Site-specific incorporation of fluorotyrosines into proteins in escherichia coli by photochemical disguise

Article abstract of DOI:10.1021/bi100013s

Fluorinated analogues of tyrosine can be used to manipulate the electronic environments of protein active sites. The ability to selectively mutate tyrosine residues to fluorotyrosines is limited, however, and can currently only be achieved through the tot

Full text of DOI:10.1021/bi100013s

Biochemistry 2010, 49, 1557–1559 1557
DOI: 10.1021/bi100013s
Site-Specific Incorporation of Fluorotyrosines into Proteins in Escherichia coli by
Photochemical Disguise†
Bryan J. Wilkins,‡ Samuel Marionni,‡ Douglas D. Young, Jia Liu,‡ Yan Wang,‡ Martino L. Di Salvo,§
Alexander Deiters,*, and T. Ashton Cropp*,‡
‡Department of Chemistry and Biochemistry, University of Maryland, College Park, Maryland 20742, §Dipartimento di Scienze
ꢀ
Biochimiche and Istituto Pasteur-Fondazione Cenci Bolognetti, Sapienza Universita di Roma, Piazzale Aldo Moro, 5-00185 Roma,
Italy, and Department of Chemistry, North Carolina State University, Raleigh, North Carolina 27695
Received January 6, 2010; Revised Manuscript Received February 1, 2010
ABSTRACT: Fluorinated analogues of tyrosine can be used
to manipulate the electronic environments of protein active
sites. The ability to selectively mutate tyrosine residues to
fluorotyrosines is limited, however, and can currently only
be achieved through the total synthesis of proteins. As a
general solution to this problem, we genetically encoded
the unnatural amino acids o-nitrobenzyl-2-fluorotyrosine,
-3-fluorotyrosine, and -2,6-difluorotyrosine in Escherichia
coli. These amino acids are disguised from recognition by
the endogenous protein biosynthetic machinery, effectively
preventing global incorporation of fluorotyrosine into
proteins.
proteins can contain a multiplicity of fluorinated amino acid
substitutions.
Several attempts have been made to address this problem of
nonexistent site selectivity. Homogeneous proteins containing
fluorotyrosines have been produced by a combination of chemi-
cal peptide synthesis and expressed protein ligation (2). This is
limited to certain locations within a protein, typically at the
C-terminus, and can be technically challenging for proteins that
are sensitive to denaturation. Unnatural amino acid mutagenesis
has been performed in an in vitro protein expression system using
amber suppressor tRNAs that are chemically aminoacylated with
fluorotyrosines (6). This approach is limited by the requirement
for a laborious synthesis of the aminoacyl-tRNA and limited
protein yields. A more versatile method would be to use in vivo
unnatural amino acid mutagenesis based on an orthogonal
aminoacyl-tRNA synthetase (aaRS)-tRNA pair enabling pro-
tein expression in E. coli with fluorotyrosine encoded by the
amber stop codon, TAG (7). In this paper, we report a general
method for this approach by temporarily masking fluorotyro-
sines from cellular metabolism using a photoremovable protect-
ing group (8-10).
We first chose to investigate the incorporation of o-nitrobenzyl-
2-fluorotyrosine (6) into proteins using a variant of the Methano-
coccus jannaschii tyrosyl-tRNA synthetase (ONB-YRS) that
was previously altered to accept the nonfluorinated amino
acid o-nitrobenzyltyrosine (5) as a substrate (11). The synthesis
of the caged fluorinated tyrosines 6-8 was achieved through
complexation of the tyrosines 2-4 with Cu2þ followed by
alkylation with o-nitrobenzylbromide [Scheme 1 and Supporting
Information (SI)], as previously reported for the conversion of 1
to 5 (11). Amino acids 2-4 were produced in enantiomerically
pure form from the corresponding fluorophenols 9-11, respec-
tively, using the enzyme tyrosine phenol lyase (TPL) (12). We
assumed that with such a small structural change to the substrate,
6 would be accepted by this enzyme and the amino acid could be
inserted into proteins using the previously described machinery.
After protein production, the caging group could be removed
by light irradiation, revealing the desired fluorinated protein.
Unfortunately, repeated attempts at producing protein [using the
gene encoding superfolder green fluorescent protein (sfGFP) (13)
having the permissive V150 codon mutated to TAG] containing 6
with ONB-YRS were unsuccessful. This disruption of substrate
specificity from a single fluorine was quite astonishing given that
2 and 3 are both substrates for the endogenous E. coli TyrRS.
Thus, we decided to create a new aaRS variant capable of
accepting 6 as a substrate. A new library was constructed in which
The fluorine atom is often considered an isosteric replacement
for hydrogen, unlikely to perturb structure, yet with a high
electronegativity that enables its application as a biological
probe. For example, the pKa of a tyrosine phenolic proton is
approximately 10, but the pKa of fluorotyrosine residues can
range from 5.2 to 9.0 depending on the extent of fluorination
(1, 2). Therefore, the acidity of individual amino acid side chains
within a protein can be precisely modulated in investigating the
participation of a given tyrosine residue in an acid-base catalysis
mechanism. The same concept applies to the redox properties of
fluorinated tyrosine residues (3). In the study of biologically
generated tyrosyl radicals, fluorotyrosines have peak reduction
potentials that range from 705 to 968 mV, depending upon the
extent and position of fluorination. This allows fluorotyrosines to
be used as comparison probes for tyrosine, which has a peak
potential of 642 mV (3).
The main limitation for the precise use of fluorinated amino
acids is the process by which they are currently introduced into
proteins. Because of the structural similarities to natural amino
˚
acids (fluorine has an only 0.15 A larger van der Waals radius
than hydrogen), fluorinated amino acids are often introduced
into proteins via global incorporation. For example, Escherichia
coli cells starved of tyrosine (1) can be grown in the presence of
2 or 3 (Scheme 1) to force incorporation of these analogues in
place of all tyrosine residues (4, 5). The obvious concern for
metabolic labeling is that there is no site control and the sample
†We thank the National Science Foundation (Grant CHE-0848398 to
T.A.C. and A.D.) and the University of Maryland (Drs. Wayne T. and
Mary T. Hockmeyer Doctoral Fellowship to B.J.W.) for financial
support of this work. A.D. is a Beckman Young Investigator and a
Cottrell Scholar.
*To whom correspondence should be addressed. A.D.: e-mail,
alex_deiters@ncsu.edu; phone, (919) 513-2958. T.A.C.: e-mail, acropp@
gmail.com; phone, (301) 405-1734; fax, (301) 314-9121.
r
2010 American Chemical Society
Published on Web 02/05/2010
pubs.acs.org/Biochemistry

1558 Biochemistry, Vol. 49, No. 8, 2010
Wilkins et al.
Scheme 1: Structures and Synthesis of Caged Fluorotyrosines
seven residues within the active site of MjYRS (14) (Y32, L65,
H70, F108, D158, I159, and Ll62) were randomized or selectively
diversified, yielding >108 variants. It should be noted that the
original ONB-YRS was derived from a library that did not
include randomized residues H70 and I159, resulting in a slightly
different starting pool of aaRSs. This new library was then
applied in a double-sieve selection experiment, based on suppres-
sion of a stop codon in the gene encoding chloramphenicol acetyl
transferase (15), in which we attempted to isolate a variant that
accepts 6 as a substrate (details provided in the SI). After multiple
rounds of selection and screening, we isolated two clones, ONB-
2FYRS-1 and -2, that grew on agar containing 90 μg/mL
chloramphenicol in the presence of 6 but only 20 μg/mL
chloramphenicol in the absence of 6 (Figure 1). This suggests
that these enzymes are selective for 6 and do not accept endo-
genous amino acids as substrates. Moreover, examination of
the active site sequences of these enzymes revealed that they
were homologous, but different from the previously evolved
enzyme, suggesting an altered substrate specificity (Table S1 of
the SI).
To verify the site-specific insertion of 6 into a protein, we chose
to overexpress sfGFP using the variant ONB-2FYRS-1, sub-
cloned into a pSUP plasmid (16) for compatibility with our
expression system. Expression of the full-length protein was
observed only when 6 was included in the growth medium at
1 mM (Figure 1). The protein yield under these conditions was
approximately 2 mg/L, which is comparable to the level of
protein expression using 5 (11). In the absence of 6, no production
of full-length protein is observed on the basis of Coomassie-
stained gels and MS. To further verify the correct protein
product, we performed an in-gel tryptic digest on the sfGFP
protein and subjected the isolated fragments to mass spectro-
metric sequencing to confirm the site-specific incorporation of
the fluorotyrosine (see SI). A comparison of the analysis of
sfGFP expressed with ONB-YRS in the presence of 5 and
expressed with ONB-2FYRS in the presence of 6 revealed
identical peptide masses for both proteins apart from the one
fragment containing the mutation which differ by m/z 18, equal
to the mass difference between fluorine and hydrogen. We were
curious if the newly developed variants, ONB-2FYRS-1 and -2,
of the M. jannaschii enzyme would have substrate specificities
that are flexible to the other fluorotyrosine analogues, 7 and 8.
Indeed, cells containing either ONB-2FYRS variant supported
production of sfGFP containing 7; however, only ONB-2FYRS-
F
IGURE 1: Characterization of evolved aaRSs. (a) Growth on glyce-
rol minimal medium with leucine (GMML) agar containing chlo-
ramphenicol (90μg/mL). (b) ExpressionofsfGFP inthe presenceand
absence of 6. (c and d) Production of sfGFP containing 7 and 8 using
(c) ONB-2FYRS-1 and (d) ONB-2FYRS-2. All protein was isolated
by Ni2þ affinity chromatography.
2 produced significant quantities of protein using 8 (Figure 1c,d).
The successful incorporation of these amino acids into sfGFP
was further verified by mass spectrometry analyses on tryptic
fragments of these proteins as described above (see SI).
We next tested whether site-specific mutations of fluorotyro-
sines could be used to alter protein function. Previously, global
mutagenesis studies have given rise to altered activity, such as
pH-rate profiles, but these experiments used proteins containing
multiple mutations. One such study used global incorporation of
3-fluorotyrosine into EGFP (4, 17), which contains 11 total
tyrosine residues, one of which, Y66, is central to the fluoro-
phore. To conduct a more precise investigation with atom-
specific resolution, we expressed and purified sfGFP with only
Y66 mutated to 6-8. This protein, unlike any previous studies,
contains only a single-residue mutation to a fluorotyrosine. In
addition, we produced a control sfGFP, which instead contains 5,
as a surrogate for the wild-type tyrosine residue. All proteins were
purified and then “decaged” by irradiation at 365 nm for 30 min,
after which we assayed the protein for fluorescence properties. In
the case of fluorophore mutation 6 or 8, containing fluorines in
the ortho position, the λem is blue-shifted (503 or 509 nm,
respectively) when compared to the tyrosine control (513 nm)
(Figure 2b). Mutations containing the meta-substituted 7 result
in a red shift to 515 nm. This demonstrates the ability of a single
fluorotyrosine residue to change the electronic properties of the
chromophore and ultimately affect the protein function.
Finally, an interesting observation is that the sfGFP bearing
6 is fluorescent at 503 nm, despite the presence of the large
o-nitrobenzyl group blocking the tyrosine hydroxyl. Fluores-
cence increases by 343% after removal of the caging group
through a brief irradiation at 365 nm (Figure 2c). This increase in
fluorescence is most likely due to the liberation of the phenolic
hydroxy group which can now delocalize a negative charge
throughout the fluorophore ring system, which is not possible
when the amino acid is caged. What is not clear is if the sfGFP
mutant is fully folded and matured prior to removal of the large

Rapid Report
Biochemistry, Vol. 49, No. 8, 2010 1559
biosynthetic machinery using a light-removable protecting
(caging) group. This provides a unique solution to the problem
of preventing global incorporation of fluorotyrosine into pro-
teins. We are currently exploring whether this approach might be
used for other fluorinated tyrosine analogues (3), in which
case this methodology will enable the precise investigation of
acid-base and redox properties of tyrosine residues.
ACKNOWLEDGMENT
We are grateful to Prof. Robert Phillips (University of
Georgia, Athens, GA) for the TPL enzyme, Prof. Peter Schultz
(The Scripps Research Institute, La Jolla, CA) for materials, and
Dr. Yiu-fai Lam (University of Maryland) for NMR assistance.
SUPPORTING INFORMATION AVAILABLE
Detailed discussion and experimentals, aaRS, and MS ana-
lyses. This material is available free of charge via the Internet at
http://pubs.acs.org.
REFERENCES
1. Brooks, B., Phillips, R. S., and Benisek, W. F. (1998) Biochemistry 37,
9738–9742.
2. Seyedsayamdost, M. R., Yee, C. S., and Stubbe, J. (2007) Nat. Protoc.
2, 1225–1235.
3. Seyedsayamdost, M. R., Reece, S. Y., Nocera, D. G., and Stubbe, J.
(2006) J. Am. Chem. Soc. 128, 1569–1579.
4. Pal, P. P., Bae, J. H., Azim, M. K., Hess, P., Friedrich, R., Huber, R.,
Moroder, L., and Budisa, N. (2005) Biochemistry 44, 3663–3672.
5. Song, L., Teng, Q., Phillips, R. S., Brewer, J. M., and Summers, A. O.
(2007) J. Mol. Biol. 371, 79–92.
6. Thorson, J. S., Shin, I., Chapman, E., Stenberg, G., Mannervik, B.,
and Schultz, P. G. (1998) J. Am. Chem. Soc. 120, 451–452.
7. Cropp, T. A., and Schultz, P. G. (2004) Trends Genet. 20, 625–630.
8. Young, D. D., and Deiters, A. (2007) Org. Biomol. Chem. 5, 999–
1005.
9. Mayer, G., and Heckel, A. (2006) Angew. Chem., Int. Ed. 45, 4900–
4921.
10. Lee, H. M., Larson, D. R., and Lawrence, D. S. (2009) ACS Chem.
Biol. 4, 409–427.
11. Deiters, A., Groff, D., Ryu, Y., Xie, J., and Schultz, P. G. (2006)
Angew. Chem., Int. Ed. 45, 2728–2731.
12. Nagasawa, T., Utagawa, T., Goto, J., Kim, C. J., Tani, Y., Kumagai,
H., and Yamada, H. (1981) Eur. J. Biochem. 117, 33–40.
13. Pedelacq, J. D., Cabantous, S., Tran, T., Terwilliger, T. C., and
Waldo, G. S. (2006) Nat. Biotechnol. 24, 79–88.
14. Kobayashi, T., Nureki, O., Ishitani, R., Yaremchuk, A., Tukalo, M.,
Cusack, S., Sakamoto, K., and Yokoyama, S. (2003) Nat. Struct.
Biol. 10, 425–432.
F
IGURE 2: Fluorescence analysis of sfGFP containing unnatural
amino acids 5-8. (a) Schematic of the GFP active site. (b) Normal-
ized fluorescence of decaged protein containing 5-8. (c) Normalized
fluorescence of protein containing 5 before and after irradiation at
365 nm for 30 min showing activation.
ONB caging group. Nevertheless, photoactivation by the re-
moval of a protecting group from the chromophore of the protein
could be used as a general method to “switch” fluorescent
proteins, and the fact that these molecules are genetically encoded
may enable the use of directed evolution to improve the photo-
switching properties.
15. Wang, L., Brock, A., Herberich, B., and Schultz, P. G. (2001) Science
292, 498–500.
16. Ryu, Y., and Schultz, P. G. (2006) Nat. Methods 3, 263–265.
17. Khan, F., Kuprov, I., Craggs, T. D., Hore, P. J., and Jackson, S. E.
(2006) J. Am. Chem. Soc. 128, 10729–10737.
In summary, we developed a general method for the site-
specific incorporation of 2-, 3-, and 2,6-fluorotyrosine into
proteins in E. coli. These near-natural amino acids are tempora-
rily disguised from recognition by the endogenous protein