, CDO research was primarily focused residue of interest. The genetic incorporation of unnatural amino
on the mouse and rat gene products (mouse and rat CDOs have acids was site specific, and all six other tyrosine residues in hCDO
identical protein sequences). Only one human CDO structure was were left unaltered. The CDO protein isolated was 100% modified
reported at 2.7Å resolution14, and it is ambiguous whether this is at position 157 with specific tyrosine analogs. This efficient genetic
a ligand-bound structure30. We determined that the ligand-free incorporation of unnatural tyrosines into a specific tyrosine residue
structure belongs to the P65 space group, with each asymmetric unit in a protein has been previously used for the study of other systems
containing one CDO molecule. The electron density of the enzyme including a nonheme Fe enzyme, ribonucleotide reductase33–35
active site shows that the obtained structures are 100% cross-linked, Of the unnatural tyrosine compounds available, F2-Tyr is the
.
with Cys93 and Tyr157 covalently connected via a thioether bond. most interesting for this study because fluorine is highly electro-
The overall structure is analogous to that of the mouse and rat negative and small in size. Thus, we expect F2-Tyr157 to mimic the
CDOs described previously7. Three water ligands were found to native human enzyme with minimal structural perturbation while
coordinate with the Fe ion in all the resting-state wild-type (WT) specifically attenuating the hydrogen bonding potential of the tyro-
CDO structures (Supplementary Fig. 2) that we obtained at vari- syl hydroxyl group because of the strong electron-withdrawing
ous resolutions, consistent with the published spectroscopic data properties of fluorine. To selectively incorporate F2-Tyr into hCDO,
from mouse and rat CDOs indicating that the iron ion exhibits a we used a mutant of a Methanocaldococcus jannaschii tyrosyl amber
pseudo-octahedral coordination geometry7,31. The first crystal suppressor tRNA(MjtRNATyrCUA)/F2-TyrRS (Y32R, L65Y, H70G,
structure, determined from Ni2+-reconstituted mouse CDO, shows F108N, Q109C, D158N, L162S) pair to target the TAG codon, as
three water ligands6, whereas the rat CDO structure and Mössbauer previously reported36. We expressed the hCDO variant bearing F2-
spectroscopy show a tetracoordinate iron center with only one or Tyr at position 157 in Escherichia coli using pEVOL-F2-TyrRS and
two water ligand(s)7,31,32. The number of water ligands is relevant to PV16-hCDO157TAG plasmids. Likewise, we also genetically incor-
the mechanistic understanding, because the first step of the catalytic porated Cl-Tyr and Cl2-Tyr into position 157 of hCDO by a similar
cycle is substrate ligation to the iron ion. The crystal structure of method using pEVOL-Cl-TyrRS and pEVOL-Cl2-TyrRS37.
NAture ChemICAL BIoLoGY | www.nature.com/naturechemicalbiology
© 2018 Nature America Inc., part of Springer Nature. All rights reserved.

NATuRe CHeMiCAl BiOlOgy
Articles
The ESI mass spectra of the WT, Cl-Tyr157, F2-Tyr157, consistent with the cross-linked peptide containing only one fluo-
and Cl2-Tyr157 variants under denaturing solution conditions rine atom (Fig. 2b).
showed an ensemble of charge states ranging from [M+8H+]+8
To confirm the location of the cross-link and displacement of a
to [M+33H+]+33. Deconvolution of these mass spectra results in fluorine atom upon cross-linking, tandem MS using LIFT technol-
experimentally determined masses for these proteins of 22,913.20, ogy39 on a MALDI–TOF/TOF mass spectrometer was conducted.
22,947.11, 22,949.16, and 22982.07Da for WT, Cl-Tyr157, F2-Tyr157 An illustration of the cross-linked peptide indicating the major
and Cl2-Tyr157 CDO, respectively (Supplementary Fig. 3). These fragment ions detected is depicted in Fig. 2a. Based on this sche-
measured mass values are consistent with the predicted masses of matic, any N-terminal fragment (b-ion, green colored) before His92
22,913.21, 22,947.17, 22,949.19 and 22,982.13Da, respectively, from of peptide A (residues Gly80 through His92) and any C-terminal
protein sequence. These masses of Cl-Tyr157, F2-Tyr157, and Cl2- ion (y-ion, orange colored) of peptide B (residues Ser158 through
Tyr157 reflect a+34, +36 or+69Da difference compared to WT Phe167) after Ser158 would have the same mass (or m/z value)
CDO, as would be expected for the substitution of one or more in the cross-linked WT and F2-Tyr157 CDO peptides. The MS
halogen atoms for hydrogen at Tyr157.
data showed that this is indeed the case. Our data also showed an
increase of +18Da in F2-Tyr157 relative to the WT peptide for any
The unnatural amino acids’ function in human CDO. The fluo- fragment containing Tyr157 (Fig. 2c). These results unambiguously
rine- and chlorine-bearing CDO variants were catalytically active, confirmed the cross-link between Cys93 and Tyr157 and estab-
with the as-isolated F2-Tyr157 variant having about 10% activity lished that Tyr157 contains one, and only one, fluorine atom in
and the Cl/Cl2-Tyr157 variants having 2% activity compared to that the mature form, as the other is displaced during the formation of
observed for WT CDO (Supplementary Fig. 4 and Supplementary the cross-link. These MS analyses further suggested that the cofac-
Table 2). The question motivating the next step of this study is tor formed in F2-Tyr157 CDO is analogous to the Cys–Tyr cofac-
whether or not the cross-linked Cys–Tyr cofactor could be formed tor found in the native protein. The fluorine atom released during
in the F2-Tyr157 and Cl2-Tyr157 CDO variants. On the basis of bond oxidative conversion was detected by 19F NMR (Fig. 2d). Because
dissociation energies, the formation of the C–S bond needed for fluorine release should be stoichiometric with the protein, more
cofactor assembly was expected to be hindered by the fluorine (but than 10,000 scans over the course of 20h on a 500MHz NMR spec-
not chlorine) substitution. Because the C–F bond is often thought to trometer were collected. Spiking in aqueous KF confirmed that
be one of the strongest single bonds (second only to the C–Si bond) the signal for the released fluorine is F– (Fig. 2d). Such a fluoride
between formally neutral atoms38, one may expect that F2-Tyr157 signal is not present in the 19F NMR spectrum of F2-Tyr157 CDO
substitution would greatly hinder, or prevent, the formation of the (Supplementary Fig. 7).
thioether bond for cofactor assembly. Oxidative formation of the
cross-link with F2-Tyr157 would require breaking a C–F bond and Achieving cofactor biogenesis in crystallo. We determined the
an S–H bond and forming a new C–S bond. Without knowing the crystal structures of F2-Tyr157 hCDO and Cl2-Tyr157 hCDO
mechanism of the cofactor formation, it was difficult to predict a (Fig. 3, Supplementary Fig. 8 and Supplementary Table 1) in the
priori whether or not the cross-linked Cys–Tyr cofactor would still absence of and in complex with l-cysteine. When F2-Tyr157 crystals
be formed in the halogen-substituted CDO variants.
were grown anaerobically, only the uncross-linked form crystallizes,
The as-isolated WT, Cl-Tyr, F2-Tyr157, and Cl2-Tyr157 CDO and we thus obtained the long-sought, 100% uncross-linked struc-
variants all exhibited two bands by SDS–PAGE, indicating that ture of CDO, refined to 2.40Å resolution (Fig. 3a). Cys93 exhibits
the substitutions were unable to abrogate cross-link formation two alternative conformations in the uncross-linked CDO struc-
(Fig. 1c). On the basis of gel analysis, we estimated that 55%, 54%, ture, with occupancy in a 73:27 ratio, and the sulfhydryl group in
41%, and 16% cross-links were present in these as-isolated CDO the major conformation points away from the iron ion. The catalytic
samples, respectively. Further processing of the as-isolated halogen- Fe(ii) ion is ligated by the Nϵ atoms of His86, His88, and His140 and
substituted CDO variants with l-cysteine and O2 led to a mainly three water molecules. The substrate-bound, uncross-linked crystal
cross-linked enzyme concomitant with the weakening of the slower- structure was refined to 2.10Å resolution (Fig. 3b). A substantial
migrating band (Supplementary Fig. 5). Upon closer analysis of the rotation of F2-Tyr157 is noted (Supplementary Fig. 9a). As shown
ESI mass spectra, we found that F2-Tyr157 CDO in the+10 charge in the wild-type structure, three water ligands are bound to the Fe
state and below contained an additional peak corresponding to ion in all ligand-free structures of the CDO variants, and only one
20Da lower mass (Supplementary Fig. 6), which fits the expected water ligand remains associated with the iron ion in the substrate-
mass of the cross-linked form of F2-Tyr157 CDO. After cross-link bound structure (Supplementary Fig. 2). The B-factor of the water
formation, the protein is more resistant to acid and organic solvent ligand changes from 15.5 in the ligand-free structure to 37.8 in the
denaturation, and thus is more compact, so it carries less protonic substrate-bound structure, indicating that the water ligand becomes
charge and appears at lower charge states.
flexible when l-cysteine binds to the metal ion.
We also obtained a crystal structure with both cross-linked and
Oxidative C–F bond cleavage during cofactor biogenesis. The gel uncross-linked forms in the same crystal structure from the as-
analysis and the ESI data suggested that the Cys–Tyr cofactor was isolated, ligand‐ free F2-Tyr157 hCDO crystalized under aerobic
formed in the CDO variants with unnatural amino acids. Next, we conditions and refined it to 1.80Å resolution. The cross-linked and
considered whether they do so by breaking the C–F or C–Cl bond uncross-linked forms of the cofactor were present in a 53:47 ratio in
or by forming a differently configured Cys-Tyr, i.e., linking the thiol this structure based on their respective occupancy (Supplementary
group to C2 or C6. To rigorously determine the nature and posi- Fig. 8a,b). The cross-linked conformation of Cys93 showed a con-
tion of the cross-link in F2-Tyr157 CDO, an MS/MS analysis was necting density to F2-Tyr157, indicating that the cross-linked
conducted (Fig. 2a).
cofactor is present. Notably, Cys93 also exhibited two alternative
After conversion to the all-cross-link form, the WT and F2- conformations in this structure. In the uncross-linked conforma-
Tyr157 CDO proteins were digested by three enzymes: trypsin, tion, the sulfhydryl group of Cys93 points away from the iron cen-
GluC, and chymotrypsin. As expected, the cross-linked peptide ter, whereas in the cross-linked form the sulfur atom of Cys93 turns
from WT hCDO was detected at m/z 3,226.4 (Fig. 2b). In samples toward the iron center by nearly 3Å. The F2-Tyr157 ring also rotated
of F2-Tyr157 CDO, no peak was detected around m/z 3,262 cor- about 46 degrees along the axis through Cβ and the OH group, with
responding to a cross-linked peptide containing two fluorine a nearly 1-Å shift toward Cys93, allowing C–S bond formation
atoms on Tyr157. Instead, a peak at m/z 3,244.5 was found, which is (Supplementary Fig. 9b). We also found a similar phenomenon in
NAture ChemICAL BIoLoGY | www.nature.com/naturechemicalbiology
© 2018 Nature America Inc., part of Springer Nature. All rights reserved.

NATuRe CHeMiCAl BiOlOgy
Articles
a
b
3,244.5
3,226.4
800
600
400
200
O
b2 b3 b4 b5 b6 b7 b8 b9 b10b11b12 b13
H
N
1,603.7
F
Peptide A
y13 y12y11y10y9 y8 y7 y6 y5 y4 y3 y2
S
1,622.7
OH
F
Peptide B
S P P F D T C* H A F
H L
N
H
y′9
O
3,180
3,220
3,260
m/z
c
1.2
y7
Peptide A
y′9
*
y4
*
Intact
cross-linked
peptide
Peptide B
0.8
Peptide B*
*
y13
y12
y5
y3
y2
y6
0.4
y8
*
b11
*
*
y9
*
*
b8
*
b13
*
b10
b9
y10 y11
b7
b12
*
b6
b3
b5
b2
b4
*
*
0.0
500
1,000
1,500
2,000
2,500
3,000
m/z
d
OH
OH
19F NMR
S
SH
F
F
F
[O]
-F-, -H+
Reaction mixture
–115
–110
–120
–125
–130
–135
–140
+ aqueous KF
–110
–115
–120
–125
19F (p.p.m.)
–130
–135
–140
Fig. 2 | mS/mS spectra of cross-linked peptides of Wt and F2-tyr157 CDo proteins and 19F Nmr detection of leaving fluoride. a, Proposed cross-link
of peptides Gly80–Phe94 and His155–Phe167 through Cys93 and F2-Tyr157. b, MS spectrum of the cross-linked peptide Gly80–Phe94/His155–Phe167,
including one carboxyamidomethyl cysteine at m/z 3,226.4 for WT and one at m/z 3,244.5 for F2-Tyr157 CDO. c, Collision-induced dissociation (CID)–
MS/MS spectrum of the precursor ions at m/z 3,226.4 and m/z 3,244.5. d, In situ detection of fluoride during cofactor biogenesis in F2-Tyr157 CDO.
19F NMR spectra of the fluoride ion in solution from the cross-linking reaction of F2-Tyr157 CDO after acid precipitation of the protein. Top, 19F NMR
spectrum after the conversion of F2-Tyr157 CDO to the all-cross-linked form after 12,480 transients. Bottom, 19F NMR spectrum after spiking in aqueous
KF to confirm that the signal is F– after 128 transients. Both spectra are referenced to internal trifluoroacetic acid. MS experiments were repeated one time,
and 19F NMR experiments were repeated two times with similar results. The F2-Tyr157 CDO was also recorded once by 19F NMR before the autocatalytic
processing reaction (a negative control), and the resulting spectrum is shown in Supplementary Fig. 7. The asterisks in a and c indicate fragment peaks in
the crosslinked peptides from F2 -Tyr157 CDO that are +18 relative to the wild-type fragments containing the tyrosine-157 and only one fluorine.
the crystal structure from the as-isolated Cl2-Tyr157 CDO crystal- for 5h, the single crystals of F2-Tyr157 hCDO showed a fully
lized under aerobic conditions (Supplementary Figs. 8c,d and 9c).
mature, cysteine-bound active site with 100% cross-link between
Taking advantage of the uncross-linked crystals, we subse- Cys93 and Tyr157 (Fig. 3c), and the structure was refined to 1.95Å
quently investigated the possible uncoupled reaction in crystallo resolution (Supplementary Table 1). An overlay of the uncross-
that leads to the oxidation of the second coordination sphere resi- linked F2-Tyr157 CDO structures with and without a substrate cys-
dues and the formation of the cofactor. After reacting the F2-Tyr157 teine bound at the active site is shown in Fig. 3d, and an overlay
variant crystals with l-cysteine (100mM) under aerobic conditions of the cross-linked and uncross-linked F2-Tyr157 CDO structures
NAture ChemICAL BIoLoGY | www.nature.com/naturechemicalbiology
© 2018 Nature America Inc., part of Springer Nature. All rights reserved.

NATuRe CHeMiCAl BiOlOgy
Articles
a
b
c
F2-Tyr157
F2-Tyr157
F2-Tyr157
Cys93
2.3 Å
Cys93
Cys93
His88
2.2 Å
2.9 Å
3.8 Å
2.7 Å
3.0 Å
2.6 Å
2.7 Å
2.6 Å
3.2 Å
3.4 Å
His88
His88
CYS
His86
His86
His86
His140
His140
His140
d
e
F2-Tyr157
F2-Tyr157
Cys93
Arg60
Arg60
WAT
WAT
His88
CYS
His88
CYS
Tyr58
Tyr58
His86
His86
His140
His140
Fig. 3 | Crystal structures of F2-tyr157 CDo. 2Fo − Fc electron density was contoured at 1.2 σrms, with potential hydrogen bond interactions (broken lines)
shown. a, The active site of the 100% uncross-linked F2-Tyr157 CDO structure in the substrate-free form. b, The active site of the 100% uncross-linked F2-
Tyr157 CDO structure in the substrate-bound form. c, The active site of the mature F2-Tyr157 CDO structure in the substrate-bound form. d, Overlaid active
site of the 100% uncross-linked F2-Tyr157 CDO structure in the substrate-free (2.40ꢀÅ resolution, gray) and substrate-bound (2.10ꢀÅ resolution, green)
forms. e, Superimposition of the substrate-bound mature F2-Tyr CDO (1.95ꢀÅ resolution, pink) and 100% uncross-linked structures (green). The
l-cysteine substrate is labeled as CYS. The omit Fo − Fc electron densities for the water ligands (WAT) and the l-cysteine are shown in Supplementary Fig. 2.
in complex with l-cysteine is shown in Fig. 3e and Supplementary a mature cofactor and bound l-cysteine substrate at the nonheme
Fig. 9d. A substantial rotation of F2-Tyr157 is noted between the Fe center, only a single conformation of the cofactor is present with
uncross-linked and mature forms of the enzyme. Cys93 shows only 100% occupancy of the chlorine (Fig. 4b). As corroborated by the
one conformation in the mature form40 but two conformations in MS analysis (Supplementary Fig. 10), the C–H bond is cleaved but
each of the uncross-linked structures. Likewise, in crystallo cofactor the C–Cl bond is not. Therefore, we conclude that the large size of
maturation was also observed in Cl2-Tyr157 CDO protein crystals.
To discern which bond would be cleaved when both C–H and
chlorine makes the C–Cl bond less accessible for oxidation.
C–Cl bonds are available, we analyzed the monochlorine-substi- Discussion
tuted Cl-Tyr157 variant by MS and determined its crystal structures The success in obtaining a purely uncross-linked crystal structure of
from the purely uncross-linked and the fully cross-linked forms CDO is an important advancement toward unfolding the cofactor
(Supplementary Table 1). In the as-isolated form, the uncross- biogenesis mechanism. Previously, the uncross-linked CDO struc-
linked and cross-linked portions are present in a similar occupancy. ture was acquired from the C93A mutant28. The key difference is
Figure 4a shows the overlay of the active site of the ligand-free that the uncross-linked structure we report is a fully capable protein
Cl-Tyr157 CDO structures in the cross-linked and uncross-linked for cofactor assembly, whereas the mutant is a permanently disabled
forms. The most revealing structural information for mechanistic variant in terms of cofactor biogenesis. In both the substrate-free
understanding is the uncross-linked form. The crystal structure and substrate-bound C93A mutant structures, a buffer-derived
shows in the uncross-linked form that equal populations of the C–H chloride ion is ligated to the iron ion, but the chloride ligation is
and C–Cl bonds of the Tyr157 are positioned toward Cys93 and not seen in our structures. Our purely uncross-linked structure
the iron ion, indicating that the C–H and C–Cl bonds have similar revealed important details for mechanistic understanding that were
opportunity for oxidative cleavage. In the fully processed form with not possible to detect in the mutant structures. Cys93 shows two
NAture ChemICAL BIoLoGY | www.nature.com/naturechemicalbiology
© 2018 Nature America Inc., part of Springer Nature. All rights reserved.

NATuRe CHeMiCAl BiOlOgy
Articles
a
b
Cl-Tyr157
Cl-Tyr157
Cys93
Cys93
WAT
WAT
WAT
His88
His88
CYS
WAT
His86
His86
His140
His140
Fig. 4 | Crystal structures of Cl-tyr157 CDo. The omit Fo − Fc electron densities (green) of the WT CDO and mutants contoured at 3 σ (green) and
6 σ (purple), respectively. a, The active site of the Cl-Tyr157 CDO structure in the cross-linked and uncross-linked forms in the absence of the substrate.
b, The active site of the substrate-bound Cl-Tyr157 CDO structure in the cross-linked form. CYS represents the l-cysteine substrate bound at the
nonheme iron center.
conformations before the cross-link reaction, and the sulfur atom of Cl-Tyr157 are equally populated toward Cys93 and the Fe ion.
of Cys93 undergoes a nearly 3-Å rotation toward the iron center However, it is the stronger C–H bond that is cleaved and the
during autocatalytic cofactor biogenesis. The findings presented in weaker C–Cl bond that is retained in the mature Cl-Tyr157 CDO
this work may be useful in considering the mechanism of the Cys– variant. These results suggest that the size of chlorine makes
Tyr biogenesis.
it more difficult for Cl-Tyr157 to position closer to a putative
The ability of CDO to catalyze aromatic C–F bond cleavage is Cys93 radical or an iron-bound oxidant relative to the fluori-
not a predictable result, as the energy of an aliphatic C–F bond nated tyrosine derivative. Because of the small size of fluorine,
(485kJmol−1) is substantially higher than that of the correspond- there is no tRNA synthetase available for producing monofluo-
ing aliphatic C–H bond (411kJmol−1)41. Because the bond energy rine-substituted tyrosine derivative. The heterogeneous substi-
of an aliphatic C–Cl bond (i.e., 327 kJmol−1) is lower than that of tution of chlorine and fluorine at the ring 3 and 5 positions of
a C–H bond, the cleavage of the C–Cl bond in Cl2-Tyr157 CDO tyrosine (3-chloro-5-fluoro tyrosine) is also inaccessible by the
is no surprise. However, it is an interesting observation that the genetic incorporation approach. Based on the current results, we
C–H bond is preferentially cleaved in mono-Cl-Tyr157 CDO even envision that the C–F bond would more likely be cleaved in a
though both C–H and C–Cl are equally populated toward Cys93 hypothetical 3-chloro-5-fluoro-Tyr157 CDO variant.
and the iron center.
The data presented in this work collectively indicate an O2-
It should be noted that the C–F bond in this case may not be dependent carbon−halogen bond cleavage by a protein-bound
as strong as those in an aliphatic molecule. However, the aromatic iron center under mild, physiologically relevant conditions. Even
C–F bond is expected to be stronger than that of the correspond- though the C–F bond is one of the strongest covalent bonds in
ing C–H bond. When a hydroxyl group is attached to the C4 ring organic chemistry, and stronger than the corresponding C–H
carbon of a tyrosine, it activates the C3 and C5 positions. We pro- bonds, CDO is able to cleave a C–F bond for the purpose of cofac-
pose that the C–F bond may be cleaved during a putative radical tor biogenesis. The cleavage of an aromatic C–F bond by nitro-
step in which Tyr157 is oxidized, either directly by the iron- gen-ligated nonheme iron model complexes has been described
bound oxidant or by a Cys93 radical (Supplementary Fig. 11). in two recent papers42,43. However, to our knowledge, the obser-
Once a tyrosyl radical is formed, it is well-known that the radical vation described here for CDO is the first example of oxidative
is delocalized to specific positions4, including the ring C3 and C–F bond cleavage by a protein-bound iron center. The activation
C5. Thus, the C–H or C–F bond activation in Tyr157 would be and functionalization of C–F bonds is an important area of inor-
easier than generally thought for a C–H or C–F bond in hydro- ganic, organic, and medicinal chemistry. In biological systems, the
carbons. This is especially true after a tyrosyl radical is formed only C–F bond cleavage previously known was the oxygen-inde-
during cofactor biogenesis.
pendent reactions catalyzed by fluoroacetate dehalogenase and
The cross-link formation is apparently less efficient for the BzCoA reductase44,45. This unexpected finding during the study of
Cl2-Tyr157 derivative than the F2-Tyr157 variant and the wild- CDO cofactor biogenesis reveals new insights into the power of
type protein. Because a C–Cl bond is less durable than a C–F the nonheme iron ion in a dioxygenase and sheds light onto the
bond, the less-efficient cross-linking of Cl2-Tyr157 CDO suggests interplay between the catalytic metal ion and the cofactor of CDO.
that there may be an additional factor governing bond cleavage. It further increases our expectation of the oxidizing power of non-
We hypothesize that the size of a chlorine atom adversely affects heme Fe enzymes. The oxidative C–F bond cleavage described in
access of the C–Cl bond. The cross-link formation in mono- this study broadens the range of fluorine chemistry and further
substituted Cl-Tyr157 CDO is as efficient as that in the native extends potential industrial applications of natural or engineered
enzyme, and in the uncross-linked structure the C–H and C–Cl Fe-dependent proteins.
NAture ChemICAL BIoLoGY | www.nature.com/naturechemicalbiology
© 2018 Nature America Inc., part of Springer Nature. All rights reserved.

NATuRe CHeMiCAl BiOlOgy
Articles
25. Heafeld, M. T. et al. Plasma cysteine and sulphate levels in patients with
motor neurone, Parkinson’s and Alzheimer’s disease. Neurosci. Lett. 110,
216–220 (1990).
26. Li, W., Blaesi, E. J., Pecore, M. D., Crowell, J. K. & Pierce, B. S. Second-sphere
interactions between the C93-Y157 cross-link and the substrate-bound Fe site
infuence the O2 coupling efciency in mouse cysteine dioxygenase.
Biochemistry 52, 9104–9119 (2013).
methods
Methods, including statements of data availability and any asso-
ciated accession codes and references, are available at https://doi.
org/10.1038/s41589-018-0085-5.
Received: 30 November 2017; Accepted: 4 May 2018;
Published: xx xx xxxx
27. Davies, C. G., Fellner, M., Tchesnokov, E. P., Wilbanks, S. M. & Jameson, G.
N. Te Cys-Tyr cross-link of cysteine dioxygenase changes the optimal pH of
the reaction without a structural change. Biochemistry 53, 7961–7968 (2014).
28. Driggers, C. M. et al. Structure-based insights into the role of the Cys-Tyr
crosslink and inhibitor recognition by mammalian cysteine dioxygenase.
J. Mol. Biol. 428, 3999–4012 (2016).
references
1. Walsh, C. T. Posttranslational Modifcation of Proteins: Expanding Nature’s
Inventory. (Roberts & CompanyPublishers, Greenwood Village, CO, 2006).
2. Walsh, C. T., Garneau-Tsodikova, S. & Gatto, G. J. Jr. Protein posttranslational
modifcations: the chemistry of proteome diversifcations. Angew. Chem. Int.
Ed. Engl. 44, 7342–7372 (2005).
29. Wang, L., Brock, A., Herberich, B. & Schultz, P. G. Expanding the genetic
code of Escherichia coli. Science 292, 498–500 (2001).
30. Driggers, C. M. et al. Cysteine dioxygenase structures from pH4 to 9:
consistent cys-persulfenate formation at intermediate pH and a Cys-bound
enzyme at higher pH. J. Mol. Biol. 425, 3121–3136 (2013).
31. Souness, R. J. et al. Mechanistic implications of persulfenate and persulfde
binding in the active site of cysteine dioxygenase. Biochemistry 52,
7606–7617 (2013).
3. Klinman, J. P. & Bonnot, F. Intrigues and intricacies of the biosynthetic
pathways for the enzymatic quinocofactors: PQQ, TTQ, CTQ, TPQ, and LTQ.
Chem. Rev. 114, 4343–4365 (2014).
4. Stubbe, J. & van Der Donk, W. A. Protein radicals in enzyme catalysis. Chem.
Rev. 98, 705–762 (1998).
5. Krebs, C., Bollinger, J. M. Jr. & Booker, S. J. Cyanobacterial alkane
biosynthesis further expands the catalytic repertoire of the ferritin-like
‘di-iron-carboxylate’ proteins. Curr. Opin. Chem. Biol. 15, 291–303 (2011).
6. McCoy, J. G. et al. Structure and mechanism of mouse cysteine dioxygenase.
Proc. Natl. Acad. Sci. USA 103, 3084–3089 (2006).
32. Tchesnokov, E. P. et al. An iron-oxygen intermediate formed during the
catalytic cycle of cysteine dioxygenase. Chem. Commun. (Camb.) 52,
8814–8817 (2016).
33. Oyala, P. H. et al. Biophysical characterization of fuorotyrosine probes
site-specifcally incorporated into enzymes: E. coli ribonucleotide reductase as
an example. J. Am. Chem. Soc. 138, 7951–7964 (2016).
7. Simmons, C. R. et al. Crystal structure of mammalian cysteine dioxygenase.
A novel mononuclear iron center for cysteine thiol oxidation. J. Biol. Chem.
281, 18723–18733 (2006).
34. Ravichandran, K. R. et al. Formal reduction potentials of difuorotyrosine and
trifuorotyrosine protein residues: defning the thermodynamics of multistep
radical transfer. J. Am. Chem. Soc. 139, 2994–3004 (2017).
35. Minnihan, E. C., Young, D. D., Schultz, P. G. & Stubbe, J. Incorporation of
fuorotyrosines into ribonucleotide reductase using an evolved, polyspecifc
aminoacyl-tRNA synthetase. J. Am. Chem. Soc. 133, 15942–15945 (2011).
36. Li, F. et al. A genetically encoded 19F NMR probe for tyrosine
phosphorylation. Angew. Chem. Int. Edn Engl. 52, 3958–3962 (2013).
37. Liu, X. et al. Signifcant expansion of fuorescent protein sensing ability
through the genetic incorporation of superior photo-induced electron-
transfer quenchers. J. Am. Chem. Soc. 136, 13094–13097 (2014).
38. O’Hagan, D. Understanding organofuorine chemistry. An introduction to the
C-F bond. Chem. Soc. Rev. 37, 308–319 (2008).
8. Whittaker, J. W. Free radical catalysis by galactose oxidase. Chem. Rev. 103,
2347–2363 (2003).
9. Cowley, R. E. et al. Structure of the reduced copper active site in
preprocessed galactose oxidase: ligand tuning for one-electron O2 activation
in cofactor biogenesis. J. Am. Chem. Soc. 138, 13219–13229 (2016).
10. Schnell, R., Sandalova, T., Hellman, U., Lindqvist, Y. & Schneider, G.
Siroheme- and [Fe4-S4]-dependent NirA from Mycobacterium tuberculosis is a
sulfte reductase with a covalent Cys-Tyr bond in the active site. J. Biol.
Chem. 280, 27319–27328 (2005).
11. Polyakov, K. M. et al. High-resolution structural analysis of a novel octaheme
cytochrome c nitrite reductase from the haloalkaliphilic bacterium
Tioalkalivibrio nitratireducens. J. Mol. Biol. 389, 846–862 (2009).
12. Hromada, S. E. et al. Protein oxidation involved in Cys-Tyr post-translational
modifcation. J. Inorg. Biochem. 176, 168–174 (2017).
13. Stipanuk, M. H., Ueki, I., Dominy, J. E. Jr., Simmons, C. R. & Hirschberger, L.
L. Cysteine dioxygenase: a robust system for regulation of cellular cysteine
levels. Amino Acids 37, 55–63 (2009).
14. Ye, S. et al. An insight into the mechanism of human cysteine dioxygenase.
Key roles of the thioether-bonded tyrosine-cysteine cofactor. J. Biol. Chem.
282, 3391–3402 (2007).
15. Dominy, J. E. Jr. et al. Synthesis of amino acid cofactor in cysteine
dioxygenase is regulated by substrate and represents a novel post-translational
regulation of activity. J. Biol. Chem. 283, 12188–12201 (2008).
16. Niewiadomski, J. et al. Efects of a block in cysteine catabolism on energy
balance and fat metabolism in mice. Ann. NY Acad. Sci. 1363,
99–115 (2016).
17. Gordon, C., Bradley, H., Waring, R. H. & Emery, P. Abnormal sulphur
oxidation in systemic lupus erythematosus. Lancet 339, 25–26 (1992).
18. Kwon, D. Y. et al. Impaired sulfur-amino acid metabolism and oxidative stress
in nonalcoholic fatty liver are alleviated by betaine supplementation in rats.
J. Nutr. 139, 63–68 (2009).
19. Olson, K. R. et al. Tiosulfate: a readily accessible source of hydrogen sulfde
in oxygen sensing. Am. J. Physiol. Regul. Integr. Comp. Physiol. 305,
R592–R603 (2013).
20. Weits, D. A. et al. Plant cysteine oxidases control the oxygen-dependent
branch of the N-end-rule pathway. Nat. Commun. 5, 3425 (2014).
21. White, M. D. et al. Plant cysteine oxidases are dioxygenases that directly
enable arginyl transferase-catalysed arginylation of N-end rule targets. Nat.
Commun. 8, 14690 (2017).
22. Ueki, I. et al. Knockout of the murine cysteine dioxygenase gene results in
severe impairment in ability to synthesize taurine and an increased
catabolism of cysteine to hydrogen sulfde. Am. J. Physiol. Endocrinol. Metab.
301, E668–E684 (2011).
23. Bradley, H. et al. Sulfate metabolism is abnormal in patients with rheumatoid
arthritis Confrmation by in vivo biochemical fndings. J. Rheumatol. 21,
1192–1196 (1994).
39. Suckau, D. et al. A novel MALDI LIFT-TOF/TOF mass spectrometer for
proteomics. Anal. Bioanal. Chem. 376, 952–965 (2003).
40. Fellner, M., Aloi, S., Tchesnokov, E. P., Wilbanks, S. M. & Jameson, G. N.
Substrate and pH-dependent kinetic profle of 3-mercaptopropionate
dioxygenase from Pseudomonas aeruginosa. Biochemistry 55,
1362–1371 (2016).
41. Cottrell, T.L. Te Strengths of Chemical Bonds. 2nd edn. (Butterworths
Scientifc, 1958).
42. Sahu, S. et al. Aromatic C-F hydroxylation by nonheme iron(IV)-oxo
complexes: Structural, spectroscopic, and mechanistic Investigations.
J. Am. Chem. Soc. 138, 12791–12802 (2016).
43. Sahu, S. et al. Direct observation of a nonheme iron(IV)-oxo complex that
mediates aromatic C-F hydroxylation. J. Am. Chem. Soc. 136,
13542–13545 (2014).
44. Chan, P. W. Y., Yakunin, A. F., Edwards, E. A. & Pai, E. F. Mapping the
reaction coordinates of enzymatic defuorination. J. Am. Chem. Soc. 133,
7461–7468 (2011).
45. Tiedt, O. et al. ATP-dependent C-F bond cleavage allows the complete
degradation of 4-fuoroaromatics without oxygen. MBio 7,
e00990–16 (2016).
Acknowledgements
The work is supported in part by the National Institutes of Health grants GM107529,
GM108988, and MH107985, the National Science Foundation grant CHE-1623856, and
the Lutcher Brown Distinguished Chair Endowment fund (to A.L.). J.W. acknowledges
the support of the National Science Foundation of China grants (91527302, 31370016,
and U1532150). The mass spectrometry facility is sponsored by the National Institutes
of Health grant G12MD007591. The MALDI-TOF and NMR spectrometers are shared
instruments sponsored by the National Science Foundation under the award numbers
#1126708 and 1625963, respectively. X-ray synchrotron data were collected at the
beamlines of the Advanced Photon Source Section 19, Structural Biology Center user
program GUP-48198, Argonne National Laboratory and at the beamline BL9-2 of the
Stanford Synchrotron Radiation Lightsource (SSRL) under the user program #5B14,
SLAC National Accelerator Laboratory. The beamline staff scientists are acknowledged
for the assistance of the remote data collections. The Advanced Photon Source is a US
Department of Energy, Office of Science User Facility operated for the DOE Office of
Science by Argonne National Laboratory under Contract No. DE-AC02-06CH11357.
SSRL is supported by the US Department of Energy, Office of Science, Office of Basic
24. Jeschke, J. et al. Frequent inactivation of cysteine dioxygenase type 1
contributes to survival of breast cancer cells and resistance to anthracyclines.
Clin. Cancer Res. 19, 3201–3211 (2013).
NAture ChemICAL BIoLoGY | www.nature.com/naturechemicalbiology
© 2018 Nature America Inc., part of Springer Nature. All rights reserved.

NATuRe CHeMiCAl BiOlOgy
Articles
Energy Sciences under Contract No. DE-AC02-76SF00515 and by the National Institutes
of Health (P41GM103393). The contents of this publication are solely the responsibility
of the authors and do not necessarily represent the official views of National Institutes of
Health or National Science Foundation.
the manuscript together with J.L. All authors contributed to data analysis and to the
writing and editing of the manuscript.
Competing interests
The authors declare no competing interests.
Author contributions
Additional information
Genetic incorporation of unnatural amino acids was performed by J.L. (cloning, protein
expression and purification, and enzyme assays). J.W. and F.L. provided TyrRS. W.P.G.
and J.L. conducted mass spectrometry analyses. J.L. obtained all protein crystals,
collected X-ray diffraction data, and interpreted and refined the structural data together
with I.S. The mechanistic models were proposed and refined by J.L., I.D., Y.W., and A.L.
Y.W. participated in the unnatural amino acid production and isolation by an enzymatic
method. D.J.W. performed the 19F NMR analysis. A.L. conceived the research and wrote
Supplementary information is available for this paper at https://doi.org/10.1038/
s41589-018-0085-5.
Reprints and permissions information is available at www.nature.com/reprints.
Correspondence and requests for materials should be addressed to A.L.
Publisher’s note: Springer Nature remains neutral with regard to jurisdictional claims in
published maps and institutional affiliations.
NAture ChemICAL BIoLoGY | www.nature.com/naturechemicalbiology
© 2018 Nature America Inc., part of Springer Nature. All rights reserved.

NATuRe CHeMiCAl BiOlOgy
Articles
Crystallization. A 2-µL aliquot of the enzyme solution (10mg/ml) was mixed
with 2µL of reservoir solution containing 0.1M MES (pH 6.5), 2M ammonium
sulfate, and 2% PEG 400 and was crystallized by the hanging-drop, vapor-diffusion
method at 22°C. The crystals grew as rods in 3 d. After soaking in a cryoprotectant
containing reservoir solution plus 20% glycerol for 0.5min, the crystal was
flash-frozen and stored by liquid nitrogen for data collection using synchrotron
radiation. The substrate-bound structures were obtained by soaking 100 mM
l-cysteine to the CDO crystals. The anaerobic crystallization for obtaining pure
uncross-linked CDO was conducted in an O2-free COY anaerobic chamber.
methods
General methods. Primers were synthetized by Integrated DNA Technologies.
Reagents were purchased from Sigma-Aldrich, New England BioLabs (NEB), and
Termo, Inc. DNA manipulations in Escherichia coli were carried out according to
standard procedures. Ampicillin (100μg/mL), and chloramphenicol (30μg/mL)
were used as antibiotics for selection of recombinant strains.
Synthesis of 3,5-difluorotyrosine. To synthesize F2-Tyr, we transformed
2,6-difluorophenol to F2-Tyr by using Citrobacter freundii (ATCC8090)
tyrosine phenol lyase (TPL) following established methods46. One liter solution
containing 30 mM ammonium acetate, 60 mM sodium pyruvate, 5 mM
β-mercaptoethanol, 40 µM PLP, and 10 mg of TPL enzyme, pH 8.0, was stirred
in the dark for 3 d at room temperature (20–22 °C). The aqueous phase was
then concentrated and purified by a C-18 column from 10% to 90% methanol,
0.1% trifluoroacetic acid (v/v) in water. To test for the presence of the unnatural
amino acid, a ninhydrin solution consisting of 0.2% (w/v) ninhydrin, 95%
(v/v) N-butanol, 0.5% (v/v) acetic acid, and 4.3% (v/v) water was prepared. The
enzymatic reaction mixture was first mixed with the ninhydrin solution at a
1:1 ratio and then spotted onto a TLC plate. A purple/pink spot indicates the
presence of the unnatural amino acid. The F2-Tyr prepared was used in the next
experiment for genetic incorporation into hCDO.
Data collection, structure determination and refinement. Crystallographic
data were acquired at a temperature of 100K at Stanford Synchrotron Radiation
Lightsource beamline BL9-2 (PDB code, wavelength (Å); 6BGM, 0.97946; 6BGF,
1.0000; 6BPU, 0.97946; 6BPV, 0.97622; 6BPW, 0.97946; 6BPX, 1.0000; 6CDH,
1.0000; 6CDN, 1.0000) and at the Advanced Photon Sources (Argonne National
Laboratory, Argonne, IL) beamline 19BM (PDB code, wavelength (Å); 6BPT,
0.97919; and 6BPS, 0.97919). All X-ray diffraction intensity data were integrated,
scaled, and merged using HKL200049. Molecular replacement was performed
with Phenix50 using the crystal structure of CDO as a starting model (Protein
Data Bank entry 2ICl)14. The final model was manually adjusted and refined with
Coot51 and Phenix. Ramachandran statistics were analyzed using MolProbity
(http://mordred.bioc.cam.ac.uk/~rapper/rampage.php). All the phi and psi
angles were located in the preferred and allowed regions without any outliers.
We generated all the molecular model figures using PyMOL (W.L. deLano, The
PyMOL Molecular Graphics System version 1.8.6.0. Schrödinger LLC, http://
www.pymol.org/; 2002).
Cloning, expression, purification and protein analysis. Expression vector
pVP16-hCDO was bought from the DNASU Plasmid Repository. Expression and
purification of His-MBP-tagged enzymes hCDO was performed as follows: the
culture was grown at 37 °C in Luria Bertani (LB) media in a baffled flask at 200
r.p.m. with the appropriate antibiotic to an optical density of 0.8 AU at 600 nm.
Isopropyl-β-thiogalactoside (IPTG) was added to a final concentration of
500 μM, and the temperature was lowered to 28 °C. After 12 h of further
incubation, the cells were harvested and resuspended in lysis buffer (300 mM
NaCl, 50 mM Tris–HCl, and 1 mM ferrous ammonium sulfate, pH 8.0) and then
disrupted by a Microfluidizer LM20 cell disruptor; the supernatant was recovered
after centrifugation (13,000g for 30 min). His-MBP-tagged protein was separated
using nickel-nitrilotriacetic acid resin. After being washed with washing buffer
(300 mM NaCl, 50 mM Tris–HCl, 20 mM imidazole, pH 8.0), protein was eluted
with elution buffer (300 mM NaCl, 50 mM Tris–HCl, 250 mM imidazole, pH
8.0). Purified protein-containing fractions were confirmed by SDS–PAGE and
then dialyzed into the storage buffer (50 mM Tris-HCl, 100 mM NaCl, 10%
glycerol, pH 8.0). The dialyzed protein was concentrated and stored at −80 °C
for further use. Protein concentration was determined by UV-vis absorbance at
280 nm (ε280 nm =25,440 cm−1M−1). For the expression of F2-Tyr157, Cl-Tyr157,
and Cl2-Tyr157 hCDO proteins, pEVOL-F2-TyrRS (or pEVOL-Cl-TyrRS and
pEVOL-Cl2-TyrRS) was co-transformed with pVP16-hCDO157TAG into BL21
(DE3). A single colony was grown overnight at 37 °C in 4 mL of LB medium.
The transformed cells were induced with 0.5 mM IPTG and 0.02% l-arabinose
at OD600 of 0.8 in the presence of 0.5 mM F2-Tyr, Cl-Tyr157, or Cl2-Tyr. After
growing for 12 h at 30 °C, the F2-Tyr157, Cl-Tyr157, and Cl2-Tyr157 hCDO
proteins were purified using the protocol described for WT hCDO. The MBP tag
was removed by cleavage with TEV protease during the purification of the MBP-
fused hCDO. The liberated native CDO and mutants were further purified by
gel-filtration chromatography in 20 mM Tris–HCl and 50 mM NaCl (pH 8.0) and
were ultrafiltrated to the required concentration for subsequent catalytic assays
and crystallization. We used the GelAnalysis (http://www.gelanalyzer.com) to
estimate the ratio of the cross-linked form.
Mass spectrometry. Solutions of the intact protein were exhaustively desalted by
filtration with 10mM ammonium acetate through centrifugal filters with 10kDa
membrane cut-off. For ESI–MS analysis, samples were diluted to approximately
2μM in a solution containing 50% methanol and 0.1% acetic acid. Mass spectra
were collected on a maXis plus quadrupole-time of flight mass spectrometer
equipped with an electrospray ionization source (Bruker Daltonics) and operated
in the positive ionization mode. Samples were introduced via syringe pump at
a constant flow rate of 3μL/min. Source parameters are summarized as follows:
capillary voltage, 3,500V with a set end plate offset of 500V; nebulizer gas pressure,
0.4bar; dry gas flow rate, 4.0L/min; source temperature, 200°C. Mass spectra were
averages of 1min of scans collected at a rate of 1 scan per second in the range of
50 ≤ m/z ≤ 3,000. Compass Data Analysis software version 4.3 (Bruker Daltonics)
was used to process all mass spectra.
To generate cross-linked peptide species within a mass range suitable for
efficient fragmentation and large coverage of the peptide sequence, a multiple
enzyme digestion approach was used as previously described by Kleffmann
and co-workers52 with minor modification. In short, the WT and F2-Tyr157
CDO proteins were digested in solution first with trypsin, followed by GluC
protease, and then chymotrypsin. Proteomics-grade enzymes were used in 1:25
enzyme:substrate ratio and incubated at 37°C with each enzyme overnight. For
MALDI–TOF/TOF mass spectrometry, a supersaturated matrix solution was
prepared by dissolving α-cyano-4-hydroxycinnamic acid (Bruker Daltonics) in a
mixture of 50% acetonitrile, 0.1% TFA in water. Reconstituted digests were spotted
onto a stainless-steel target using the sandwich method in which 1μL of matrix
solution was spotted and dried, followed by 1μL of sample solution, and another
1μL of matrix solution. All MALDI–TOF/TOF mass spectra were collected on an
ultrafleXtreme MALDI–TOF/TOF mass spectrometer equipped with a smartbeam
II laser (Bruker Daltonics) in the positive ionization and reflector modes. The
instrument was calibrated using the peptide calibration standard mixture (from
Bruker Daltonics containing angiotensin II, angiotensin I, substance P, ACTH clip
1-17, ACTH clip 18–39, and somatostatin 28); and acquisition optimized in the
mass range from 500 to 4,500m/z. Approximately 3,000–8,000 shots were acquired
per MS and LIFT–MS/MS spectrum using 1,000Hz acquisition speed. FlexAnalysis
3.3 software (Bruker Daltonics) was used for data processing.
Catalytic assay & maturation assay. l-Cysteine was dissolved in hCDO storage
buffer. Freshly prepared hCDO and l-cysteine solutions were incubated in a
37°C water bath for 5min. Appropriate amounts of these two solutions were well
mixed to initiate the reaction in a 37°C water bath. The activity assay protocol
for hCDO was derived from that previously described for CDO15,47,48. A typical
assay was conducted in a total volume of 300μL as follows: 0.2−0.5μM as-purified
hCDO, 50mM MES buffer, pH 6.1, 0.3mM ammonium iron sulfate, 62.5μM
bathocuproine disulfonate and varying concentrations of l-cysteine (0–25mM).
The reaction was initiated by addition of cysteine with shaking to ensure proper
oxygenation using a shaker (220 r.p.m.). The hCDO-catalyzed reaction was
quenched by the addition of 2μL HFBA and centrifugation for removal of
precipitated proteins. All aliquots were immediately ultrafiltered through Microcon
(Millipore, Bedford, MA) 10,000Da molecular-weight cut-off ultrafiltration tubes.
The flow-through was kept in an ice water bath for further high-performance
liquid chromatography analysis. To detect the dioxygenase product CSA, reactions
aliquots were applied to a reverse-phase column (Thermo InertSustain C-18
column, 100×4.6mm, 5μM) with a mobile phase of 20mM sodium acetate, 0.6%
methanol, 0.3% heptofluorobutyric acid, pH 2.0. Curves of initial reaction velocity
versus cysteine starting concentration were fitted with the Michaelis–Menten
equation in OriginPro (OriginLab, Northampton, MA) to determine kcat and Km
values. The maturation assay was performed in 100mM Tris–HCl (pH 8.0) buffer
with 25 mM l-cysteine and 75µM enzyme at 37°C. The SDS–PAGE was replicated
at least three times by the independent experiments to ensure reproducibility.
Nuclear magnetic resonance spectroscopy. 19F NMR spectra were recorded on
an Agilent (Billerica, MA) DD2 500 MHz spectrometer equipped with a 5-mm
inverse detect probe at 300 K. Spectra were recorded in 90/10 buffer/D2O and
referenced to internal trifluoroacetic acid (−76.5 p.p.m.). One-dimension 19
F
spectra (s2pul) were recorded with 5 s relaxation delay, 64 k data points, and
multiplied with an exponential function for a line-broadening of 5 Hz before
Fourier transformation. All NMR data were processed using MestReNova
NMR v11.0.3 software. The condition for conversion of F2-Tyr157 CDO to the
all cross-linked form was 50 mM pH 6.5 MES buffer, 25 mM L-cysteine, and
300 µM protein.
Statistics and reproducibility. All experiments, including the assays and
SDS–PAGE, were obtained from replicated independent experiments
to ensure reproducibility. The representative results, the numbers of
independent experiments are labeled in the figure legends or associated text in
Supplementary Table 2.
NAture ChemICAL BIoLoGY | www.nature.com/naturechemicalbiology
© 2018 Nature America Inc., part of Springer Nature. All rights reserved.

NATuRe CHeMiCAl BiOlOgy
Articles
Reporting Summary. Further information on experimental design is available in
48. Stipanuk, M. H., Dominy, J. E. Jr., Ueki, I. & Hirschberger, L. L. Measurement
of cysteine dioxygenase activity and protein abundance. Curr. Protoc. Toxicol.
38, 6.15.11–16.15.25 (2008).
49. Otwinowski, Z. & Minor, W. Processing of X-ray difraction data collected in
oscillation mode. Methods Enzymol. 276, 307–326 (1997).
50. Adams, P. D. et al. PHENIX: a comprehensive Python-based system for
macromolecular structure solution. Acta Crystallogr. D Biol. Crystallogr. 66,
(213–221 (2010).
51. Emsley, P. & Cowtan, K. Coot: model-building tools for molecular graphics.
Acta Crystallogr. D Biol. Crystallogr. 60, 2126–2132 (2004).
52. Klefmann, T., Jongkees, S. A. K., Fairweather, G., Wilbanks, S. M. &
Jameson, G. N. L. Mass-spectrometric characterization of two
posttranslational modifcations of cysteine dioxygenase. J. Biol. Inorg. Chem.
14, 913–921 (2009).
the Nature Research Reporting Summary linked to this article.
Data availability. Structure factors and coordinates for the crystal structures
solved in this work have been deposited in the Protein Data Bank under accession
codes 6BGM, 6BGF, 6BPU, 6BPV, 6BPW, 6BPX, 6CDH, 6CDN, 6BPT, and 6BPS.
references
46. Seyedsayamdost, M. R., Yee, C. S. & Stubbe, J. Site-specifc incorporation of
fuorotyrosines into the R2 subunit of E. coli ribonucleotide reductase by
expressed protein ligation. Nat. Protoc. 2, 1225–1235 (2007).
47. Arjune, S., Schwarz, G. & Belaidi, A. A. Involvement of the Cys-Tyr cofactor
on iron binding in the active site of human cysteine dioxygenase. Amino
Acids 47, 55–63 (2015).
NAture ChemICAL BIoLoGY | www.nature.com/naturechemicalbiology
© 2018 Nature America Inc., part of Springer Nature. All rights reserved.