Halogenation of tyrosine perturbs large-scale protein self-organization [original]

nature communications

Article https://doi.org/10.1038/s41467-022-32535-2

Halogenation of tyrosine perturbs large-

scale protein self-organization

Huan Sun

1,2

,HaiyangJia

1,3

, Olivia Kendall

2,4

, Jovan Dragelj

Vladimir Kubyshkin

, Tobias Baumann

, Maria-Andrea Mroginski

Petra Schwille

& Nediljko Budisa

2,5

Protein halogenation is a common non-enzymatic post-translational mod-

iﬁcation contributing to aging, oxidative stress-related diseases and cancer.

Here, we report a genetically encodable halogenation of tyrosine residues in a

reconstituted prokaryotic ﬁlamentous cell-division protein (FtsZ) as a platform

to elucidate the implications of halogenation that can be extrapolated to living

systems of much higher complexity. We show how single halogenations can

ﬁne-tune protein structures and dynamics of FtsZ with subtle perturbations

collectively ampliﬁed by the process of FtsZ self-organization. Based on

experiments and theories, we have gained valuable insights into the

mechanism of halogen inﬂuence. The bending of FtsZ structures occurs by

affecting surface charges and internal domain distances and is reﬂected in the

decline of GTPase activities by reducing GTP binding energy during poly-

merization. Our results point to a better understanding of the physiological

and pathological effects of protein halogenation and may contribute to the

development of potential diagnostic tools.

The lifespan of living cells and organismsis inﬂuenced by the quality of

their proteins. The primary sequence of a protein is one of the most

important determinants of protein folding and ﬁnal conformation as

well as biochemical activity, stability, and half-life. Enzymatic and non-

enzymatic post-translational modiﬁcations (PTMs) greatly expand the

structural and functional space of proteins, as well as the proteome as

a whole. Many protein PTMs produce the non-enzymatic attachment

of speciﬁc chemical groups to amino acid side chains, e.g., glycation,

nitrosylation, oxidation/reduction, acetylation, succination, and halo-

genation. From a structural perspective, these modiﬁcations usually

alter the overall tertiary fold, often destabilize proteins, and may result

in cleavage of the protein backbone or protein aggregation1.Sucha

process is considered as protein damage and has a particularly pro-

nounced deteriorating impact in the quality of individual proteins, but

also on protein complexes and cellular activities that require protein

self-assembly.

Halogenation is a typical pathological modiﬁcation that can occur

in the context of oxidative stress in the environment. Oxidative halo-

genation hasbeenshown to impair proper cellularfunctions and cause

severe long-term health problems such as aging, cancer, or even

death2–4. The aromatic amino acid tyrosine is a major target for oxi-

dativehalogenation in proteins.Halogenation atthis residue canoccur

endogenously; for example, hypochlorous acid generated by the

myeloperoxidase-hydrogen peroxide-chloride system of phagocytes

has been shown to be the major source of 3-chlorotyrosine formation

in cells and tissues5–7. In addition, the widespread use of halogenated

aromatic compounds inagriculture, dye, chemical and pharmaceutical

industries8,9raises major concerns about acquired halogenation, as

reactive halogen species might be the direct source of halogenation of

both enzymatic and non-enzymatic origin10. For this reason, the study

of halogen modiﬁcations and their role in protein-based systems is of

enormous importance in understanding the cellular mechanisms of

Received: 8 March 2022

Accepted: 4 August 2022

Check for updates

School of Chemistry and Chemical Engineering, Beijing Institute of Technology, Beijing 100081, PR China.

Technical University of Berlin, Müller-Breslau-Str.

10, D-10623 Berlin, Germany.

Max Planck Institute of Biochemistry, Am Klopferspitz 18, D-82152 Martinsried, Germany.

University of Edinburgh, David

Brewster Road, King’s Buildings, Edinburgh EH9 3FJ, UK.

University of Manitoba, 144 Dysart Rd., R3T 2N2 Winnipeg, MB, Canada.

e-mail: [email protected];andrea.mroginski@tu-berlin.de;schwille@biochem.mpg.de;nediljko.budisa@umanitoba.ca

Nature Communications | (2022) 13:4843 1

1234567890():,;

oxidative damage. Such information provides insights that might lead

to novel strategies in biomedical approaches that prevent and heal

proteome damage.

The emerging technology of genetic code expansion has enabled

the co-translational and site-speciﬁc incorporation of diverse non-

canonical amino acids (ncAAs) that confer versatile physicochemical

and biological properties to proteins11,12. Genetic introduction of

selective chemical modiﬁcations into a protein of interest provides a

powerful approach to characterize the structural and biochemical

consequences of the modiﬁcations13,14. By mimicking PTMs, the site-

speciﬁc incorporation of the ncAAs can provide information on how

the position, density, and distribution of protein modiﬁcations perturb

protein structures and functions on a small scale. Co-translational

modiﬁcation of target proteins with oxidized ncAAs at deﬁned posi-

tions has already proven to be a useful tool to study the role of protein

nitration15 or oxidation16 at speciﬁc positions. For example, nitrotyr-

osine has been genetically incorporated in superoxide dismutase from

mitochondria to encode the protein oxidative damage17. The effect of

protein modiﬁcations in collective intramolecular processes18 in pro-

tein complexes assembly19,20 and in cells and tissues15,21,22 have also

been studied. For instance, modiﬁcations such as acetylation20 and

methylation19 havebeenusedtodeterminetheeffectsofmodiﬁcations

on nucleosome complexes assembly and cellular transcriptional

responses. Nonetheless, there is still no insight into how oxidative

modiﬁcationscouldactonlargescaleproteinassemblies.Inparticular,

the ampliﬁcation effects during maturation of complex structures are

still unclear. The high cellular noiseand the extreme complexity of the

system under study pose a major challenge.

To address this knowledge gap, we here employ a reconstituted

bacterial minimal system of the cytoskeleton to gain insight into the

effects of halogenation on the dynamic process of molecular self-

organization. Since cytoskeleton components are major actors in the

cellular lifecycle, a perturbation or disruption of its functions would

contribute signiﬁcantly to aging5,23. The minimal system studied here24

is a one-component biological system reconstituted in vitro from

scratch with the puriﬁed FtsZ protein (the ftsZ gene product), the

known prokaryotic homologue of the eukaryotic protein tubulin

(Fig. 1a). The protein has been shown to be sensitive to halogenating

chemicals25 and its activity is sensitive to modiﬁcations at individual

sites26. As an essential part of the bacterial division ring, known as “Z

ring”, FtsZ has shown intriguing self-organization when reconstituted

in vitro on biological membranes. Itpolymerizes into dynamic vortices

by circular treadmilling dynamics fueled by GTP hydrolysis27.These

properties of FtsZ make it a promising candidate for studying the

collective behaviors of PTMs in vitro. By combining experiments and

theory, we demonstrate how incorporation of halogenated tyrosine

analogues (HYs) at a key position of FtsZ perturbs its structural fea-

tures and GTPase activity. Consequently, the precise and highly tai-

lored perturbations of structures and dynamics can be collectively

ampliﬁed through protein self-organization (Fig. 1b). Finally, we could

elucidate the detrimental effects of halogenated tyrosine on collective

FtsZ self-organization behavior in the reconstituted minimal cell divi-

sion system and further interpret how the change at a single site poi-

sons the system globally.

Results

Co-translational incorporation of halogenated tyrosine

analogues

In living systems, oxidation of tyrosine residue can often lead to the

formation of halogenated residues known as HYs3:3-chloro-tyrosine

(ClY), 3-bromo-tyrosine (BrY), 3-iodo-tyrosine (IY), 3,5-dichloro-tyr-

osine (Cl

Y), 3,5-dibromo-tyrosine (Br

Y), and 3,5-diiodo-tyrosine

Y)28–31. Halogenation affects several key molecular properties of the

parent residue: molecular volume, pK

of the side-chain group, and

hydrophobicity(Fig. 1c). We studied the experimental pK

values of the

phenolic group in free amino acids, and found that the value depends

primarilyon thenumber of halogen atomsintroduced to the molecule:

Tyr (pK

9.9) > ClY ≈BrY ≈IY(pK

8.3)>Cl

Y≈Br

Y≈I

Y(pK

6.5).Next,

we examined the hydrophobicity of the amino acid residues using

lipophilicity measurements using a recently developed method32.The

evaluation of the residue lipophilicities is complicated because the

ionization of the side chainis also perturbed by halogenation. Thus, we

examined the experimental distribution coefﬁcient (logD) in a range of

pH values (pH 6–9) to illustrate possible changes in the ionization

state. We found that the experimental lipophilicity values led to a

rather complex outcome: while each newly introduced halogen atom

increased the hydrophobicity of the side-chain, the concomitant

decrease of pK

facilitated the transition to the deprotonated form,

and thus lowered the hydrophobicity. This complex interplay between

halogenation and the properties of the tyrosine residue makes it rather

difﬁcult to make accurate predictions without analyzing the actual

experimental protein data. Since the microenvironment of the protein

affects the pK

transition of the tyrosine side chain, halogenation may

reduce overall hydrophobicity in one position, where pK

falls below

the pH, but elevate the hydrophobicity at another position, where pK

stands above pH of the medium.

Thus,to examine the roles of modiﬁcations in proteins, we set out

to produce corresponding proteins containing above-mentioned

modiﬁcations and study their properties. We employed the stop

codon suppression technique, in which so-called orthogonal pairs are

used. An orthogonal pair consists of an aminoacyl-tRNA synthetase

(aaRS), which selects the amino acid of interest from the pool of

intracellular substances, and a transfer-RNA (tRNA), which carries the

amino acid to the ribosome and incorporates it into the sequence in

response to an in-frame stop codon. Few methods have been devel-

oped for incorporation of HYs into proteins33–38, however, none of

them was sufﬁcient to incorporate the full set of analogues selected for

this study because they are typically being restricted to 3-halo-

tyrosines only.

To selectively incorporate the HYs into the protein of interest, we

constructed a gene library based on Methanocaldococcus jannaschii

TyrRS (MjTyrRS) by randomizing eight active-site residues of the

aminoacyl-tRNA synthetase (detailed mutations see Supplementary

Table 1). The MjTyrRS library was then selected by one-round positive

selection14,39 (Supplementary Fig. 1) against HYs. For selection, we used

chloramphenicol acetyltransferase with two amber codons and a small

ubiquitin-like modiﬁer tagged superfolder green ﬂuorescent protein

(sfGFP) with one in-frame amber stop codon for readout. After a single

round selection, the MjTyrRS mutant B48RS (Y32G, L65E, H70G,

F108Q, Q109C, D158A, L162N) was selected as a candidate with high

selectivity towards several HYs: ClY, Cl

Y, BrY, Br

Y, and IY. Another

mutantC64RS(Y32V,L65E,H70G,F108N,D158A,L162C)wasselected

for incorporation of I

Y (Supplementary Fig. 2). Incorporation of HYs

into sfGFP using the mutants B48RS (for ClY, Cl

Y, BrY, Br

Y, and IY)

and C64RS (for I

Y) was also conﬁrmed by mass spectra of the intact

protein obtained after electrospray ionization (Supplementary Fig. 3).

Incorporation efﬁciency was also assessed by comparing the ﬂuores-

cence intensity of sfGFP containing HYs (Supplementary Fig. 2).

Markedly, in the expression experiments, the mutant B48RS showed

an almost equal efﬁciency with Cl

Y, BrY, Br

Y, and IY, which in each

case was more than 100-fold higher than the control without HYs

additions (Supplementary Fig. 2).

Single-site halogenation suppresses protein activity

After establishing the tool for HYs incorporation, we next prepared

FtsZ variants containing the analogues. Within the native asymmetric

unit FtsZ, it adopts two distinct conformational states40–43 with its two

subdomains acting independently. Therefore, positions localized at or

near the interface may dramatically affect overall conformation

transitions44,45.WeincorporatedClY,Cl

Y, BrY, Br

Y, IY and I

Yat

Article https://doi.org/10.1038/s41467-022-32535-2

Nature Communications | (2022) 13:4843 2

position Tyr222, which is the only tyrosine structurally close to the

boundary between the N-terminal domain and C-terminal domain. The

position is located on the outer surface of the protein and is exposed

to solvent, allowing us to mimic natural halogenation progress that is

modiﬁed by external natural modiﬁers. In addition, this position is

known to be sensitive to PTMs26,46, making it a good candidate for

investigation of halogenations.

To visualize the FtsZ dynamic patterns in vitro, we designed a

truncated FtsZ(1-366) variant fused with yellow ﬂuorescent protein

(YFP) and a membrane-targeting sequence (mts)47 (Fig. 1a). The FtsZ-

YFP-mtsvariantswereexpressedinE. coli harboring plasmid-borne

B48RS and C64RS orthogonal pairs in media supplemented with HYs.

Fidelity of the HYs incorporation was veriﬁed by mass-spectra analysis

(Fig. 1d), which demonstrated that the FtsZ protein isolates were

Interface

FtsZ_CTD

YFP

MTS

FtsZ_NTD

X=Cl, Br, I

Collective integrationSelf-organization

V<V0

68000 68200 68400

0.0

0.2

0.4

0.6

0.8

1.0

Molecular weight (Da)

Relative Intensity

Halogenated

FtsZ

010203040506070

100

Time(min)

Normalized light scattering(%)

ClY

Cl2Y

BrY

Br2Y

NHAc

CO2CH3

Cl Cl

NHAc

Br Br

NHAc

I I

6789678967896789678967896789

logD

pH pH pH pH pH pH pH

ClY Cl2YBrY IY Br2Y

pKa

Vol.（Å3)8.3

133.6

6.5

150.1

8.3

139.3

9.9

116.6 8.3

146.5

6.4

161.5 6.5

175.7

ClY Cl2YBrY IY Br2Y

Br Br

Cl Cl OH

68600

ClY Cl2YBrY IY Br2Y

Rate of GTP hydrolysis(pi/min)

GTP GDP

Fig. 1 | Genetic incorporation of halogenated tyrosine analogues in FtsZ based

constructs. a Schematic illustration of the FtsZ construct (FtsZ-YFP-mts): FtsZ

(1–366, split blue cylinder), yellow ﬂuorescent protein (YFP, yellow star) and a

membrane target sequence (mts) (grey triangle). V and V

indicate the treadmilling

speed. bSchematic model for inhibition of FtsZ ring formation by halogenated

residues. cMolecular properties of the tyrosine analogues as dependent from

halogenation. Experimental lipophilicity was determined in methyl

N-acetylaminoacetates against buffers at different pH values, pK

of the phenolic

side-chain wasexperimentallydetermined in free amino acids, molecular volume is

calculated for free amino acids. dChemical structures of the halogenated tyrosine

analogues incorporated into FtsZ and the deconvoluted ESI-MS spectra. The

expected and observed masses are listed in Supplementary Table 3. eGTPase

activities of wild type FtsZ-YFP-mts (Y) and halogenated FtsZ(Y222X)-YFP-mts

(X = ClY, BrY, Br

Y, I

Y, Cl

Y, and IY). Data from two independent replicates are

shown as means with standard deviations. fDynamic light scattering of wild type

FtsZ-YFP-mts and halogenated FtsZ-YFP-mts variants. Source data of c–fare pro-

vided as a Source Data ﬁle.

Article https://doi.org/10.1038/s41467-022-32535-2

Nature Communications | (2022) 13:4843 3

homogeneously labelled with HYs at position 222. Deconvoluted mass

spectra revealed the masses expected for the wild type and ClY, Cl

BrY, Br

Y, IY and I

Y containing FtsZ-YFP-mts variants (Fig. 1d, Sup-

plementary Fig. 4, Supplementary Table 3). The mass shifts caused by

ClY, Cl

Y, BrY, Br

Y, IY and I

Y were +33.31, +69.33, +78.40, +158.70,

+125.33 or +251.88Da, respectively, compared with the wild type pro-

tein. This result would be expected for the substitution of one or two

hydrogen atoms with halogens at Tyr222.

Tofurtherevaluate the inﬂuence of the site-speciﬁc halogenation,

we examined the GTPase activity and polymerization rate in the pro-

tein samples. Curiously, we found that even a single hydrogen-to-

halogen atom exchange was sufﬁcient to alter the enzymatic activity.

Indeed, we found that GTPase activity was lowered in all cases by the

presence of HYs. Particularly low activities were observed for Cl

Y, IY,

and I

Y containing variants, which showed a decrease in GTP hydro-

lysis rate by up to 70% (Fig. 1e). In addition, light scattering results

showed that all halogenated proteins exhibited suppressed poly-

merization rates compared with wild type (Fig. 1f). The decreased

GTPase activity and polymerization rates may be caused by reduced

GTP hydrolysis as a result of molecular perturbations produced by

halogenation. We infer that the molecular perturbations could result

from the changes of key molecular properties demonstrated in Fig. 1c,

such as side-chain acidity, molecular volume, and hydrophobicity. As

mentioned before, the pK

values of HYs at pH7.5 were all lower than

that of tyrosine and the lipophilicity was higher than that of tyrosine,

which could increase the acidity of tyrosine side-chain and hydro-

phobicity, leading to suppressed GTPase activities. In addition, halo-

genation can enlarge the volumes of the amino acids, and thus might

disrupt the protein structures.

Halogenations of tyrosine affect protein assembly and dynamics

We next examined the ring formation behavior of halogenated FtsZ in

live E. coli cells (Supplementary Fig. 5) using ﬂuorescence confocal

microscopy. To this end, we examined microscope images of cells

overexpressing the FtsZ-YFP-mts constructs. We observed that with

the wild type protein, multiple complete rings were relatively evenly

distributed along the cell length (Supplementary Fig. 5a), consistent

with previous reports48,49. In contrast, the cells containing

FtsZ(Y222ClY)-YFP-mts produced only a single ring in the center

(Supplementary Fig. 5d). Conversely, the BrY and Br

Ycontaining

variants formed multiple interconnected short helices, densely dis-

tributed along the cells (Supplementary Fig. 5 b, e). FtsZ(Y222Cl

Y)-

YFP-mts and FtsZ(Y222IY)-YFP-mts produced incomplete bands orarcs

that could be parts of helix or spiral and ring fragments (Supplemen-

tary Fig. 5c, f). Finally, no ring formation was observed in the cells

overexpressing I

Ymodiﬁed protein (Supplementary Fig. 5g). Such

severe changes of FtsZ architectures may be caused by the altered

structures of FtsZ proteins during halogenations, which requires fur-

ther structural analysis. To investigate the dynamics of FtsZ assembly

in vivo, we performed ﬂuorescence recovery after photobleaching

(FRAP) analysis on theseFtsZ architectures. In FRAP analysis, a portion

of a ﬂuorescent FtsZ pattern is photo-bleached with a focused laser

beam and the recovery of the ﬂuorescence signal caused by replace-

ment of subunits outside the photobleached region is measured. We

found that the ﬂuorescence recovery of wild type FtsZ-YFP-mts was

faster than that of the halogenated FtsZ-YFP-mts (Supplementary

Fig. 5h–i), which is consistent with the trend of GTPase activity shown

in Fig. 1e. All of the above results are sufﬁcient to ascertain effects on

both structures and dynamics of FtsZ resulting from single site halo-

genation. However, because of the physiological complexity and high

background of endogenous FtsZ, it is not yet possible to deduce a

speciﬁc underlying molecular mechanism and the speciﬁc effects of

the different types of halogenations from these cellular results.

To better understand the architecture and dynamics of the ﬁla-

ment network generated by halogenated FtsZ, we reconstituted the

protein variants on supported lipid bilayers (SLBs) and quantiﬁed their

self-organized patterns using a total internal reﬂection ﬂuorescence

microscope (TIRFM) (Fig. 2a, Supplementary Movie 1). First, SLB with

negatively charged lipid composition was prepared in a home-made

microscope chamber. Then FtsZ was introduced on the SLB and self-

assembly of FtsZ was initiated by addition of GTP. In our experiments,

we found that the wild type FtsZ-YFP-mts could be recruited to the

membrane, where it can promptly self-organize into a homogeneous

treadmilling ring pattern (Fig. 2b). The variants containing ClY, BrY,

Y, and I

Y were also able to form rings, albeit with heterogenous

ring morphology (Fig. 2b, c). The average diameters of the rings

formed by FtsZ(Y222ClY)-YFP-mts (0.62 ± 0.15 μm) were 22.5% smaller

than those formed by the wild type protein (0.80 ± 0.18 μm).However,

in the case of FtsZ(Y222BrY)-YFP-mts, FtsZ(Y222Br

Y)-YFP-mts and

FtsZ(Y222I

Y)-YFP-mts, the ring diameters were 11.3%, 15.0% and 28.8%

larger than in the wild type, respectively. Moreover, FtsZ variants

modiﬁed with closely analogues ClY and BrY, characterized with

similar side-chain acidity and molecular volume (Fig. 1c), exhibitedself-

organization into signiﬁcantly distinct sizes of ring patterns. This fact

indicates that even subtle volume differences (6 Å3) may be ampliﬁed

by large-scale protein self-organization, resulting in 43.5% enlarged

ring diameters. Overall, single atom modiﬁcations allowed us to pre-

cisely ﬁne-tune the protein-assembly progress, and the changes are

even visible through ﬂuorescence microscopy.

Inthefollowingstep,weexaminedthetreadmillingdynamics

50 of

FtsZ. Treadmilling is a GTPase dependent process that results from

polymerization at one end and depolymerization at the other end of a

polymer chain51. Compared with the more noisy and complex in vivo

approaches (Supplementary Fig. 5), we expected treadmilling

dynamics to provide a clearer illustration of the protein damage, since

this assay is performed purely in vitro. Since the deﬁciency in the

GTPase activity(Fig.1d)suggestsa strongperturbationofthisdynamic

process from halogenation, we quantiﬁed the vortices’velocities by

the slope of kymographs47 (Fig. 2d) generated along the cir-

cumference. Results showed lower velocities in all halogenated pro-

teins, as expected. The halogenations decrease the mean rotation

velocities down to 48.0% compared to the wild type FtsZ-YFP-mts

(21.63 ± 5.11 nm s−1)(Fig.2eandSupplementaryTable4).

In two cases, the presence of HYs resulted in full suppression of

the ring pattern formation in vitro. We did not observe any distinctive

ring formation produced by protein variants containing Cl

YandIY.

These proteins display ﬁber-like and highly meshed ﬁlament patterns

in the entire membrane area (Fig. 2b). Then we analyzed the curvature

of FtsZ ﬁlaments with stretching open active contours (SOAX)52.

Compared to the wild type FtsZ-YFP-mts, FtsZ ﬁlaments containing

Y and IY displayed decreased curvatures indicating that the two

types of halogenations may seriously bend the phenotype of FtsZ

ﬁlaments and further inhibit the ring pattern formation (Fig. 2h).

Instead of ring velocity, we analyzed their turnover rate by FRAP to

better understand their assembly dynamics. With the wild type FtsZ-

YFP-mts, a relatively fast recovery was observed with a half-time

4.97 ± 0.24 s (n= 6). Conversely, with Cl

Y and IY containing proteins

the half-time was nearly doubled, indicating slower dynamics (Fig. 2f, g

and Supplementary Table 4). The lowered dynamics comes in agree-

ment with the in vivo results (Supplementary Fig. 5h–i). To better

understand theprominentchangesexhibitedbyIYandCl

Ycontaining

proteins, we compared the molecular properties of the amino acids.

The side-chain acidity measured in free amino acids showed distinctly

different pK

values to those of tyrosine: 9.9 in tyrosine, 8.3 in IY, and

6.5 in Cl

Y.Atthesametime,themolecularvolumeoftheanalogues

was very similar, differing by only 3.6 Å3(Fig. 1c). In addition, we

compared the polarity of the side-chains in model compounds. The

halogenated amino acids IY and Cl

Y exhibited identical lipophilicity in

non-ionized form (measured at pH 6) and both were more hydro-

phobic than any amino acid in the canonical protein repertoire (Fig. 1c).

Article https://doi.org/10.1038/s41467-022-32535-2

Nature Communications | (2022) 13:4843 4

The high similarity in molecular volume and hydrophobicity of the side

chains of IY and Cl

Y may explain the similar behavior of the FtsZ-YFP-

mts constructs with these residues. Overall, these results clearly indi-

cate that halogenation not only perturbs enzymatic activity, but that

the associated effects on structure and dynamics are clearly down-

stream of the self-organization process.

Partial halogenation disrupts ring pattern formations

To investigate the collective progress of halogenated FtsZ integration

in vitro, we performed a self-organization assay by mixing the FtsZ

chimera bearing HYs with wild type protein. We selected two variants:

thewildtypeastheoneexhibitingnativeself-assemblyandthe

Y-containing protein as the one unable to form rings. The proteins

were mixed in different proportions, with FtsZ(Y222Cl

Y)-YFP-mts

provided at 25%, 50%, and 75%, yielding a total protein concentration

of 0.5 µM. The formation of FtsZ ring pattern was examined on SLB

under TIRF-monitoring (Fig. 3a, Supplementary Movie 2). FtsZ archi-

tectures and dynamics, such as ring diameters, velocity and ﬁlament

curvatures were analyzed. We observed that the density of the multiple

homogeneously distributed rings decreased upon addition of the

halogenated sample. The density values were 1.44 (±0.12) × 105µm−2,

1.01 (±0.04) × 105µm−2,and5.28(±0.21)×10

4µm−2upon the presence

of 0%, 25%, and 75% of the Cl

Ymodiﬁed variant, respectively

(Fig. 3b–c). By analyzing the ring morphology and treadmilling

020

40 60 80 100

Time(s)

0.0

0.2

0.4

0.6

0.8

Relative fluorescence intensity

Cl2Y

-5s 0s 5s 15s

Cl2Y

CH2

I2Y

ClY

BrY

CH2

Br Br

Br2Y

0.0 0.4 0.8 1.2 1.6 2.0

0.0

0.1

0.2

Ring diameter

(μm)

0.0 0.4 0.8 1.2 1.6 2.0

0.0

0.1

0.2

0.3

0.4

Relative frequency

0.0 0.4 0.8 1.2 1.6 2.0

0.0

0.1

0.2

0.3

0.4

Relative frequency

0.0 0.4 0.8 1.2 1.6 2.0

0.0

0.1

0.2

0.3

0.4

Relative frequency

0.0 0.4 0.8 1.2 1.6 2.0

0.0

0.1

0.2

0.3

0.4

Relative frequency

0.3

0.4

Relative frequency

1 μm

1 min

360

010203040

0.0

0.1

0.2

0.3

Relative frequency

010203040

0.0

0.1

0.2

0.3

Relative frequency

010 20 30 40

0.0

0.1

0.2

0.3

Relative frequency

010203040

0.0

0.1

0.2

0.3

Relative frequency

010203040

0.0

0.1

0.2

0.3

Relative frequency

Ring velocity

(nm/s)

N=149 N=115

N=150 N=93

N=150 N=101

N=150 N=93

N=149 N=104

Y BrY Cl2YI2YIYClY Br2Y

cde

xxx

Lipid FtsZ

Mg2+

GTP

Chamber

Coverslide

TIRF

SLB

0.00 0.05 0.10 0.15

0.0

0.1

0.2

0.3

Relative frequency

YCl2YIY

0.00

0.05

0.10

0.15

Curvature (1/μm)

1/R

Cl2Y

Half time (s)

p1p1

Article https://doi.org/10.1038/s41467-022-32535-2

Nature Communications | (2022) 13:4843 5

velocity, we found that the ﬁlament curvature gradually decreased

upon addition of the halogenated protein. A gradual enlargement in

ring sizes and a slowdown in ring dynamics were observed as well

(Fig. 3d). Finally, the increased proportion of halogenated FtsZ resul-

ted in self-organization patterns that became increasingly similar to

those of pure Cl

Y containing FtsZ, with a highly meshed ﬁlament

pattern (Fig. 3b). These observations conﬁrm that halogenation on

FtsZ at Y222 position impairs the assembly of protoﬁlaments from

monomeric proteins and ultimately inhibits Z-ring formation.

Structure modelling elucidates the effects of halogenation

To better understand how the subtle changes in halogenated protein

structure suppress protein activity and how the effects are further

ampliﬁed by the self-organization process, we constructed structural

models for the wild type FtsZ monomer and dimer, as well as for the

FtsZ(Y222Cl

Y) mutant including a GTP molecule at the interface and

in the upper monomer binding site. Starting coordinated were gen-

erated from a homology model of the wild type FtsZ with the I-TASSER

suite53 using the M. jannaschii FtsZdimer structure (PDB code: 1W5A)40

(Fig. 4a, Supplementary Fig. 6) as template. Compared to the pub-

lished structures of the two monomer conformations41, the structural

models were identiﬁed to be in the closed state i. e., a conformation

preferred in solution. Thus, the FtsZ dimer models represent the

nucleation of a ﬁlament. Structures containing the mutated residues

FtsZ(Y222Cl

Y)weremodelledbymodifyingthetyrosineresidueinthe

homology model. Brieﬂy, chlorine positions weretaken by aligningthe

modelled residue with a crystal structure of a halogenated tyrosine

(PDB: 4NX2)54 (Fig. 4b) and partial charges for the backbone were

calculated (Supplementary Tables 5–6). Detail description of the

molecular modeling protocol is found in supplementary information.

These models of wild type and halogenated FtsZ monomers and

dimers, were subjected to 500ns molecular dynamics (MD) simula-

tions (see Supplementary Methods) to gain insight into protein

dynamics and GTP binding.

When aligning the halogenated and wild type monomer models

after 500 ns, we did not observe drastic structural changes (Fig. 4d).

On the contrary, all models were stable throughout the simulations:

the root-mean square deviation (RMSD) of the monomer backbone

remained below 3 Awhen the disordered C-terminal tail and

N-terminus were excluded (Fig. 4d–f). Halogenation may lead to more

subtle changes while maintaining structural stability. To investigate

this possibility, we analyzed the electrostatic potential surface (EPS)

using the APBS PyMOL plugin55. After mutation, the surface charges of

the protein mainly change around the GTP binding pocket and the

N-terminal domain interface in the dimer structure (Supplemen-

tary Fig. 7).

Because GTP forms a large part of the dimer interface and GTP

hydrolysis is key to monomer dissociation, we postulated that halo-

genation may affect GTP binding and thus cause the altered dynamics.

Therefore, we calculated the solvation binding energies of the FtsZ-

GTP interactions throughout the last 100 ns of the MD trajectories

(Supplementary Fig. 8, Fig. 4c). We found that both monomer and

dimer models of FtsZ(Y222Cl

Y) exhibited stronger GTP binding

energy than thewild type. A stronger GTP-monomer association could

increase the likelihood of polymerization; by lowering the free energy

barrier of nucleation when two monomers come together in the pre-

sence of GTP. Furthermore, we hypothesize that the stronger asso-

ciation leads to tighter binding and inhibits the conformational

dynamics required for GTPase activity, thereby increasing the activa-

tion energy of hydrolysis.

The dimer simulations showed overall stability, but closer exam-

ination revealed that subtle conformational changes within the

monomer units were ampliﬁed when the monomers interacted with

each other (Fig. 4g–j). In particular, when the N-terminal domain

(residues 379–561) of the upper monomer was aligned to the energy

minimized model, a signiﬁcant bending of the overall wild type dimer

structurewasdetected(Fig.4g), while less conformational distortion is

predicted for the FtsZ(Y222Cl

Y) variant (Fig. 4h). This result is con-

sistent with experimental observations that the Cl

Ymodiﬁcation

reduces the curvature of the ﬁlament and thus inhibits the ring for-

mation (Fig. 2). Comparison of the C-terminal domain of the lower

monomer with the N-terminal domain of the upper monomer revealed

that the FtsZ(Y222Cl

Y) monomers remained in a similar conformation

and quickly reached thermodynamic equilibrium withlittle variation in

RMSD, while RMSD in the case of wild type gradually increased from 2

Ato 5 Aand reached equilibrium at ~300 ns (Fig. 4i). This observation

suggests that a slower adaptation to favorable conformations was

necessary for the wild type. In addition, it agrees well with previous

studies that reported the existence of two conformations of the

monomer, emphasizing that a transition between the two states is

important for treadmilling41,42. However, the slower adaptation pre-

dicted for the wild type FtsZ was not detected for the FtsZ(Y222Cl

dimer on the basis of RMSD evolution, indicating that conformation

dynamics were inhibited by the halogenation. As consequence, the

structure of the halogenated mutant appeared to be more rigid than

the wild type, thereby leading to weaker bending. This can be

explained by subtle conformation changes in the monomer, such as

the reorganization of hydrogens bonds and other non-covalent inter-

actions due to the altered electrostatics around the introduced mod-

iﬁcation (Supplementary Fig. 7), which further causes larger

conformational differences during protein assembly (Fig. 4j).

Discussion

In conclusion, we successfully developed a platform for site-speciﬁc

incorporation of various halogenated tyrosine analogues into recom-

binantly expressed proteins with high ﬁdelity and efﬁciency in E. coli.

Reprogrammed translation resulted in fairly clean biosynthesis of

halotyrosine-containing protein samples without the need for a rather

harsh and nonspeciﬁc PTM, such as treatment of proteins with hypo-

halous acids. To elucidate the collective effects of halogenations, we

Fig. 2 | Halogenation affects FtsZ patterning and treadmilling dynamics.

aSchematic illustrating FtsZ self-organization on model membrane monitored by

TIRF. bRepresentative cytoskeletal pattern of wild type FtsZ-YFP-mts (Y) and

halogenatedFtsZ(Y222X)-YFP-mts(X=ClY,BrY, Br

Y,I

Y,ClY,andIY)onsupported

membrane (0.5 μMproteins,4mMGTPand1mMMg

2+). Scale bar: 3 μm. The

experiment was performed three times under identical conditions. cRing size

distributions of wild type FtsZ-YFP-mts and halogenated FtsZ-YFP-mts. D indicates

the ring diameter. ∗Analysis of Variance (ANOVA) one-way statistical test

=6.16×10

−18;p

=4.91×10

−5,p

=1.28×10

−7;p

=1.18×10

−20). dRepresentative

kymograph along the circumference of the vortices formed by wild type FtsZ-YFP-

mts and halogenatedFtsZ-YFP-mts. The respective slopes (red lines) correspond to

the treadmilling velocity (V) of the vortices. eVelocity distributions for wild type

FtsZ-YFP-mts and halogenated FtsZ-YFP-mts. Vindicates the ring velocity. ∗Analysis

of Variance (ANOVA) one-way statistical test (p

=9.57×10

−18;p

=2.01×10

−13,

=6.08×10

−21;p

=4.92×10

−39). fFilament curvature distributions for wild type

FtsZ-YFP-mts (n=1700ﬁlaments) and FtsZ(Y222 Cl

Y/IY)-YFP-mts (n=700ﬁla-

ments). Wild type FtsZ-YFP-mts (Y) is shown in black, FtsZ(Y222Cl

Y)-YFP-mts

(Cl

Y) is red and FtsZ(Y222IY)-YFP-mts is blue (IY). 1/R represents the ﬁlament

curvature. ∗Analysis of Variance (ANOVA) one-way statistical test (p

=6.00×10

−25;

=3.19×10

−4). Box plots in f: the lines represent medians, box limits represent

quartiles 1 and 3, whiskers represent 1.5 × interquartile range and points are out-

liers. gSnapshots and hﬂuorescence recovery curves for wild type FtsZ-YFP-mts

and FtsZ(Y222Cl

Y)-YFP-mts after photo-bleaching. Scale bar: 3 μm. Inset: The half-

life of ﬂuorescence recovery. Data from three independent replicates are shown as

means with standard deviations. ∗Analysis of Variance (ANOVA) one-way statistical

test (p

=1.28×10

−5).The solidcurves in cand erepresent the Gaussian ﬁt. The solid

curves in frepresent the extreme ﬁt. Source data of c,fand hare provided as a

Source Data ﬁle.

Article https://doi.org/10.1038/s41467-022-32535-2

Nature Communications | (2022) 13:4843 6

demonstrated an in vitro protein self-organization assay that can

amplify the structuralperturbations caused by PTMs atthe atom-scale

and allow us to detect the structural changes with optical microscopy.

Through the in vitro protein assay and theoretical structure simula-

tion, we discovered that even one or two newly introduced halogen

atoms could readily alter the enzymatic activity of the protein, amplify

or propagate their effects through protein-protein interactions, and

subsequently produce global effects on protein pattern formation and

dynamics. The in vitro methodology used here is not limited to a

quantitative and conceptual understanding of halogenation, but could

also be applied to other types of post-translational modiﬁcation

occurring in natural proteins. Furthermore, as a kind of ﬁne-tuning

tools56 our halogenated tyrosine enables the study of complex

protein systems in vitro and in vivo with residue precision. The dif-

ferent variants of halogen atoms and modiﬁed positions will allow

production of a large repertoire of new modiﬁcation functions in the

future. Speciﬁcally, HYs are typical products of the oxidation of the

tyrosine residue in proteins, ultimately contributing to aging

processes3. The site-speciﬁcmodiﬁcation of protein structures and

activities in the presence of HYs should therefore enable a much better

understanding of the oxidative damage-related diseases such as aging

and cancer. Accumulation of this information will further help improve

diagnostic methods and therapeutic interventions for patients in

the future57–59.

0.00 0.05 0.10 0.15

0.0

0.1

0.2

0.3

Relative frequency

Curvature (1/μm)

0.00

0.05

0.10

0.15

1/μm

0255075

100

Cl2Y(%)

0% ClY

25% ClY 50% ClY 75% ClY 100% ClY

Protein mixing

02550

75 100

1/R

0.0 0.4 0.8 1.2 1.6

Ring diameter

(μm)

0.2

0.1

0.0

0.2

0.1

0.0

0.2

0.1

0.0

0.2

0.1

0.0

Relative Frequency

0% ClY

25% ClY

50% ClY

75% ClY

010 20 30 40

0.0

0.1

0.2

0.0

0.1

0.2

0.0

0.1

0.2

0.0

0.1

0.2

Relative Frequency

Ring velocity

(nm/s)

n=150

n=200

n=202

n=154

n=237

0255075

100

Cl2Y(%)

Ring density(10⁴/mm²)

100

Fig. 3 | Partial halogenated FtsZ disrupts dynamic pattern formations of wild

type FtsZ. a Schematic representation of the self-organization assay with wild type

FtsZ-YFP-mts (Y) and FtsZ(Y222Cl

Y)-YFP-mts (Cl

Y) at certain proportions.

bRepresentative cytoskeleton images of wild type FtsZ-YFP-mts and

FtsZ(Y222Cl

Y)-YFP-mts at certain proportions on supported membrane (0.5 µM

proteins, 4 mM GTP and 1 mM Mg2+). Scale bar: 5 μm. The experiment was per-

formed three times under identical conditions. cRing densities for wild type FtsZ-

YFP-mts mixed with FtsZ(Y222Cl

Y)-YFP-mts. Data from three independent repli-

cates are shown as means with standard deviations. dDistributions of curvature

(left), ring size (middle), and ring velocity (right) of wild type FtsZ-YFP-mts mixed

with FtsZ(Y222Cl

Y)-YFP-mts in certain proportions. Curvatures in d. were calcu-

lated for the mixture of ring and ﬁlaments, representing the overall curvatures of

the bulk reaction. The solid curves in d(left panel) represent the extreme ﬁt for the

histograms (n

100

=829,n

= 781, n

=312,n

=217,n

= 1475). The histograms of

the curvatures are not shown. Box plots in d(inset ofleft panel):the lines represent

medians, box limits represent quartiles 1 and 3, whiskers represent 1.5 × inter-

quartile range and points are outliers. The solid curves in the middle and right

panels represent the Gaussian ﬁt. Source data of c,dare provided as a Source Data

ﬁle. D,Vand 1/R in Fig. 3indicate the ring diameter, velocity and curvature

separately.

Article https://doi.org/10.1038/s41467-022-32535-2

Nature Communications | (2022) 13:4843 7

Methods

Materials

Primers were synthetized by Integrated DNA Technologies.

Sequences of genes used in this research can be found in the sup-

plementary information. All the plasmids used in this study can be

found in Supplementary Table 8. Unless indicated, all standard

chemicals were purchased from Sigma-Aldrich, New England

BioLabs (NEB), and Thermo, Inc. Halogenated Tyr analogs (HYs)

were purchased from Sigma-Aldrich, abcr GmbH, and TCI: 3-chloro-

L-tyrosine (ClY, 97%, CAS 7423-93-0), 3,5-dichloro-L-tyrosine (Cl

≥98.0%, CAS 15106-62-4), 3-bromo-L-tyrosine (BrY, 97%, CAS 38739-

13-8), 3,5-dibromo-L-tyrosine (Br

Y, ≥96.0% CAS 300-38-9), 3-

iodo-L-tyrosine (IY, 95%, CAS 70-78-0), 3,5-diiodo-L-tyrosine

Y, ≥98.0%, CAS 18835-59-1)

11.2o

17.9o

0100 200 300 400 500

Time (ns)

RMSD (Å)

ClY

0 100 200 300 400 500

Time (ns)

RMSD (Å)

ClY

Monomer

Dimer

ClY

Mon-GTP Dimer-GTP

-36

-46

-133

Average binding energies (KJ/mol)

Dimer (Y) Dimer (ClY)

NTD

CTD

Loop

Y222

GTP

abc

def

ghi

14.5

10.2

5.4 5.1

4.3

10.2

Q332

K154

K154 T151

T151

Cl2Y222

Y222

Cl2Y

Monomer Dimer

Cl2Y

Cl2

Cl2Y

Filament

27.9 nm 17.9 nm

Monomer

Cl Cl

CE1

CD2

CD1

HD1 HD2

HB1

HB2

HB3

Cl2

CE2

Cl1

524K

271L

524K

271L 271L

Fig. 4 | Structure simulations of halogenated FtsZ. a Monomer structure of wild

type FtsZ. The main protein body is shown in cartoon representation, GTP and

Tyr222 are shown in licorice representation. The NTD is colored red, CTD is blue

and the central H7 helix is shown in green. bCharge calculation of 3,5-dichloro-

tyrosine. cAverage binding energies calculated according to the protocol descri-

bed in Methods section using the Adaptive Poisson-Boltzmann Solver (APBS)

software with time frames from the last 100ns of MD simulations. Two sets of

interactions were analyzed: Mon-GTP describes the interaction between monomer

body and Mg2+-GTP; Di-GTP illustrates the interactions between dimer body and

Mg2+-GTP at dimer interface. dStructural alignment of FtsZ(Y222Cl

Y) (Red) with

the wild type FtsZ (cyan). eOverlay of the mutant sites of wild type FtsZ(cyan) and

FtsZ(Y222Cl

Y) (red), showing the conformational changes induced by the

Y222Cl

Ymutation.fTime evolution of the backbone RMSD of FtsZ monomer

(backbone atoms only) after alignment to the energy-minimized FtsZ(Y222Cl

model. Residues 13to 316 areshown, excluding the disordered regions of sequence

(head and C-terminal tail). g,hThe bending motion of dimer models. Dimer

alignments of gwild type FtsZ (cyan) and hFtsZ(Y222Cl

Y) (red) with their energy-

minimizedmodels(grey).iTimeevolutionofthebackboneRMSDoftheFtsZdimer

when aligned to same selection of reference dimer structures, after energy mini-

mization using AMBER software66 in the dielectric constant (ε=4).Residues13–316

are shown for each monomer. jThe scheme illustrates protein self-organization

amplify the changes of structure by site-speciﬁc halogenation. Source data of fand

iare provided as a Source Data ﬁle.

Article https://doi.org/10.1038/s41467-022-32535-2

Nature Communications | (2022) 13:4843 8

One-round positive selection of MjTyrRS variants

For one-round positive selection (Supplementary Fig. 1a), a dual-

reporter selection plasmid pPAB26 (Supplementary Fig. 1c) was con-

structed, which encodes both small ubiquitin-like modiﬁer tagged

superfolder green ﬂuorescent protein (SUMO-sfGFP) with amber

codon at Arg2 position driven by T5 promoter (can elicit gene

expression in DH10b) and chloramphenicol acetyltransferase reporter

cassettes containing the TAG stop codon at Gln98, Asp18160.The

experiment design implies that thesynthetase-dependent suppression

of amber stop codons (TAGs) should result in the cell survival and

production of a ﬂuorescence signal at the same time. The Methano-

caldococcus jannaschii-Tyrosyl-tRNA-Synthetase (MjTyrRS) was used

in this study. MjTyrRS gene library was constructed on pBU18’1GK

plasmid (pBU18’1GK_MjTyrRS_library). One round-positive selection

was performed in freshly prepared electro-competent DH10b carrying

the selection plasmid pPAB26 by transformation of 100 ng MjTyrRS

gene library (Supplementary Fig. 1b and Supplementary Table 1)61.

Cells were grown on new minimum medium (NMM62)agarplates

(24 cm× 24 cm) with 1 mM Br

YorCl

Y, 100 µg/mL Ampicillin (Amp)

(for propagation of the library plasmid), 50 µg/mL Kanamycin (Kan)

(for maintaining positive selection plasmid), 70 μg/mL chlor-

amphenicol (Cm) (for positive selection pressure) and 0.5 mM iso-

propyl β-D-1-thiogalactopyranoside (IPTG) under 37°C for 24h, while

the control plate did not contain the Br

YorCl

Y. After selection, more

ﬂuorescent colonies were observed on the plate in the presence of

YorCl

Ycompared to the control. Green ﬂuorescent colonies from

the Br

YorCl

Y containing plate were randomly selected and subse-

quently screened by streaking on NMM plates supplemented with Cm

(70 μg/mL) with or without 1 mM Br

YorCl

Y. 37 out of 100 clones

showed Br

Y-dependent survival and green ﬂuorescence. Meanwhile,

27 out of 66 colonies can grow in the presence of Cl

Yandgenerate

green ﬂuorescence (Supplementary Fig. 2a). The MjTyrRSs from active

clones (abbreviated as BRS and CRS) were sequenced for analysis of

the mutations in the synthetase. As a result, 9 unique synthetase

(Supplementary Table 2) mutants were obtained after comparing the

sequences of all the synthetases screened against Cl

YandBr

Y. Nine

selected clones containing unique synthetase genes of interest were

directly applied to a 96-well ﬂuorescence assay to test the incorpora-

tion efﬁciency or poly-speciﬁcity of screened MjTyrRSs. The selected

MjTyrRSs speciﬁc for HYs were then cloned into pULTRA63 for halo-

genated FtsZ production.

Analysis of HYs incorporation through intact cell ﬂuores-

cence assay

The intact cell ﬂuorescence assay was performed to approximately

compare the Br

YandCl

Yincorporationefﬁciency of screened

MjTyrRSs as well as to estimate their promiscuity against other Tyr

analogues.Since the gene expressionof SUMO-sfGFP wasdriven byT5

promoter on the one round positive selection plasmid, SUMO-sfGFP

can be directly expressed in DH10b strain. Then clones containing the

9 unique MjTyrRSs (pBU18’1GK plasmid) and positive selection plas-

mid (pPAB26_cat (Q98TAG, D181TAG) MjtRNATyrCUA-his-SUMO-

sfGFPR2TAG-strep), which selectively grew on the plates in the pre-

sence of Br

YorCl

Y were directly cultured in LB medium with Amp

and Kan at 37 ˚C, overnight. Next day, an overnight culture was

inoculated in a well at a ratio of 1:100 containing 300 μLTBmedium

supplemented with 50 μg/mL Kan,100 μg/mL Amp, 0.5 mM IPTG as

well as 1 mM HYs while the corresponding control lacked the HYs. Cells

were grown in 96 well plates (Ibidi) with orbital shaking (2mm

amplitude) for 24 h at 37 °C covered with a gas-permeable foil (Sigma

Aldrich, Taufkirchen, Germany). The optical density OD

600

along with

the ﬂuorescence of the bacterial cultures were directly measured via

bottom reading using excitation and emission wavelengths of

481 ± 4.5 nm and 511 ± 10 nm, respectively, and a ﬁxed manual gain of

85. After 24 h incubation, the optimal density and ﬂuorescence was

measured via top reading after the gas-permeable foil was removed.

Each measurement was done in triplicate.

SUMO-sfGFP protein expression and puriﬁcation

The clones containing the pBU18’1GK_B48RS (for ClY, Cl

Y, BrY, Br

and IY incorporation) or pBU18’1GK_C64RS (for I

Yincorporation)

plasmids together with positive selection plasmid were directly cul-

tured inLB mediumwithAmp and Kanat37 °C,overnight.1:50or 1:100

of preculture and was transferred to TB medium. Afterwards, culture

was shaken at 37 °C until OD

600

reaching around 1. Then 1 mM HYs

were added directly into culture before the target gene induction.

The culture was shaken continuously for another 30–40 min to

allow dissolving and cellular uptake of HYs. Protein expression was

induced by 1 mM IPTG for overnight at 37 °C. The SUMO-sfGFP was

puriﬁed by Ni-NTA afﬁnity chromatography (Immobilized metal ion

chromatography: IMAC).

Living E. coli imaging and sample preparations

Prior to imaging, all cells were grown from a single colony in LB media

overnight at 37°C. For our default slow growth condition, cells were

then diluted in M9 minimal media supplemented with 0.4% Glucose,

thiamine and biotin, and proper antibiotics for plasmids maintenance.

Then cells were grown at 37°C until OD

600

between 0.2–0.5. The

expression of wild type FtsZ was induced with 20 μM IPTG, grown at

room temperature for 3h. For halogenated FtsZ variants expression,

HYs (2mM) were added directly into culture before the target gene

induction. Then the culture was shaken continuously for another

30 min to allow dissolving and cellular uptake of the HYs. Then halo-

genated FtsZ was induced with 20 μMIPTGat37°Cfor3h.Afterwards,

1 ml cells were collected and wash twice with PBS buffer and ﬁnally

resuspended in PBS. The cells were dropped on the Poly-D-lysine

coated coverslips and imaged with confocal microscopy. FRAP in

Supplementary Fig. 5h–i was achieved by focusing a pulsed white light

laser to a diffraction-limited spot on the specimen for an exposure of

≈200 ms. Confocal ﬂuorescent images of the cells were acquired

before and after photobleaching on a Leica SP8 confocal microscope

equipped with an 100x HC PL APO oil objective (NA 1.49). Images of

cells were obtained every 3 s for 30 s.

FtsZ protein expression and puriﬁcation

The selected synthetase B48RS and C64RS were cloned into pUL-

TRA plasmid. This highly efﬁcient suppressor plasmid, pULTRA,

harbors a single copy each of the tRNA and aaRS expression

cassettes that exhibits higher suppression activity than its

predecessors63. Wild type FtsZ-YFP-mts was cloned in pET-11b

expression vector and transformed into E. coli strain BL21(DE3).

Proteins were induced by 0.5 mM IPTG at 20 °C for overnight. The

gene of FtsZ(Y222TAG)-YFP-mts containing an amber codon at Y222

position was cloned on pET-28a expression vector and co-

transformed with pULTRA_B48RS_tRNATyrCUA (for ClY, Cl

Y, BrY,

Y, and IY incorporation) or pULTRA_C64RS_tRNATyrCUA (for I

incorporation) into RF1 free BL21(DE3) (B95.ΔA)64. A preculture

(1–50 mL LB with 25 mg/mL Kan, 100 mg/mL Spec and 1% glucose)

was grown overnight. 1:50 or 1:100 of preculture and was trans-

ferred to TB medium. Afterwards, culture was shaken at 37 °C until

600

reaching around 1. Then 1 mM HYs were added directly into

culture before the target gene induction. The culture was shaken

continuously for another 30–40 min to allow dissolving and cellular

uptake of HYs. Protein expression was induced by 0.3 mM IPTG for

3.5 h at 37 °C. Then FtsZ chimeric proteins were puriﬁed as pre-

viously described65.Brieﬂy, protein was precipitated from the

supernatant by adding 30% ammonium sulphate at 4 °C. After-

wards, the precipitate was shaken for 20 min at 4 °C with slow

speed. After centrifugation and resuspension of the pellet, the

protein was puriﬁed by anion exchange chromatography using a

Article https://doi.org/10.1038/s41467-022-32535-2

Nature Communications | (2022) 13:4843 9

5 ml Hi-Trap Q-Sepharose column (GE Healthcare, 17515601). Purity

of the protein was conﬁrmed by SDS-PAGE and mass spectrometry.

Mass-spectrometry

Mass-spectrometry Intact mass measurements of puriﬁed proteins

were performed by electrospray LC-MS on an Agilent 6530 QTOF

instrument coupled with an Agilent 1260 HPLC system after external

calibration. 80–100 μL of a protein solution with a concentration

around 0.1 mg/mLwasprepared.Sampleswereinfusedata ﬂow rate of

0.3mLmin

−1onto a gradient from 5% acetonitrile 0.1% formic acid in

water to 80% acetonitrile 0.1% formic acid in water through a C5 col-

umn, 2.1 × 100 mm, 3 micron (Supelco analytical, Sigma-Aldrich) over

20 minutes. The protein was ionized via electrospray ionization (ESI).

Spectra deconvolution was performed with Agilent MassHunter Qua-

litative Analysis software (v. B.06.00, Bioconﬁrm Intact mass module)

employing the maximum entropy deconvolution algorithm.

GTPase activity assay of FtsZ-YFP-mts

GTPase activities of FtsZ-YFP-mts were determined by measuring

released inorganic phosphate using BIOMOL®GREEN assay (Enzo Life

Sciences USA). GTP hydrolysis reaction was performed in poly-

merization buffer (50mM Tris/HCl, 300mM KCl, 5 mM Mg2+,pH7.5)

using FtsZ-YFP-mts at 5 μM with 1 mM GTP. Reactions were performed

every 20 s for a total time of 120 s. After 30 min of incubation with

BIOMOL®GREEN at room temperature, the absorbance of the samples

at 620 nm was monitored by TECAN plate reader. The phosphate

release concentrations were calculated based on a phosphate

standard curve.

Dynamic light scattering (DLS)

DLS of FtsZ was measured by a Protein Solutions DynaPro MS/X

instrument (Wyatt) at 25°C using 90°light scattering cuvette. FtsZ was

added in SLB buffer to a ﬁnal concentration of 12.5µΜ,andsamples

were measured using a ﬂuorometer cuvette with a 1-cm path length.

Prior to measurements, samples were centrifuged for 10 min at 105×g.

Excitation and emission wavelengths were set to 350nm, with slit

widths of 5 nm. Data was collected for 5 min to get baseline. After-

wards, 4 mM GTP was added to achieve a ﬁnal reaction volume of

300 μl. The reaction mixture was gently stirred and returned to

chamber for data collection for at least 70min.

Small unilamellar vesicles (SUVs) preparation

1,2-dioleoyl-sn-glycero-3-phosphocholine (DOPC):1,2-dioleoyl-sn-gly-

cero-3-phospho-(1’-rac-glycerol) (DOPG), 70:30mol % mixture, was

dissolved in chloroform in a quartz container, then dehumidiﬁed

under a gas nitrogen stream. Chloroformtraces were further removed

through desiccation (1 h). Afterwards, the lipid ﬁlm was hydrated to a

ﬁnal concentration of 4 mg/ml in supported lipid bilayers (SLB) buffer

(50mM Tris-HCl at pH 7.5, 150mM KCl), and incubated at 37°C for

30 min. The lipid ﬁlm was then completely resuspended by vortexing

rigorously to obtain multilamellar vesicles of different sizes. This

mixture was then placed in a bath sonicator where shear forces help to

reduce the size of the vesicles, giving rise to small unilamellar vesicles

(SUVs). The SUV dispersion were stored at −20 °C as 20 µl aliquots.

Supported lipid bilayer (SLB) preparation

Glass coverslips (1.5#, 24 x 24 mm) were cleaned by piranha solution

overnight, followed by extensive washing with milliQ H

O. Then glass

coverslips were blown dry with compressed air. The reaction chamber

was prepared by attaching a plastic ring on a cleaned glass coverslip

using ultraviolet glue (Thorlabs No. 68). For supported lipid bilayer

formation, the SUV dispersion was diluted in SLB buffer to 0.5mg/ml,

of which 75 µl was added to the reaction chamber. Adding CaCl

to a

ﬁnal concentration of 3 mM induced fusion of the vesicles and the

formation of a lipid bilayer on coversilde. After 20 min of incubation at

37°C, the sample was rinsed with 2ml pre-warmed SLB buffer.

Self-organization assays

FtsZ-YFP-mts was added to the reaction buffer above the supported

lipid membrane in the chamber. The ﬁnal volume of a sample was

approximately 250 μl. FtsZ-YFP-mts was added with a ﬁnal con-

centration of 0.5 μM. Polymerization was induced by adding

4mMGTP.

Total internal reﬂection ﬂuorescence microscope (TIRFM)

imaging

All experiments were performed on a WF1 GE DeltaVision Elite Total

internal reﬂection ﬂuorescence microscope (TIRFM, GE Healthcare

Life Sciences, Germany) equipped with an OLYMPUS 100× TIRF

objective (NA 1.49). The UltimateFocus feature of DeltaVision Elite

maintains the focus plane constant in time. FtsZ-YFP-mts was excited

with a 488 nm diode laser (10 mW before objective). Fluorescence

imaging was performed using a standard FITC ﬁlter set. Images were

acquired with a PCO sCMOS 5.5 camera (PCO, Germany) controlled by

the softWoRx Software (GE Healthcare Life Sciences, Germany). For

time-lapse experiments, images were acquired every 3 s with a 0.05 s

exposure time and light illumination shuttered between acquisitions.

Ring velocity analysis and processing

FtsZ ring velocities were analyzed according to the published

approach47. Brieﬂy, image analysis was carried out in MATLAB 2016s

(MATLABandImageProcessing andComputerVisionToolbox Release

2016a, The MathWorks, Inc., Natick, Massachusetts, USA) and pro-

cessingwithFiji/ImageJ(1.53f51).Imagescorrespondedtoanaverageof

5–10 frames from a time-series experiment. For the kymograph ana-

lysis, time-series acquisitions were ﬁltered using a standard mean ﬁlter

(2 pixel) and were drift corrected (multistackreg plugin). A MATLAB

script allowed the user to deﬁnearingbyprovidingtwocoordinates.

Every ring was automatically ﬁtted to a circle with radius r. Then, three

trajectories corresponding to three concentric circles having radii r,

r+1,andr−1 pixels were determined. At this point, the script read the

time-series data and calculated a kymograph for each time point and

trajectory. To automatically calculate the slope, the kymograph

was smoothed with a Savitzky-Golay ﬁlter of order 2 and enhance

its contrast using a contrast-limited-adaptive-histogram-equalization

(CLAHE) routine (MATLAB). Next, using Fourier analysis from previous

study47, the characteristic frequency for the patterns on the kymo-

graph was found. Finally, the slope corresponded to the change in

phase at this frequency. Quality criteriawere properly chosen to reject

low-quality regions over the kymograph. To synchronize time-lapse

acquisitions, the initial frame (time 0) was deﬁned when surface mean

intensity was around 200 A.U.

Fluorescence recovery after photobleaching data (FRAP)

Fluorescence recovery after photobleaching data (FRAP) on the SLB

were evaluated by choosing two separately circle areas (r=4µm). One

circle was taken as a reference. Another one was photobleached by 20

iterations of 488nm, laser under 100% laser power.Theirﬂuorescence

recovery in the green channel was monitored. Intensity traces were

collected using Fiji, corrected for photobleaching and normalized with

the pre-bleaching intensity and the reference intensity. The ﬂuores-

cence recovery curves were then ﬁtted with mono-exponential equa-

tion using Originpro (v_2017 & v_2019):

Y=Aekt +B,ð1Þ

where Yrepresents ﬂuorescence intensity at time t,Aand Bare para-

meters, and krepresents the rate constant. The half-life, t

1/2

,was

Article https://doi.org/10.1038/s41467-022-32535-2

Nature Communications | (2022) 13:4843 10

determinedusingthefollowingequation:

t1=2=ln2k1ð2Þ

Side-chain acidity

10–20 mg of an amino acid, 50 mg of KH

was dissolved in 15 ml of

water and titrated by sodium hydroxide solution to distinct pH values.

500 μl were taken to NMR tubes, 50 μl deuterium oxidewas added and

1H NMR spectra were measured using W5 pulse tray water suppression

at 500MHz frequency. The chemical shifts were plotted against pH

and analyzed using Henderson-Hasselbalch equation to deliver the pK

values. Experimental error ±0.15.

Ammonium pK

: Y 9.44, ClY 9.50, Cl

Y9.40,BrY9.52,Br

Y 9.47, IY

9.37, I

Y9.38.

Phenolic pK

: Y 9.90, ClY 8.31, Cl

Y 6.52, BrY 8.29, Br

Y 6.42, IY

8.30, I

Y6.45.

Lipophilicity analysis

The lipophilicity of the amino acid residues was measured against

150 mM buffers as described32.Moredetailscanbefoundinthesup-

plement methods. N-Acetyl-tyrosine methyl ester and N-acetyl-3-iodo-

tyrosine methyl ester were reported earlier, and other compounds

were synthesized according to the protocol. The distribution coefﬁ-

cients are provided in Supplementary Table 7.

Statistics and reproducibility

Unless otherwise mentioned, all measurements were performed for at

least three independent experiments. The numbers of experimental

repeats are indicated in the respective ﬁgure legends. The statistical

tests were analyzed by ANOVA one-way statistical tests (signiﬁcance

level, 0.05; mean comparison, Tukey; tests for equal variance, Levene;

power analysis, actual power). All pvalues are given in the respective

ﬁgure captions.

Reporting summary

Further information on research design is available in the Nature

Research Reporting Summary linked to this article.

Data availability

All data used in this paper are available at Figshare through the iden-

tiﬁer [https://doi.org/10.6084/m9.ﬁgshare.20368626.v1]orfromthe

corresponding authors upon request. Source data are provided with

this paper. All the structures used in this study are publicly available

under the PDB accession codes 1W5A (FtsZ dimer, MgGTP soak), 4NX2

(Crystal structure of DCYRS complexed with DCY) and 3GFD (Crystal

structure of IYD bound to FMN and mono-iodotyrosine MIT) Source

data are provided with this paper.

Code availability

No custom code was used in this research. The previously published

code of ring velocity analysis is available upon request from author

(Prof. Petra Schwille, schwille@biochem.mpg.de, PLoS biology, 2018,

16(5): e2004845).

References

1. Harmel, R. & Fiedler, D. Features and regulation of non-enzymatic

post-translational modiﬁcations. Nat. Chem. Biol. 14,

244–252 (2018).

2. Pimviriyakul, P., Wongnate, T., Tinikul, R. & Chaiyen, P. Microbial

degradation of halogenated aromatics: molecular mechanisms and

enzymatic reactions. Microb. Biotechnol. 13,67–86 (2020).

3. Feeney, M. B. & Schöneich, C. Tyrosine modiﬁcations in aging.

Antioxid. Redox Signal. 17,1571–1579 (2012).

4. Olaniran,A.O.&Igbinosa,E.O.Chlorophenols and other related

derivatives of environmental concern: properties, distribution and

microbial degradation processes. Chemosphere 83,

1297–1306 (2011).

5. Santos, A. L. & Lindner, A. B. Protein posttranslational modiﬁcations:

roles in aging and age-related disease. Oxid. Med. Cell. Longev.

2017,1–19 (2017).

6. Santra,M.,Dill,K.A.&deGraff,A.M.Proteostasiscollapseisa

driver of cell aging and death. Proc. Natl Acad. Sci. U. S. A. 116,

22173–22178 (2019).

7. Liguori, I. et al. Sarcopenia: assessment of disease burden and

strategies to improve outcomes. Clin. Inter. Aging 13, 913 (2018).

8. Arora,P.K.,Srivastava,A.,Garg,S.K.&Singh,V.P.Recentadvances

in degradation of chloronitrophenols. Bioresour. Technol. 250,

902–909 (2018).

9. Arora, P. K., Sasikala, C. & Ramana, C. V. Degradation of chlorinated

nitroaromatic compounds. Appl.Microbiol.Biotechnol.93,

2265–2277 (2012).

10. Sadowska-Bartosz, I., Ott, C., Grune, T. & Bartosz, G. Posttransla-

tional protein modiﬁcations by reactive nitrogen and chlorine spe-

cies and strategies for their prevention and elimination. Free Radic.

Res. 48,1267–1284 (2014).

11. Ros, E., Torres, A. G. & de Pouplana, L. R. Learning from Nature to

Expand the Genetic Code. Trends Biotechnol. 39,460–473 (2021).

12. Chin, J. W. Expanding and reprogramming the genetic code. Nature

550,53–60 (2017).

13. Davis, L. & Chin, J. W. Designer proteins: applications of genetic

code expansion in cell biology. Nat. Rev. Mol. Cell Biol. 13,

168–182 (2012).

14. Hoppmann, C. et al. Site-speciﬁc incorporation of phosphotyrosine

usinganexpandedgeneticcode.Nat. Chem. Biol. 13,

842–844 (2017).

15. Porter, J. J. et al. Genetically encoded protein tyrosine nitration in

mammalian cells. ACS Chem. Biol. 14,1328–1336 (2019).

16. Moroder, L. & Budisa, N. Synthetic biology of protein folding.

ChemPhysChem 11, 1181–1187 (2010).

17. Neumann, H., Hazen, J. L., Weinstein, J., Mehl, R. A. & Chin, J. W.

Genetically encoding protein oxidative damage. J. Am. Chem. Soc.

130,4028–4033 (2008).

18. Wolschner, C. et al. Design of anti-and pro-aggregation variants to

assess the effects of methionine oxidation in human prion protein.

Proc. Natl Acad. Sci. U. S. A 106, 7756–7761 (2009).

19. Nguyen,D.P.,GarciaAlai,M.M.,Kapadnis,P.B.,Neumann,H.&

Chin,J.W.GeneticallyencodingNϵ-methyl-l-lysine in recombinant

histones. J. Am. Chem. Soc. 131,14194–14195 (2009).

20. Elsässer,S.J.,Ernst,R.J.,Walker,O.S.&Chin,J.W.Geneticcode

expansioninstablecelllinesenables encoded chromatin mod-

iﬁcation. Nat. Methods 13,158–164 (2016).

21. Wang, L. Engineering the genetic code in cells and animals: biolo-

gical considerations and impacts. Acc. Chem. Res. 50,

2767–2775 (2017).

22. Ernst, R. J. et al. Genetic code expansion in the mouse brain. Nat.

Chem. Biol. 12,776–778 (2016).

23. Wang, B. et al. The oxidized thiol proteome in aging and cataractous

mouse and human lens revealed by ICAT labeling. Aging Cell 16,

244–261 (2017).

24. Jia, H. & Schwille, P. Bottom-up synthetic biology: reconstitution in

space and time. Curr. Opin. Biotechnol. 60,179–187 (2019).

25. Läppchen, T. et al. Probing FtsZ and tubulin with C8-substituted

GTP analogs reveals differences in their nucleotide binding sites.

Chem. Biol. 15,189

–199 (2008).

26. Sun,H.,Jia,H.,Ramirez‐Diaz,D.A.,Budisa,N.&Schwille,P.Fine-

Tuning Protein Self‐Organization by Orthogonal Chemo-

Optogenetic Tools. Angew. Chem. Int. Ed. 133,4551–4556 (2021).

Article https://doi.org/10.1038/s41467-022-32535-2

Nature Communications | (2022) 13:4843 11