Absorption, Excitation And Emission Spectra
At first, we carried out a rational mutagenesis and obtained mSAASoti variants with M163A and F177S substitutions based on the experience of obtaining biphotochromic proteins [20], [15]. After protein isolation and purification, we characterized their spectral and physicochemical properties (λex/λem, ε, pKa). Both these mutants were shown to have a blue shift of the absorption maxima, 13 nm in the case of M163A and 8 nm in the case of F177S, while the emission maxima remained unchanged (Fig. 1A,B). The excitation spectrum of all the studied mutants is not only broader compared to mSAASoti, but also highly structured (Fig. 1C, D).
As a result of these substitutions (Table 1), the extinction coefficients at neutral pH decreased, which may be caused by the pKa shift to more alkaline region and the broadening of the spectra. The pKa values of the red forms in the case of M163A and F177S shifted to the more alkaline pH.
Table 1
Spectral properties and relative switching rate constants (S) of the obtained mutant forms of SAASoti.
Mutant form | Green form λex/λem, nm | Red form λex/λem, nm | ε (G/R) /1000 M− 1*cm− 1 | pKa G/R | Photoswitching S (= koff (s− 1) / koff (s− 1)`) |
Green form | Red form |
mSAASoti [3] | 509/519 | 578/589 | 75/24 | 6.3/6.6 | 1.0 | – |
C21N [23] | 509/519 | 579/590 | 82/25.4 | 6.4/7.5 | 1.0 | – |
M163A | 496/519 | 560/587 | 62/0.3* | 6.7/7.5 | 13.1 | 2.3 |
M163T | 498/515 | 565/582 | 56/37 | 6.6/7.4 | 19.1 | 14 |
C21N/M163A | 497/517 | 561/587 | 53/18 | 6.5/7.3 | 12.8 | – |
C21N/M163T | 498/516 | 565/580 | 53.5/12.6* | 5.9/7.2 | 33.0 | 15 |
C21N/M163G | 496/516 | 558/581 | 50/5.3* | 5.6/7.0 | 22.0 | n.a. |
C21N/M163S | | | | | 19.3 | |
C21N/M163L | | | | | 21.3 | |
F177S | 501/519 | 560/588 | 51/1* | 6.8/7.8 | 9.8 | 3.9 |
C21N/F177A | 503/518 | 566/588 | 42/3* | 6.3/7.7 | 10.3 | n.a. |
C21N/F177N | 503/517 | 564/587 | 55/4* | 6.2/7.8 | 9.7 | n.a. |
C21N/F177T | 506/518 | 573/588 | 66/4.5* | 6.2/7.6 | 6.7 | n.a. |
C21N/F177S | | | | | 9.0 | |
– no photoswitching observed; n.a. – measurements were not carried out; the (*) symbol in the ε values of the red form is the actual maximum value, obtained without taking into account photodestruction during photoconversion; colored in grey are data obtained for colonies. |
In addition to spectral changes, there were changes in the fluorescence lifetime, a slight decrease from 3.3 to 3.0 ns was observed for F177S, and a bi-exponential dependence of fluorescence attenuation with lifetimes of 2.4 and 1.2 was observed for M163A (Table 2).
Table 2
Fluorescent lifetime values of the M163A, F177S, and mSAASoti proteins measured in 20 mM TrisHCl, 150 mM NaCl, pH 7.4 buffer.
| mSAASoti | M163A | F177S |
τ, ns ± | 3.30 ± 0.03 | τ1 2.35 ± 0.02 τ2 1.23 ± 0.04 A1/A2 = 1.7/1 | 3.048 ± 0.008 |
The origin of absorption spectra variations.
We performed molecular dynamic simulations with QM/MM potentials to explain the origin of broadening of absorption and excitation band maxima of mSAASoti variants compared with the mSAASoti. It was already shown for the fluorescent proteins with GFP-type chromophore that the value of bond length alternation (BLA, difference between lengths of C–C bonds in the C–C = C bridge of the chromophore, Fig. 2) correlates with the maximum of the absorption band [21], [22]. Structures with predominance of resonance form with BLA > 0 are characterized by a larger energy gap between the ground and excited electronic states and shorter wavelength of transition. We deduce that the distribution of the BLA should correlate with the absorption band shape. To study this we performed QM/MM MD simulations of two model systems, mSAASoti and its M163A variant. BLA distributions are shown on Fig. 2. The main difference is observed for the weights of the fractions characterized by larger BLA values and consequently corresponding to the shorter wavelengths in the absorption spectrum. The overall distribution is wider for the mSAASoti M163A that is also in agreement with experimental observations. Thus, changes of the absorption band shape can be explained by the overall change of the influence of the entire protein on the chromophore group pronounced in the change of distribution between two resonance forms of the negatively charged chromophore. We also assume that the structuring of the excitation spectrum in Figs. 1C,D, as well as the change in the lifetimes in the excited state (Table 2), is associated with the existing BLA distribution, with each of the forms in this distribution having almost the same absorption has different fluorescence quantum yields (Fig. 1C ,D)
Photoswitching
Only the green form of the wild type SAASoti protein can be photoswitched to the dark state [4], whereas in the case of M163A and F177S SAASoti variants we observed photoswitching of both green (Fig. 3A) and red forms (Fig. 3B). To analyze the photoswitching, we irradiated purified proteins in solution with 470 nm light (167 mW/cm2) for the green form and 550 nm (300 mW/cm2) for the red form. Photoswitching kinetics is generally described with a bi-exponential model (Eq. 1),
$$I\left(t\right)={A}_{1}*\text{exp}\left(-{k}_{1}*t\right)-{A}_{2}*\text{exp}\left(-{k}_{2}*t\right)+c$$
1
where the first component describes switching, and the second component is responsible for green fluorescence intensity increase at the fixed wavelength, but in the case of M163A there is only one component (k1). Photoswitching rate of the red M163A and F177S variants can be described by a mono- and biexponential function (Eq. 2), respectively, and the second exponent in the case of F177S may be related to the photodestruction of the sample (Fig. 3C) as it is evident from multiple cycles of photoswitching in contrast to M163A (Fig. 3D).
$$I\left(t\right)={A}_{1}*\text{exp}\left(-{k}_{1}*t\right)+{A}_{2}*\text{exp}\left(-{k}_{2}*t\right)+c$$
2
To compare the rate constants in the photoswitching reaction between the mutants, we will use a special S value (Table 1, the last two columns), calculated as mean koff = (A1*k1 + A2*k2)/(A1 + A2) normalized to the corresponding mean koff` of the wild type mSAASoti (Supplementary Table S1, Figure S1). In general, green form photoswitching increased by 13.1 and 9.8 times for M163A and F177S, respectively (Table 1).
During the cycles of sequential photoswitching with 470 nm light with the regeneration of the fluorescent form with 400 nm light (10 mW/cm2) of the same sample, it was found that M163A substitution leads to a more photostable variant, which, apparently, is due to the exclusion of the photooxidation stage of methionine (Fig. 3C, D) [4].
It was shown in the previous study, that C21N substitution resulted in the allosteric regulation of the chromophore pKa in mSAASoti [23]. As in the case of other proteins of this group, 163 and 177 positions turned out to be sensitive to the photoswitching for mSAASoti, but their substitutions to alanine and serine greatly affect the red form (significant pKa shift, small extinction coefficient, SI, Figure S2). For this reason, we conducted site-saturated mutagenesis using degenerate primers, and also obtained C21N/M163X and C21N/F177X mSAASoti clones (X states for any amino acid residue). The most promising colonies were sent for sequencing to determine the appropriate substitution. Among them C21N/M163A, C21N/M163G, C21N/M163T and C21N/F177A, C21N/F177N, C21N/F177T mSAASoti proteins were also isolated and purified. We illuminated the colonies expressing different mutant forms with 470 nm light and recorded emission spectra over the time. As it can be seen from Fig. 4A, C21N/M163T and C21N/M163G mutant forms showed the maximum photoswitching rate and the greater degree of phototransformation (highest contrast between the on- and off- states). M163 replacement also led to the increased stability when repeating on-off cycles (Fig. 4C). The red M163A mSAASoti switches to the dark state only 2.3 times faster than the green mSAASoti form, and the red C21N/M163A mSAASoti is not photoswitchable as well as the original protein (Fig. 4B). The normalized photoswitching rates of the red C21N/M163T (Fig. 4B) and M163T form were almost identical (15 times higher than that of the green form mSAASoti, Table 1). As it can be seen from Fig. 4C, mutants with M163X substitution are more photostable when moving to subsequent photoswitching cycles in comparison with F177X.
On-off-on photoswitching cycles measured for the green form of different SAASoti mutant variants and registered as emission spectra during 470 nm (150 s) and 400 nm (5 s) illumination (С).
The Origin Of Photoswitching In Fluorescent Proteins
We performed dynamic network analysis to find out the features that are responsible for the ability of the cis-trans chromophore isomerization. To do this, we obtained a set of models with both green and red forms of the chromophore in the anionic cis-form for the mSAASoti and its C21N, C21N/M163T, M163A, F177S variants. We simulated 300 ns trajectories for each model system. For all considered systems photoswitching is observed for the green form and in the red form it is observed only in variants with substitutions at either 163 or 177 positions. Dynamic network analysis reveals groups of residues with the correlated motion and we determine the residues that belong to the same group with the phenyl fragment of the chromophore (Fig. 5). According to our calculations, the systems that can undergo isomerization have a distinctive feature: phenyl fragment of the chromophore and the His68 residue next to the chromophore belong to the same community. Importantly, this criterion works for both red and green forms despite the considerable differences in this region. In red form both π-systems of the His68 residue side chain and a green form of the chromophore are united to the extended conjugated π-system.
We examined neighboring residues, that may be responsible for the ability to isomerize. But we found no similarities between considered systems in this respect. Even in structurally similar mSAASoti and its C21N variant residues located close to the phenyl part of the chromophore, Phe177, Met163 and Ile161, act differently. In red form the motion of the phenyl part of the chromophore is correlated with the Phe177 in the mSAASoti and with the Met163 and Ile161 in its C21N variant. Similarly, involvement of the Met163 and Ile161 is observed for the F177S variant and is not observed for the species with the substitution at 163rd position.