PNA-Induced Assembly of Fluorescent Proteins Using DNA as a Framework

Controlled alignment of proteins on molecular frameworks requires the development of facile and orthogonal chemical approaches and molecular scaffolds. In this work, protein−PNA conjugates are brought forward as new chemical components allowing efficient assembly and alignment on DNA scaffolds. Site-selective monomeric teal fluorescent protein (mTFP)−peptide nucleic acid (PNA) (mTFP-PNA) conjugation was achieved by covalent linkage of the PNA to the protein through expressed protein ligation (EPL). A DNA beacon, with 6-Fam and Dabcyl at its ends, acts as a framework to create an assembled hetero-FRET system with the mTFP-PNA conjugate. Using fluorescence intensity, frequency domain lifetime measurements, and anisotropy measurements, the system was shown to produce FRET as indicated by decreased donor intensity, decreased donor lifetime, and increased donor anisotropy. Extension of the DNA scaffold allowed for the assembly of multiple mTFP-PNA constructs. Efficient formation of protein dimers and oligomers on the DNA-PNA frameworks could be shown, as visualized via size exclusion chromatography (SEC) and electrophoresis (SDS-PAGE). Assembly of multiple proteins in a row induced homo-FRET for the mTFP-PNA’s assembled on the DNA scaffolds. The oligonucleotide framework allows an induced and controllable assembly of proteins by fusing them to PNAs directed to align on DNA scaffolds.


■ INTRODUCTION
Controlled assembly is of great importance in the modulation, localization, and interactions of proteins. For example, assembly of proteins plays an important role in vital biological process such as signal transduction, 1−3 transcription, 4 and formation of fibers and filaments. 5−7 Well-defined synthetic protein assemblies allow for molecular insights and targeting of biological processes. 8−17 Nevertheless, the molecular characteristics and diversity of protein chemistry makes designing accurate and specific self-assembled systems based on proteins a challenging endeavor. DNA possesses unique recognition capabilities as a scaffold to address precise and programmed self-assembled nanostructures. 18,19 The combination of the unique programmability of DNA-directed self-assembly and the versatility of functional proteins can be provided by semisynthetic DNA−protein conjugation techniques. DNA−protein conjugates have been used as controllable templates to mimic self-assembled protein complexes in biological systems. 19−21 For example, DNA−protein conjugates were applied in assembly of enzymatic networks, 22,23 highly ordered protein arrays, 16,24 and biomolecular delivery systems. 25 Despite the wide development of DNA-directed assembly systems, the high negative charge density of DNA may limit its practical application as a scaffold in self-assembled protein systems. Peptide nucleic acid (PNA) is a DNA mimic in which the regular phosphodiester backbone has been replaced by a pseudopeptide skeleton. 26,27 PNA binds to the complementary DNA and RNA through Watson−Crick base pairs with high affinity and sequence specificity. 28 PNA provides better stability than DNA duplexes, even for short sequences, and has higher mismatch sensitivity. 26,27 Moreover, unlike the DNA backbone, the pseudopeptide skeleton can be altered by replacing glycine with different amino acides, such as lysine or cysteine, to provide reactive amine or thiol groups to conjugate the PNA to proteins. 27 Expressed protein ligation (EPL) is a semisynthetic version of native chemical ligation enabling the site-selective attachment of synthetic constructs to proteins. 29 EPL has been used previously for the construction of a variety of peptide− protein 30,31 and protein−PNA conjugates. 32 EPL typically makes use of a recombinant protein containing a thioester or cysteine at the C-or N-terminus. 33−35 A protein with a Cterminal thioester provides a conjugation site for a PNA with an N-terminal cysteine, enabling a ligation via a native peptide bond. 36,37 Applying EPL to fluorescent proteins (FP) to create FP-PNA conjugates (Scheme 1) provides an ideal system for studying controlled protein assembly: (i) it provides precise control over the exact composition and structure of an inducibly assembled system following addition of a DNA template; (ii) it allows fundamental insights into the behavior of proteins in self-assembled architectures to be gained by observing the photophysical behavior of assembled FPs. Two relevant photophysical strategies that can be employed to observe template directed assembly of FP-PNA monomer units are: (i) assembly along a complementary DNA and detection via Forster resonance energy transfer (FRET) among like fluorophores (homo-FRET) (Scheme 2); and (ii) assembly with a DNA beacon and detection via conventional FRET (hetero-FRET) (Scheme 3). DNA beacons have been widely used for sequence recognition in biology, chemistry, medical sciences, and biotechnology since they were reported in 1996. 38 A variety of such beacons based on chimeric DNA-PNA and purely PNA beacons have been reported to provide stronger binding to templates. 39 This study reports a programmable model system allowing precise control over the assembly of proteins directed by the molecular recognition capabilities of PNA-DNA architectures. EPL allowed the creation of well-defined FP-PNA monomer units (Scheme 1). Subsequent assembly was directed by engineered DNA templates (Schemes 2 and 3). Unlabeled DNA templates enabled assembly of FPs into protein dimer and tetramer homo-FRET systems, which were studied using fluorescence anisotropy. A complementary DNA beacon hetero-FRET system, enabling single binding detection, was studied with fluorescence intensity and lifetime measurements. The photophysical methods, combined with mass spectrometry, SDS-PAGE, and size exclusion chromatography experiments, highlight the inducible formation of supramolecular protein assemblies on a predetermined scale.
Protein Expression and Purification. Monomeric teal fluorescent protein (mTFP) with an N-terminal 6-His-tag was expressed in E. coli using plasmid pHT581 possessing a Cterminal engineered intein and chitin binding domain (CBD), analogues to previously described compounds. 41 A thioester group was introduced at the C-terminus of the protein through cleavage of the intein using MESNA, and the resulting protein was purified using chitin binding affinity chromatography. Chitin resin (New England Biolabs) was transferred to a 10 mL disposable chromatography column (Biorad) and equilibrated with 10 volumes of washing buffer (Na 2 HPO 4 (19.3 mM), NaH 2 PO 4 (25 mM), NaCl (500 mM), EDTA (0.5 mM) in distilled deionized water. Cleavage of the intein-tag fused to the C-terminus of mTFP was performed through incubation of loaded proteins with 20 mL elution buffer (Na 2 HPO 4 (23.2 mM), NaH 2 PO 4 (25 mM), NaCl (100 mM), EDTA (0.5 mM), MESNA (400 mM) in ddH 2 O, pH 7.5) in the dark with slow shaking overnight at room temperature. Afterward, the cleavage buffer was replaced by storage buffer (Na 2 HPO 4 (19.3 mM), NaH 2 PO 4 (25 mM), NaCl (50 mM), EDTA (0.1 mM), in ddH 2 O (pH 7). The collected proteins were concentrated by centrifugation at 3700 g using centrifugal filter tubes (Amicon 10000 MW, Millipore) for 10−16 min several times to remove the MESNA. The resulting mTFP was further purified using His-tag affinity chromatography. To prepare the Ni-NTA resin, it was washed with 2−3 column volumes of distilled water. Nickel sulfate solution (50 mM) was added to the column followed by washing with distilled water and His-tag buffer (Na 2 HPO 4 (19.3 mM), NaH 2 PO 4 (25 mM), NaCl (50 mM), EDTA (0.1 mM) in ddH 2 O, imidazole (20 mM), pH 7). After loading of the protein, the column was washed again with 10 mL of His-tag washing buffer and the proteins eluted using 4 mL His-tag elution buffer (Na 2 HPO 4 (19. The product was isolated using centrifugal filter tubes (Amicon 10000 MW, Millipore) at 13 000 rpm. The retained portion was resuspended in 70 mM TCEP, incubated, and centrifuged several times to reduce disulfide bond generated at the ligation site. The mTFP-PNA conjugate was subsequently stored at 4°C (short times) or −20°C (long-term storage). Products were analyzed by MALDI-TOF MS and yields quantified by UV−vis spectrophotometry. Ligation samples were diluted and their absorbance at 260 nm measured to estimate PNA concentration. The extinction coefficient of PNA was estimated based on the published values for the molar extinction coefficients of PNA bases. 37 Pure mTFP absorbance at 260 nm was measured and subtracted from the mTFP-PNA absorbance.
Hybridization and Assembly of mTFP-PNA with DNA. Two DNA sequences (DNA1 5′TGCATGGATCTGC-ATG3′and DNA2: 5′TGCATGGATCTGCATGGATC-TGCATGGATCTGCATG3′) (Life technologies corporation, USA) were used to make assembled mTFP-PNA dimers and tetramers, respectively (Note: The PNA:DNA mediated protein assemblies are referred to by the number of proteins assembled; e.g., a dimer consists of two proteins and the DNA scaffold and a tetramer of four proteins and the DNA scaffold). The hybridization of DNA to mTFP-PNA was determined by SDS-PAGE electrophoresis. 1.25 μM DNA1 and 0.62 μM DNA2 were titrated with different concentrations of mTFP-PNA (1.25−2.5 μM and 0.62−2.5 μM for dimer and tetramer assembly, respectively) in phosphate buffer (100 mM, NaCl (200 mM), pH 7) with 2 h incubation at room temperature. The results were assessed by SEC-HPLC at 214 and 462 nm using phosphate buffer (150 mM, pH 7) as mobile phase.
Photophysical Measurements. Different concentrations of DNA1 and DNA2 (0−2.5 μM) were added to 2.5 μM mTFP-PNA and the emission intensity and anisotropy recorded between 480 and 530 nm using a fluorimeter equipped with polarizers (Cary Eclipse; Varian). Fluorescence lifetimes were measured using a spectroscopic fluorescence lifetime imaging system operating in the frequency domain 42 with rhodamine 6G as a lifetime reference. Energy transfer efficiency (E) can be estimated based on R 0 (the Forster distance in which the transfer efficiency E is 50%) and r (distance between fluorophores) (eq 1). 43 (1) In the case of heterotransfer this can be assessed by lifetimes or intensity where τ 1 and I 1 are the lifetime and intensity of the donor in the absence of acceptor, and τ 2 and I 2 are the lifetime and intensity in the presence of the acceptor. When frequency domain lifetime measurements are used, the measured phase, ϕ i , and modulation, m i , may be presented in polar coordinates.
In polar coordinate analysis, the fraction of mTFP-PNA in the hybridized form (α) can be calculated based on the following equation: where (x 1 , y 1 ), (x 2 , y 2 ), and (x i , y i ) are the coordinates in the absence of acceptor, the fully hybridized form, and mixture of free and hybridized forms, respectively. Lifetime does not change in the case of homotransfer. As a result the fluorescence anisotropy is a useful indicator of assembly. To a first approximation, the anisotropy of an oligomer of fluorophores undergoing homo-FRET, r oligomer , is the anisotropy in a monomer, r monomer , divided by the number of fluorophores, n, in the oligomer.
This approximation only applies when the interfluorophore distance is <0.8 R 0 . In more complex cases, treatment of homo-FRET is more difficult, depending on the cluster size, the orientation between fluorophores, and interfluorophore distances. 44, 45 Competing with homo-FRET depolarization are the opposing effects of greater molecular weight. This phenomenon can be understood through the Perrin equation: 43 where τ is the fluorescence lifetime, θ is the rotational correlation time, and r and r 0 are measured and initial anisotropy, respectively. The rotational correlation time depends on viscosity, η, the molecular weight, M, the specific volume and hydration, ν ̅ + h, the gas constant, R, and the temperature, T.
Assuming all other parameters are equal (temperature, viscosity, etc.), the rotational correlation time scales with the mass indicating that as mass increases, the measured r will approach r 0 . was chosen for its greater brightness and photostability when compared to CFP variants and its ability to form good FRET pairs with the YFP variants, mKate, mKO2, and tdTomato. 46−50 The mTFP was expressed with a thioester group at the C-terminus by the intein based method and purified using chitin and His-tag affinity chromatography. Mass spectrometry identified the integrity of the protein (measured m/z 27 969 ± 6 Da, calculated m/z 27 965 Da) and SDS-PAGE (28 kDa) showed the purity of the resulting mTFP-thioester.
Kinetic Studies of Ligation to Form mTFP-PNA. The reaction of 100 μM purified mTFP-thioester group with excess (400 μM) cysteine-PNA1 was studied in the presence and absence of 50 mM MPAA catalyst at pH 7 and variable incubation times (20 min to 18 h) at room temperature. Mass spectrometry results showed that the MPAA catalyst was required for the ligation of PNA to mTFP to proceed. The hydrolyzed mTFP exhibited a peak at m/z 27 833 ± 7 Da (calculated m/z 27 844 Da) and the mTFP-PNA product at m/ z 29 560 ± 6 Da (calculated m/z 29 555 Da). An additional peak at m/z 31 293 ± 7 Da (calculated m/z 31 286 Da) indicated partial double PNA attachment to mTFP. Incubation with the reducing agent TCEP followed by centrifugation assisted dialysis efficiently removed the second attached PNA; since native mTFP does not feature a cysteine, the second PNA was most probably attached via disulfide bond formation to the newly introduced SH group at the ligation site. HPLC analysis of the final product confirmed purity and integrity of the mTFP-PNA with a yield near 100%.
Effect of Catalyst and PNA Concentration on mTFP-PNA Ligation Yield. To elucidate the effect of catalyst concentration on mTFP-PNA ligation, varying concentrations of MPAA (5− 200 mM) were added to the ligation reaction of the PNA to the fluorescent protein. All components were mixed at pH 7 and incubated for 60 min. The conversion was studied by spectrophotometry using the estimated absorbance of the mTFP-PNA at 260 nm as an indicator of the ligated PNA concentration and it was compared with the calculated absorbance of 100% ligation of PNA to mTFP (1:1 ratio) to obtain the final percent yield. The results confirmed >90% yield. The yield increased monotonically with added catalyst up to 100 mM (Figure 1). Based on the data, 50 mM catalyst was selected for subsequent ligation experiments. The PNA:mTFP ratio was optimized by adding different concentrations of PNA (400−600 and 700 μM) to the ligation solution. The spectrometry data showed that a high ratio of PNA:mTFP resulted in more evidence of a second attached PNA at 60 min ligation time. Therefore, an optimized ratio of 4:1 was used in all ligation experiments.
The SEC-HPLC was calibrated with molecular weight standard and the chromatogram of mTFP-PNA showed a single peak (8.7 min) when monitored at 462 nm (maximum absorption of mTFP) and 214 nm (general absorption of all components), in line with the expected molecular mass of a monomeric construct. A solution featuring a 2:1 ratio of mTFP-      Figure 5). 52,53 These data are consistent with partial unfolding of the Beacon in the presence of PNA2 and with the production of an assembled system of mTFP-PNA:Beacon:PNA2 in which there is energy transfer from mTFP to 6-FAM.
Assembly of mTFP-PNA with DNA Beacon. In this hetero-FRET system, the modification and assembly of mTFP were accompanied by a small increase in mTFP fluorescence anisotropy. The anisotropy of mTFP was initially 0.32 and increased after PNA1 ligation to 0.33. Subsequent hybridization of mTFP-PNA1 to the DNA beacon yielded a further increase in r to 0.35. The calculated correlation times (eq 7) for mTFP-PNA1 (θ 1 ) and mTFP-PNA1:DNA (27 μM DNA) (θ 2 ) were 18 and 25 ns, respectively, using experimentally determined average lifetimes and r 0 = 0.39 from previous studies of fluorescent proteins. 54 These estimated rotational correlation times are in the typical range of monomeric fluorescent FPs in buffer. 36 The observed 27% increase in θ when mTFP-PNA1 (29 560 Da) formed the mTFP-PNA1:Beacon assembly (33 241 Da) was consistent with the 12% increase in mass. These results were interpreted as reinforcing the intensity and polar plot analyses which together proved the ligation and assembly strategy was successful and the assemblies accessible to photophysical analysis.
Anisotropy Measurements of Homo-FRET System Demonstrating Dimer and Oligomer Assembly. Anisotropy changes of the system were studied by adding a range of DNA1 and DNA2 concentrations (0−2.5 μM) to 2.5 μM of mTFP-PNA. Maximal anisotropy values of 0.34 (DNA1) and 0.36 (DNA2) were observed when the ratio of mTFP-PNA to DNA1 and DNA2 (monomer) was at 1:1. These were slightly higher than observed for free mTFP-PNA (0.32) and consistent with a slight increase in the rotational correlation time in the larger assemblies and similar to values measured in the presence of DNA beacon. A homo-FRET system was prepared by adding mTFP-PNA to a complementary DNA template (DNA1 or DNA2) which produced a decrease in measured anisotropy. Titration of mTFP-PNA with DNA1 resulted in a gradual decrease of anisotropy compared to mTFP-PNA:DNA1 monomer ( Figure 6). The maximum decrease of 45% was observed in the presence of 2:1 ratio of mTFP-PNA to DNA1, close to the predicted 50% (eq 6). Thereafter, the anisotropy gradually increased on addition of more mTFP-PNA ( Figure  6).
In template directed oligomer formation via DNA2, a 15% decrease in anisotropy was obtained with a 2:1 ratio of mTFP-PNA:DNA2. This value is significantly different than predicted by eq 6. However, DNA2 has 4 locations to bind, and with two mTFPs attached, the distance between monomer units is expected to range from 3.3 to 10 nm. The larger distances are The assemblies illustrated at points a, b, and c represent the stoichiometry. The mixtures of assembled species at these locations are more complex. For example, the 2:1 species can exist in 3 forms and a randomly assembled system will include some amount of 1:1, 2:1, 3:1, and 4:1 species. As with the dimer system ( Figure  6) the trend in anisotropy follows the extent of homo-FRET which tracks the extent of assembly. Homo-FRET is maximal (minimum anisotropy) at the equivalence point (4:1 mTFP-PNA to DNA2). Addition of further template after equivalence is reached causes the anisotropy to increase until all mTFP-PNAs are in 1:1 complexes. Note that the 1:1 complex of mTFP-PNA to DNA2 has a higher anisotropy (d) than free mTFP-PNA. outside 0.8 R 0 44,55 resulting in less change in anisotropy than seen in the DNA1 assembly. Similarly, when the ratio of mTFP-PNA to DNA2 was 3:1 a 44% decrease was observed. The lowest anisotropy (0.082) was obtained with a 4:1 ratio. The resulting template directed tetramer assembly was in reasonable agreement with the predictions of eq 6 ( Figure 7).

■ CONCLUSION
In summary, the self-assembled fluorescent protein model system based on PNA conjugates demonstrates a new way to take control of protein assembly. The unique structure and characteristics of PNA and the capability of studying the current model system with fluorescence measurements make it a promising system to study protein assemblies under controlled conditions. This study demonstrates that PNA can efficiently conjugate to proteins with high specificity via expressed protein ligation. Taking advantage of the precise recognition characteristics of PNA for complementary oligonucleotides, directed assembly proceeded conveniently to the formation of dimers and higher oligomers. The assemblies gave clear and readily interpreted photophysical signatures creating a practical tool to study the behavior of protein aggregates. By fusing EPL enabled FPs to targets of interest, such as receptors, control of both assembly and visualization is provided following addition of template to PNA ligated proteins. This approach could be extended to study aggregation behavior in both biochemical and cellular environments. Further work could readily extend this approach from homodimers and oligomers to binary, ternary, and higher oligomer systems containing any number of different dyes or fluorescent proteins in precisely engineered arrangements.

Notes
The authors declare no competing financial interest.

■ ACKNOWLEDGMENTS
Z.G., Q.H., and L. B. acknowledge funding from NanoSci-E+ to the NanoActuate consortium. In the UK, NanoActuate is administered by EPSRC as EP/H00694X/1. In The Netherlands, NanoActuate is funded through STW grant 11022-NanoActuate. Also, we thank H. D. Nguyen and D. T. Dang for plasmid pHT581 and help with protein expression.