Red-shifting mutation of light-driven sodium-pump rhodopsin

Microbial rhodopsins are photoreceptive membrane proteins that transport various ions using light energy. While they are widely used in optogenetics to optically control neuronal activity, rhodopsins that function with longer-wavelength light are highly demanded because of their low phototoxicity and high tissue penetration. Here, we achieve a 40-nm red-shift in the absorption wavelength of a sodium-pump rhodopsin (KR2) by altering dipole moment of residues around the retinal chromophore (KR2 P219T/S254A) without impairing its ion-transport activity. Structural differences in the chromophore of the red-shifted protein from that of the wildtype are observed by Fourier transform infrared spectroscopy. QM/MM models generated with an automated protocol show that the changes in the electrostatic interaction between protein and chromophore induced by the amino-acid replacements, lowered the energy gap between the ground and the first electronically excited state. Based on these insights, a natural sodium pump with red-shifted absorption is identified from Jannaschia seosinensis.


1
Replies to the comments of Reviewer 1

Reviewer #1 (Remarks to the Author):
The authors report on the engineering of a red-shifted Na rhodopsin. Super important as a tool for optogenetics.
Dear Dr. Oded Béjà, Thank you for your quite positive comments (italic black) on our work. We revised the manuscript according to your comments.

Remarks:
The first sentence of the introduction should be in plural. RhodopsinS.
We appreciate your careful reading and the manuscript was revised according to your suggestion.
It would be extremely important to mark the P219 S254 positions in figure S1. And to also include it that figure the sequence of JsNaR.

Oded Beja
Thank you for the fruitful suggestion. We marked those residues and included the sequence of We thank the reviewers for his/her detailed review and positive comments (italic black) that, "The light-driven sodium pump rhodopsin is a potential optogenetic tool therefore this result may lead to future useful applications. In general, this work comprises useful information for color tuning of other rhodopsins." and for his/her criticism, "Red-shifting Mutation of Light-driven Sodium Pump Rhodopsin" by Inoue et al. is a solid experimental work but it lacks sufficient novelty which is a reason for publishing it in Nature Communications. Indeed, color tuning is quite an old story in rhodopsin field. Indeed, in the introduction of J. Am. Chem. Soc. 2006, 128, 10808-10818 the authors systemize major ideas of color tuning which were already discussed in literature and experimentally proven: "Several mechanisms for color tuning in these systems have been proposed: (i) coplanarization of the ring-chain system and further distortion of the chromophore structure; (ii) electrostatic interaction of the chromophore with ionic, polar and polarizable groups of the protein environment; and (iii) a change in the interactions between the chromophore and its complex counterion." Color tuning was also discussed in The ISME Journal (2007), 48-55, in  As this reviewer suggested there are many earlier researches on the color tuning rule of rhodopsins. However, we consider that the most outstanding finding in our work is "color tuning without impairing the biological function". Although the earlier works suggested by the reviewer focused on the color tuning mechanism of rhodopsins (except for the work in The ISME Journal (2007) which investigated geographical distribution of green and blue absorbing proteorhodopsins and is not related to the molecular mechanism of color tuning), they were not concerned about the functional alteration by mutations. Both the P219T and S254A mutations reported in our work significantly shift the absorption while retaining transport as efficient as that of the wildtype. We believe that the discovered mutations are applicable to many other types of ion-transporting rhodopsins to develop new optogenetic tools, and difficult to predict based on the insights reported before. In addition, we revealed that the mutation of Pro219 is naturally occurring in Jannaschia seosinensis to adapt favorable sun-light wavelength without loss of the function.
Therefore, we believe this work would attract broad interest from not only biophysical but also optogenetics, microbial, evolutional researchers. In order to more clearly show this point, we added the following sentence in the last paragraph of the manuscript: Page 17, The molecular-level design of rhodopsins regulating the absorption wavelength of the retinal chromophore without impairing the transport activity is still difficult compared to the prediction of the only absorption wavelength. Thus, our findings expand the experimental basis useful to establish the artificial design of functional molecules useful for optogenetics application.
Also, since we consider the previous works mentioned by the reviewer would be insightful for the readers, we cited these works in the revised manuscript: Page 4-5, line 100-102 The O-H bearing Ser141 and Thr142 neighbouring to the -ionone ring in BR were also suggested to contribute to the red-shift of the absorption by structural, mutational and theoretical studies 38-40 .

6
Replies to the comments of Reviewer 3 As we report in the following, the reviewer comments (italic black) have been carefully considered and the corresponding point-to-point answers are given below. Below, we also report on the corresponding changes to the main text and supporting information that, we believe, should satisfy the reviewer's requests.
Reviewer #3 ( We are grateful to the reviewer for his/her remark and questions. We agree with the reviewer that the ARM label is non-standard and that the sentence referring to a qualitative and quantitative expla nation is confusing and needs to be changed. With regard to this second point, we meant that, as documented and reported in the present manuscript, the type of QM/MM models used in our research can reproduce the observed trend in wavelength of the absorption maxima and, in turn, can be used as an analytical tool to evaluate, for each mutation, the magnitude of the electrostatic effect associated with each given amino acid replacement.
In order to make the abstract more readable and avoid confusion, the original sentence: "...QM/MM models generated with ARM protocol showed qualitatively and quantitatively that the changes in the electrostatic interaction between retinal and the mutated residues lowered the energy gap between the ground and the first electronically excited state...." has been replaced with the following sentence: Page 2, line 34-37 "...QM/MM models generated with an automated protocol showed that the changes in the electrostatic interaction between protein and retinal chromophore induced by the amino acid replacements, lowered the energy gap between the ground and the first electronically excited state...." 3. p2 line 42: please reformulate: "and so on".
Thank you for this suggestion. The sentence has been changed to: Page 2, line 42-43 "…Microbial rhodopsins are photoreceptive membrane proteins widely distributed among diverse eubacteria, archaea, eukaryotic algae, fungi and giant viruses. 1,2 …" 4. p.3 line 71: please provide a reference that S1 is more sensitive to the charge perturbations.
The original sentence led to a misinterpretation. In fact, we are not aware of any study showing that the S 1 , rather than the S 0 state, of the retinal chromophore is more sensitive to and external charge perturbation. We just wanted to stress the fact ( The following reference has been repositioned: The original reference 60 (PNAS (2006) We have considered the point raised by the reviewer and clarified it. In fact, the considered mutations does distort the chromophore with respect to the WT geometry. On the other hand, we find that, as stressed in the manuscript, such distortions blue-shift the absorption. However, the electrostatic interactions introduced by the same mutation red-shift it at a larger and dominating extent.
The following parts have been improved for clarifying the matter: Page 11, line 259-263 "...The constructed QM/MM models allowed to disentangle the electrostatic and steric effects responsible for the observed ΔΕ S1-S0 red-shifted values. To do so we computed the ΔΕ S1-S0 values of the retinal chromophore in isolated condition (i.e. removing the protein part from the model while keeping the chromophore geometry fixed at its equilibrium geometry in the protein environment)...." Page 11, line 382-387 "...Interestingly, it was found that the mutation-induced changes in the retinal geometry cause a blue-shift with respect to the WT. However, as shown in Table 1 (right column), we also found that a red-shifting electrostatic contribution imposed by the protein environment dominates, thus quenching th e chromophore steric effect and resulting in a net red-shifted change, with the strongest effect observed for the P219T/S254A mutant...." The idea is to analyze the tuning associate to a mutation by hypothesizing that it can be seen as a sum of electrostatic (point charges) and steric (chromophore and cavity geometrical distortion). We determine the electrostatic effect at the mutant equilibrium geometry by switching off the protein charges to isolate the geometrical effect.
In order to clarify the meaning of our analysis and avoid confusion, the original sentence: "... In addition, using the QM/MM models, a quantitative explanation regarding the red -shift effect observed in the mutants, with respect to the WT, can be provided. We can evaluate the specific effect of each mutation, by setting the point charges of the side-chain under investigation to zero and then use the same QM/MM models to re-compute the vertical excitation energy (ΔΕ off )....." has been replaced with the following sentence: Page 12, line 281-283 "…Using the same QM/MM models, it is also possible to investigate the role played by each protein amino acid residues in the described red-shifting relative to the WT form. In fact, one can set the point charges of each specific residue to zero and then re-compute the ΔΕ S1-S0 value (ΔΕ off )...." 8. The experimental value of KR2 P219G is 535 nm (Page 13, line 309 in the original manuscript).
Therefore the amount of the shift from the  max of KR2 WT is reproduced reasonably well by the calculation. In the original manuscript, we claimed that the spectrum of KR2 P219G is shown in Fig. 7a.
However, due to an error in the production of the figure, that was not the case. Therefore, we added the spectrum of KR2 P219G into Fig. 7a.

Figure caption of Figure 7a
The absorption spectrum of KR2 wildtype and P219G mutant are shown by the magenta dotted line and cyan solid line, respectively.
Revised Figure 7 9. We thank the reviewer for his/her remark. We agree that the all-trans chromophore in vacuum and in its equilibrium geometry absorbs around 600 nm. Such observed absorption (see for instance  Table 1 of the original manuscript, we discovered a typographical error in compiling the table. In fact, the correct computed red-shifted excitation energy values are only 6-12 kcal/mol red-shifted with respect to the protein values and not 12-20 kcal/mol as originally reported. Therefore, we modified Table 1 in the main text which now includes the correct values yielding wavelengths around 620 nm due to the fact that the in Vacuum chromophores are not relaxed being taken with their protein optimized geometries. Revised P219T/S254A 51.5 (-3.7) 45.9 (+2.7) +5.6 (-6.5) Fig. 6: A positive dipole is drawn from the oxygen of the hydroxide to the beta-ionine unit. However, the oxygen is more electronegative (e.g. -0.66e in the CHARMM force field) than the beta-ionone unit. The more plausible explanation is that the positive charge on the beta-ionone in S1 is stabilized by the negative charge on the Thr unit. Please comment and revise model and discussion accordingly.

I do not agree on the explanation of the shifts given
We are indebted to the reviewer for his/her very helpful comment. Indeed, regrettably, the interpretation of the generated dipole was not correct for P219T and P219T/S254A. Following the suggestion in point 16 of the reviewer criticism, we have now modified Figure 6 to provide a better and qualitatively correct visualization of the retinal and residues and we increased the quality of the figure. The new version of Figure 6 is reported below (see point 16 for the inserted changes).

Is the amount of charge that moves to the beta-ionone unit the same in WT and all mutants?
We analysed the total positive charge on the C11H-C12H-C13(Me)-C14H-C15H-NHR moiety (i.e., relevant to the C11=C12 isomerization) of the retinal chromophore for the ground (S 0 ), fist (S 1 ) and second (S 2 ) singlet states. Also, we performed the same analysis for the charges corresponding to the shorter C13(Me)-C14H-C15H-NHR moiety (i.e., relevant to the C13=C14 isomerization, see values in parenthesis). The corresponding data show data there is not a large difference in the charge movement between KR2 WT and its mutants. These data have now been inserted in the Supporting Information file as Supplementary  (2016)), the QM/MM models constructed with the ARM protocol are a gas-phase and globally uncharged monomer protein models, which do not incorporate the protein environment (membrane and solvent). As demonstrated in the mentioned paper, the effect of the protein environment is indirectly incorporated with the introduction of Cland Na + counterions on the cytoplasmic and external faces of the rhodopsin structure. Of course, such only apparently basic models have been benchmarked. In fact, it has been shown that they can reproduce the observed trends in excitation energies with, most frequently, a systematic blue shifted error of 2-3 kcal/mol.
In order to clarify the meaning of our analysis and avoid confusion, we have improved the following text in the Methods section: Page 18-19, line 452-458 "...While ARM models are basic gas-phase and globally uncharged monomer models constructed starting from an X-ray crystallographic structure or comparative model, the benchmark shows that ARM models reproduce the experimental λ max values with a few kcal/mol discrepancy and that can reproduce trends in λ max . Accordingly, in the present contribution, the rhodopsin models are used to study the molecular mechanics determining the λ max changes after having been validated by reproducing the observed λ max ...." 14. p. 30 figure 1: please clearly indicate the location of S254.
We now show Ser254 in Figure 1   The residues mutated to the identical ones as occurring in Chrimson and nine further screened residues are coloured in orange and green, respectively. The C  atoms are shown as spheres for Gly residues. Ser254 near the retinylidene moiety (in yellow) is coloured in cyan.
15. p. 34: figure 5b is difficult to understand. Please move this to the SI. This data should also be plotted rather than listing so many values.
We thank the reviewer for suggesting how to improve Figure 5. We decided to modified Figure 5 in the following way: (i) Figure 5a remains as previously presented, (ii) we created two new figures, Figure   5b and 5c, in which we show, for each mutant, the difference in bond lengths and dihedral angles relative to the corresponding KR2 WT values. The corresponding numerical data are then collected in two new supplementary, Tables 3 and 4, in the Supporting Information.
Notice that we also improved the quality of the figures (600 DPI) and the absolute width for a single-column figure of 1040 pixels wide. The new supplementary Tables 3 and 4 and the modified Figure   5 are reported below: