Experimental Characterization of In Silico Red-Shift-Predicted iLOVL470T/Q489K and iLOVV392K/F410V/A426S Mutants.

iLOV is a flavin mononucleotide-binding fluorescent protein used for in vivo cellular imaging similar to the green fluorescent protein. To expand the range of applications of iLOV, spectrally tuned red-shifted variants are desirable to reduce phototoxicity and allow for better tissue penetration. In this report, we experimentally tested two iLOV mutants, iLOVL470T/Q489K and iLOVV392K/F410V/A426S, which were previously computationally proposed by (KhrenovaJ. Phys. Chem. B2017, 121 ( (43), ), pp 10018-10025) to have red-shifted excitation and emission spectra. While iLOVL470T/Q489K is about 20% brighter compared to the WT in vitro, it exhibits a blue shift in contrast to quantum mechanics/molecular mechanics (QM/MM) predictions. Additional optical characterization of an iLOVV392K mutant revealed that V392 is essential for cofactor binding and, accordingly, variants with V392K mutation are unable to bind to FMN. iLOVL470T/Q489K and iLOVV392K/F410V/A426S are expressed at low levels and have no detectable fluorescence in living cells, preventing their utilization in imaging applications.

Introduction

Fluorescent proteins (FPs) have revolutionized cell biology by enabling researchers to investigate dynamic cellular processes in real time.[1] The most widely used FP is the green fluorescent protein (GFP) and its spectrally shifted and/or engineered variants,[1] which have been optimized, for instance, to mature more rapidly (sfGFP)[2] or fluoresce more brightly (mNeonGreen).[3] Despite their usefulness, FPs of the GFP family have some limitations: they depend on molecular oxygen for chromophore formation, which impedes their use to visualize processes under hypoxic or anoxic conditions,[4−6] are relatively large (∼25 kDa), which can be problematic in some applications,[7,8] and are mostly sensitive to pH.[9] While pH-resistant GFP variants have been engineered,[9] oxygen dependence and size are mostly unchangeable features. FMN-binding fluorescent proteins (FbFPs) can overcome these limitations and thus are valuable alternatives to GFP and its variants.[10,11] FbFPs consist of a small light oxygen voltage (LOV) domain (∼11 kDa) emitting green fluorescence when light in the UV-A-blue range reaches the noncovalently bound FMN chromophore.[12]

One prominent FbFP is iLOV, a derivative of the LOV2 domain of Arabidopsis thaliana.(7) A mutation in a key cysteine residue and several rounds of DNA shuffling led to enhanced fluorescence emission.[7] While FbFPs have been used in molecular imaging for more than a decade,[13] they suffer from the autofluorescence signal of flavin molecules in the cell and relatively weak fluorescent intensity, issues which have not been improved until very recently.[14,15] Engineering a red-shifted iLOV would bring several advantages, for instance, lower phototoxicity and deeper tissue penetration. Moreover, a red-shifted FbFP variant could be used orthogonally to other FbFPs and would potentially allow for multicolor imaging and FRET-based biosensors.

Based on the observation that the red fluorescent proteins Rtms5[16] and mKeima[17] possess positively charged residues in close proximity to the chromophore, Khrenova and colleagues applied quantum mechanics/molecular mechanics (QM/MM) simulations and proposed that the iLOVQ489K mutant would have a ∼50 nm red shift in its excitation and emission maxima compared to wild-type (WT) iLOV.[18] Their rationale was that introducing a positively charged amino group at position 489 next to the chromophore would stabilize the π-electron system of the FMN in the excited singlet state. However, later Davari and colleagues computationally and experimentally showed that K489 is mostly populated in an open conformer and flipped away, which is far from the chromophore and in fact iLOVQ489K is blue-shifted.[19] In a follow-up study, Khrenova and colleagues applied the second round of QM/MM calculations and found other mutations that were predicted to have more stabilized lysine residues next to the chromophore compared to iLOVQ489K, which led them to propose iLOVL470T/Q489K and iLOVV392K/F410V/A426S as mutants with ∼50 nm red shift for both excitation and emission spectra[20] (Figure ). As a shift of such magnitude would open up new applications for iLOV and FbFPs in general, we experimentally tested these two mutants in this short report by measuring the absorption, excitation, and emission spectra and calculating the quantum yield and brightness of the purified proteins.

Figure 1

Model structures of iLOVL470T/Q489K and iLOVV392K/F410V/A426S mutants. Cartoon and surface representations of the iLOVL470T/Q489K (A) and iLOVV392K/F410V/A426S (B) mutants. Positions of mutations are shown in cyan. The structures (yellow) were generated using the Swiss-Model web-server[21] and the FMN (red) was placed by aligning the model structures to the WT iLOV X-ray structure (PDB id: 4EEP). Figures were generated using PyMOL.[22]

Results

iLOVV392K, iLOVL470T/Q489K, and iLOVV392K/F410V/A426S are not Detectable in Living Cells

To experimentally test the in silico predictions of Khrenova and colleagues that the double L470T/Q489K and triple V392K/F410V/A426S mutations would lead to a red shift in the excitation and emission spectra of iLOV, we expressed the iLOVL470T/Q489K and iLOVV392K/F410V/A426S mutant proteins in Escherichia coli cells and analyzed their fluorescence. We additionally studied the single-point mutant iLOVV392K to assess whether adding a single positive residue in close proximity to the chromophore was sufficient for the red shift. While for the positive controls (GFP and WT iLOV), we could detect fluorescence at the microscope; this was not the case for any of the mutants (Figure ). We then expressed the same iLOV constructs in mammalian cells, but again no fluorescence was observed except for WT iLOV (Figure S1).

Figure 2

Analysis of the fluorescence of iLOV mutants in E. coli. Representative bright-field (top) and GFP channel (bottom) images of E. coli Rosetta (DE3) pLysS cells transfected with the indicated constructs. The scale bar for all micrographs is 5 μm. Nontransformed cells were used as the negative control. Cells transformed with a plasmid expressing GFP and WT iLOV, respectively, served as positive controls.

One possibility for the lack of fluorescence could be limited protein expression. Alternatively, the proteins might be well expressed but might be unable to bind the FMN chromophore. Another possibility is that, while FMN can still bind to iLOV, the fluorescence quantum yield is significantly lower. Interestingly, only the pellets of E. coli cells overexpressing GFP and WT iLOV appear green to the naked eye (Figure S2). Thus, we aimed to test whether the constructs were expressed at sufficiently high levels, FMN was bound to the proteins, and the fluorescence had sufficient quantum yield.

iLOVL470T/Q489K and iLOVV392K/F410V/A426S Mutants are Expressed at Low Levels Likely Due to Misfolding

To better understand the reasons for the lack of fluorescence, proteins produced from E. coli cells were analyzed using an SDS gel (Figure ). While the WT and the iLOVV392K mutants showed only single bands at around 15 kDa, higher-molecular-weight bands (∼70–80 kDa) were observed for the iLOVL470T/Q489K and iLOVV392K/F410V/A426S mutants (Figure ). Moreover, the iLOVL470T/Q489K mutant had very low concentrations.

Figure 3

SDS gel of purified proteins. The indicated amounts of proteins were diluted in 10 μL of storage buffer, boiled, and run on a 10% SDS gel. The expected size of all proteins is 15.1 kDa. Pictures of gels were taken with a BIO-RAD ChemiDoc XRS+.

To investigate whether the higher-molecular-weight (MW) bands observed in the SDS gels for the double and triple mutants were constituted by the iLOV mutant themselves or other proteins, which co-purified with the mutants, mass spectrometry analysis (MS) of these bands was performed. The results indicated that the high MW bands corresponded to the chaperones GroEL (57.34 kDa) and DnaK (also known as Hsp70; 69.13 kDa) running together with the iLOV mutants (15.1 kDa) leading to a total weight of 72.44 and 84.23 kDa, respectively. GroEL and DnaK have been previously detected in other studies with purified proteins from E. coli.(23) Being chaperones, they might be associated with newly synthesized proteins to help them fold properly. Interestingly, visual inspection of the purified proteins showed that only the WT protein fluoresced in a way detectable by the naked eye (Figure S3). To further clarify this observation, optical spectroscopy with the purified proteins was conducted.

Spectroscopic Analyses Indicate that the V392K Mutation Leads to Loss of FMN Binding

Next, we measured the absorption spectrum of all purified iLOV proteins between 250 and 800 nm. Surprisingly, the mutants harboring the V392K mutation, namely, iLOVV392K and iLOVV392K/F410V/A426S, did not show the typical FMN peak at 450 nm. Considering that the iLOVV392K mutant was purified as a rather pure protein without contamination according to the SDS gel (Figure ), these results suggested that mutation to V392K prevents correct folding of the protein and/or its binding to FMN, both essential for fluorescence. This conclusion would be in line with our cell and protein pellet observations indicating no fluorescence in cells with V392K mutation (Figures S2 and S3). In contrast to the QM/MM predictions of Khrenova and colleagues,[20] iLOVV392K/F410V/A426S did not bind FMN, thus showing no fluorescence, and both excitation and emission spectra of iLOVL470T/Q489K were slightly blue-shifted by ∼2 nm (Figure ).

Figure 4

Absorption, excitation, and emission spectra measurements of iLOV mutants. For all purified proteins, absorption spectra were measured between 250 and 800 nm. The excitation and emission spectra measurements were recorded between 250 and 500 nm and 480 and 700 nm, respectively.

Finally, we determined the quantum yield and the brightness of the WT iLOV and the iLOVL470T/Q489K mutant. The results for WT iLOV were in line with the previous reports.[11] Interestingly, despite the low-expression levels and the co-purification with the chaperone (Figures and 3), iLOVL470T/Q489K exhibited a slightly higher quantum yield and brightness compared to WT iLOV (Table ).

Table 1

Quantum Yield and Brightness Measurements of WT iLOV and iLOVL470T/Q489Ka,b

	WT	iLOVL470T/Q489K
quantum yield (Φs)	0.39 ± 0.06 (literature 0.34[11])	0.47 ± 0.07 (0.40 ± 0.07)
brightness	4875 ± 750 (4250)	5875 ± 781 (5000 ± 313)

Numbers in parentheses are calculated by using the literature quantum yield of WT iLOV.

Linear regression plots used to calculate quantum yields are given in Figure S4.

Discussion

In this study, we set out to analyze the excitation and emission spectra of two iLOV mutants, iLOVL470T/Q489K and iLOVV392K/F410V/A426S, which were computationally predicted to be red-shifted by about 50 nm.[20] We found that the iLOVV392K/F410V/A426S mutant does not bind to FMN and requires chaperones to fold. Our further analysis of iLOVV392K indicates that the V392K mutation causes a loss of FMN binding. Furthermore, the iLOVL470T/Q489K mutant is slightly blue-shifted in both excitation and emission spectra. Initially, Khrenova and colleagues predicted the single mutant iLOVQ489K to be shifted in excitation and emission maxima by 52 and 97 nm, respectively;[18] however, when experimentally measured, the excitation and emission maxima exhibited rather a blue-shift of about 10 nm.[19] This highlighted a disagreement between the predictions and the experimental validations. Even though the QM/MM modeling approach was subsequently updated,[20] our results argue that it is still not able to completely capture the complexity of the interaction between the iLOVL470T/Q489K mutant and its chromophore. Nevertheless, the iLOVL470T/Q489K mutant has an ∼20% increased brightness and quantum yield compared to WT iLOV. These results indicate that L470 and Q489, and probably some other amino acids in the close vicinity, can influence the optical properties of the FMN cofactor in the required direction.

Several studies recently also provided a molecular characterization of the spectral effects of mutations on iLOV to gain a mechanistic understanding.[24−26] Röllen and colleagues have shown that iLOVV392T/Q489K fluorescence has a slight red-shift with an emission spectrum maximum of 502 nm.[26] Similar mutations on iLOV homolog protein CagFbFPI52T/Q148K with the emission maximum of 504 nm and another blue-shifted variant CagFbFPQ148K with the emission maximum of 491 nm were used in fluorescence microscopy experiments and were successfully spectrally separated.[26] While iLOVV392T/Q489K (and its homolog CagFbFPI52T/Q148K) was mutated at the same sites as in our study, the V392T mutation may not disrupt FMN binding to iLOV as V392K does. Additionally, while individual mutations such as Q489K may not lead to a red shift alone, they might do so when combined with other mutations.

Previous efforts to improve the properties of iLOV for imaging purposes resulted in the generation of phiLOV, an iLOV derivative with superior photostability, thus solving one of the major drawbacks of WT iLOV.[27] Furthermore, iLOV variants with improved brightness recently have been reported.[14,15] It remains to be seen whether future mutagenesis efforts to red shift iLOV can be combined with these more stable and brighter versions of iLOV.

We believe that the quantification of the fluorescence of large libraries of iLOV mutants would be desirable, as these data (including negative ones, equally informative) could be fed to machine learning algorithms likely to improve the understanding of the residues impacting the spectral properties of iLOV. Previously, similar approaches were successfully applied for improving the brightness of avGFP[28] and YFP.[29]

Material and Methods

Plasmid Construction

Constructs were generated using classical restriction enzyme cloning or golden gate cloning. The ilov gene was amplified from iLOV-N1, which was a gift from Michael Davidson (Addgene plasmid #54673; http://n2t.net/addgene:54673; RRID:Addgene_54673), and cloned into pcDNA3.1(+) using HindIII and EcoRI restriction sites, yielding plasmid PL01. PL01 was used as a template to clone all iLOV variants used in this study. Single-point mutations Q489K and V392K were created via overhang PCR yielding PL02 and PL04, respectively. The L470T mutation was introduced into PL02 using golden gate cloning. Amplicons were digested with HindIII and BpiI, or BpiI and EcoRI, respectively, and cloned into pcDNA3.1(+) yielding PL03. The A426S mutation was inserted into PL04 by addition of an XbaI restriction site into the coding sequence and cloned into pcDNA3.1(+) yielding PL05. The F410V mutation was introduced using golden gate cloning. Amplicons were digested with HindIII and BpiI or BpiI and EcoRI, respectively, and cloned into pcDNA3.1(+) yielding PL06.

For expression in and purification from E. coli, the ilov, ilov, ilov, and ilov genes were amplified and then cloned into pET28a using NdeI and XhoI restriction sites, yielding PL07–PL10. The sequences of all constructs were verified using Sanger sequencing.

Bacterial Cell Culture and Transformation

The bacterial strain used in this study was E. coli Rosetta (DE3) pLysS. iLOV constructs were transformed into chemically competent cells via heat shock (42 °C, 1.5 min), plated on lysogeny broth (LB) agar plates containing 0.05 mg/mL kanamycin, and incubated overnight at 37 °C. For expression, 5 mL of liquid LB was inoculated with a single colony from a fresh LB agar plate and incubated overnight at 37 °C and 200 rpm. The next day, the cultures were used to inoculate 1 L of LB with a starting OD600 of 0.1. All liquid media were supplied with 0.05 mg/mL kanamycin.

Protein Expression and Purification

Each pET28a plasmid harboring one iLOV construct was transformed into E. coli Rosetta (DE3) pLysS cells and cultures were grown at 37 °C and 220 rpm until they reached an OD600 of 0.4. IPTG (isopropyl-β-d-thiogalactopyranosid) was then added to the culture to a final concentration of 1 mM, and the culture was further grown at 18 °C for 16 h. Afterward, cells were harvested via centrifugation at 5000g for 30 min. The pellets were resuspended in a lysis buffer (50 mM KH2PO4, pH 7.5, 300 mM NaCl and 10 mM imidazole, pH 8.0) containing one tablet of cOmplete protease inhibitor cocktail (Roche) and lysed by sonication. The lysate was clarified by centrifugation at 20,000g for 30 min at 4 °C and loaded onto an IMAC nickel column (1 mL) using the Bio-Rad NGC automated liquid chromatography system. The column was washed with a wash buffer (same as the lysis buffer but with 20 mM imidazole and 10% glycerol) and eluted with an elution buffer (same as the lysis buffer with further addition of 10% glycerol and 500 mM imidazole). Finally, the elution buffer was replaced with a storage buffer (50 mM HEPES-KOH, pH 7.25, 150 mM KCl, 10% glycerol, 0.1 mM EDTA, pH 8.0) using a P-6 desalting column (10 mL).

Fluorescence Microscopy

Fluorescence microscopy was performed on a Zeiss Axio Observer Z1/7 (Carl Zeiss, Germany) inverted wide-field microscope, equipped with a Colibri 7 LED light source, an α Plan-Apochromat x 100/1.46 oil DIC (UV) M27 objective, filter sets 38 HE (ex. 450–490, dichroic beamsplitter495, em. 500–550; sfGFP), and an Axiocam 506 Mono camera. Six microliters of the culture (OD600 = 1.5) was loaded on a glass slide.

SDS Gel and Mass Spectrometry Analyzes of Purified iLOV Proteins

The concentration of purified proteins was determined using NanoDrop One (Thermo Scientific). For performing SDS-PAGE, protein samples were set to contain 0.1, 0.5, 1, and 2 μg, mixed with the Laemmli buffer, and incubated at 95 °C for 15 min. After incubation, the samples were spun down and 20 μL was loaded on a pre-cast 12% agarose gel (MiniProtean TGX gel, BIO-RAD). SDS-PAGE was performed at 130 V for 40 min. Afterward, gels were stained for 1 h using ReadyBlue Protein Gel Stain (Sigma-Aldrich) and destained with ddH2O overnight. Images were taken with the ChemiDoc XRS+ system (BIO-RAD). Protein bands analyzed using mass spectrometry were excised, treated with 10 mM dithiothreitol and 10 mM iodoacetamide, and subsequently digested with trypsin. The peptide solution was then separated on a Waters Acquity I-class UPLC in positive HD-MSE mode using a Waters Peptide CSH C18 column (2.1 mm × 150 mm, 1.7 μm particle size). A gradient from 1 to 40% ACN/0.1% formic acid (v/v) in water/0.1% formic acid (v/v) was utilized at a starting temperature of 80 °C and a dissolving temperature of 400 °C with a gas flow rate of 800 l/h. Spectra obtained by the separation were analyzed by matching with the UniProt database.

Spectroscopy Analyzes

All optical spectra were measured at room temperature in a fluorescence cuvette with a 1 cm path length (Art. No. 105–250–15–40, Hellma Analytics). Samples (volume 100 μL) with a concentration of 500 μg/mL were dissolved in a storage buffer (50 mM HEPES-KOH, pH 7.25, 150 mM KCl, 10% (v/v) glycerol, 0.1 mM EDTA). Absorption spectra were recorded from 250 to 800 nm (UV-2450 Spectrophotometer, Shimadzu). Fluorescence measurements were performed using a Luminescence Spectrometer (LS 55, Perkin Elmer). Fluorescence excitation spectra were recorded from 250 to 500 nm at a fixed emission wavelength of 496 nm. Emission spectra were measured from 480 to 700 nm at an excitation wavelength of 450 nm. The quantum yields were determined with the comparative method[30,31] using FMN dissolved in the storage buffer as the reference. For the linear regression, the integrated emission (480–700 nm) was plotted against the absorbance at 450 nm. The quantum yield was then calculated using the following equation:where m indicates the slope from the linear regression and η the refractive index of the sample or the reference. The fraction of refractive indices equals 1 since the same buffer was used. For the quantum yield, Φr of FMN 0.24 was used.[32] The brightness was calculated as the product of the determined quantum yield and the extinction coefficient of free FMN as 12500 M–1 cm–1.[33]