Structural Investigations of the Light-Driven Sodium-Pumping Rhodopsin KR2

Structural investigations of the light-driven sodium-pumping rhodopsin KR2

Von der Fakultät für Georessourcen und Materialtechnik der Rheinisсh -Westfälischen Technischen Hochschule Aachen

zur Erlangung des akademischen Grades eines

Doktors der Naturwissenschaften

genehmigte Dissertation

vorgelegt von

M. Sc. Kirill Kovalev

aus Yaroslavl, Yaroslavl region, Russland

Berichter: Herr Univ.-Prof. Dr. Valentin Gordeliy Herr Prof. Dr. Georg Büldt Herr Prof. Dr. Ernst Bamberg Herr Prof. Dr. Martin Engelhard

Tag der mündlichen Prüfung: 07.12.2020

Diese Dissertation ist auf den Internetseiten der Universitätsbibliothek online verfügbar

Table of contents

Table of contents 2

Main results 5

Publications 6

List of Abbreviations 8

1. Introduction 9

1.1. Microbial rhodopsins 9

1.2. Identification of the light-driven Na+ pumps 12

1.3. Biological role(s) of NaRs 14

1.4. Functional and spectroscopic features of NaRs 15

1.5. NDQ motif of NaRs 17

1.6. Oligomeric state of NaRs 20

1.7. High-resolution structures of NaRs 22

1.8. Structures of the KR2 in the intermediate states 26

1.9. Mechanism of light-driven Na+ pumping 26

1.10. Functional conversions of NaRs 29

1.11. NaRs as potential optogenetic tools 31

2. Materials and Methods 33

2.1. Materials 33

2.1.1. Organisms 33

2.1.2. Vectors 33

2.1.3. Genes 33

2.1.4. Oligonucleotides 33

2.1.5. Chemicals for molecular biology 33

2.1.6. Crystallization 33

2.1.7. Crystal harvesting tools 34

2.2. Methods 35 2

2.2.1. Cloning 35

2.2.2. Protein Expression and Purification 37

2.2.3. Measurements of pumping activity in E. coli cells 37

2.2.4. Liposome preparation 38

2.2.5. Measurements of the pumping activity in liposomes 38

2.2.6. Oligomeric state analysis by size exclusion chromatography 38

2.2.7. Crystallization details and crystals preparation 39

2.2.8. Time-resolved visible absorption spectroscopy on KR2 crystals 39

2.2.9. Spectroscopic characterization and accumulation of the intermediate state in KR2 crystals 40

2.2.10. Acquisition and treatment of diffraction data 40

2.2.11. Serial millisecond crystallography data collection and processing 41

2.2.12. Structure determination and refinement 42

2.2.13. Molecular dynamics simulations 42

3. Results and Discussion 44

3.1. High-resolution structure of KR2 in the ground state 44

3.1.1. Crystallization of KR2 under physiological conditions 44

3.1.2. Crystal structure of the pentameric KR2 in the ground state 44

3.1.3. Interprotomer contacts in KR2 complex 48

3.1.4. Comparison with known KR2 structures 50

3.1.5. The second Na+ ion identified at the KR2 surface in the ground state 54

3.1.6. Structural switches in KR2 upon pH decrease 55

3.1.7. Effects of dehydration on KR2 crystals 58

3.1.8. The structure of the monomeric form of KR2 at different pH 59

3.1.9. Structures of K+-pumping mutants of KR2 60

3.1.10. Role of the KR2 pentamerization on the function of the protein 64

3.2. Crystal structure of KR2 in the O-state 68

3.2.1. Accumulation and cryo-trapping of the O intermediate state in KR2 crystals 68 3

3.2.2. Determination of the crystal structure of the O-state of KR2 71

3.2.3. The retinal binding pocket of KR2 in the O-state 73

3.2.4. Transient Na+ ion-binding site inside the KR2 protomer 75

3.2.5. Structure of the KR2 protomer in the O-state 79

3.2.6. Crystal structure of the ground and the O-states of KR2 at room temperature 83

3.2.7. Conformational switches guide Na+ uptake and release in KR2 86

3.2.8. Na+ translocation pathway 91

3.3. Molecular mechanism of light-driven Na+ pumping 99

3.4. Outlook 102

4. References 103

5. Appendix 113

Acknowledgments 119

Abstract 120

Main results

In the framework of the present dissertation, the structure of the biologically relevant pentameric form of KR2 was obtained under physiological conditions. The structure revealed a large polar water-accessible cavity in the core of the KR2 protomer in the ground state of the protein photocycle. The results were published in Science Advances in 2019.

Based on the first result, the structure of the functionally key O intermediate state of KR2 photocycle was obtained by using three alternative techniques. First, the intermediate state was accumulated and cryo-trapped in KR2 crystals upon their illumination with 532 nm laser following by flash-cooling in the cryo-stream. Second, the structure of the O-state was obtained at room temperature using data collection upon continuous light illumination of large single crystals of KR2. Third, the data on the O-state were collected using serial crystallography with the stream of KR2 microcrystals, injected into the X-ray beam of the synchrotron source with simultaneous illumination of the stream by 532 nm laser. All three structures are similar and demonstrate a transient Na+ ion-binding site in the core of KR2 protomer.

Based on the structural data we suggest that the mechanism of a light-driven Na+ pumping is likely a chimera of the relay mechanism of proton translocation and passive diffusion of the Na+ ions through the cavities inside KR2. The results were published in Nature Communications in 2020.

The structural studies of KR2 are still ongoing. Particularly, the next goal is to obtain the structures of the early K, L, and M states of the pentameric KR2 using time-resolved crystallography at synchrotrons and X-ray free-electron laser sources.

Publications

T Varaksa, S Bukhdruker, I Grabovec, E Marin, A Kavaleuski, A Gusach, K Kovalev et al. (2020) Metabolic fate of human immunoactive sterols in Mycobacterium tuberculosis. bioRxiv. https://doi.org/10.1101/2020.07.07.192294.

N Maliar*, K Kovalev* et al. (2020) Crystal structure of the N112A mutant of the light-driven sodium pump KR2. Crystals. Volume 10, Issue 6, Pages 1-15

K Kovalev, R Astashkin et al. (2020) Molecular mechanism of light-driven sodium pumping. Nature Communications. 11, 2137.

A Remeeva, V Nazarenko, I Goncharov, A Yudenko, A Smolentseva, O Semenov, K Kovalev et al. (2020) Effects of proline substitutions on the thermostable LOV domain from Chloroflexus aggregans. Crystals. Volume 10, Issue 4 (256).

K Kovalev*, D Volkov*, R Astashkin*, A Alekseev* et al. (2020) High-resolution structural insights into the heliorhodopsin family. PNAS. Volume 117, Issue 8, Pages 4131-4141.

D Zabelskii*, A Alekseev*, K Kovalev* et al. (2020). Viral channelrhodopsins: calcium- dependent Na+/K+ selective light-gated channels. bioRxiv. https://doi.org/10.1101/2020.02.14.949966.

A Vlasov*, K Kovalev*, S-H Marx*, E Round* et al. (2019) Unusual features of the c-ring of F1FO ATP synthases. Scientific Reports. Volume 9, Issue 1 (18547).

A Gusach, A Luginina, E Marin, R Brouillette, É Besserer-Offroy, J-M Longpré, A Ishchenko, P Popov, N Patel, T Fujimoto, T Maruyama, B Stauch, M Ergasheva, D Romanovskaia, A Stepko, K Kovalev, et al. (2019) Structural basis of ligand selectivity and disease mutations in cysteinyl leukotriene receptors. Nature Communications. Volume 10, Issue 1 (5573).

D Bratanov*, K Kovalev* et al. (2019) Unique structure and function of viral rhodopsins. Nature Communications. Volume 10, Issue 1 (4939).

K Kovalev*, V Polovinkin* et al. (2019) Structure and mechanisms of sodium-pumping KR2 rhodopsin. Science Advances. Vol. 5, no. 4, eaav2671.

V Nazarenko, A Remeeva, A Yudenko, K Kovalev et al. (2019) A thermostable flavin-based fluorescent protein from: Chloroflexus aggregans: A framework for ultra-high resolution structural studies. Photochemical and Photobiological Sciences. Volume 18, Issue 7, Pages 1793-1805.

O Volkov*, K Kovalev*, V Polovinkin*, V Borshchevskiy* et al. (2017) Structural insights into ion conduction by channelrhodopsin 2. Science. Vol. 358, Issue 6366, eaan8862.

I Melnikov, V Polovinkin, K Kovalev et al. (2017) Fast iodide-SAD phasing for high-throughput membrane protein structure determination. Science Advances. Volume 3, Issue 5, e1602952.

V Shevchenko*, T Mager*, K Kovalev*, V Polovinkin* et al. (2017) Inward H+ pump xenorhodopsin: Mechanism and alternative optogenetic approach. Science Advances, Volume 3, Issue 9, 1603187.

V Borshchevskiy, E Round, Y Bertsova, V Polovinkin, I Gushchin, A Ishchenko, K Kovalev et al. (2015) Structural and functional investigation of flavin binding center of the NqrC subunit of sodium-translocating NADH: Quinone oxidoreductase from vibrio harveyi. PLoS ONE. Volume 10, Issue 3, e0118548.

I Gushchin, V Shevchenko, V Polovinkin, K Kovalev et al. (2015) Crystal structure of a light- driven sodium pump. Nature Structural&Molecular Biology. Volume 22, Issue 5, Pages 390-396.

*These authors contributed equally to the work

List of Abbreviations

MR - microbial rhodopsin

MP - membrane protein

HsBR - bacteriorhodopsin from Halobacterium salinarum

PR - proteorhodopsin

RSB - retinal Schiff base

PDB - protein data bank

FTIR - fourier-transform infrared spectroscopy

RR - resonance Raman spectroscopy

NaR - light-driven sodium-pumping rhodopsin

KR2 - light-driven sodium pump from Krokinobacter eikastus

IPTG - isopropyl β-D-1-thiogalactopyranoside

CCCP - carbonyl cyanide m-chlorophenyl hydrazone

TPP+ - tetraphenylphosphonium bromide

GPCR - G protein-coupled receptor

NMR - nuclear magnetic resonance

HS-AFM - high-speed atomic force microscopy

SEC - size exclusion chromatography

OD600 - optical density at 600 nm wavelength

LCP - lipidic cubic phase

SBC1 - Schiff base cavity 1

SBC2 - Schiff base cavity 2

IUC - ion-uptake cavity pIRC1 - putative ion-release cavity 1 pIRC2 - putative ion-release cavity 2

WT - wild type

1. Introduction

1.1. Microbial rhodopsins

Microbial rhodopsins (MRs), also known as type-1 rhodopsins, are light-driven integral membrane proteins (MPs), found in archaea, bacteria, simple eukaryotes and also in viruses. They were shown to be major contributors to the solar energy captured in the ocean1. MRs possess numerous diverse functions, such as active and passive translocation of ions, sensory and enzymatic activities, etc. MRs share similar topology, forming a bundle of seven transmembrane (TM) helices named A to G connected with 3 extracellular and 3 intracellular loops2,3 (Figure 1.1.1a). The light sensitivity of MRs is provided by the prosthetic group retinal, covalently bound via protonated retinal Schiff base (RSB) to the lysine residue, located in the middle of the helix G. In the dark (ground) state MRs typically possess all-trans retinal in their core. Upon the absorption of the light photon, the retinal cofactor of MRs isomerizes to 13-cis configuration, following the thermal reisomerization back to the all-trans form (Figure 1.1.1b). The changes in the chromophore trigger the structural rearrangements of the surrounding protein, which leads to the appearance of the chain of metastable spectrally distinct intermediate states of MR. Each state is characterized by a specific maximum absorption wavelength. Finally, when the retinal is fully- relaxed to its original configuration, the protein returns to its ground state, and the reaction could be repeated in the same manner. Thus, MR works in a cycle, called photocycle (Figure 1.1.1c). Importantly, each photocycle exactly repeats the previous one. The intermediate states of the MR photocycle are named with capital letters I to O.

Figure 1.1.1. Characteristics of MRs. a. Overall architecture of HsBR (PDB ID: 1C3W). Retinal cofactor is colored teal. Hydrophobic/hydrophilic membrane core boundaries are shown with black

9 horizontal lines. BC-loop containing a β-sheet is colored orange. b. Scheme of retinal isomerization from all-trans to 13-cis configuration in HsBR. The image from 4. c. Photocycle of HsBR. The image from 5.

The first and at the moment the most studied MR is a light-driven outward H+ pump bacteriorhodopsin from archaeon Halobacterium salinarum (HsBR)6. It was found almost fifty years ago, in 1971 by D. Oesterhelt and W. Stoeckenius6. Later in 1980 the second MR was found in the same organism7,8. This was a light-driven inward Cl- pump halorhodopsin (HsHR). Originally, this protein was suggested to be a Na+ pump, but it was soon shown that it pumps not cations, but anions8. Two more MRs, sensory rhodopsins I and II (HsSRI and HsSRII, respectively) became the next discoveries in the rhodopsins field of research. Thus, at the end of the XXI century there were only a few and exclusively archaeal MRs known and it seemed that the rhodopsins era is nearly over.

The game-changing discovery of the first light-driven H+ pump in marine bacteria, proteorhodopsin (PR) by O. Beja and colleagues in 1999 by using metagenomics served as a foundation of the new era of rhodopsins in the last 20 years9. Soon after that two channelrhodopsins (CrChR1 and CrChR2) were found in algae Chlamydomonas reinhardtii10,11. In 2005 it was shown that neuronal activity could be controlled in vitro with light. Thus, the discovery of CrChR2 led to the most significant biotechnological application of the proteins - optogenetics12.

Figure 1.1.2. Phylogenetic tree of microbial (type-1) rhodopsins. Type-1 rhodopsins can be roughly divided into seven major clades: i) Bacteria- and Archaea-rhodopisns, that includes proton pumps and proteorhodopsins (i.e. blue- and green- absorbing proteorhodopsins, BPR and GPR, respectively), xanthorhodopsins (XR), sodium- and chloride pumps (NaR and ClR, respectively). ii) Rhodopsins encoded by Halobacteriales archaea. Inward pumps xenorhodopsins appear related to Halobacteriales sensory rhodopsins, although they were identified in five different bacterial phyla. iii) Channel rhodopsins encoded by single cell eukaryotes. The lineage includes mono/divalent cation- (i.e., CrChR2) and anion-channelrhodopsins (ACRs, such as GtACR1). iv) Histidine-kinase rhodopsins (i.e., HKR1). v) Two types of viral rhodopsins encoded by viruses that prey on unicellular algae. vi) Heliorhodopsins (HeR) that differ from other type-1 rhodopsins (and type-2) for the opposite orientation in the plasma membrane (N-termini facing the cytoplasm, and extracellular C-termini). vii) Schizorhodopsins (SzR), placed between HeR and the other type- 1 rhodopsins. These rhodopsins are encoded by a particular group of Archaea (Asgardarchaeaota) that has been hypothesized to occupy a basal position of the eukaryote tree of life and share common features with both type-1 and HeR rhodopsins.

The development of optogenetics required the expansion of the set of known MRs. Due to the simultaneous progress of functional metagenomics, many new MRs were found during the last

10 years (Figure 1.1.2). Among them, there are bacterial cation and anion pumps (NaRs and ClRs)13,14, anion channelrhodopsins (ACRs)15, rhodopsins from giant viruses16, inward proton pumps xenorhodopsins (XeRs)17,18 and schizorhodopsins (SzRs)19, enzymorhodopsins20, and heliorhodopsins (HeRs)21.

Such discoveries and consecutive accurate characterization of MRs with new properties is a challenging problem. Nowadays, the number of research groups working on rhodopsins is constantly increasing. In some cases it results in the appearance of inconsistencies between the results obtained by different teams. The first and most important example is HsBR. Indeed, despite almost fifty years of intensive studies of this protein, its mechanism of work remain under debates22. Unfortunately, similar situation exists in the case of some other proteins. This work is dedicated to one of such examples, KR2, a member of recently found family of light-driven Na+ pumps.

1.2. Identification of the light-driven Na+ pumps

In 2012 eight strains of Flavobacteria were found to contain PR-like genes23. Expression of these genes allowed the cells to create transmembrane potential high enough for ATP synthesis upon light illumination. Soon after that, in 2013, flavobacterium Krokinobacter eikastus was shown to contain two PR-like genes13. While the existence of the first gene likely coding an outward proton pump (KR1) was in line with the previous study, the expression of the second gene resulted in unusual effects. It was shown, that at 48 h of K. eikastus cell growth the pH of the surrounding solution was increasing upon light illumination (Figure 1.2.1a). The effect was not altered much by the addition of the protonophore carbonyl cyanide m-chlorophenyl hydrazone (CCCP) and was eliminated by the addition of ionophore tetraphenylphosphonium bromide (TPP+), which is not typical for the H+-pumping rhodopsins. This second PR-like gene was isolated and the protein, named KR2, was expressed heterologously in E.coli cells for the functional studies (Figure 1.2.1b). Investigations of pH changes upon light illumination in the suspension of E.coli cells expressing KR2 depending on the composition of the surrounding solution showed that the protein is capable of transporting Na+, Li+, and protons in the absence of other cations. It was also shown that KR2 cannot pump larger cations, such as K+. These results were presented in 2013 together with the first spectroscopic characterization and intensive mutational analysis of the protein13. Thus, KR2 became the first known Na+-pumping rhodopsin (NaR).

Figure 1.2.1. Functional activity of KR2. a. Pumping tests in native K. eikastus cell suspensions. Cells were collected at 48h of the growth phase. b. Pumping tests in E. coli cells suspension expressing KR2. Image from 13.

This was an unexpected finding since scientific society mostly did not believe in the possibility of MR to pump cations other than protons. The primary reason is that the core of MRs is formed by a protonated RSB, and there should be a conflict of the simultaneous presence of two positive charges nearby, as the transporting substrate is usually bound in the active center of the transporter in the ground state. Another reason was in the fundamental difference between an H+ and non-H+ cation transport. Indeed, in the latter case, the protein cannot utilize Grotthuss or tunneling mechanisms for ion transport, which means that the molecular mechanism of the light- driven Na+ pumping cannot be solely understood in the framework of classical H+ pumps, such as HsBR. However, already the first study showed that Na+ does not bind near the RSB in the ground state of KR2, which made the Na+ pump a unique active transporter.

Soon after the discovery of KR2, in 2014, the second NaR was functionally and spectroscopically characterized. This was rhodopsin from Gillisia limnaea, named GLR24. Overall, GLR shares the same functional properties as those of KR2. However, unlike KR2, GLR was shown not to pump H+ in the absence of Na+ ions. Also in 2014, third NaR was found in flavobacterium Nonlabens marinus S1-08T (NM-R2) and was functionally characterized14. Surprisingly, a unique Cl- pump (NM-R3) was also found in the genome of this bacteria14. At the beginning of 2015, next Na+-pumping proteorhodopsin from Dokdonia sp. PRO95 was described25. Later in 2015, a NaR was identified in Indibacter alkaliphilus (IAR) and its Na+-

13 pumping activity was verified in vitro using the protein reconstituted into large unilamellar vesicles (LUVs)26. NaRs were also found in Nonlabens sp. YIK1127, Croceitalea dokdonensis28, and Nonlabens dokdonensis29. Electrophysiology of twelve NaRs was studied in 2017, which identified one particular NaR from Salinarimonas rosea DSM21201 (SrNaR) with unusual kinetics and functional behavior30.

Despite constantly increasing number of identified microbial rhodopsin genes NaRs at the moment form a relatively small family. Namely, there are 76 proteins in the NaR family including those found in terms of Tara oceans expedition31. Since KR2 is the most characterized member of the NaR family and is the only structurally described, we will mostly focus on this rhodopsin. However, we will refer to other NaRs where needed.

1.3. Biological role(s) of NaRs

The biological role of light-driven Na+ pumps remains unknown. The main version is the involvement of the NaRs in the creation of Na+ gradients across the cell membrane13. These gradients could be utilized by the cell to perform the import of nutrients, or to synthesize ATP using Na+ ATP-synthases32,33. Na+ bioenergetics is considered to be the primary one since the first membranes in the course of evolution were very leaky due to high permeability to H+ 34. On the contrary, Na+ diffusion through the membrane is more difficult. Therefore, it is hypothesized that the first organisms might utilize Na+ gradients for their functions.

In terms of this concept, NaRs are proposed as possible ancestors for other MRs and also for class A G protein-coupled receptors (GPCRs) (Figure 1.3.1)35. Indeed, it became known that members of the class A GPCRs contain the well-conserved inner Na+ binding site. It was thus speculated that this site might originate from that of NaRs. Based on this suggestion, the Na+ binding site in KR2 was predicted to be located in the cytoplasmic inner part of the protein near W215 residue.

Figure 1.3.1. Proposed in 35 scheme of the evolution of MRs and GPCRs from NaRs. The scheme illustrates the proposed order of appearance of functions in evolution as a series of gains and losses. The image from 35.

Later in 2019, it was further suggested the class A GPCRs could have resting Na+ pumping activity36. This finding was also supported by the certain similarity of the KR2 and class A GPCRs, described in 35. Thus, the studies of KR2, and NaRs, in general, are of high importance not only for MRs field of research but also for the understanding of their evolution and fundamental mechanisms of other MPs.

1.4. Functional and spectroscopic features of NaRs

As was already mentioned, KR2 is able to pump Na+, Li+, and H+ in the absence of these ions. This is valid for some other NaRs. However, two proteins, GLR and Dokdonia sp. PRO95 rhodopsin lacks the H+-pumping activity24,25. Thus, these rhodopsins are highly-selective to

15 cations other than H+. The comparison of the sequences of KR2 and Dokdonia sp. PRO95 rhodopsin, together with the mutational analysis indicated that the most probable reason for the difference in ion selectivity between these two proteins is in the mutation at the position of S253 in KR237. In Dokdonia sp. PRO95 rhodopsin, serine is replaced by cysteine. The same cysteine was found in the case of GLR.

Spectroscopy provided the first understanding of mechanisms of NaRs. Their typical maximum absorption wavelengths are within 520-530 nm range13,24. Thus, the maximum absorption wavelength of KR2 under physiological conditions, similar to those in the ocean (pH 8.0, 500 mM NaCl), is around 524 nm13 (Figure 1.4.1a). Upon acidification of the surrounding media, the peak of the absorption spectra of KR2 is red-shifted presumably due to the protonation of the RSB counterion, D116 in KR2. The maximum absorption wavelength of the protein was originally shown not to depend notably on the cation composition of the surrounding solution. At the same time, Fourier-transform infrared (FTIR) spectroscopy showed Na+ binding to the protein13. Combining the two latter observations, Inoue and colleagues suggested that Na+ binds to KR2 but distant from the RSB region13. The difference FTIR spectra of KR2 mutants indicated the involvement of BC-loop and R109 and N112 residues in the formation of the ion-binding site in the ground state of the Na+ pump.

Time-resolved visible absorption spectroscopy revealed the photocycle of KR2, both solubilized in detergent and reconstituted into lipids13. The photocycles were similar, with only small variations. KR2 undergoes four main intermediate states, K, L, M, and O, where only the formation of the O-state is sensitive to the Na+ concentration (Figure 1.4.1b). In fact, two regimes of KR2 work were identified. The first regime corresponds to the Na+-pumping activity. It is characterized by the pronounced accumulation of the red-shifted O intermediate state, the rise of which accelerates upon increasing of Na+ concentration. The second regime corresponds to the H+-pumping mode of the NaR, and appear at low concentrations or complete absence of Na+. The formation of the K and L states are similar in both regimes, while the accumulation of the O-state is decreased dramatically in the H+-pumping form of KR2. The photocycle in this case is also much slower. The results indicated that the Na+ is likely uptaken by KR2 with the rise of the O- state (Figure 1.4.1b). Since it is not bound near the RSB in the ground state, the Na+ release should be associated with the O-to-ground transition. Thus, Na+ binding in the core of KR2 occurs transiently in the late O-state. The L- and the M- states of the rhodopsin photocycle coexist, with only a small fraction of the M-state, corresponding to the deprotonated RSB. The same findings on NaR photocycle were shown in the thorough spectroscopic study of GLR24. 16

Figure 1.4.1. Spectroscopy of KR2. a. Absorption spectra of KR2 in the presence (red) and absence (blue) of Na+. Image from 13. b. Photocycle of KR2 in the Na+-pumping mode. Reproduction from 13.

In 2015 the kinetic analysis of KR2 at different Na+ concentrations showed that the protein uptakes Na+ and H+ in a competitive manner38. Moreover, the rate constant of H+ uptake is more than 103 times larger than that of Na+. However, taking into account that under physiological conditions [Na+] is approx. 107 higher than [H+], KR2 pumps almost exclusively Na+ in the native host.

1.5. NDQ motif of NaRs

The functional properties of MRs are considered to be dictated by the features of their amino acid sequences. In terms of function, aside from the key lysine in the helix G, the most critical residues of rhodopsins are typically located in the helix C. Particularly, three residues, called a motif, are considered to be closely related to MR’s function. The motif is a set of the key functional amino acids, located at the positions analogous to D85, T89, and D96 in HsBR (Figure 1.5.1). In HsBR, D85 serves as a proton acceptor from the RSB, while D96 is a proton donor39. T89 stabilizes the RSB in the process of photocycle and is also believed to be a part of a transient chain of hydrogen bonds from the RSB to proton acceptor D85 in the process of L-to-M transition40. In KR2, the corresponding positions are occupied by N112, D116, and Q123 triad (Figure 1.5.1). Hence, all members of the NaR family have a characteristic NDQ motif in the TM helix C.

Figure 1.5.1. Amino acid motifs of HsBR (left) and KR2 (right). Retinal cofactor is colored teal.

The most significant feature of NaRs is the shift of the aspartate residue one α-helix turn closer to the RSB compared to HsBR and many other MRs. Indeed, aspartate at the position of T89 in HsBR is only found in NaRs. Therefore, unlike in HsBR, where the connection between the RSB and proton acceptor D85 in the ground state is mediated by a water molecule41, the RSB and D116 of KR2 are located close and are directly hydrogen-bonded13. The substitution of the aspartate at the 116th position of KR2 resulted in a complete loss of the Na+-pumping activity in KR213. Furthermore, the photocycle of the D116N mutant of KR2 lacks the blue-shifted states (L and M), which likely indicates that deprotonation of the RSB does not occur in KR2-D116N13. D116 is thus considered as a primary proton acceptor from the RSB, similar to D85 in HsBR. The existence of the strong hydrogen bond between the RSB and D116 in the ground state of KR2 was demonstrated by resonance Raman (RR) and FTIR spectroscopy42, solid-state nuclear magnetic resonance (NMR)43 and X-ray crystallography methods44,45. The hydrogen bond was also found in the K, L, and O states, and is only absent in the M-state42,46. This is not surprising since the deprotonation of the RSB to D116 occurs also with the formation of the M-state, thus the RSB- D116 pair is neutralized at that moment.

Another feature of NaRs is the absence of charged amino acids at the cytoplasmic part of the protein. Indeed, the negatively charged aspartate, found in HsBR, is substituted by glutamine in KR2. The Q123 residue of KR2 was shown to provide optimal interhelical interactions for 18 effective Na+ transport47. Thus, Q123A/V/D/E mutants of KR2 preserve Na+-pumping activity, although it is lowered in these proteins13,48,49. Q123A/V also demonstrate much slower kinetics.

NDQ motif is considered to be a determinant of light-driven Na+ pumping49. However, the increasing number of known NDQ rhodopsins genes showed that the NaR family might have several branches (Figure 1.5.2). The major branch (29 members) is very similar to KR2, while another group (18 members) has several important differences. One of the representatives of this another branch, SrNaR, also demonstrated very slow kinetics and corresponding weak Na+- pumping activity30. Time-resolved spectroscopy of SrNaR showed that the photocycle of the rhodopsin is different from that of KR230. Therefore, there might be a chance that not all MRs with NDQ motif functions as Na+ pumps. Further experiments are needed to test this hypothesis.

Figure 1.5.2. Phylogenetic tree of NaRs. KR2 and SrNaR are colored red.

1.6. Oligomeric state of NaRs

The oligomeric state is an important characteristic of MRs50. It is well-known that MRs in most cases form oligomers in the lipid membrane (Figure 1.6.1). Often it is difficult or even impossible to address the oligomerization of rhodopsin in vivo; however some other methods, such as high-speed atomic force microscopy (HS-AFM)51, circular dichroism (CD) spectroscopy52,53, size-exclusion chromatography (SEC)45, native blue gel electrophoresis54, electron microscopy (EM)55, macromolecular crystallography (MX)45,56, small-angle X-ray and neutron scattering (SAXS and SANS, respectively)57 etc may help to study the quaternary structures of MRs.

Figure 1.6.1. Oligomeric assembly of microbial rhodopsins. Surface representation of oligomers of different classes of microbial rhodopsins: BR represents bacteriorhodopsins (PDB ID: 1C3W41); HR and NTQ rhodopsin represent chloride-pumping rhodopsins from archaea and bacteria, respectively (PDB IDs: 1E1258 and 5ZTK59); XeR represents inward proton pumps xenorhodopsins (PDB ID: 6EYU18); ChR represents CCRs and ACRs (PDB ID: 6EID60); PR represents proteorhodopsins (PDB IDs: 4KLY61 and 4JQ661); NaR represents Na+-pumping rhodopsins (PDB ID: 6REW45); VirR1 and VirR2 represent viral rhodopsins of group 1 and 2, respectively (PDB IDs: 6JO062 and 6SQG56); HeR represents heliorhodopsins (PDB ID: 6SU363).

The oligomeric state of KR2 was not studied before 2015, when some of its first crystal structures first demonstrated pentamers of the protein44. The SEC profiles also suggested pentameric assembly of KR2 in detergent micelles at neutral pH and in the presence of Na+. By analogy with PRs, which also may form pentamers, and taking into account some structural features of pentameric KR2, which will be described below, it was thus suggested that the protein is organized in pentamers in vivo64. In 2018 it was directly shown by HS-AFM that KR2 forms pentamers in lipid membranes at both acidic and neutral pH values51 (Figure 1.6.2). Additional studies of the oligomerization of KR2 were performed in terms of this thesis. For instance, we showed that KR2 pentamers dissociate into monomers in the detergent micelles upon the acidification of the surrounding media45. Recently, it was shown that the pentamers of solubilized KR2 dissociate also upon a decrease of the concentrations of Na+ and K+ ions in surrounding solution65.

Figure 1.6.2. AFM images of KR2 at different pH values. a. HS-AFM image of KR2 in lipids at pH 8.0. b. HS-AFM image of KR2 in lipids at pH 4.3. Magenta and cyan arrows indicate the N- terminal (extracellular) and the C-terminal (cytoplasmic) faces of the pentamers, respectively. Image from 51.

It should be noted that the amino acid residues, responsible for the pentamer formation in KR2 are highly-conserved within NaR family. Although the oligomeric state of any other NaR was not addressed, it is believed that all light-driven Na+ pumps form pentamers in vivo.

Unfortunately, the functional relevance of the pentameric assembly of KR2 remained poorly understood. This is one of the major obstacles in the understanding of the mechanisms of KR2 work. It is due to the significant differences in the inner organization of the protein in

21 monomeric and pentameric forms, which is one of the topics of the present thesis and will be described in detail in the Results and Discussion section. Importantly, there is no much information on the functional relevance of MRs oligomerization in general. For instance, HsBR forms trimers in the membrane66; however, the monomeric form of HsBR was also shown to be functional67. The unequivocal conclusion on the importance of KR2 pentamerization also requires the pumping activity tests of the monomeric form of the protein, which is problematic. Indeed, for the functional studies, the protein should be reconstituted into lipid membranes, where it always forms pentamers51. The more indirect methods, such as spectroscopy and kinetics analysis are also difficult since KR2 dissociates into monomers in detergent only in the absence of Na+ or at acidic pH when the photocycle of the protein is altered. At the moment, the direct demonstration of Na+ pumping was performed only with pentameric KR2. The only effort to assess the functionality of the monomer was reported in 2020 and was also indirect, where authors studied the photocycle of the monomeric form of KR2 fixed in crystals68. For that, the crystals containing one KR2 monomer per asymmetric unit were grown at acidic pH and then were soaked in the solution with neutral pH with a sufficient amount of Na+ ions. The results of the study are contradictory. On the one hand, the photocycle of monomeric KR2 in soaked crystals seems to resemble that of the protein in the membrane. On the other hand, the authors described two distinct O-states, O1 and O2, never mentioned previously. Moreover, the structures of these O1 and O2 states differ from that of the O- state of pentameric KR2, as will be shown in the Results and Discussion section. Thus, the identity of the photocycles of monomeric KR2 in crystals and pentameric KR2 in lipid membranes is questionable. Consequently, the study of pentameric protein is the first priority for the understanding of the molecular mechanism of light-driven Na+-pumping.

1.7. High-resolution structures of NaRs

Unfortunately, at the moment, the high-resolution structural data on NaRs are limited to only that of KR2. The results presented in this thesis are the major contributors to the understanding of the 3D architecture of KR2 and will be described in detail in the Results and Discussion section of the work. Therefore, we only briefly summarize the current situation below.

The first structures of KR2 were obtained already in 2015, two years after its discovery, simultaneously by two research groups44,48 (Figure 1.7.1). The group from Japan presented two structures of KR2 in the monomeric form at 2.3 Å resolution (PDB IDs: 3X3B, 3X3C)48. Our team presented three structures: one monomeric and two pentameric forms at 1.45, 2.2 and 2.8 Å, respectively (PDB IDs: 4XTL, 4XTN, 4XTO)44. Only one of the five presented models was 22 obtained at physiological pH value (PDB ID: 3X3C), while the rest four corresponded to pH below 5.6 (Figure 1.7.1). Unfortunately, even in the case of pH 8.0, the structure was obtained using crystals, initially grown at pH 4.0, and soaked in the buffer solution with neutral pH48. Moreover, the structure demonstrated a monomeric form of KR2 and did not reveal Na+ binding site, proposed in the original work on KR213. In opposite, the Na+ binding site was demonstrated by our group in monomeric and two pentameric structures. It turned out that the Na+ ion is located on the KR2 surface and is coordinated by two neighboring protomers of the rhodopsin pentamer.

Figure 1.7.1. Crystal structures of KR2 obtained in 2015. Na+ pumping indicates the range of pH values where KR2 was directly shown to work as a Na+ pump.

Interestingly, four out of five presented models were different from each other, particularly the organization of the RSB region, and also the position of the helix C varied in the structures. The differences were analyzed in our work in 201664. It was found that in the monomeric form of KR2 at acidic and neutral pH values N112 side chain is oriented inside the protomer and is hydrogen-bonded to D116 and S70 residues. A similar configuration was found in all five protomers of the pentameric 4XTO model, obtained at pH 5.6. On the other hand, another conformation, where N112 is oriented towards an oligomerization interface was observed in protomers A and D of the pentameric model 4XTN at pH 4.9. This orientation of N112 creates a large cavity near the RSB. Therefore, the conformation was named expanded64 (Figure 1.7.2). Consequently, another conformation was called compact64 (Figure 1.7.2). It was suggested that the expanded conformation allows enough space for Na+ ion passage during the KR2 photocycle. Hence, the expanded conformation was speculated to be more biologically relevant64.

At that time nobody thought that it will take another four years to obtain the first biologically relevant structure of the protein. In 2019, we reported the structure of KR2 under physiological conditions in its ground Na+-pumping state45. The structure showed that all five protomers within pentameric assembly are in the expanded conformation. The compact conformation was speculated to appear only in the non-physiological monomeric form of KR2 or in pentameric protein upon D116 protonation. It was also noticed that dehydration of KR2 crystals might lead to the appearance of the compact conformation, which presumably occurs in the case of 4XTO model.

Figure 1.7.2. The architecture of KR2 and its conformations. Cavities inside the protein are shown with the pink surface. Retinal cofactor is colored teal. Key hydrogen bonds are shown with black dashed lines. Gray arrow indicates the putative Na+ ion pathway.

Nevertheless, despite several remarkable differences in the protein structures in monomeric and pentameric forms, several features of the KR2 also are common in all the existing models (Figure 1.7.2). First, there is a big water-filled cavity at the cytoplasmic part of the protein, which protrudes from the protein surface almost to the RSB, being separated from it only by the Q123 residue, which is part of the NDQ motif. Second, there is a short N-terminal α-helix, capping the 24 inside of the protein at the extracellular side and thus creating a putative ion-release cavity near E11, E160, and R243 residues. The mutational analysis showed that the helix is crucial for Na+ pumping, although the E11A, E160A, and R243A mutations did not abolish the ion-pumping ability of the protein44,48.

Currently, the most relevant high-resolution structure of KR2 is 6YC369, which corresponds to the pentameric form of the protein at pH 8.0 at 2.0 Å resolution (Table 1.7.1). It showed that the rhodopsin has a third key element: a big polar cavity near the RSB (Schiff base cavity 1), which is filled with 4 water molecules in the ground state. As it was already mentioned, the Schiff base cavity 1 (SBC1) is present only in the expanded conformation.

Oligomeric Resolution, PDB ID Conformation Remarks form Å 3X3B monomeric 2.3 Compact pH 4.0 pH 8.9, soaked 3X3C monomeric 2.3 Compact from pH 4.0 Iodide-bound, 5JRF monomeric 2.5 Compact pH 4.3 4XTL pentameric 1.45 Compact pH 4.3

4XTN pentameric 2.2 Expanded/Compact pH 4.9 pH 5.6, 4XTO pentameric 2.8 Compact dehydrated 6REW pentameric 2.2 Expanded pH 8.0 pH 8.0, 6RF0 pentameric 3.0 Compact dehydrated pH 8.0, 6RF1 pentameric 2.8 Expanded recovered after dehydration pH 6.0, soaked 6REX pentameric 2.7 Expanded from pH 8.0 pH 5.0, soaked 6REZ pentameric 2.6 Expanded/Compact from pH 8.0 pH 6.0, soaked 6RF5 monomeric 2.3 Compact from pH 4.3 pH 8.0, soaked 6RF6 monomeric 1.8 Compact from pH 4.3 pH 8.9, soaked 6RF7 monomeric 2.6 Compact from pH 4.3 6YC3 pentameric 2.0 Expanded pH 8.0

Table 1.7.1. Description of the KR2 structures, deposited to the Protein Data Bank. The most relevant model (PDB ID: 6YC369) is shown with bold font.

1.8. Structures of the KR2 in the intermediate states

Recently, the results of two studies were published, describing the structures of the intermediate states of KR2 photocycle68,69. In our work, which is a subject of the present thesis, we showed the first structure of the pentameric form of KR2 in the O-state. For that, we used three complementary approaches. First, the intermediate state was accumulated and cryo-trapped in KR2 crystals upon their illumination with 532 nm laser following by flash-cooling in the cryostream. Second, the structure of the O-state was obtained at room temperature using data collection upon continuous light illumination of large single crystals of KR2. Third, the data on the O-state were collected using serial crystallography with the stream of KR2 microcrystals, injected into the X-ray beam of the synchrotron source with simultaneous illumination of the stream by 532 nm laser. All three structures were similar and revealed a transient Na+ binding site in the core of the rhodopsin protomer. They also showed the rearrangements in the extracellular side of the protein, which are likely associated with the preparation of the Na+ pathway inside KR2 for ion release.

Another study described the structures of five intermediate states of monomeric KR2, starting from femtoseconds up to several milliseconds. For that, the authors performed time- resolved serial femtosecond crystallography (TR-SFX) at X-ray Free-Electron Laser (XFEL) in Villigen, Switzerland (SwissFEL). The structures showed the evolution of retinal configuration and surrounding protein conformation. While the changes are minor in the K-state, similar to that in the case of HsBR, they are increasing in the L/M and O-states. Particularly, the hydrogen bond between the RSB and D116 is indeed absent in the M-state, when the retinal is in the 13-cis + configuration. The structures of the O1 and O2 states showed two consecutive transient Na ion- binding sites at the extracellular side of the protein. As was already mentioned, the sites are different from those in the pentameric form of KR2 in the O-state. This contradiction is believed to originate from the difference in the mechanisms of protein work in the monomeric and pentameric forms.

1.9. Mechanism of light-driven Na+ pumping

Since NaRs are the only known active light-driven non proton cation transporters, the revealing of the principles of their work is the sole way to determine the molecular mechanism of 26 light-driven Na+ pumping across the cell membrane. Na+ transport is a fundamental biological process. First of all, it is obligatory for many organisms to maintain the certain vital salt composition of the cytoplasm. It also allows the creation of the membrane Na+ gradients, necessary for the nutrients uptake, flagella motion, and ATP synthesis.

The mechanism of light-driven Na+ pumping is also interesting and important for understanding of Na+ transport in general since it is a fundamental biological process. Indeed, the principles of NaRs’ work cannot be solely understood in terms of active proton transporters, such as HsBR. H+ pumps may utilize Grotthuss and tunneling mechanisms for the translocation of the ion, which are restricted in the case of Na+ ions. Moreover, H+ is translocated through the central part of rhodopsin via its association with the RSB, e.g. RSB protonation. Upon the light photon absorptions, the retinal isomerizes and RSB deprotonates to the extracellular (in the case of outward H+ pumps) or cytoplasmic (in the case of inward H+ pumps) side of the protein. After that, new H+ comes from the opposite side to reprotonate the RSB. This allows effective H+ permeation through the core of MRs, and also provides unidirectional ion transport. However, such a mechanism cannot be used in the case of Na+ ions, since it cannot bind to the RSB. Moreover, while the RSB is protonated (positively charged), it strongly repels Na+. Therefore, the determination of the mechanisms of NaRs’ work was the most important and challenging question since their discovery in 2013.

The first version of the mechanism was proposed in the original work in 2013. However, high-resolution structures of KR2 were required to present more certain mechanism. Following the obtaining of the first structures of the protein in 2015 by two groups, two different mechanisms were then suggested. Kato and colleagues based on the flip of D116 side chain, which was found after crystals soaking at neutral pH. Importantly, Kato and colleagues only analyzed the monomeric form of KR2. The authors proposed that in the ground state, D116 is oriented towards the RSB and is directly hydrogen-bonded to it. Upon light photon absorption, the H+ is translocated from the RSB to D116 and the residue flips to the S70 and N112 residues. This was speculated to open the way for Na+ passage through the large cavity in the cytoplasmic part of KR2 and its binding near N112 and D251 in the O-state. Then, together with the retinal relaxation, the photocycle ended with Na+ release to the extracellular space.

Since we did not detect any flips of the D116 side chain, but instead noticed the existence of two principal conformations of KR2, expanded and compact, another version of the mechanism was suggested in 201544 and developed in 201664. The proposed mechanism was based on the

27 structural switch from the compact to the expanded conformation of KR2 (Figure 1.9.1). Particularly, it was suggested that in the ground state pentameric KR2 is in the compact conformation. D116 is hydrogen-bonded to the protonated RSB, but is also hydrogen-bonded to N112 and S70 residues, which is possible in the pentameric, but not shown for the monomeric form of KR2. Upon light illumination, retinal isomerizes from all-trans to 13-cis configuration. The H+ is translocated from the RSB to D116 and the pair is neutralized in the M-state. Upon M- to-O transition Na+ ion enters the protein via a large cavity at the cytoplasmic part and binds in the central region of the rhodopsin. For the binding, N112 was believed to flip outside of the protomer and face the oligomerization interface. Thus, it was speculated that the expanded conformation appears in the O-state to allow enough space for Na+ binding. Then, the expanded conformation is switched back to the compact, the Na+ ion is released from the protein and KR2 returns to its ground state.

Figure 1.9.1. The scheme of mechanism of KR2 work proposed by Gushchin and colleagues in 2016. The figure from 64.

However, both proposed mechanisms were far from reality. Our works in 2019 and 2020 on the structures of the pentameric form of KR2 under physiological conditions in the ground and O-states allowed us to suggest a more accurate and complete model of the rhodopsin work. This is the main result of the present thesis; thus, the mechanism is described in very detail in the Results and Discussion section. In brief, it was shown that KR2 is in the expanded conformation in the ground state. In that configuration, D116 is hydrogen-bonded to the RSB and S70, but not N112, which is oriented to the oligomerization interface. Upon light photon absorption and following retinal isomerization, the RSB deprotonates to D116 with the rise of the M-state. The Na+ is uptaken through the cytoplasmic cavity during M-to-O transitions. For the binding of Na+ inside the protomer, KR2 switches from the expanded to the compact conformation. In the O-state, Na+ is bound to D116, S70, and N112 residues, and the overall conformation of the protein is similar to compact. Then, upon O-to-ground transition, the reverse compact-to-expanded conformational switch guides Na+ release likely through the aqueous basin in the central part of KR2 pentamer.

Last but not least, the authors of the recent work on TR-SFX of KR2 also proposed the mechanism of KR2 work. Importantly, the work only relates to the monomeric form of the protein. In brief, the proposed by Skopintsev and colleagues mechanism is similar to that proposed by Kato and colleagues in 2015. The main exception is that Skopintsev and colleagues showed that D116 is always oriented towards S70 and N112 in the course of KR2 photocycle, but is directly hydrogen-bonded to the RSB in the ground, K, L, and O-states68. Na+ ion is bound inside the protein in the two consecutive O-states, O1 and O2. The binding site in the O1 state (site I) is formed by N112 and D251 residues, as predicted earlier. Another binding site in the O2 state (site II) is located near E11, E160, and R243 triad, as also speculated previously. During O1-to-O2 transition Na+ ion is translocated from the site I to site II and is released to the extracellular space from site

II upon O2-to-ground transition.

1.10. Functional conversions of NaRs

The ability of NaRs to pump cation other than H+ created a great interest for the engineering of new NaR-based light-driven pumps. First attempts were made in 2015 when K+-pumping variants of KR2 were designed44,48. These were mutants of N61 and G263 residues of KR2, located at the cytoplasmic side of the protein and forming a large putative ion-uptake cavity. The most promising variants were G263F, N61P, G263W, and N61P/G263W mutants. However, all these

29 mutants showed significant residual Na+-pumping activity. The N61P/G263W double mutant was the only to demonstrate stronger K+ than Na+ pumping activity.

In the end of 2015, the N61L/G263F mutant of KR2 was demonstrated to pump Cs+ ions70. This was the first demonstration of MR to actively transport such large cations. This also was an important finding for the structural studies of KR2, since Cs+ is much easier to detect directly in the electron density maps. This was an advantage for the future time-resolved crystallography of NaRs, where the occupancy of the intermediate state is usually low and there is a problem of distinguishing between Na+ ions and water molecules.

In 2016 the successful functional conversions of KR2 into H+ and Cl- pumps were demonstrated71,72. For the Na+-to-H+ pump conversion, the NDQ motif of KR2 was mutated to DTE, the motif of H+-pumping PRs. To further increase the efficiency of engineered KR2-based H+ pump the authors introduced additional D102N mutation to destroy the inter-protomer Na+- binding site. For the Na+-to-Cl- pump conversion, the NDQ motif was replaced by NTQ motif of bacterial light-driven Cl- pumps. Surprisingly, the sole D116T mutation (NDQ ￫NTQ) was not enough for the creation of Cl- transport ability in the case of KR2. The authors supplemented the motif replacement with D102N and F72G mutation to mimic Cl- pumps. Finally, D116T/D102N/F72G mutant of KR2 showed strong Cl- pumping activity. On the other hand, all the effort of the research group to convert H+ or Cl- pumps into Na+ pumps failed, which suggests that NaRs were accurately optimized in course of evolution and one has to consider more parameters for the rational engineering of non-proton cation pumps.

An important effort was made by Vogt and colleagues to engineer light-gated Na+ and K+ channels based on KR273. The authors showed that mutations of the RSB counter-ion complex of KR2 lead to the appearance of passive K+ conductance by the protein. Particularly, D251E and L75K mutants were shown to be leaky pumps with ability to weak non-selective passive ion transport. On the other hand, the R109Q mutant of KR2 turned out to be a more selective K+ channel with the residual Na+-pumping activity. The introduction of additional S70A mutation to KR2-R109Q led to almost complete loss of residual Na+-pumping activity. In fact, the KR2- R109Q-S70A mutant showed strong photocurrents indicating the passive conductance of both Na+ and K+. Unfortunately, the effective channeling was observed exclusively at alkaline extracellular pH values (pH 10), which limits the application of the engineered variants to use in optogenetics.

1.11. NaRs as potential optogenetic tools

As it was already mentioned, MRs are the core of the optogenetics. In terms of optogenetics, NaRs are unique potential inhibitory instruments. Indeed, their high selectivity to Na+, but not to K+ and Ca2+, allows the most native manipulations of living cells without the creation of undesirable side effects. Indeed, NaRs do not affect the pH of the cells and Cl- compositions and thus are an invaluable alternative for most commonly used for neural silencing H+ and Cl- pumps.

The first demonstration of NaRs’ optogenetic application was made in 201548. Kato and colleagues showed expression of KR2 in mammalian cells, however, the targeting of the protein to the plasma membrane was not optimal. Nevertheless, expression of KR2 in rat cortical neurons and also Caenorhabditis elegans neurons and voltage-clamp recordings from the cells showed effective hyperpolarization of the membrane and consequent inhibition of neural spikes.

A step to the improvement of the KR2 targeting was made in 2016 when a chimeric NaR composed of helix A of IaNaR and B-G helices of KR2 was shown to slightly improve the protein delivery to plasma membrane74. However, the targeting remained weak compared to other inhibitory optogenetic tools.

In 2017, 12 members of the NaRs family were electrophysiologically investigated in mammalian cells (ND7/23)30. The study showed that 4 out of 12 proteins were able to generate photocurrents strong enough for neuronal spiking inhibition. These 4 proteins were further tested in dissociated hippocampal neuron culture and proved to be potential silencers. Unfortunately, their efficiency remained more than 3 times lower than that of other inhibitory optogenetic tools, an enhanced H+ pump from archaeon Halorubrum sodomense (AR3).

In 2019, an enhanced variant of KR2 (eKR2) was engineered, which showed much improved targeting to the plasma membrane of mammalian cells75. It also demonstrated the generation of sufficient photocurrents for effective inhibition of action potential. Moreover, it was shown that the photocurrent is carried by Na+ ions, but not H+ or other ions. Using eKR2, authors of 75 showed that pumping of H+ is negligible even at low pH values.

Another important property of MRs in terms of optogenetics is the maximum absorption wavelength, as it determines the depth of light penetration inside the tissue. This is a key determinant of the effectivity of an optogenetic instrument. The more the maximum absorption wavelength is the deeper light can pass inside the organs; however, in NaRs this wavelength is 31 relatively short (520-530 nm). Hence, the engineering of NaRs variants with a red-shifted peak of the spectrum is necessary for their use in neuroscience and medicine.

The red-shifted mutants of KR2 were designed and reported in 201976. The authors showed that double S254A/P219T mutation leads to the 40-nm redshift of the maximum absorption wavelength of KR2 without impairing its functional properties. However, also in 2019, it was shown that the S254A mutant (also 20-nm red-shifted compared to the wild type protein) of KR2 has additional weak K+-pumping activity. Unfortunately, the authors of 76 have not studied the K+- pumping activity of S254A/P219T as they used the CsCl solution for the validation of H+-pumping in KR2 mutants, but not KCl as in previous works. Although the K+-pumping activity is minor in S254A mutant, it could have significant effects when using in neurons.

One particular light-driven Na+ pump, SrNaR, was demonstrated to have an action spectrum with the maximum absorption wavelength at around 550 nm30. However, the kinetics of this rhodopsin is slow (1 s) and the formation of the O-state is negligible. Therefore, further investigations of red-shifted KR2 variants are required.

2. Materials and Methods

2.1. Materials

2.1.1. Organisms

For the cloning Escherichia coli cells strain TOP10 (F- mcrA Δ(mrr-hsdRMS-mcrBC) φ80lacZΔM15 ΔlacX74 nupG recA1 araD139 Δ(ara-leu)7697 galE15 galK16 rpsL(StrR) endA1 λ-) were used (Invitrogen, USA).

For the expression of target proteins, E. coli cells strain C41 (DE3) (F - ompT gal dcm hsdSB (rB- mB-)) were used (Lucigen, USA).

2.1.2. Vectors

For the expression of the proteins and cloning pET15b vector was used (Novagen, USA).

2.1.3. Genes

The encoding gene sequence for the KR2 gene (UniProt ID: N0DKS8) was synthesized de novo by MWG, Ebersberg, Germany. The gene was codon-optimized for the expression in E. coli system.

2.1.4. Oligonucleotides

Synthetic oligonucleotides were purchased at MWG, Ebersberg, Germany.

2.1.5. Chemicals for molecular biology

Non-organic salts and acids were bought from Sigma-Aldrich, Applichem, and Merck. Detergents and lipids were bought from Sigma-Aldrich, Affymetrix, and Avanti Polar Lipids.

2.1.6. Crystallization

Sodium malonate for KR2 crystallization was bought from Hampton Research, USA. Host lipid for in meso crystallization, monoolein (MO), was bought from Nu-Chek Prep, USA.

96-well plates for protein crystallization were bought from Marienfeld, Germany. The 100 μl gastight syringes for mesophase mixing were bought from Hamilton, USA. 96-well plates for the precipitant kits were bought from Sigma-Aldrich, USA.

The preparation of the optimization precipitant kits was performed using Formulator robotic system (Formulatrix, USA). The dispensing of the nanovolumes of mesophase and precipitant solution was performed using NT-8 robotic system (Formulatrix, USA).

2.1.7. Crystal harvesting tools

Microloops for crystal harvesting, carbide scribes for well opening, and magnetic wand for microloop fixation were bought from Mitegen, USA.

2.2. Methods

2.2.1. Cloning

The chemically competent E. coli TOP10 and C41 cells were prepared using high- efficiency transformation protocol77 with the following modifications. Briefly, several colonies from a freshly stroke plate were placed into the flask containing 40 ml SOB medium (2% tryptone,

0.5 yeast extract, 10mM NaCl, 10mM KCl, 20mM MgCl2). The flask was intensively shaken (200- 300 rpm in Infors Multitron incubator) at 18 °C for 24-36 hours until the culture was grown to

OD600 of 0.5-0.6. Washing step in filter-sterilized TB buffer (10mM HEPES, 55mM MnCl2,

15mM CaCl2, 250mM KCl, pH 6.7) was repeated twice, incubating cells in the buffer on ice for 15 minutes each time. Then, the cells were resuspended in 2 ml of TB buffer, subsequently, 140 µl of DMSO was added in two equal steps with 10 minutes incubation between them. Finally, the cells were 100 µl aliquoted in 1.5 ml centrifuge tubes and shock frozen in liquid nitrogen. The frozen cells were kept at −80 °C for further use.

For transformation a plasmid solution (2 µl) or ligation mixture (6 µl) was added to the 100 µl of frozen cells and kept on ice for 1 hour. Then the heat pulse was applied 42∘С for 1 minute without shaking. After that the cells were kept on ice for additional 10 minutes. Then 700 µl of SOC medium (SOB + 20 mM glucose) was added and the tube was incubated for 45 minutes at 37∘С with moderate shaking. After that the tube was spun at 3000×g for 2 minutes, 750 µl of supernatant was removed and the pellet resuspended in the residual medium. Finally, the suspension was spread on the agar plate with the selective antibiotic (1% tryptone, 0.5% yeast extract, 1% NaCl, 1% glucose, 1.5% agar) and incubated overnight at 37∘С.

For commercially available competent cells the manufacturers’ protocol was followed.

For plasmid DNA isolation the separate colony was grown overnight in 10 ml of LB medium (1% tryptone, 0.5% yeast extract, 1% NaCl, pH 7.0) in a small flask at 37°C with shaking (180 rpm). The bacteria were collected by centrifugation at 5000×g for 10 minutes at 4°C and resuspended in 250 µl Buffer A1 from commercial NucleoSpin® Plasmid purification kit (Macherey-Nagel, Düren, Germany). Next, the bacteria were processed by SDS/alkaline lysis (addition of 250 µl Buffer A2). Highsalt Buffer A3 (300 µl) was added to neutralize the lysate and to create appropriate conditions for DNA binding to the silica membrane of the spin column. After centrifugation at 10000×g for 10 minutes at 4°C the clear supernatant was loaded onto a NucleoSpin® Plasmid spin column. Contaminations like salts and macromolecular cellular 35 components were removed by simple washing with 600 µl of AW wash buffer. Final washing with 600 µl ml with ethanol containing Buffer A4 was done. The silica membrane was dried by centrifugation at 10000×g for 1 minute. The plasmid DNA was eluted in 40 μl of slightly alkaline Buffer AE (5 mM Tris-HCl, pH 8.5) and stored at −20°C.

DNA fragments were amplified in PCR, mainly following the manufacturers guidelines. 50 μl of the reaction mixture contained 10 ng of DNA matrix, 0.5 μmol of each primer, 10 μl of Phusion HF buffer, 200 μmol of each dNTPs, 3% DMSO, 1 U of Phusion Hot Start II DNA Polymerase (Thermo, USA).

The PCR implementation protocol was different for two cloning strategies: the extension time varied from 45 seconds for short amplified DNA fragments up to 3 minutes for circular PCR, when complete plasmid was amplified.

PCR products and plasmids were digested using desired digestion enzymes by incubation for 1 h at 37°C. Typically, FastDigest®, Thermo digestion enzymes were used in a reaction mix of 300-1000 ng of DNA and corresponding amount of FastDigest® Green buffer. Digestion was terminated by incubation for 5-10 minutes at 60−80∘С. Plasmids, further used for ligation, were supplemented with 1 U of thermosensitive alkaline phosphatase (Thermo, USA) for 15 minutes at 37°C after digestion process termination.

For blunt ends ligation of circular PCR products, the ends phosphorylation was performed. The reaction mixture contained 500-1000 ng of DNA, 1mM ATP, the appropriate amount of FastDigest® Green buffer and 10 U of T4 polynucleotide kinase (Thermo, USA).

The PCR products were analyzed by horizontal agarose gel-electrophoresis. DNA probes, mixed with DNA loading buffer (Thermo, USA) in 1:5 dilution, were loaded onto 0.7-1.5% agarose gels running in TAE buffer (40mM Tris-base, 1mM EDTA, pH 8.0 with glacial acetic acid). Staining of DNA bands was done by adding GelRed (Biotium) in 1:20000 dilution to TAE buffer. The DNA fragments were separated by applying 9 V/cm voltage to the gel. The bands of interest were cut out and purified with NucleoSpin® Gel and PCR Cleanup kit (Macherey-Nagel, Düren, Germany). Briefly, the agarose gel slice was melted in a high-salt Buffer NT and applied to a NucleoSpin® Gel and PCR Cleanup silica column followed by centrifugation and a subsequent washing with ethanol-containing Buffer NT3. The DNA was eluted in slightly alkaline Buffer NE (5 mM Tris-HCl, pH 8.5).

T4 DNA Ligase (Thermo, USA) was used for the ligation of DNA fragments. 100 ng of the vector DNA were mixed with an excess amount of insert DNA (3-5 times molar excess), the appropriate amount of ATP and DTT containing buffer and 1 U of T4 DNA Ligase. The mixture was incubated overnight at 20∘С and then used for transformation or stored at −20°C. For blunt end ligation additional 5% of PEG4000 was added to the reaction mixture and incubation temperature was lowered to 16°C.

2.2.2. Protein Expression and Purification E. coli cells of strain C41(DE3) (Lucigen) were transformed with the KR2 expression plasmid. Transformed cells were grown at 37 °C in shaking baffled flasks in an autoinducing

78 medium, ZYP-5052 containing 100 mg/L ampicillin, and were induced at optical density OD600 of 0.7–0.9 with 1 mM isopropyl β-d-1-thiogalactopyranoside (IPTG) and 10 μM all-trans-retinal. 3 h after induction, the cells were collected by centrifugation at 4,000g for 20 min. Collected cells were disrupted in M-110P Lab Homogenizer (Microfluidics, USA) at 25000 psi in a buffer containing 20 mM Tris-HCl pH 8.0, 5% glycerol, 0.5% Triton X-100 (Sigma-Aldrich, USA) and 50 mg/L DNase I (Sigma-Aldrich, USA). The membrane fraction of cell lysate was isolated by ultracentrifugation at 90000 g for 1 h at 4º C. The pellet was resuspended in a buffer containing

50 mM NaH2PO4/Na2HPO4 pH 8.0, 0.1 M NaCl and 1% DDM (Anatrace, Affymetrix, USA) and stirred overnight for solubilization. The insoluble fraction was removed by ultracentrifugation at 90000 g for 1 h at 4º C. The supernatant was loaded on Ni-NTA column (Qiagen, Germany) and

KR2 was eluted in a buffer containing 50 mM NaH2PO4/Na2HPO4 pH 7.5, 0.1 M NaCl, 0.5 M imidazole and 0.1% DDM. The eluate was subjected to size-exclusion chromatography on 24 ml Superdex 200i column (GE Healthcare Life Sciences, USA) in a buffer containing 50 mM

NaH2PO4/Na2HPO4 pH 7.5, 0.1 M NaCl, 0.05% DDM. Protein-containing fractions with the minimal A280/A525 absorbance ratio were pooled and concentrated to 60 mg/ml for crystallization.

2.2.3. Measurements of pumping activity in E. coli cells

E. coli cells of strain C41(DE3) (Lucigen) were transformed with the KR2 expression plasmid. Transformed cells were grown at 37 °C in shaking baffled flasks in an autoinducing medium, ZYP-5052 containing 100 mg/L ampicillin, and were induced at optical density OD600 of 0.7–0.9 with 1 mM isopropyl β-d-1-thiogalactopyranoside (IPTG) and 10 μM all-trans-retinal. 3 h after induction, the cells were collected by centrifugation at 4,000g for 10 min and were washed three times with unbuffered salt solution (100 mM NaCl, and 10 mM MgCl2) with 30-min intervals between the washes to allow an exchange of the ions inside the cells with the bulk. After that, the

37 cells were resuspended in 100 mM NaCl solution and adjusted to an OD600 of 8.0. The measurements were performed on 3 ml of stirred cell suspension kept at 1 °C. The cells were illuminated for 5 min with a halogen lamp (Intralux 5000-1, VOLPI) and the light-induced pH changes were monitored with a pH meter (LAB 850, Schott Instruments). Measurements were performed with the addition of 30 μM of protonophore carbonyl cyanide 3-chlorophenylhydrazone (CCCP).

2.2.4. Liposome preparation

Phospholipids (asolectin from soybean, Sigma-Aldrich) were dissolved in CHCl3

(chloroform ultrapure, PanReac AppliChem) and dried under a stream of N2 in a glass vial. The residual solvent was removed using a vacuum pump overnight. The dried lipids were resuspended at a final concentration of 1% (w/v) in 0.15 M NaCl supplemented with 2% (w/v) sodium cholate. The mixture was clarified by sonication at 4°C, and KR2 was added at a protein/lipid ratio of 7:100 (w/w). The detergent was removed by overnight stirring with detergent-absorbing beads (Amberlite XAD-2, Supelco). The mixture was dialyzed against 0.15 M NaCl (adjusted to the desired pH) at 4°C for 1 day (four 200-ml changes) to obtain a certain pH.

2.2.5. Measurements of the pumping activity in liposomes

The measurements were performed on 2 ml of stirred proteoliposome suspension at 0°C. Proteoliposomes were illuminated for 18 min using a halogen lamp (Intralux 5000-1, Volpi) and then were kept in the dark for another 18 min. Changes in pH were monitored using a pH meter (Lab 850, SCHOTT Instruments). Measurements were repeated for different starting pH and in the presence of 30 μM CCCP under the same conditions.

2.2.6. Oligomeric state analysis by size exclusion chromatography

For the studies of the oligomeric form of KR2 and its mutant forms H30K, H30L, Y154F, and D116N in detergent micelles the concentrated to 60 mg/ml proteins were diluted to a final concentration of 1 mg/ml in the buffer solutions, containing 100 mM Na-K/Pi of pH 4.3, 6.0, or 8.0, 200 mM NaCl and 0.1% DDM. The samples were then equilibrated with the solution using dialysis against the same buffer supplemented with 0.1% DDM. The buffer:sample volume ratio was 100:1. The buffer was changed at least 4 times during dialysis. For the dialysis, the membrane with 14 kDa pores were used. The dialysis lasted for 5 days for complete equilibration of the samples. After the dialysis, the samples were centrifuged at 10000×g for 10 min at 4 °C for the sedimentation of the denatured protein. The pellet was removed prior to size exclusion

38 chromatography (SEC). SEC was performed using 24 ml Superdex 200i columns. The study of the wild type KR2 at different pH values was performed using another similar column from that used in other studies, which explains the difference in the elution volume of the wild type protein at pH 8.0.

2.2.7. Crystallization details and crystals preparation

The crystals were grown using the in meso approach79,80, similarly to our previous works81. The solubilized protein in the crystallization buffer was added to the monooleoyl-formed lipidic phase (Nu-Chek Prep, USA). The best crystals were obtained using the protein concentration of 25 mg/ml. The crystals of the monomeric and pentameric forms of the WT, G263F, S254A, D116N, H30A proteins were grown using the precipitate 1.0 M sodium malonate pH 4.6 and 1.2 M sodium malonate pH 8.0, respectively (Hampton Research, USA). The crystals of the monomeric form of H30K and Y154F mutants were grown using 1.2 M sodium malonate pH 8.0 as a precipitant solution. Crystallization probes were set up using the NT8 robotic system (Formulatrix, USA). The crystals were grown at 22 °C and appeared in 2-4 weeks. Before harvesting, the crystallization drop was opened and covered with 3.4 M sodium malonate solution with the same pH value to that of the precipitant solution to avoid dehydration. All crystals were harvested using micromounts (MiTeGen, USA) and were flash-cooled and stored in liquid nitrogen for further crystallographic analysis.

2.2.8. Time-resolved visible absorption spectroscopy on KR2 crystals

The laser flash photolysis was performed similar to that described by Chizhov et al with minor differences. The excitation system consisted of Nd:YAG laser Q-smart 450 mJ with OPO Rainbow 420-680 nm range (Quantel, France). For the experiments the wavelength of the laser was set 525 nm. Microcrystals of KR2 in the lipidic cubic phase were plastered on the 4x7 mm cover glass. The thickness of the slurries was adjusted in order to give a sufficient signal. The glass with crystal slurries was placed into 5x5 mm quartz cuvette (Starna Scientific, China) filled with the buffer solution containing 3.4 M sodium malonate pH 8.0 and thermostabilized via sample holder qpod2e (Quantum Northwest, USA) and Huber Ministat 125 (Huber Kältemaschinenbau AG, Germany). The detection system beam emitted by 150W Xenon lamp (Hamamatsu, Japan) housed in LSH102 universal housing (LOT Quantum Design, Germany) passed through pair of Czerny–Turner monochromators MSH150 (LOT Quantum Design). The received monochromatic light was detected with PMT R12829 (Hamamatsu). The data recording subsystem represented by a pair of

DSOX4022A oscilloscopes (Keysight, USA). The signal offset signal was measured by one of the oscilloscopes and the PMT voltage adjusted by Agilent U2351A DAQ (Keysight).

2.2.9. Spectroscopic characterization and accumulation of the intermediate state in KR2 crystals

Absorption spectra of KR2 in solution were collected using the UV-2401PC spectrometer (Shimadzu, Japan). The spectroscopic characterization of O-state build-up in KR2 crystals was performed at the icOS Lab located at the ESRF82. The same set up was established at the P14 beamline of the PETRAIII synchrotron source (Hamburg, Germany) for accumulation of the O- state in crystals for X-ray diffraction data collection. Also the same accumulation procedure was applied to crystals at icOS and P14 beamline. Briefly, UV-visible absorption spectra were measured using as a reference light that of a DH-200-BAL deuterium-halogen lamp (Ocean Optics, Dunedin, FL) connected to the incoming objective via a 200 µm diameter fiber, resulting in a 50 um focal spot on the sample, and a QE65 Pro spectrometer (Ocean Optics, Dunedin, FL) connected to the outgoing objective via a 400 µm diameter fiber. The actinic light comes from a 532 nm laser (CNI Laser, Changchun, P.R. China) coupled to a 1000 µm diameter fiber which is connected to the third objective whose optical axis is perpendicular to those of the ingoing and outgoing objectives. Ground states spectra (100 ms acquisition time averaged 20 times) were collected on crystals flash-cooled in liquid nitrogen and kept under a cold nitrogen stream at 100 K. In order to maximize the population of the O-state, a crystal was put under constant laser illumination at 100 K, the nitrogen stream was then blocked for 2 seconds, then the laser was switched off once the crystal is back at 100 K. For the accumulation of the O-state laser power density of 7.5 mW/cm2 at the position of the sample was used. The mean size of the crystals was 200x100x30 μm3 (Supplementary Fig. 22). The plate-like crystals were oriented so that the largest plane (200x100 μm2) was as perpendicular to the laser beam. The laser beam was focused to the size of 500x500 μm2 (1/e2). A UV-visible absorption spectrum was then recorded to show the red-shifted absorption maximum characteristic of the O-state of KR2. The crystals with the accumulated intermediate state were then stored in liquid nitrogen and transported to the PETRAIII, Hamburg, Germany for the X-ray experiments and showed the same structure as that obtained using crystals with the O- state, accumulated directly at the P14 beamline of PETRAIII.

2.2.10. Acquisition and treatment of diffraction data

X-ray diffraction data of the wild type KR2 (PDB IDs: 6REW, 6REX, 6REZ, 6RF5, 6RF7, 6RF6, 6RF1, 6RF0) and S254A (PDB IDs: 6RF4, 6RFB), G263F (PDB IDs: 6RFC, 6RF3), H30K

(PDB ID: 6RFA), Y154F (PDB ID: 6RF9), D116N (PDB IDs: 6YBY, 6YBZ), and H30A (PDB ID: 6YC1) mutants were collected at the beamlines ID23-1, ID29, ID30A-3, and ID30b of European Synchrotron Radiation Facility (ESRF), Grenoble, France, using a PILATUS 6M and EIGER X 16M detectors, respectively. X-ray diffraction data of the KR2 in the ground (PDB IDs: 6YC3, 6YC2) and O-states (PDB IDs: 6XYT, 6YC4) at 100 and 293 K (room temperature, RT) were collected at the P14 beamline of the PETRAIII, Hamburg, Germany, using EIGER 16M detector. For the collection of the X-ray diffraction data at RT the crystals in the cryoloop were placed on the goniometer of P14 beamline and maintained in the stream of the humid air (85% humidity). The stream of humid air was provided by the HC humidity controller (ARINAX, France). For activation of the proteins in crystals and obtaining the structure of the O-state at RT the laser flash was synchronized with the X-ray detector. The laser was illuminating the crystal only during X-ray data collection to avoid drying and bleaching. The crystals were rotated during the data collection and laser illumination. Diffraction images were processed using XDS83. The reflection intensities of the monomeric forms of KR2 and its mutant were scaled using the AIMLESS software from the CCP4 program suite84. The reflection intensities of all the pentameric forms were scaled using the Staraniso server85. There is no possibility of twinning for the crystals. For the both structures of KR2 at RT diffraction data from three crystals was used (Appendix Table 5.3). For the ground state structure of the wild type KR2 under physiological conditions the data from 20 single crystals were used. In all other cases, diffraction data from one crystal was used. The data statistics are presented in the Appendix in Tables 5.1-5.4.

2.2.11. Serial millisecond crystallography data collection and processing

Serial millisecond crystallography data of the O-state of KR2 (PDB ID: 6YC0) was obtained at RT at BL13-XALOC beamline of ALBA (Barcelona, Catalunya) using a PILATUS 6M detector working at 12.5 Hz and a 40 x 60 µm sized beam at 12.6KeV (1.4.1012 ph/s). For that purpose, a LCP stream of protein microcrystals was injected into the focus region using a LCP injector86 placed at 45º of the diffractometer table- with the help of an ÄKTA pump flowing at 1 μl/min and a constant Helium supply (10-14 psi) yielding an extrusion speed of 30 nl/min for 100 μm capillary. A Roithner laser source (RLTMLL-532-100-5) working at 20 mW was used for protein activation.

A total number of 1,208,640 detector images were collected and processed with CrystFEL87 (version 0.8.0) without any additional modification. Among all images collected, 350,862 were identified as potential crystal hits with more than 30 Bragg peaks with SNR=3.5, --

41 threshold=12, --highres=2.5 using peakfinder8 algorithm as implemented in CrystFEL, corresponding to an average hit rate of 29%. The overall time of data collection from a sample with a total volume of 120 μl was about 36 hours.

Data were processed using CrystFEL. For peak finding, we used peakfinder8 and min-snr=3.5, threshold=12. For indexing, indexers dirax88, xds83, taketwo89 (in that order) were used, with -- multi option enabled. Integration was performed using --int-radius=3,4,5. Data were merged using process_hkl with --min-res=3.3, --push-res=3.0 and--symmetry=mmm. This yielded a dataset with 131,872 indexed images with 136,656 crystals, corresponding to 39% average indexing rate. Among these images, 38761 were merged together after process_hkl rejection. Among indexers, dirax was the most successful one, providing 77,681 indexed crystals. Initial geometry, provided by the beamline staff, was optimized with detector-shift and geoptimiser, as described in 90.

2.2.12. Structure determination and refinement

Initial phases for the pentameric structure of KR2 under physiological conditional (PDB ID:

6REW) were successfully obtained in the C2221 space group by molecular replacement (MR) using MOLREP91 using the chain A of the 4XTN structure as a search model. Initial phases for the other pentameric structures of KR2 and its mutants were successfully obtained in the C2221 space group by MR using MOLREP using the 6REW structure as a search model. Initial phases for monomeric forms of KR2 and its mutants were successfully obtained in the I222 space group by MR using the 4XTL structure as a search model. The initial MR models were iteratively refined using REFMAC592, PHENIX93 and Coot94.

2.2.13. Molecular dynamics simulations

The simulation system consisted of a KR2 pentamer in the O-state with sodium ions in SBC2 and cocrystallized water molecules. The proteins was embedded in a POPC bilayer (256 lipids) and then solvated with TIP3P water with a Na+/Cl− concentration of 150 mM using the CHARMM- GUI web-service95. The simulation box contained 94817 atoms in total. All ionizable amino acids were modeled in their standard ionization state at pH 8, including D116 and D251 which were modeled charged.

The CHARMM-GUI recommended protocols were followed for the initial energy minimization and equilibration of the system. The atoms of protein and lipids in the system were subjected to a harmonic positional restraint and 5000 steps of steepest descent minimization followed by two 25 ps equilibration steps in the NVT ensemble using Berendsen thermostat and one 25 ps and three 42

50 ps equilibration steps in the NPT ensemble using Berendsen thermostat and barostat. During all equilibration steps, the force constants of the harmonic positional restraints were gradually reduced to zero. The system was further equilibrated for 10 ns in the NPT ensemble with Nose- Hoover thermostat and Parrinello−Rahman barostat, which were also used for the further production simulations. The temperature and pressure were set to 303.3 K and 1 bar with temperature and pressure coupling time constants τt = 1.0 ps−1 and τp = 0.5 ps−1, respectively. All MD simulations were performed with GROMACS version 2018.196. The time step of 2 fs was used for all the simulations except for the early steps of equilibration. The CHARMM36 force field97 was used for the protein, lipids, and ions. Parameters for retinal bound to lysine were adapted from98.

In order to investigate the putative sodium translocation pathways, metadynamics approach (metaMD) was employed99. This method is based on biasing of the potential surface via addition of repulsive functions (“hills”, typically Gaussians) which force the investigated molecular system to explore its configurational space broader and faster than in a regular unbiased MD simulation. Since the correct sampling of protein degrees of freedom relevant to the ion release at the late stages of the photocycle appears problematic, the performed metadynamics simulations allowed us to reveal possible sodium unbinding pathways rather than quantitatively assess the free energies changes associated with them. We used the PLUMED plugin for GROMACS to perform metaMD simulations100. The projections of the vector connecting the sodium ion and its original position onto x, y and z directions were used as the collective variables (i.e., 3 CVs were used). In order to prevent sodium passage back to the cytoplasmic side, we applied a flat-bottom potential (k=1000 kJ/mol/nm2) in the normal to the membrane direction, which discouraged ion moving towards the cytoplasmic side. Also, harmonic restraints were applied to all protein C� atoms above the C� atom of K255 (in the direction of CP) to prevent the overall motion of the protein complex. The deposition rate for hills was 0.5 ps; the width and height of deposited hills were equal to 0.05 nm and 1 kJ/mol, respectively. The simulations were continued until the exit of ion from the protein interior was observed (typically, during 5-20 ns). We have carried out 10 metaMD runs in total, 2 replicates for each of the 5 protomers of the KR2 pentamer.

3. Results and Discussion

3.1. High-resolution structure of KR2 in the ground state

3.1.1. Crystallization of KR2 under physiological conditions

To obtain the structure of KR2 in its biologically relevant conformation, we crystallized the protein using in meso approach using Na malonate as a precipitant solution. Particularly, 1.7 M of Na malonate pH 8.0 was used in the case of the best-diffracting crystals. The crystals appeared within two to four months and were red, plate-like with the average size of 200x100x30 μm3. For the validation of the KR2 state in the crystals, the visual absorption spectrum was measured both at 100 K and at room temperature (RT), 293K. In all cases, before the opening of the crystallization well, it was covered with 3.4 M sodium malonate pH 8.0 to avoid dehydration of the crystals. In the case of studies at 100K, the crystals were harvested into microloops and flash-freezed in the liquid nitrogen for the studies. In the case of studies at RT, the crystals were harvested into microloops and immediately placed onto goniometer in the stream of humid air with the humidity of 80% to avoid dehydration. The spectrum of the KR2 in crystals has a maximum absorption wavelength of 525 nm, which corresponds to that in the solution and in lipids (Figure 1.4.1). As a further crystallographic analysis of the crystals revealed KR2 pentamer within an asymmetric unit, and the crystals were grown at pH 8.0 and high Na+ concentration, we considered that the crystals contain KR2 is its biologically relevant conformation. We also called it Na+- pumping state of KR2.

3.1.2. Crystal structure of the pentameric KR2 in the ground state

For the structural investigations, cryo-freezed crystals were exposed to the X-ray beam of the synchrotron sources (European Synchrotron Radiation Facility, ESRF, Grenoble, France, and PETRAIII, European Molecular Biology Laboratory, Hamburg, Germany) and the diffraction data were collected. Importantly, the crystals of pentameric form of KR2 grown at pH 8.0 diffracted anisotropically. Therefore, to improve the quality of the data and resulting resolution of the final structure KR2, we have collected several datasets and merged some of them. For that, the correlation between datasets was calculated based on the CC1/2 parameter. The procedure allowed us to select 20 datasets for merging. Next, the crystallographic data were scaled and merged using Staraniso server85 to account for the anisotropy of the data. In the case of KR2, processing of the data using Staraniso helped us to improve the electron density maps and Rwork/Rfree factors significantly. The highest resolution cutoff of the merged dataset was determined to be 2.2 Å by

44 the CC1/2 parameter. The phasing problem was solved using the molecular replacement method in Molrep of the CCP4i suite. For the search, chain A of the 4XTL model of KR2 was used. The initial model was then iteratively refined using Refmac5, Phenix, and Coot. The data collection, treatment, and structure refinement statistics are shown in Table 5.1.

Thus, we have solved the structure of the biologically relevant pentameric form of KR2 at 2.2 Å. In opposite to the previous results, all the five protomers in the asymmetric unit have the same conformation, similar to that of protomers A and D of the 4XTN model44 (Figure 3.1.2.1)

Overall architecture of the KR2 protomer is similar to that observed earlier. It consists of seven TM α-helices and a small N-terminal α-helix, capping the inside of the protein. Retinal cofactor is in the all-trans configuration and covalently attached to K255 of the helix G via the RSB. Each protomer is in the expanded conformation, with five major inner cavities (Figure 3.1.2.1).

Figure 3.1.2.1. Structure of KR2 protomer under physiological conditions. a. Overall side view of the protomer. b. Detail view of the cytoplasmic ion-uptake cavity (IUC). c. Detail view of the central region and Schiff base cavities 1 and 2 (SBC1 and SBC2, respectively). d. Detail view of the extracellular part. Hydrophobic/hydrophilic membrane core boundaries are shown with black horizontal lines. Retinal cofactor is colored teal. Cavities were calculated using HOLLOW and are shown with pink surfaces.

The first cavity is located at the cytoplasmic part (Figure 3.1.2.1). It protrudes from the protein surface at the hydrophilic part of the lipid bilayer towards Q123. It therefore creates a pore

45 from the cell cytoplasm directly to the Q123 residue, filled with numerous water molecules. Q123 is further separated from the RSB by the hydrophobic residues V67 and L120 (Figure 3.1.2.1). Mutagenesis showed that the cavity might serve as an ion-uptake vestibule, as it is one of the determinants of the ion selectivity of KR2. Particularly, the mutations of G263 and N61 residues, forming the cavity, lead to the appearance of K+-pumping ability of the rhodopsin. Importantly, in the pentameric form of KR2 the pore entrance directly face the cytoplasm of the cell, while in the monomeric form of the protein it is buried in the lipid bilayer due to the overall orientation of the protomer (Figure 3.1.2.2). Since Q123 is located at the level of hydrophobic/hydrophilic border of the protein surface, the location of the pore entrance is crucial for effective uptake of Na+ ions. Thus, this is an evidence that pentameric assembly of KR2 is important for Na+-pumping. Hereafter, we name the cavity in the cytoplasmic part of KR2 as an ion-uptake cavity (IUC).

Figure 3.1.2.2. The entrance pore to the ion-uptake vestibule of KR2. a. The pore is buried in the lipid bilayer in the monomeric form. The uptake of Na+ is hampered. b. The pore is exposed to the cytoplasm and is more accessible for Na+. c. Detailed view of the entrance pore. Hydrophobic/hydrophilic membrane core boundaries were calculated using HOLLOW101 and are shown with blue disks. The inside of the pore is colored red.

Second and third, there are two polar cavities in the RSB region. The first is located between the RSB and R109, surrounded by W113, S70, D116, D251, N112, R109, and L74. We named it the Schiff base cavity 1 (SBC1). It is filled with four water molecules (Figure 3.1.2.1). Three of them (Wat419, Wat428, Wat429) are well-ordered and were observed previously, unlike the fourth molecule Wat519. In the new structure, the presence of w519 is indicated by strong 4.0 σ signal at the difference electron densities (Figure 3.1.2.3). However, the molecule is poorly- ordered, and its position slightly varies between the protomers. The second cavity (Schiff base

46 cavity 2, SBC2) is located close to SBC1, between R109, D251, and T248 residues and contains one water molecule (Figure 3.1.2.1).

The RSB and the D116 side chain are within 2.8 Å from each other and are hydrogen- bonded. The S254 side chain is connected to the hydrogen bonding network through Y218 and D251 side chains and therefore, may play an important role in protein function (Figure 3.1.2.1).

Figure 3.1.2.3. Electron density maps of the water molecules in the SBC1. a. 2Fo-Fc electron density map around water molecule 519. The map is contoured at the level of 1.5 σ. Hydrogen bonds are shown with dashed black lines. b. Fo-Fc difference electron density map of KR2 model at pH 8.0 without Wat519 refined against experimental data. The map is contoured at the level of 4.0 σ. The density has spherical shape and is located at reasonable distances from neighbor water molecules and residues.

Fourth, a small cavity is placed in the extracellular part of the protein (Figure 3.1.2.1). Hydrogen bonding network protrudes from the RSB directly to this cavity via D116, D251, R109, Q244, N106, and water molecules. The cavity was previously suggested to be a part of the ion- release pathway. It is separated from the bulk by the E11, E160, and R243 triad, which was also believed to be involved into Na+-release to the extracellular space. Hereafter, we will refer to this cavity as a putative ion-release cavity 1 (pIRC1).

Fifth, the pentameric assembly of KR2 creates an additional cavity at the extracellular part of the protomer (Figure 3.1.2.4). The cavity protrudes from the central region of the pentamer to Q78 residue. Importantly, the inside of the KR2 protomers is thus separated from the extracellular bulk by the only side chain of Q78. Thus, we suggest that the additional cavity (we named it putative ion-release cavity 2, pIRC2) also could be involved into the Na+ release from the protein.

Figure 3.1.2.4. Putative ion-release cavity 2 (pIRC2) of the ground state of KR2. A. Side section view of KR2 pentamer. The red ellipse-like shape contours the pore in the protein surface leading to the pIRC2 from the concave aqueous basin formed in the central pore of KR2 pentamer. Membrane core boundaries are shown with black lines. B. Section view from the extracellular side at the level of the pIRC2. pIRC2 of each protomer is formed by helices B, C and BC loop and A’ from adjacent protomer. The pIRC2 from the section A is also contoured with red line. Helices are signed with capital bold letters.

3.1.3. Interprotomer contacts in KR2 complex

Each pair of KR2 protomers interacts through TM helices B, C, D, A’ and B’ via an extended hydrogen bonds network (Figure 3.1.3.1). The connection between protomers at the cytoplasmic side is mediated exclusively by water molecules (w506’, w515’, w488’, and w434) (Figure 3.1.3.1). On the opposite, the hydrogen bond network at the extracellular side consists of two regions (Figure 3.1.3.1). There is a tight connection between the main chain oxygen atom of H30’ and the side chain oxygens of N112 and Y108 through w401. The side chain of H30’ is hydrogen-bonded to those of Y154 from helix D and Q26’ from the same helix A’. Q26’ interacts further with E22’. E22’ also interacts with N105, N7, and the backbone of N-terminus of the neighboring protomer via water molecules w411, w412, w448’, and w491. In the central part of the pentamer N81 side chain interacts with Q80’, S29’, T83’, T33’, and Y36’ through several water molecules. Backbone atoms of N-terminus and BC’-loop are also hydrogen-bonded via w514’, stabilizing the structure. Additionally, Y108 is directly hydrogen-bonded to S29’.

Figure 3.1.3.1. Oligomerization contacts in KR2. a. View at the KR2 pentamer from the cytoplasmic side. b. View at the KR2 pentamer from the extracellular side. c. Detail view of the contacts at the cytoplasmic side. d-e. Detail view of the contacts at the extracellular side.

An important part of the interprotomer connection is the Na+ ion-binding site, comprising D102 and Y25’ side chains, F86’ and T83’ main chain oxygens and w420’ and placed at the extracellular side of the protein interface. The structure of this external Na+ binding site resembles that observed earlier (Figure 3.1.3.1).

3.1.4. Comparison with known KR2 structures

The structure of pentameric KR2 at neutral pH differs notably from those obtained with the crystals grown at acidic pH. The structure is also very different from that obtained with the crystals grown at pH 4.0 and soaked at pH 8.0-9.0 (Figure 3.1.4.1). It means that, indeed, soaking of the crystals, initially grown at low pH, at pH 8.0-9.0 cannot transfer KR2 into its biologically relevant conformation. This likely means that this conformation is not compatible with the monomeric form of the protein. A striking difference between monomeric and pentameric forms is the different orientation of the protein relative to the membrane plane. In the monomeric form, the protein is tilted about 70° to the membrane plane. In opposite, the TM helices of the protein in the pentameric form are oriented nearly perpendicular to the membrane plane (Figure 3.1.4.1). This results in a significant difference in the positions of the IUC relative to the hydrophobic/hydrophilic membrane boundary. In pentameric KR2, the entrance of the cavity is placed in the hydrophilic part. In opposite, in monomeric KR2, it is expected to be buried into the hydrophobic part (Figure 3.1.2.2a and 3.1.4.1).

Figure 3.1.4.1. Comparison of different KR2 structures. a. Structure of chain A of pentameric Na+ pumping form (expanded conformation, pH 8.0, PDB ID: 6REW) is shown in yellow. b. chain E of 4XTN model (compact conformation, pH 4.9) is shown in salmon. c. 4XTL model (compact conformation, pH 4.3) is shown in light blue. d. 3X3C model (compact conformation, pH 7.5-8.5) is shown in green. Red contoured arrows show the important change in the position of the ion uptake cavity relative to the cytoplasmic side in the case of pentameric form. The hydrophobic membrane core boundaries are shown with the black lines. The cavities are colored pink. Cofactor retinal is colored dark green.

The structure of the pentameric form of KR2 at pH 8.0 allowed the reinterpretation of the results obtained by solid-state NMR, time-resolved absorption spectroscopy, and other methods 50 available at the moment. The structure showed that the orientation of D116 is uniform within the pentameric assembly under physiological conditions, which is in agreement with the observed homogeneity of the retinal binding pocket at pH 8.0. Table 3.1.4.1 shows that in pentameric KR2 at pH 8.0, the oxygen atom of the side chain of D116 is within 2.8 Å from the RSB. In monomeric KR2 at neutral pH, two conformations were found in the first study in 2015 with a distance of 2.5 Å and 3.3 Å, respectively. Recently reported a new model of monomeric KR2 at pH 8.0 at 293 K, however, demonstrated the organization of D116-RSB pair similar to that in the expanded conformation of pentameric KR2 (PDB ID: 6TK6). It also showed a direct hydrogen bond between D116 and the RSB. The distance between them, in this case, was 3.0 Å. In pentameric models at pH 4.9 and 5.6 (PDB IDs: 4XTN and 4XTO, respectively), this distance varies within protomers from 2.9 Å to 3.9 Å.

PDB pH Oligomeric state Distance, Å

8.0 2.8

6.0 Pentameric 2.9

5.0 3.2 This 8.9 3.3 work 8.0 3.2 Monomeric 6.0 3.7

4.3 3.6

4XTO 5.6 3.3-3.9 (compact) Pentameric 4XTN 4.9 2.9 (expanded) -3.5 (compact)

4XTL 4.3 3.6

3X3B 4.0 Monomeric 3.7

3X3C 7.5-8.5 2.5-3.3

6TK7 4.4 Monomeric 4.2

6TK6 8.0 Monomeric 3.0

Table 3.1.4.1. Distance between D116 oxygen and RSB nitrogen atoms in KR2 models.

Also, in contrast to the previously obtained structures, the orientations of the two key residues, N112 and L74, are identical in all five protomers within polymeric assembly and is the same as that of the expanded conformation of KR2. Thus, the distance between the Cα atoms of N112 and D251 is 12.2 Å, allowing more space for Na+ passage through the protein. The side chain of N112 is stabilized by a hydrogen bond with water molecule w401, present only in pentameric form. The water molecule is hydrogen-bonded to the main chains oxygens of Y108 and H30’ of the oligomerization interface (Figure 3.1.4.2).

Figure 3.1.4.2. Comparison of the RSB regions of different KR2 structures. a. Chain A of pentameric Na+ pumping form (expanded conformation, pH 8.0) is shown in yellow. b. Chain E of 4XTN model (compact conformation, pH 4.9) is shown in salmon. c. 4XTL model (one of two closely related to compact conformation, pH 4.3) is shown in light blue. d. 3X3C model (closely related to compact conformation, soaked at pH 8.0-9.0) is shown in green. Red dashed ellipse shows the double conformation of the D116 side chain. Red contoured arrows show the important displacement of the N112-L74 pair (colored teal). Helix A’ of nearby protomer and fragments of 52 lipid molecules are shown in orange for pentameric and monomeric models, respectively. The cavities are colored pink. Cartoon representation of helix A is hidden for clarity. The cofactor retinal is colored dark green.

Consequently, the SBC1 suggested to be the region of the transient Na+ binding site in the O-state, is large in the case of the expanded conformation of KR2. In contrast, it is completely absent in the KR2 structure, obtained from crystals grown at pH 4.0 and soaked at pH 8.0-9.0 (Figure 3.1.4.2). In opposite, SBC2 is slightly enlarged in these structures.

Figure 3.1.4.3. Comparison of extracellular regions of different KR2 structures. a. Structure of chain A of pentameric Na+-pumping form (expanded conformation, pH 8.0, PDB ID: 6REW) is shown in yellow. b. chain E of 4XTN model (compact conformation, pH 4.9) is shown in 53 salmon. c. 4XTL model (compact conformation, pH 4.3) is shown in light blue. d. 3X3C model (compact conformation, pH 7.5-8.5) is shown in green. The hydrophobic membrane core boundaries are shown with the black lines. The cavities are colored pink. Cofactor retinal is colored dark green.

The extracellular parts of the expanded and compact conformations in pentameric models of KR2 are similar (Figure 3.1.4.3). On the contrary, there is a significant difference between pentameric and monomeric forms of KR2 in this part of the protein, mainly caused by the displacement of the helix C and the N-terminal helix. In particular, the pIRC1 is larger in the case of the monomeric form, and positions of the E11, Q157 and E160 side chains differ considerably.

3.1.5. The second Na+ ion identified at the KR2 surface in the ground state

Soon after we solved and published the crystal structure of KR2 pentameric Na+-pumping state under physiological conditions at 2.2 Å resolution, we improved the resolution of the model to 2 Å. Therefore, we obtained a more complete structure of the KR2 resting state at 2 Å resolution (Table 5.3).

While the inner region of the protein protomers is the same as described above, we identified numerous of water molecules at the KR2 surface. Moreover, we observe a second Na+ binding site at the pentamer surface (Figure 3.1.5.1). The additional Na+ ion is coordinated by the main chain oxygen of S100 residue and 5 water molecules with the mean Na-O distance of 2.5 Å (Figure 3.1.5.1). The Na+ was not identified at lower resolution, and its B-factor is considerably higher than that of the previously reported Na+ (52 and 21 Å, respectively). As will be described further, we believe that this loosely bound Na+ might be released to the extracellular bulk from the surface upon the O-to-ground transition and substituted by the ion transported in the current cycle.

Figure 3.1.5.1. Additional Na+ identified at the KR2 interface in the ground state. A. Overall view of KR2 pentamer in the ground state from the extracellular side. Na+ bound at the protein + surface are shown with purple spheres. B. Zoomed-in view of the Na -binding sites. 2Fo-Fc electron density maps around the Na+ and interacting residues and water molecules are shown with black mesh and are contoured at the level of 1.2 σ. Distances between Na+ and nearby oxygens are shown with black dashed lines. Distances are in Å. C. Surface of KR2 pentamer near Na+. Putative ion-release cavity 2 (pIRC2), which will be described in detail further, is labeled. Black arrows indicate putative relay pathway of the Na+ release.

3.1.6. Structural switches in KR2 upon pH decrease

As it was mentioned above, previously reported structures of pentameric KR2 were obtained at low pH values and showed various conformations of the protein. These conformations are also different from those found in the structure of KR2 at pH 8.0. Therefore, it is logical to suggest that there are structural rearrangements in KR2 upon pH decrease. Moreover, these changes could be associated with the functional switches of the protein. Indeed, it was shown that KR2 pumps Na+ ions when their concentration is much higher than that of H+, which is at least at pH 6.4-8.0. This means that under physiological conditions, KR2 is in the Na+-pumping mode. However, no information on Na+ transport by KR2 at acidic pH is available in literature. To address this question, we studied the pumping activity of KR2 reconstituted into lipid vesicles at pH 8.0, 6.0, and 4.3 in the presence of 200 mM of NaCl in the same manner, as described elsewhere. The data showed that KR2 can pump Na+ ions at pH 6.0 and 8.0, and likely pumps H+ at pH 4.3 (Figure 3.1.6.1). The absolute value of the pH changes upon illumination is deficient in the case of pH 4.3. However, the addition of the photophore CCCP decreased the ∆pH value in that case, which is a characteristic feature of the H+, but not the Na+ pump. Therefore, we assume that KR2 possesses a weak H+-pumping activity at acidic pH. Consequently, we define the low pH KR2 form as the H+-pumping form.

Figure 3.1.6.1. Activity tests of KR2 at different pH. Pumping activity of KR2 reconstituted into lipid vesicles. The unbuffered solutions contain 100 mM NaCl (black, dashed) and 100 mM NaCl and 30 uM CCCP (black, solid). Starting pH was adjusted by dialysis against 100 mM NaCl unbuffered solution with needed pH. The liposomes were illuminated for 600 s (yellow area on the plots).

Also, the optical properties of the protein change with the decrease of pH, and structural rearrangements in the retinal binding pocket were reported. Structural insights into the nature of transitions in KR2 associated with the acidification of the surrounding media may help to better understand the mechanism of light-driven Na+ pumping. To discover the dependence of KR2 conformations on the pH, we have determined crystal structures of KR2 using crystals initially grown at pH 8.0 and soaked for 48 hours in buffer solutions with pH 4.3, 5.0, 6.0, 8.0, and in the buffer solution described in 48.

After soaking at pH 6.0, crystals remain red and diffract to 2.7 Å. The structure of KR2 at pH 6.0 is the same as that at pH 8.0 (Figure 3.1.6.2). This result is in accordance with solid-state NMR structural studies of the retinal binding pocket of KR2, where no changes were observed between pH 6.0 and 8.0.

Figure 3.1.6.2. KR2 crystals soaking. a-d. Crystals grown at pH 4.3, contain protein monomer in the asymmetric unit. e-h. Crystals grown at pH 8.0, contain protein pentamer in asymmetric unit. Size of KR2 crystals, grown at pH 4.3 and 8.0 were 50-100 μm and 100-200 μm respectively.

Unfortunately, we could not determine the structure of the pentameric form of KR2 at pH 4.3. In course of soaking, the crystals changed their color from red to purple (absorption maximum at 550 nm) (Figure 3.1.6.2). However, they lost diffraction quality after the soaking procedure. To decrease the stress on crystals, we soaked them in pH 4.3 with a very gradual exchange of the buffer during 48 hours. Nevertheless, even such gentle manipulations led to a drop in diffraction quality to 7 Å, and the mosaicity became extremely high. Thus, it was decided to soak crystals at a higher pH.

The structure of KR2 obtained with the crystals soaked at pH 5.0 was solved to 2.6 Å. Crystals also changed their color from red to purple after soaking (Figure 3.1.6.2). The location of most of the residues remains the same as in the model at pH 8.0. However, the region comprising residues 109-115 of helix C is slightly displaced. The distance between the D116 and the RSB is increased to 3.2 Å when at pH 8.0 it is only 2.8 Å. The weakening of D116-RSB interactions upon pH decrease was also reported in the solid-state NMR study of the retinal binding pocket of KR243. In that study, the effect occurred at pH around 5 and was assigned to the protonation of D116. Importantly, the change in the protonation state of D116 could be indicated by the shift of absorption maximum of KR2 in crystals from 528 to 550 nm (Figure 3.1.6.2). At the same time, L74 and N112 side chains occupy two alternative conformations. The first is identical to that in the expanded, while the second is similar to that of the compact conformation of KR2 (Figure 3.1.6.3). Consequently, the SBC1 is reduced in volume in a fraction of KR2 at pH 5.0 (Figure 3.1.6.3).

Figure 3.1.6.3. Conformational rearrangements in KR2 upon pH decrease. a. The expanded conformation of KR2 at pH 8.0. b. Co-existence of the expanded and the compact conformations in 1:1 ratio in pentameric KR2 at pH 5.0. c. Examples of 2Fo-Fc maps near L74-N112 pair at pH 5.0. The maps are contoured at the level of 1.0 σ.

In summary, according to our functional and structural analysis of KR2 at different pH, the rhodopsin acts as an H+ pump at acidic pH, and under the same conditions, a compact conformation of KR2 appears. Thus, we suggest that the compact conformation corresponds to the H+-pumping mode. Hence, we conclude that the expanded-to-compact conformational switch may result in the corresponding Na+-to-H+ pump functional switch in KR2.

3.1.7. Effects of dehydration on KR2 crystals

Interestingly, a previously published structure of the pentameric form of KR2 at pH 4.9 (PDB ID: 4XTN) contains protomers in both compact and expanded conformations, when the structure at pH 5.6 (PDB ID: 4XTO) contains only protomers in the compact conformation. This finding is in conflict to what we observed in the present study of pH influence on KR2 conformations (see previous section). Therefore, we decided to investigate the molecular basis of this inconsistency. We assume that this discrepancy is mainly due to the crystals dehydration during the long crystallization process (two to four months). Indeed, the crystallization drop was not covered with additional buffer solution during crystals harvesting procedure in our previous work. To check this hypothesis and resolve the controversy, we harvested dehydrated crystals of 58

KR2 initially grown at pH 8.0, which changed their color from red to purple with time without the addition of the buffer solution. After that, we added a buffer solution to the same already opened crystallization well and let it equilibrate for 30 minutes to rehydrate the crystals. The color of the crystals turned back to the red within 1 minute. Finally, we harvested the rehydrated crystals and solved the structure of both pentameric “dry” and “wet” forms at 3.0 Å and 2.6 Å, respectively. The structure showed that in the “dry” form, all five KR2 protomers within the pentamer are in the compact conformation, while after soaking, they switched back to the expanded one (Figure 3.1.7.1). The structure of the “dry” form is almost identical to the deposited earlier (PDB ID: 4XTO). These observations stress the importance of humidity control during crystal growth as well as during their harvesting. Importantly, this also showed that the 4XTO model should not be used for further analysis of KR2 since it represents the dehydrated non-functional form of the protein.

Figure 3.1.7.1. KR2 “dry” and “wet” forms. a. Crystals of pentameric KR2 grown at pH 8.0, wizened and changed their color from red to purple with drying after 6 months in the crystallization plate, revealing “dry” conformation of the protein. b. The same wizened crystals after soaking in 3.4 M sodium malonate pH 8.0, revealing the “wet” conformation of the protein. c. Side view of the aligned protomers (helices B and C) of the “dry” and “wet” KR2 models. N112 and L74 side chain positions are shown with sticks. d. 2Fo-Fc electron density maps are shown for N112 and L74 side chains in “dry” and “wet” conformations. Maps are contoured at the level of 1.5 σ. e. Detailed view of the two protomers aligned. Key amino acid side chains in KR2 are shown with sticks. Helix A is hidden for clarity.

3.1.8. The structure of the monomeric form of KR2 at different pH

To resolve the existing controversy of structural data reported in 2015, we also studied the structures of the monomeric form of KR2 at different pH. Therefore, we determined crystal structures of KR2 using crystals, initially grown at acidic pH 4.3. These crystals were blue-colored under the initial crystallization conditions, similar to our previous work (Figure 3.1.6.2). Before

59 fishing, crystals were soaked for 48 hours in buffer solutions with pH 4.3, 6.0, 8.0, and in the buffer solution, described in 48.

All structures are nearly identical to the existing model of the monomeric form of protein (PDB ID: 4XTL). There are no differences in the location of the residues important for protein functioning. We did not observe the expanded conformation in the monomeric form of KR2. Interestingly, we observed no flip of D116 at neutral pH even at 1.8 Å, described in 48. We suggest that the flip of D116 found by Kato and colleagues could be explained by the presence of radiation damage artifacts in the electron density maps, similar to that found earlier in the case of bacteriorhodopsin102,103. Indeed, in that study strong negative electron densities around D116 residue were observed; however, no complementary positives densities were found. The appearance of negative electron densities might also be caused by the protein disturbance due to soaking of the crystals initially grown at low pH.

The flip of D116 was also not identified in the recent crystal structure of the monomeric form of KR2 obtained at 293 K at SwissFEL (PDB ID: 6TK6)68. Moreover, in that work, the authors did not observe the flip even during the photocycle. This additionally supports the idea that the second conformation of the D116 side chain is an artifact of the crystals’ manipulations or X-ray diffraction experiments.

Interestingly, due to the minor reorientation of the RSB in the structure of monomeric KR2 at pH 8.0, the RSB-D116 distance is shortened in this case to 3.2 Å (PDB ID: 6RF6). This supports that D116 becomes deprotonated upon soaking at neutral pH and is able to form a direct hydrogen bond with the RSB even in monomeric model of the protein. In support of this finding, the 6TK6 structure of the monomeric KR2 at 293 K showed even closer relative location of D116 and the RSB within the 3.0 Å (Table 3.1.4.1).

3.1.9. Structures of K+-pumping mutants of KR2

To better understand the principles of KR2 functioning and ion selectivity, verify the proposed key determinants of light-driven Na+ pumping, and check whether they are universal also for pumping of other cations, we decided to study the structures of K+-pumping mutants of KR2.

For that, we selected S254A and G263F mutants of KR2. Indeed, mutational analysis of the NaR indicated that the S254A mutant of the protein can pump not only H+ and Na+ but also K+ ions (Figure 3.1.9.1). The spectroscopic analysis indicated the red-shift of maximum absorption wavelength from 525 to 545 nm in the case of that mutant. G263F was also shown to pump K+. 60

We produced and crystallized S254A and G263F mutants and solved their ground state structures in monomeric and pentameric forms.

Figure 3.1.9.1. Activity tests of KR2 K+ pumping mutants. Pumping activity of KR2 and its mutants in E. coli cells suspension. The solutions contain 100 mM NaCl or KCl (black, dashed) and 100 mM NaCl or KCl and 30 uM CCCP (black, solid). The pH of the starting solutions was around 7. The cells were illuminated for 300 s (yellow area on the plots).

The structures of the monomeric form of the mutants were determined at 2.0 and 2.1 Å in the case of G263F and S254A, respectively, using the crystals grown at pH 4.3. The structures are nearly the same as that of the monomeric wild type KR2. Structures of the pentameric form of the mutants were determined using crystals grown at pH 8.0. The resolution for the models of both mutants was 2.4 Å. The structures are very similar to that of the pentameric KR2 at pH 8.0. The differences were found exclusively in the regions of amino acid substitutions (Figure 3.1.9.2). In the case of G263F mutant, the IUC is significantly decreased and became separated from the solvent by the bulky side chain of F263 in both the monomeric and pentameric forms (Figure 3.1.9.3).

Figure 3.1.9.2. Structural alignment of K+ pumping KR2 mutants with the model of WT protein. a. Alignment of G263F (orange) and WT monomeric forms (PDB 4XTL, light blue). b. Alignment of G263F (light pink) and WT pentameric forms (yellow). The cavities inside WT protein are shown and colored pink. c. Alignment of S254A (cyan) and WT monomeric forms (PDB 4XTL, light blue). d. Alignment of S254A (purple) and WT pentameric forms (yellow).

Figure 3.1.9.3. Structures of K+-pumping KR2 mutants. a. The cytoplasmic region of the G263F monomeric form. b. The cytoplasmic region of the G263F pentameric form. c. Retinal region of S254A monomeric form. d. Retinal region of S254A pentameric form. The cavities are colored pink. The cofactor retinal is colored dark green.

Importantly, we observed the presence of a large SBC1 filled with four water molecules in the structures of the pentameric form of the mutants. On the contrary, the cavity is absent in the monomeric form of both mutants due to the rearrangement of the N112-L74 pair in the same manner, as described for the wild type protein (Figure 3.1.9.4).

Figure 3.1.9.4. Schiff base cavity 1 in K+ pumping mutants of KR2. a. The monomeric blue form of G263F. b. The pentameric red form of G263F. c. The monomeric blue form of S254A. d. The pentameric red form of S254A. The cavities are colored pink. Cofactor retinal is colored dark green.

These findings showed that the presence of large SBC1 is a universal feature of both Na+- and K+-pumping variants of KR2 in the biologically relevant form. Thus, this is an additional evidence that SBC1 is one of the key determinants of light-driven cation non-H+ pumping.

3.1.10. Role of the KR2 pentamerization on the function of the protein

Previous sections showed that the structure of KR2 depends significantly on the oligomeric state. Our main observation was that in the pentameric form, KR2 has a large polar cavity in the RSB region, which, as we believe, is important for the Na+ passage through the rhodopsin.

However, the functional assays of KR2 were only performed using pentameric protein, since KR2 forms pentamers in lipid membranes. To determine whether the pentameric organization is vital for KR2 functioning one needs to study pumping activity of the monomeric form of KR2 or other NaRs. Unfortunately, all our efforts to reconstitute the protein into the lipid membrane in the monomeric form failed. Therefore, we decided to use another approach and mutate the oligomerization interface to break the pentameric assembly and check the functional activity of the resulting protein variants. It was shown that the alterations of the oligomerization interface in the region of H30 residue affect the oligomeric assembly and change protein selectivity and RSB- counterion interaction. Consequently, we produced H30L, H30K, and Y154F mutants of KR2, which were aimed at breaking the direct hydrogen bond between H30 and Y154’. The residues are located in the core of the oligomerization interface and therefore stabilize the KR2 pentamer.

SEC profiles of purified KR2 in detergent at different pH values are shown in Figure 3.1.10.1a. They demonstrate that the pentameric complex of the wild type protein is disrupted upon acidification of the surrounding media. This effect qualitatively corresponds to that in the crystals. Nevertheless, it is clear that KR2 is pentameric at neutral pH both in detergent and in lipid environments. In contrast, at pH 8.0 only a negligible fraction of the H30K and Y154F mutants is organized into pentamers when KR2-H30L is notably unstable and has a large peak corresponding to the protein aggregates (Figure 3.1.10.1b). The absence of pentameric organization was recently demonstrated also for Y154A (in detergent micelles), but, surprisingly, not for the H30A mutant.

Figure 3.1.10.1. Effects of mutations of KR2 at the oligomerization interface. a. SEC profiles of wild type KR2 at pH 4.3, 6.0, and 8.0. b. SEC profiles of wild type KR2 and its mutants at pH 8.0. Protein with an initial concentration of 70 mg/ml was dissolved in a buffer solution with 0.1% DDM to a final concentration of 1-2 mg/ml and incubated for 72 hours. c. The Na+-pumping activity of KR2 and its mutants measured in E. coli cells suspension. The solutions contain 100 mM NaCl (black, dashed) and 100 mM NaCl and 30 uM CCCP (black, solid). The cells were illuminated for 300 s (yellow area on the plots). d. Ribbon representation of the structural 65 alignment of H30K (green), Y154F (orange) mutants and wild type (yellow) of KR2. Side chains at positions 30 and 154 are shown in sticks. The hydrophobic membrane core boundaries are shown with the solid horizontal lines. e,f. Y154F and H30K mutation region aligned with the wild type protein, respectively. 2Fo-Fc electron density maps are shown for the structures of the mutants and are contoured at the level of 1.5 σ (Y154F) and 1.0 σ (H30K).

Next, we studied the pumping activity of the mutants in E. coli cells suspension as described elsewhere. The tests showed that Na+-pumping activity is dramatically decreased in the cases of H30K and Y154F while it is completely absent for H30L (Figure 3.1.10.1c).

We then produced the Y154F and H30K mutants and crystallized them at pH 8.0. The structures of the mutants were determined at 1.8 and 2.2 Å, respectively. The proteins were in the monomeric form in crystals and the space group is the same as that of the monomeric form of wild type KR2 at acidic pH. Both structures are the same as that of the monomeric form of KR2, except for the mutation region (Figure 3.1.10.1d-f).

We also produced the H30L mutant of KR2. However, it was unstable during the purification. Crystallization trials were set and small crystals of the H30L mutant were obtained. Unfortunately, the diffraction quality was low, and we could not collect a complete dataset. Further optimization of the crystallization conditions is required. The experiments are ongoing.

These results support the existence of a strong correlation between the oligomeric state and ion-pumping activity of KR2. Indeed, the obstruction of the pentameric assembly by the introduction of point mutation of the inter-protomer interface of KR2, observed both in detergent and in crystals, leads to the decrease or complete absence of Na+ transport in the corresponding mutants. Thus, we suggest that pentamerization is necessary for the efficient Na+-pumping activity of KR2.

Another indirect evidence of the importance of the pentameric assembly of KR2 is the absence of any rearrangements in the structures of monomeric H30K and Y154F mutants compared to the monomeric form of wild type protein. Indeed, the functional properties are notably altered. If one considers the monomeric form as a functional unit, at least minor differences between the structures of the wild type and mutant proteins are expected. However, they were not found even at high resolution. This means that either the monomeric form does not represent the functional state of KR2 or that the differences occur in the course of the photocycles of the proteins

66 and are not present in their ground states. In the latter case, further structural studies of the intermediate states of both the wild type and mutant forms of KR2 are required.

3.2. Crystal structure of KR2 in the O-state

3.2.1. Accumulation and cryo-trapping of the O intermediate state in KR2 crystals

Next, based on the established protocols of KR2 crystallization and stable protein production, we aimed at obtaining the structures of the KR2 in its intermediate states. The structures are obligatory for the understanding of the molecular mechanism of light-driven Na+ pumping. Since the binding of the Na+ ion in the core of the KR2 protomer was expected to occur in the late red-shifted O-state, we decided to start the work on this intermediate (Figure 3.2.1.1).

First, we have performed time-resolved visible absorption spectroscopy of the KR2 crystals to verify that the photocycle of the protein is not altered by the crystal contacts. For that, we used the slurries of KR2 microcrystals, placed onto the cover glass, and put into the cuvette filled with 3.4 M sodium malonate pH 8.0 to avoid dehydration of the crystals. The experiments showed that similar to the protein in detergent micelles and lipids, crystallized KR2 also forms characteristic K-, L/M- and O-states (Figure 3.2.1.1).

Figure 3.2.1.1. Spectroscopy of KR2 in crystals. a. The scheme of the KR2 photocycle indicates that Na+ binding occurs transiently in the red-shifted O-state. b. UV–visible absorption spectra measured in crystallo at 100 K of the ground state (red) and the O-state (blue) of KR2 (insets: photos of the same frozen KR2 crystal in the cryoloop before and after laser illumination, near the corresponding spectra). c. Difference spectrum calculated between the blue and red spectra shown in b. d. Photocycle of KR2 reconstituted in DOPC. e. Time traces of absorption changes of KR2 crystals at 410 (black), 480 (light blue), 530 (green), and 610 nm (red) probe wavelengths. Black lines indicate fitting lines based on the sequential kinetic model shown in f. Photocycle of KR2 in crystals, determined in the present work.

Second, we optimized the procedure of the accumulation of the O-state in the crystals at the icOS station of the ESRF, Grenoble, France. For that, large but thin crystals of KR2 were harvested with microloops and froze into liquid nitrogen. The microloops were then placed onto the goniometer in the cryostream at 100 K for the spectroscopy studies and laser illumination. Using the traditional approach, we illuminated the crystal in the loop with 532 nm laser. The cryostream was blocked for a short period during laser exposure and released back before turning off the laser (Figure 3.2.1.2). This procedure allows accumulation and trapping in crystals of the dominant intermediate of the protein photocycle104. We have optimized the laser power and the time of cryostream blocking to obtain the major fraction of the O-state in crystals. As a result, we succeed in the accumulation of almost 100% of the O-state in the KR2 crystals. For that, the laser power density of 7.5 mW/cm2 at the position of the sample and blocking of the cryostream for 1 s were used.

Figure 3.2.1.2. Scheme of the cryo trapping procedure of the O-state in KR2 crystals. The plate-like crystal in the loop was oriented perpendicular to the 532-nm laser beam for more uniform activation of the proteins inside the crystals. The manual blocking was performed for 1 second by the plate-like thin piece of plastic.

The mean size of the crystals was 200x100x20 μm3 (Figure 3.2.1.3). The plate-like crystals were oriented so that the largest plane (200x100 μm2) was as perpendicular to the laser beam. The laser beam was focused to the size of 500x500 μm2 (1/e2). The use of higher laser power density resulted in the non-reversible protein bleaching and/or sample drying. A UV-visible absorption spectrum was then recorded to show the red-shifted absorption maximum characteristic of the O-

70 state of KR2. The pure O-state in KR2 crystals has a maximum absorption wavelength of 602 nm (Figure 3.2.1.1). This is 36 nm more red-shifted than that observed in solution. We suggest that the reason for such a difference could be a cryo-freezing of the crystals.

Figure 3.2.1.3. Example of KR2 crystals used in present work. The mean size of the crystal is 200x100x30 μm3. The same result of the O-state trapping procedure was reproduced with approximately 30 crystals of KR2, the same as presented in the figure.

3.2.2. Determination of the crystal structure of the O-state of KR2

The crystals with the accumulated O intermediate state were stored in liquid nitrogen and transported to the P14 beamline of PETRAIII, Hamburg, Germany, for the X-ray experiments. The crystals were exposed to X-rays, and the diffraction data were collected at 100 K using traditional techniques. In the case of the crystals with the cryo-trapped O-state, brought from the icOS station, the best diffraction was at the level of 2.8 Å. We suggest that the relatively low resolution was a result of the use of thin KR2 crystals suitable for spectroscopy studies. For obtaining the higher- resolution structure of the O-state, we have constructed a set up for the illumination of the KR2 crystals with 532 nm laser directly at P14 beamline of PETRAIII. The scheme of the set up is shown in Figure 3.2.1.2. The laser power density was adjusted to a level of 7.5 mW/cm2. For the trapping of the O-state at P14 beamline, we used the same protocol as that optimized at icOS.

Finally, the best dataset was collected at 2.1 Å resolution. As in the case of the ground state of KR2, the data were scaled and merged using the Staraniso server to accurately account for the

71 anisotropy. The crystal symmetry and lattice parameters are the same as described previously for the ground state of the protein, with one KR2 pentamer in the asymmetric unit. The phasing problem was solved using the molecular replacement method with Molrep of the CCP4i suite. The structure of the ground state of pentameric KR2 (PDB ID: 6REW) was used for the search of the initial model. The initial model was then iteratively refined using Refmac5 and Coot. Data collection and processing and structure refinement statistics are shown in Appendix Table 5.3.

The data analysis revealed the presence of solely the O-state in the structure. This is in line with the spectroscopy data, described in the previous section. The 100% occupancy of the intermediate state in crystals, together with the 2.1 Å resolution of the final model, resulted in the resolving of the exact conformations of the retinal cofactor, key functional residues, and numerous water molecules in the O-state of the KR2 photocycle. It should be also mentioned that this is the first case when intermediate state was accumulated with 100% occupancy.

The structure demonstrates notable rearrangements compared to the ground state of KR2 (Figure 3.2.2.1). The root mean square deviation (RMSD) between the backbone atoms of the pentamers and protomers of the ground (PDB ID: 6REW) and the O-states are 0.55 and 0.53 Å, respectively. The main changes occur in the extracellular parts of the helices B and C, which are shifted by 1.0 and 1.8 Å, respectively (Figure 3.2.2.1). Helices A, D, and G are also displaced by 0.7 Å in the extracellular regions (Figure 3.2.2.1).

Figure 3.2.2.1. Structural alignment of KR2 protomers in the ground (yellow) and the O- (blue) states. A. Overall alignment. N-terminal α-helix and N terminus are colored blue. BC loop, containing the β-sheet, is colored orange. B. Enlarged view of the most notable rearrangements in

72 the protomer backbone. The retinal cofactor is colored teal. Membrane core boundaries are shown with black lines. Helices are indicated with capital letters. The shifts of extracellular parts of helices B and C are demonstrated with black arrows.

3.2.3. The retinal binding pocket of KR2 in the O-state

The polder electron density maps built around the retinal cofactor strongly suggest its all- trans configuration in the O-state distorted around C14 atom (Figure 3.2.3.1). Indeed, the fitting of the electron density maps with either 13-cis or the mixture of all-trans/13-cis retinal results in the appearing of strong negative peaks of the Fo-Fc difference electron maps at the level higher than 3 σ. Moreover, our data show that 13-cis configuration would result in the steric conflict between

C15 and C20 atoms of the retinal and W113 and W215 residues, respectively. Therefore, the all- trans retinal was modeled into the final structure of the O-state.

Figure 3.2.3.1. Examples of electron densities of the KR2 O-state. a. Polder maps for K255 and retinal cofactors of all five protomers of KR2 O-state structure. Maps are contoured at the level of

4.0 σ. b. 2Fo-Fc electron density maps around K255 and retinal cofactor contoured at the level of

1.5 σ. c. 2Fo-Fc and Fo-Fc electron density maps built using the data of the O-state of KR2 at 100K at 2.1 Å and 6 models with 100/0, 80/20, 60/40, 40/60, 20/80 and 0/100 proportions of all-trans/13- cis retinal configurations ratios. 2Fo-Fc maps are contoured at the level of 1.5 σ and are shown with gray mesh. Difference negative Fo-Fc maps are contoured at the level of 3 σ and are shown with red mesh. d. Steric conflict and inadequately short distances between retinal C15 and C20 atoms to the nearby residues W113 and W215 when fitting the data with 13-cis retinal. Retinal and K255

73 are colored teal. C14, C15 and C20 atoms of retinal are indicated. The lengths of the hydrogen bonds are shown with bold italic numbers and are in Å.

The positions of the residues comprising the retinal pocket, particularly W113, D251, D116, I150, Y218, and W215 are correspondingly shifted relative to those in the ground state (Figure 3.2.3.2). Surprisingly, in contrast to recent FTIR data, we found all-trans configuration of the retinal in the O-state. Indeed, in ref. 42 the configuration was suggested to be 13-cis in the O- state of NaRs42. At the same time, our finding is in line with the data on another light-driven sodium pump GLR, published in 2014, where a distorted all-trans configuration of retinal in the O-state was reported24. In ref 42 the authors found a broad peak at 940 cm-1 in the FTIR spectrum of the O-state of KR2 containing 12, 14-D2 retinal, which they interpreted as 13-cis configuration of the retinal. Our data demonstrate that retinal is all-trans in both the ground and the O-state; however, it is bended notably around C14 atom only in the intermediate, but not in the ground state (Figure 3.2.3.2). Also, retinal is almost fully planar in the O-state, while it is a little curved in the ground state. It is possible that the bent of the retinal near C14 atom in the O-state may result in the appearance of the peak at 940 cm-1 described in ref 42. We also cannot exclude that the inconsistency of the results on the retinal configuration in the O-state may originate from the different conditions and protein environment during the experiments.

Figure 3.2.3.2. Retinal binding pockets of ground (yellow) and O (blue) states. A, B. View from the cytoplasmic side. C, D. View from the side of the β-ionone ring of the retinal molecule. Retinal cofactor is colored teal.

3.2.4. Transient Na+ ion-binding site inside the KR2 protomer

The crystal structure of KR2 in the O-state at 2.1 Å reveals the transient Na+ ion-binding site near the RSB, comprised of S70, N112, and D116 side chains and the backbone oxygen of V67 residue (Figure 3.2.4.1). The mean distance between the Na+ ion and the coordinating oxygen atoms is 2.3 Å, which is characteristic for Na+, but not to larger cations, such as K+ (Figure 3.2.4.1). Interestingly, the organization of the site is also suitable for Ca2+ binding; however, it might be prevented by the electrostatic conflict because of the larger positive charge of Ca2+ compared to Na+.

The importance of S70, N112, and D116 residues for KR2 Na+-pumping activity has been confirmed by previous mutational analysis. Moreover, N112 and D116 are part of the NDQ motif characteristic for all NaRs. The absence of aspartate at the 116th position of the protein leads to the loss of Na+ pumping activity by the rhodopsin. Thus, D116 is critical for protein functioning. N112 determines the ion selectivity of KR2. S70A and S70T mutations dramatically decreases the Na+- pumping activity of KR2105.

Figure 3.2.4.1. Transient Na+ ion-binding site in the core of KR2 protomer.

Hydrophobic/hydrophilic membrane core boundaries are shown with blue disks. 2Fo-Fc electron density maps around Na+ binding site are contoured at 1.5 σ and are shown with gray mesh. Distances between Na+ ion and coordination oxygens are shown with black dashed lines.

The S70-N112-D116 (S-N-D) triad, completely conserved within NaRs and comprising the core of the Na+ ion-binding site in the middle part of the protein, is very similar in respect to the residues composition and their relative location to the central gates (CGs) of channelrhodopsins (Figure 3.2.4.2). Indeed, the CGs of the wild-type cation channelrhodopsin-2 (CrChR2)60, chimeric engineered cation channelrhodopsin (C1C2)106, and also natural anion channelrhodopsin (GtACR1)107,108 are composed of the S63-N258-E90, S102-N297-E129, and S43-N239-E68 triads (Figure 3.2.4.2). Although the residues are located in the different helices compared to the KR2, they form similar overall configuration. CGs serve as constriction sites in the central part of the channels and are important for the ion selectivity. Therefore, this is an evidence that the transient Na+ binding site is important for KR2 selectivity. One the other hand, one cannot exclude that CGs may act as transient ion-binding sites in channelrhodopsins during ion translocation.

Figure 3.2.4.2. Na+ binding site of KR2 and central gates of channelrhodopsins. A. Structural alignment of KR2 (present work, blue), CrChR2 (PDB ID: 6EID, orange), C1C2 (PDB ID: 3UG9, brown) and GtACR1 (PDB ID: 6CSM, green). B. Transient sodium binding site in KR2. C. Central gate of CrChR2. D. Central gate of C1C2. E. Central gate of GtACR1.

Moreover, as was already mentioned in the Introduction section, in 2015 the evolutionary relationship of microbial light-driven Na+ pumps and class A GPCRs was proposed, based on the existing structures of the ground state of KR2. Presented here structure of the O-state with Na+ bound inside the protein molecule supports the high similarity of the Na+ binding sites in KR2 and GPCRs. Namely, in both cases, the sites are located near helices B and C (TM2 and TM3 in GPCRs) and formed by the similar to the S-N-D triad of KR2 set of residues (D95, N131, and S135 in the case of human δ-opioid receptor109) (Figure 3.2.4.3). This opens the way for a more accurate analysis of the interconnection between two highly important families.

Last but not least, general feature of the Na+ ion-binding site inside KR2 are also very similar to that of another type of Na+ pumps - Na+ ATP synthases. For instance, it is almost

+ 110 identical to that of the c11-ring of the Na ATP synthase from Ilyobacter tartaricus (Figure 3.2.4.3).

Figure 3.2.4.3. Na+ ion-binding sites of different families of membrane proteins. Na+ ions are shown with purple spheres. Distances are shown with black dashed lines and are in Å.

The location of the transient Na+ ion-binding site in the O-state of KR2 photocycle is similar to that of the Cl- ion-binding sites in the ground state of archaeal and bacterial light-driven Cl- pumps (Figure 3.2.4.4). Namely, in Cl--pumping rhodopsin from Nonlabens marina S1-08 (ClR, Nm-R3), the anion is coordinated by N98 and T102 residues of the characteristic NTQ motif59. N98 and T102 of ClR are analogous to the N112 and D116 of the NDQ motif of light- driven Na+ pumps (Figure 3.2.4.4). 78

Figure 3.2.4.4. Na+ and Cl- binding sites of the ion-pumping bacterial rhodopsins. A. O-state of KR2 (present work). B. The ground state of the chloride-pumping Nonlabens marinus S1-08 rhodopsin (PDB ID: 5ZTK). The retinal cofactor is colored teal. Water molecules are shown with red spheres. Na+ is shown with a purple sphere. Cl- ion is shown with a green sphere. Distances are shown with black dashed lines. The length of the D116-RSB hydrogen bond is shown with bold italic numbers and are in Å. Helix A is hidden for clarity.

3.2.5. Structure of the KR2 protomer in the O-state

The all-trans retinal configuration in the O-state of KR2 obtained in the present work means that relative locations of the protonated RSB and D116 side chain are similar to those in the ground state (PDB ID: 6REW). The protonated RSB is hydrogen-bonded to D116, and the distance between them is 2.9 Å in the O-state (Figure 3.2.5.1a). The existence of this hydrogen bond is supported by time-resolved resonance Raman spectroscopy. Indeed, in ref 42 the authors reported that C=N stretching frequencies of the RSB are very similar between the ground state (1640 cm-1) and the O-state (1642 cm-1). The C=N stretching frequency is a sensitive marker for the hydrogen bond strength of the protonated RSB. The similar frequencies of the C=N stretching mode support that relative locations of the protonated RSB and D116 side chain are similar to those in the ground state.

While in the ground state, KR2 is in the expanded conformation, in the O-state, the side chain of N112 is flipped towards S70 and D116. Therefore, the overall conformation is similar to that of the compact one of KR2 (Figure 3.2.5.1b). This is also evidenced by the disappearance of SBC1 and enlargement and elongation of SBC2 in the O-state (Figure 3.2.5.1). In the O-state, the SBC2 is filled with 2 water molecules.

Figure 3.2.5.1. RSB region of the KR2 in different states. a. O-state of KR2 (present work). b. The compact conformation of KR2 (PDB ID: 4XTN, chain ‘I’). Cavities are shown with pink surfaces. The retinal cofactor is colored teal. Water molecules are shown with red spheres. Na+ is shown with a purple sphere. Hydrogen bonds are shown with black dashed lines. The lengths of the D116-RSB hydrogen bond are shown with bold italic numbers and are in Å. Helices C and G are indicated with capital bold letters. Helices A and B are hidden for clarity.

Four water molecules, which fill the SBC1 in the ground state (Figure 3.2.5.2), are displaced in the O-state. Two of them are placed in the small transiently-formed cavity in the intermediate near S70 at the pentamerization interface; one remains at the same place and is coordinated by N112 and D251; and the last one is relocated to the SBC2 near L75 and R109 at the inner extracellular part of the protein (Figure 3.2.5.1, 3.2.5.2). In course of Na+ uptake and formation of the O-state, the side chains of L74 and N112 residues flips synchronously to avoid the steric conflict of these two residues. This observation demonstrates that the compact conformation of KR2 occurs not only in the ground state of the protein upon dehydration or acidification of the surrounding media. The compact conformation seems, therefore, to be vital for KR2 functioning.

Figure 3.2.5.2. Overall comparison of the ground and O-states of KR2. a, d. Side view of the KR2 protomer in the ground (yellow, PDB ID: 6REW) and O- (blue, present work) states. b, e. View from the side of the helices A and B. Membrane hydrophobic/hydrophilic boundaries are shown with the black lines. The membrane boundary at the extracellular side is located at two levels for the inner and outer parts of the KR2 pentamer, respectively. Helices A and B face the concave aqueous basin, formed in the central pore of the pentamer and helices C–G face the lipid bilayer, surrounding the pentamer. Water molecules are shown as yellow and blue spheres for ground and O-state, respectively. Helices A and B are hidden for clarity. c, f. Detailed view of the RSB region of the ground and the O-state of KR2. Cavities (ion-uptake cavity — IUC; the Schiff base cavities 1 and 2 — SBC1 and SBC2, respectively; putative ion-release cavities 1 and 2 — pIRC1 and pIRC2, respectively) inside the protein are shown in pink and marked with red labels. Retinal cofactor is colored teal. Water molecules are shown with red spheres. Na+ ion is shown with a purple sphere. Hydrogen bonds involving S70, N112, D116, D251, and RSB are shown with black dashed lines. The lengths of the shown hydrogen bonds are shown with bold italic numbers and are in Å. Helix A and SBC2 are hidden for clarity.

As the rearrangement of L74-N112 pair is the most significant event during the expanded- to-compact conformational switch in KR2, the study of the role of the pair in the protein functioning is critical. The role of N112 has already been investigated by extensive mutational analysis. It was shown that the KR2 variants with the substitution of the asparagine at the 112th position of KR2 by the glycine, aspartate, serine, and threonine retain Na+-pumping activity. However, even in those cases, the activity was lowered notably compared to the wild type protein. The substitutions of the N112 to other amino acid residues resulted either in the loss of only Na+- pumping regime with the conservation of the H+-pumping ability or in the complete loss of any transport activity. To study the role of L74 residue, we performed additional mutational analysis. Particularly, we introduced L74A mutation into the KR2 gene and expressed the protein in E. coli cells for functional studies. The tests indicated that L74A substitution dramatically decreases the pumping activity of the protein (Figure 3.2.5.3). Hence, this additionally supports the importance of the flipping motion of L74 and therefore conformational compact-to-expanded and expanded- to-compact switches for Na+ pumping by KR2.

Figure 3.2.5.3. E. coli activity tests of KR2 and its mutants. pH changes upon illumination in the media containing KR2-expressing E. coli cells. The solutions contain 100 mM NaCl and 30 µM CCCP (magenta). The cells were illuminated for 300 s (light area on the plots).

3.2.6. Crystal structure of the ground and the O-states of KR2 at room temperature

It is well-known that freezing of the protein crystals may affect the conformation of the molecule. To verify that in our case freezing does not affect the KR2 conformations in the ground and the O-state we collected diffraction data on KR2 at 293 K using single crystals placed in the stream of humid air (relative humidity of 85%) and solved the structures of the dark (ground) and illuminated (49% ground : 51% O-state) states of the protein at 2.5 and 2.6 Å, respectively (Appendix Table 5.3). We also used serial millisecond crystallography (SMX) approach to collect the data at 2.5 Å on the steady-state activated KR2 using microcrystals of the protein injected into the X-ray beam of the synchrotron source in the stream of mesophase. The data allowed us to solve the steady-state-SMX activated structure of KR2 at 2.7 Å (Appendix Table 5.3). To activate the proteins in crystals, they were illuminated continuously by 532 nm laser during X-ray data collection. This approach allows the accumulation of the dominant intermediate of protein photocycle111, which in the case of KR2 is the O-state. Analysis of the electron density maps and occupancies refinement indicated that such procedure results in the 51 and 50 % occupancy of the O-state in case of single-crystal and SMX approaches, respectively.

Figure 3.2.6.1. Structural alignment of KR2 protomers at 100K and 293 K. A. Overall alignment. N-terminal α-helix and N terminus are colored blue. BC loop, containing the β-sheet, 83 is colored orange. Membrane core boundaries are shown with black lines. KR2 ground state model at 100 and 293 K are colored yellow and red, respectively. Retinal cofactor is colored teal. B. Enlarged view of the most notable rearrangements in protomer backbone. Helices are indicated with bold capital letters. Residues comprising the E-F and C-D loops are labeled. The shift of the helix E is shown with the black arrow and the distance is indicated.

Overall, the structures of the ground and the O-states of KR2 at 100 and 293 K are nearly identical (Figure 3.2.6.1, 3.2.6.2). The RMSD between the structures of the ground state of KR2 at 100 and 293 K is 0.2 Å, and between those of the O-state is also 0.2 Å. Comparison of the KR2 structures identified slight shifts of the positions of the E-F and F-G loops and also cytoplasmic parts of helices E and F by only 0.4 Å (Figure 3.2.6.1). The orientations of all key residues and the arrangement of water molecules inside the protein are similar in models at 100 and 293 K (Figure 3.2.6.2).

Figure 3.2.6.2. RSB region of the ground and the O-states of KR2 at 100 K and 293 K. A. The ground state of KR2 at 100 K (PDB ID: 6REW). B. The ground state of KR2 at 293 K (present work). C. The O-state of KR2 at 100 K (present work). D. The O-state of KR2 at 293 K obtained using single-crystal crystallography (present work). E. The O-state of KR2 at 293 K obtained using serial crystallography (present work). Retinal cofactor is colored teal. Water molecules are shown with red spheres. Na+ is shown with a purple sphere. Hydrogen bonds are shown with black dashed lines. Helix A is hidden for clarity. F, G. Electron density maps of the Na+ binding site in the O- + state at 293 K (black, 2Fo-Fc at the level of 1.0 σ; green, polder difference maps omitting Na are contoured at the level of 4.0 σ and 3.0 σ for the single crystal (F) and serial crystallography (G), respectively).

3.2.7. Conformational switches guide Na+ uptake and release in KR2

The similarity of protein conformation in the O-state to the compact is intriguing. However, it could easily be understood. For that, one need to consider the relative location of the protonated RSB and Na+-D116 pair. In the O-state the overall distribution of the charges in the core of the protein is nearly identical to that of the KR2 with protonated D116 (Figure 3.2.7.1). The protonation of D116 occurs naturally upon acidification of the surrounding media. The compact conformation of KR2 were also found only at low pH values. As we suggested earlier, the compact conformation may appear in response to the D116 neutralization45. To understand better the nature of conformational switches in KR2 and the influence of D116 protonation on the protein conformation, we decided to study KR2-D116N, which mimics the WT protein with fully protonated D116. We solved its structure under physiological conditions (pH 8.0) in the pentameric form at 2.35 Å.

Figure 3.2.7.1. The scheme of charge distribution in the RSB region in different KR2 structures. Blue and red are for the positively and negatively charged elements, respectively. The expanded conformation appears in response to the total neutral charge of the D116-RSB pair. The compact state appears in response to the addition of a positive charge (either H+ or Na+) to the D116-RSB pair.

Confirming our hypothesis, the structure of KR2-D116N shows that introduction of asparagine at the position of D116 led to the flip of the side chains of N112 and L74 residues compared to the ground state of the wild type KR2. The SBC1 is absent in the mutant (Figure 3.2.7.2a). Thus, the structure of D116N is very similar to that of the O-state (RMSD 0.2 Å) and also to the compact conformation (RMSD 0.3 Å) of the wild type protein (Figure 3.2.7.2b). However, Na+ ion is not found inside the protomers of the mutant and the relative orientation of the RSBH+ and N116 is altered (Figure 3.2.7.3).

Figure 3.2.7.2. Comparison of the ground structures of WT KR2 and D116N mutant. a. RSB region in the ground state of WT KR2 (expanded conformation, PDB ID: 6REW). b. RSB region in the ground state of KR2-D116N (similar to compact conformation, present work). Cavities are shown with pink surfaces. Retinal cofactor is colored teal. Water molecules are shown with red spheres. Na+ is shown with a purple sphere. Hydrogen bonds are shown with black dashed lines. The lengths of the D116-RSB hydrogen bond are shown with bold italic numbers and are in Å. Helices C and G are indicated with capital bold letters. Helices A and B are hidden for clarity.

Particularly, the RSBH+ forms two alternative conformations, and the hydrogen bond between the RSBH+ and N116 is absent. Indeed, the electron densities around retinal and K255 in the D116N mutant suggest the coexistence of two orientations of the RSBH+ (Figure 3.2.7.3). In one of them, similar to the O-state and compact conformation of KR2, RSBH+ is pointed towards N116. However, there is no hydrogen bond between them, and the distance between RSBH+ and N116 is 4.3 Å. Likely, there is a hydrogen bond to N112 in this conformation. In the second conformation, RSBH+ is shifted closer to D251 (3.1 Å) and may form a hydrogen bond with the residue (Figure 3.2.7.2). In both conformations, retinal remains in the all-trans configuration (Figure 3.2.7.3). Importantly, such an organization of the RSBH+ region is in line with the existing spectroscopic and NMR data on D116N mutant43. Indeed, it was shown, that the hydrogen bond 87 between the RSBH+ and D116- exists in only a fraction of this KR2 variant. Moreover, as it was demonstrated by the NMR studies, the RSBH+ is in multiple conformations in D116N43. Therefore, we refined our crystallographic data of the mutant with both alternative RSBH+ orientations in the final model (Figure 3.2.7.3).

Figure 3.2.7.3. Examples of electron densities of the KR2-D116N. A. Polder maps for K255 and retinal cofactors of all five protomers of KR2-D116N structure. Maps are contoured at the level of 4.0 σ. B. 2Fo-Fc electron density maps around K255 and retinal cofactor contoured at the level of 1.5 σ. Retinal and K255 are colored teal. Hydrogen bond between the RSB and D251 is shown with black dashed lines. The distance between the RSB and nearest atom of N116 is shown with a black arrowed line. The lengths of the hydrogen bond and the distance are shown with bold italic numbers and are in Å.

We also observed that D116 protonation destabilizes the pentameric assembly of KR2. Indeed, as we showed previously, the oligomeric state of the KR2 is pH-dependent in the detergent micelles and also in the crystals grown from the lipidic cubic phase. KR2 is organized into 88 pentamers at pH values higher than 6-6.5, while the monomers are found at acidic pH. The basis of pentamer disruption at low pH remains unclear. One of the hypotheses is the influence of pH on the rechargeable residues of the oligomerization interface, such as H30. It was demonstrated that pentameric assembly is disturbed in H30K and H30L mutants. However, the H30A mutant remained pentameric. This suggests another mechanism of oligomer dissociation. As it is shown, with the pH decrease not only the pentameric assembly is affected, but also protonation of D116 occurs. For the wild type protein, the shift from the pentameric to monomeric state appears detergent at pH 5-6, which is also close to the pKa of D116. Consequently, it is natural to suggest that the protonation of D116 may influence the oligomeric state of the protein and lead to the pentamer dissociation. To check the hypothesis, we studied oligomerization of the D116N in detergent micelles using SEC. The mutant D116N imitates the protein with protonated D116 at all pH values. We showed that at pH 8.0, where KR2 forms pentamers, D116N is observed in several oligomeric states. Notable portions of both smaller and bigger oligomers are present in the solution. As follows from the analysis of the D116N oligomerization dependence on pH, the smaller oligomers dominate at pH 6, but some intermediate-sized oligomers (smaller than pentamers of the wild type protein) appear as pH is lowered to 4.3 (Figure 3.2.7.4). Altogether, this allows us to suggest that the protonation of D116 and presumably binding of Na+ near D116 in the O-state affect the oligomeric state of KR2 and is one of the driving forces of the destabilization of the KR2 pentamer. Consequently, this additionally supports the fact that KR2 protomers are distorted in the O-state, and the pentamerization interface could also be affected. Thus, pentameric assembly not only plays a key role in the organization of the expanded conformation, important for the Na+ release but also stabilizes the proteins in the O-state with Na+ bound in close proximity of the RSBH+. Summarizing, we suggest that neutralization of D116 is the key determinant of the formation of the compact conformation, which explains their structural similarity.

Figure 3.2.7.4. Size exclusion chromatography profiles of D116N mutant of KR2. Protein with an initial concentration of 70 mg/ml was dissolved in buffer solution containing 200 mM NaCl with 0.1% DDM to a final concentration of 1 mg/ml and dialyzed against 100x volume of the buffer of needed pH with no less than 5 times substitution of the outer buffer solution during at least 72 hours.

The structures of KR2 O-state and D116N mutant, together with previously described pH dependence of the KR2 organization, allow us to suggest that the compact conformation might stabilize neutralized RSB counterion and correspondingly neutral transiently formed Na+-D116- pair during the photocycle.

The structure of the O-state of KR2 showed that our suggestion that the SBC1 in the expanded conformation surrounded by R109, N112, W113, D116, and D251 might be a transient Na+ binding site in an intermediate state of the protein photocycle is not accurate. In fact, in opposite to our initial hypothesis, not the expanded, but the compact conformation is critical for transient Na+ binding inside the protomer. Our work shows that Na+ binds distant from R109 and D251 (Figure 3.2.5.1a). Since the release of Na+ occurs upon the O-to-ground state transition, which structurally corresponds to compact-to-expanded switch, we suggest that the expanded conformation is important for the ion release to the extracellular space. Importantly, Na+ uptake 90 and release are guided by the switch from the expanded to the compact and then again back to the expanded conformations, respectively.

3.2.8. Na+ translocation pathway

We next analyzed the organization of the cytoplasmic and extracellular inner parts of the protein molecule in the O-state. Although the central part of KR2 protomer is altered in the O- state, the organization of the cytoplasmic ion uptake region remains the same to those in the ground state (Figure 3.2.8.1b). This fact is not surprising. Since the passage of Na+ through the cytoplasmic part does not proceed in the ground state, and is already happened in the O-state, the organization of this region might easily be restored already in the O-state. Moreover, in the O-state Na+ should not have a way to flow back to the cytoplasm. Therefore, the Na+ pathway, presumably formed in the M-state of KR2 photocycle, connecting the IUC with the RSB, should not be optimized for facilitated Na+ diffusion in both the ground and the O-states. Thus, it is expected that the initial conformation of the S64-Q123 pair, separating the IUC from the RSB, is similar in the ground and the O- states.

Figure 3.2.8.1. Ion uptake and release pathways of KR2. a. Section view of KR2 pentamer in the membrane. Concave aqueous basin facing the extracellular space is indicated by the black line. Only one protomer is shown in cartoon representation. Membrane core boundaries are shown with black lines. b. Structural alignment of the cytoplasmic parts of the ground (yellow) and O- (blue) states of KR2. Water molecules are shown with yellow and blue spheres for the ground and O- state, respectively. c. Detailed view of the extracellular side of KR2 in the ground state. d. Detailed view of the extracellular side of KR2 in the O-state. Cavities inside the protein are shown in pink and marked with red labels. Protein surface concavity from the aqueous basin at the extracellular side is colored gray. Retinal cofactor is colored teal. Water molecules are shown with red spheres. 91

Na+ ion is shown with a purple sphere. N-terminal α-helix is colored blue. BC loop is colored orange. H30′ of adjacent protomer is colored with dark-green. Helices A, F, and G are hidden for clarity. Gray arrows identify putative ion uptake and two ion release pathways.

The extracellular part of KR2 protomer in the O-state differs from that in the ground state. Particularly, the region between the R109-D251 pair and Q78 residue is rearranged in the O-state. Surprisingly, the organization of the E11-E160-R243 cluster and pIRC1 in the ground and O-states is identical. The only difference is that the water molecule, placed near the β-ionone ring of the retinal in the ground state, is relocated to the pIRC1 in the O-state. However, this replacement does not affect the conformation of the surrounding residues and the size and the shape of the cavity. This observation is intriguing, since this region was suggested to be a part of the Na+ release pathway to the extracellular bulk during the direct O-to-ground transition. One can expect that the pathway should be prepared for such Na+ passage in prior to the release, e.g. already in the O-state. The absence of the disturbance in this region might mean that the energy stored in the not fully relaxed all-trans retinal in the O-state is enough for direct release of the Na+ to the bulk without any transient binding sites. Another possibility is that there might be another ion release pathway in KR2.

The organization of the extracellular part of KR2 in the O-state suggests that the second hypothesis is more probable. First, it is supported by the mutational analysis, indicating that E11A, E160A or R243A mutations do not abolish Na+-pumping activity; however, they affect the stability of the proteins. Second, the other (alternative) putative way for Na+ release might go through the elongated in the O-state SBC2 to the bulk near the interprotomeric Na+ ion at the surface of KR2 in both the ground and O-states. This pathway is constricted with the only side chain of Q78 residue (Figure 3.2.8.1d) and protrudes from the inner region between Q78, N106, and R109 to the pIRC2 between helices B and C, BC loop, and helix A’ of adjacent protomer at the extracellular side (Figure 3.2.8.2). The pIRC2 is directly connected to a concave aqueous basin facing the extracellular solution, formed in the central pore of the KR2 pentamer. pIRC2 is surrounded by Q78, N81, S85, D102, Y108, and Q26’ residues and filled with water molecules in both the ground and O-states (Figure 3.2.8.2). These residues and waters are rearranged notably in the O-state. For instance, the side chain of N81 is flipped towards H30’ of the adjacent protomer (Figure 3.2.8.2). Consequently, additional water molecule appears in the pIRC2 in the O-state. This water is coordinated by hydrogen bonds with E26’, H30’, and N81 (Figure 3.2.8.2). The positions of Q78 and Y108 are also altered in the O-state (Figure 3.2.8.1d, 3.2.8.2).

Figure 3.2.8.2. Putative ion-release cavity 2 (pIRC2) of the ground and O-states of KR2. A. Structural alignment of the ground (yellow, PDB ID: 6REW) and the O- (blue, present work) states. B. Detail view of the pIRC2 in the ground state. C. Detail view of the pIRC2 in the O-state. Na+ at the extracellular surface of KR2 pentamer are shown with purple spheres. Adjacent protomer is colored green. Hydrogen bonds are shown with black dashed lines. The length of the hydrogen bonds are shown with bold italic numbers and are in Å. Water molecules are shown with small spheres and colored yellow and blue in section A and red in sections B and C.

To validate our suggestion about an alternative ion release pathway, we conducted 10 short metadynamics simulations starting from the Na+-bound conformation (Figure 3.2.8.3). The simulations showed that the positively charged R109 side chain forms a barrier for sodium exiting SCB1, and changes its position to allow sodium passage. This observation is similar to that found in HsBR, where analogous arginine in the helix C (R82) also serves as a barrier for protons in the extracellular part of the protein39. R82 in HsBR flips towards extracellular space in course of the protein photocycle, similar to that found in our simulations of KR2. The R109 role as a barrier for

Na+ ions in KR2 is supported also by the mutational analysis, since R109Q mutation results in the appearance of passive conductance in the rhodopsin73. Upon passing near R109, Na+ either exits the protein via pIRC2 (8 simulations out of 10) or proceeds towards pIRC1 (2 simulations). In the first scenario, the ion is quickly released towards the aqueous basin in the middle of KR2 pentamer in the vicinity of another ion found at the interface between the protomers and is sometimes observed to replace it. In the second scenario, the ion samples different locations around the E11- E160-R243 triad and is later released via N106 and Q157 on the outer side of the pentamer.

Figure 3.2.8.3. Simulated trajectories of Na+ release identified by molecular dynamics simulations. A. Density surface corresponding to the volume accessible to Na+. B. Positions of sodium taken every 100 ps. Each trajectory is shown in a different color; trajectories where sodium exits via pIRC1 are shown in yellow and orange. Two protomers of KR2 pentamer are rendered as purple and green ribbons. X-ray orientations of key amino acid side chains are shown.

We also found that the organization of the pIRC2 region is identical in the O-state and KR2-D116N (Figure 3.2.8.4a). It means that the rearrangements on the surface of the KR2 could be allosterically coupled to the redistribution of the charges in the RSB region. Such long-distance interactions between the RSB region and protein surface were already studied for the WT and H30A mutant of KR2. Moreover, the presence of the allosteric communication between the interprotomer Na+ ion-binding site and the RSB hydrogen bond already in the ground state was recently shown. Thus, structural rearrangements of the RSB-counterion pair upon Na+ release and corresponding compact-to-expanded conformational switch might affect the interprotomer Na+ binding site, promoting Na+ unbinding from the site, observed in metadynamics simulations.

Figure 3.2.8.4. Putative ion-release cavity 2 (pIRC2) of the D116N and H30A mutants of KR2. A. Structural alignment of the O-state of the WT KR2 (blue, present work) and the D116N ground state (pink, present work). The conformations are nearly identical. B. Structural alignment of the ground states of the WT KR2 (yellow, PDB ID: 6REW) and conformation A of H30A mutant (H30AA, light green, present work). C. Structural alignment of the ground states of the WT KR2 (yellow, PDB ID: 6REW) and conformation B of H30A mutant (H30AB, light green, present work). D. Structural alignment of the ground states of the conformations A and B of H30A mutant.

Additional water molecules, appearing in the structure of the H30AB are identified as wB’ and wB’’. E. Detail view of the pIRC2 of the H30AA. F. Detail view of the pIRC2 of the H30AB. + pIRC2 is notably enlarged in the H30AB. Na at the extracellular surface of KR2 pentamer are shown with purple spheres. Adjacent protomer is colored dark green. Water molecules are shown with small spheres and colored blue and pink in section A, yellow and light green in sections B and C and red in sections D-F.

To gather more details about long-distance interactions between the protein core and surface, we produced and crystallized H30A mutant of KR2 and solved its structure in the pentameric form at pH 8.0 at 2.2 Å.

The overall structure of this mutant is nearly identical to that of the ground state of WT protein (RMSD 0.15 Å) (Figure 3.2.8.5). The organization of their inner cytoplasmic, central, and extracellular parts is exactly the same in the proteins. It should be noted that in contrast to the previous FTIR experiments13, in H30A mutant we found Na+ ion bound at the oligomerization interface. The organization of the ion-binding site is the same as that of the wild-type protein. We

95 suggest that the difference in the FTIR and X-ray crystallography data might originate from the differences in the protein environments during the experiments.

Figure 3.2.8.5. Structural alignment of KR2 protomers in different states. A. Overall alignment. N-terminal α-helix and N terminus are colored blue. BC loop, containing the β-sheet, is colored orange. Membrane core boundaries are shown with black lines. B. Enlarged view of the most notable rearrangements in protomer backbone. Helices are indicated with capital bold letters. The ground state of KR2 (the expanded conformation, PDB ID: 6REW) is colored yellow. O-state of KR2 (present work) is colored light-blue. KR2-D116N (present work) is colored pink. KR2- H30A (present work) is colored green. The ground state of KR2 (the compact conformation, PDB ID: 4XTN, chain ‘I’) is colored orange. The ground state of monomeric KR2 (PDB ID: 4XTL) is colored cyan.

In the KR2-H30A, the region of H30 is altered (Figure 3.2.8.4). Particularly, the bulky H30 side chain, absent in the mutant, is replaced by two additional water molecules (wB’ and wB’’) (Figure 3.2.8.4F). In H30A mutant we found double conformation of Y108. While the first is the same to that in the ground state of the wild type protein, the second was not identified in any other structures of KR2 or its variants (Figure 3.2.8.4). Importantly, it was shown that H30A is more selective to Na+ and almost does not pump H+13. Since the only existing differences in the structures of the wild type protein and the mutant occur near Q78, N81, Y108, H30’ and pIRC2,

96 we suggest that this region is important for cation selectivity. This additionally supports the hypothesis that this region is a part of ion release pathway.

To verify the suggested ion-release pathway we engineered several functional mutants of KR2 and test their Na+-pumping activity in E. coli cells suspension (Figure 3.2.5.3). The results showed that Q78 is the key residue, which is likely to act as a gate for Na+. Indeed, the Q78A mutant, similar to Q123A, has a weak Na+-pumping activity; however, it is decreased notably in comparison to the wild type protein. The flipping motion of the Q78 might play a critical role for Na+ passage, since Q78L mutant remains almost fully functional. On the other hand, the blocking of Q78 motion (Q78Y,W mutations) resulted in a dramatic decrease in the Na+-pumping activity. Another interesting finding was that the Y108A mutant almost fully lost its pumping ability.

Therefore, we suggest that the Na+ translocation pathway propagates from IUC to pIRC2 via a chain of polar inner cavities, modified during photocycle. IUC and pIRC2 are separated from the inside of KR2 by two weak gates in the cytoplasmic and extracellular parts of the protomer, formed by Q123 and Q78, respectively. This makes the KR2 ion pathway similar to that of the CrChR2 (Figure 3.2.8.6). However, unlike in CrChR2, KR2 has the Na+ binding site in the central region near the RSB. Interestingly, the O-state is not formed in KR2 variants with engineered passive conductance, such as R109Q112. Thus, it could be suggested that the transient Na+ binding site is also absent in these mutants. This additionally highlights the importance of the site for the unidirectional transport of the cation, distinguishing light-driven pumps from light-gated channels.

Figure 3.2.8.6. Comparison of ion pathways in different classes of microbial rhodopsins. A. The O-state of the KR2 (present work). B. The ground state of the HsBR (PDB ID: 1C3W41). C. The ground state of the CrChR2 (PDB ID: 6EID60). Cavities are shown with a pink surface. Gray lines indicate membrane hydrophobic/hydrophilic boundaries. Gray arrows indicate ion translocation pathways. In the case of KR2 1 is for the Na+ uptake and 2 is for Na+ release. In the case of HsBR 1 is for proton translocation from the RSB to D85, 2 is for proton release from E194- 97

E204 pair to extracellular space, 3 is for proton translocation from D96 to the RSB, 4 is for the D96 reprotonation from the cytoplasmic space and 5 is for the proton relocation from D85 to E194- E204 pair. In the case of CrChR2 gray arrow indicates the ion pathway through the channel.

3.3. Molecular mechanism of light-driven Na+ pumping

Crystal structures and spectroscopy data of the O-state and D116N and H30A mutants of KR2 together with available literature data allow us to propose a mechanism of protein functioning (Figure 3.3.1). We suggest that it has three main steps.

Figure 3.3.1. Proposed Na+ pumping mechanism. A schematic side section view of the KR2 pentamer is shown. Membrane core boundaries are shown with black lines. Cavities are demonstrated as white ellipses. An enlarged view of the RSB region is shown in the left part of the pentamer. The 13-cis configuration of the retinal cofactor is modeled manually for the schematic representation. Na+ is shown with violet spheres. Black arrows indicate proposed Na+ uptake and release pathways. Violet arrows indicate rearrangement of the N112 side chain during expanded-to-compact and back compact-to-expanded switches. Small gray arrows indicate the translocation of the hydrogen from the RSB to D116 during the formation of the M-state and following reprotonation of the RSB from the D116 in the M-to-O transition. The retinal cofactor is colored teal. Waters are shown with red spheres. Hydrogen bonds in the RSB region are shown with black dashed lines.

Step 1. In the ground state KR2 is in the expanded conformation. The large polar SBC1 is filled with four water molecules. The RSB is protonated and hydrogen-bonded to its deprotonated 99 counterion D116. The gate near Q123-S64 at the cytoplasmic part of the protein separates the IUC from the RSB region. After the absorption of the light photon, the retinal isomerizes from all-trans to 13-cis configuration. The red-shifted K-state appears in nanoseconds, followed by the formation of the coexisting L and M intermediates in about 30 μs. The proton is translocated from the RSB to the D116 with the formation of the M-state. Consequently, the hydrogen bond between the RSB and D116 is absent in this intermediate. The gate near Q123-S64 also opens in this step.

Step 2. With the decay of the M-state and the rise of the O-state Na+ flows inside the KR2 molecule from the cytoplasm. It passes the gate near Q123-S64 and neutralized RSB-D116 pair and binds between S70, N112 and D116 residues. Na+ uptake presumably causes the proton translocation from D116 back to the RSB with the restoration of the RSB-counterion hydrogen bond, which prevents the Na+ backflow to the cytoplasmic side. The cytoplasmic part of KR2 and Q123-S64 is also restored to its ground conformation at this step. Na+ binding also results in the flip of the N112 side chain for the stabilization of Na+-D116- pair. L74 reorients synchronously with N112 to avoid the steric conflict between these residues. Therefore, the compact conformation of KR2 appears at this step. Retinal is already in the all-trans conformation in the O-state; however, it is not fully- relaxed to its initial configuration.

Step 3. With the decay of the O-state retinal returns completely to its ground configuration. The expanded conformation of KR2 is formed, which causes the flip of the N112 residue towards the pentamerization interface. This reorientation of N112 opens the way for Na+ release from the transient ion-binding site. The Na+ release occur through the prepared in the O-state pathway at the extracellular part of KR2 protomer. We suggest that the pathway is constricted only by the Q78 side chain and ion release proceeds towards the cavity, formed by N81, Y108, and H30’ of adjacent protomer (pIRC2). The Na+ release might proceed using the relay mechanism, similar to Grotthuss mechanism for proton transport in HsBR. This mean that the Na+ ion, released from the KR2 protomer in the current working cycle, replaces the ion, bound at the protein oligomerization interface. This allows lowering the energy barriers for the facilitation of the Na+ release. The E11- E160-R243 cluster and the pIRC1 cavity, suggested earlier as a part of Na+ release pathway, thus may be involved mainly in protein stabilization.

We also suggest thar water molecules, found in the SBC1 in the ground state of KR2 photocycle might be involved in the Na+ hydration during its transitions inside the protein.

Last but not least, in the absence of Na+ KR2 acts as a H+ pump with the significantly altered photocycle and the absence of the pronounced O-state. The long-living L/M/O-like state decays 100 slowly in the H+-pumping mode. We note that this is in agreement with the suggested mechanism of Na+ pumping. We suggest that the formation of the K-, L- and M-states is the same for the Na+- and H+-pumping modes. However, in the absence of Na+ the ion does not flow diffusely to the central region in the M-state when the RSB is neutral. Consequently, the RSB is presumably reprotonated from the aqueous phase in the UIC at the cytoplasmic side of the protein and therefore no rearrangements occur in the region of D116. Then the slow relaxation of the retinal to its original all-trans configuration triggers the H+ release from the D116 to the extracellular side and the protein returns to the ground state.

101

3.4. Outlook

The presented work describes the ground and O-states of the biologically relevant pentameric form of light-driven Na+ pump KR2. However, the structures of the rest intermediate states, such as K, L, and M are missing. Although the structures of all intermediate states were recently published for the monomeric form of KR2, the present work showed that mechanisms of light-driven Na+ pumping in monomeric and pentameric forms of KR2 are notably different. There are only little doubts that KR2 is pentameric in vivo. Therefore, to decipher the natural mechanism of light-driven Na+ pumping future plans are to solve the structures of KR2 in the K, L, and M states. For that, one needs to use TR-SFX at XFELs. Importantly, the L/M state appears at μs time scale and thus could also be captured at modern synchrotron sources, like Extremely Bright Source (EBS), ESRF, Grenoble.

Another part of the future work on KR2 is the determination of the key elements for ion selectivity in the protein. The transient Na+ ion-binding site in the O-state of KR2 is optimized for the Na+. On the other hand, G263F and G263W mutations, located far from the site, provide KR2 with the ability to effectively pump K+. The mechanism of these conversions remains unknown. Therefore, it is planned to study the structures of the intermediate states of KR2 mutants to reveal the principles of ion selectivity in the protein.

Last but not least, we intend to investigate structurally and functionally other NaRs, since there could be a significant difference between the members of the family. Particularly, the structure of SrNaR (or similar protein) should be interesting, since the protein has different spectroscopic properties from KR2. It is also important to understand the evolutionary pathways of NaRs. For this, it is planned to solve the structures of NaRs from different domains of life. It might be critical for the understanding of mechanisms of light-driven non proton cation pumping.

102

4. References

1. Gómez-Consarnau, L. et al. Microbial rhodopsins are major contributors to the solar energy captured in the sea. Sci. Adv. (2019) doi:10.1126/sciadv.aaw8855.

2. Henderson, R. & Unwin, P. N. T. Three-dimensional model of purple membrane obtained by electron microscopy. Nature 257, 28–32 (1975).

3. Gushchin, I. & Gordeliy, V. Microbial Rhodopsins. in 19–56 (2018). doi:10.1007/978-981-10-7757-9_2.

4. Jin, Y. et al. Bacteriorhodopsin as an electronic conduction medium for biomolecular electronics. Chem. Soc. Rev. 37, 2422–2432 (2008).

5. Mak-Jurkauskas, M. L. et al. Energy transformations early in the bacteriorhodopsin photocycle revealed by DNP-enhanced solid-state NMR. Proc. Natl. Acad. Sci. 105, 883–888 (2008).

6. Oesterhelt, D. & Stoeckenius, W. Rhodopsin-like protein from the purple membrane of Halobacterium halobium. Nature. New Biol. 233, 149–152 (1971).

7. Matsuno-Yagi, A. & Mukohata, Y. Two possible roles of bacteriorhodopsin; a comparative study of strains of Halobacterium halobium differing in pigmentation. Biochem. Biophys. Res. Commun. (1977) doi:10.1016/0006-291X(77)91245-1.

8. Schobert, B. & Lanyi, J. K. Halorhodopsin is a light-driven chloride pump. J. Biol. Chem. 257, 10306–10313 (1982).

9. Beja, O. et al. Bacterial rhodopsin: Evidence for a new type of phototrophy in the sea. Science 289, 1902–1906 (2000).

10. Nagel, G. et al. Channelrhodopsin-1: A light-gated proton channel in green algae. Science 296, 2395–2398 (2002).

11. Nagel, G. et al. Channelrhodopsin-2, a directly light-gated cation-selective membrane channel. Proc. Natl. Acad. Sci. (2003) doi:10.1073/pnas.1936192100.

12. Boyden, E. S., Zhang, F., Bamberg, E., Nagel, G. & Deisseroth, K. Millisecond- timescale, genetically targeted optical control of neural activity. Nat. Neurosci. (2005) doi:10.1038/nn1525.

103

13. Inoue, K. et al. A light-driven sodium ion pump in marine bacteria. Nat. Commun. 4, (2013).

14. Yoshizawa, S. et al. Functional characterization of flavobacteria rhodopsins reveals a unique class of light-driven chloride pump in bacteria. Proc. Natl. Acad. Sci. U. S. A. (2014) doi:10.1073/pnas.1403051111.

15. Govorunova, E. G., Sineshchekov, O. A., Janz, R., Liu, X. & Spudich, J. L. Natural light-gated anion channels: A family of microbial rhodopsins for advanced optogenetics. Science (2015) doi:10.1126/science.aaa7484.

16. Yutin, N. & Koonin, E. V. Proteorhodopsin genes in giant viruses. Biol. Direct (2012) doi:10.1186/1745-6150-7-34.

17. Inoue, K. et al. A natural light-driven inward proton pump. Nat. Commun. (2016) doi:10.1038/ncomms13415.

18. Shevchenko, V. et al. Inward H + pump xenorhodopsin: Mechanism and alternative optogenetic approach. Sci. Adv. (2017) doi:10.1126/sciadv.1603187.

19. Schizorhodopsins: A family of rhodopsins from Asgard archaea that function as light-driven inward H+ pumps | Science Advances. https://advances.sciencemag.org/content/6/15/eaaz2441.

20. Avelar, G. M. et al. A Rhodopsin-Guanylyl cyclase gene fusion functions in visual perception in a fungus. Curr. Biol. 24, 1234–1240 (2014).

21. Pushkarev, A. et al. A distinct abundant group of microbial rhodopsins discovered. Nature (2018) doi:10.1038/s41586-018-0225-9.

22. Wickstrand, C., Dods, R., Royant, A. & Neutze, R. Bacteriorhodopsin: Would the real structural intermediates please stand up? Biochimica et Biophysica Acta - General Subjects (2015). doi:10.1016/j.bbagen.2014.05.021.

23. Yoshizawa, S., Kawanabe, A., Ito, H., Kandori, H. & Kogure, K. Diversity and functional analysis of proteorhodopsin in marine Flavobacteria. Environ. Microbiol. 14, 1240– 1248 (2012).

24. Balashov, S. P. et al. Light-driven Na+pump from Gillisia limnaea: A high-affinity Na+binding site is formed transiently in the photocycle. Biochemistry (2014) 104 doi:10.1021/bi501064n.

25. Bertsova, Y. V., Bogachev, A. V. & Skulachev, V. P. Proteorhodopsin from Dokdonia sp. PRO95 is a light-driven Na+-pump. Biochem. Mosc. 80, 449–454 (2015).

26. Li, H., Sineshchekov, O. A., Da Silva, G. F. Z. & Spudich, J. L. In vitro demonstration of dual light-driven Na+/H+ pumping by a microbial rhodopsin. Biophys. J. 109, 1446–1453 (2015).

27. Kwon, Y. M., Kim, S.-Y., Jung, K.-H. & Kim, S.-J. Diversity and functional analysis of light-driven pumping rhodopsins in marine Flavobacteria. MicrobiologyOpen 5, 212– 223 (2016).

28. Kwon, S.-K., Lee, H. G., Kwak, M.-J. & Kim, J. F. Genome sequence of the marine flavobacterium Croceitalea dokdonensis DOKDO 023 that contains proton- and sodium-pumping rhodopsins. Mar. Genomics 26, 1–3 (2016).

29. Zhao, H. et al. Coexistence of light-driven Na+ and H+ transport in a microbial rhodopsin from Nonlabens dokdonensis. J. Photochem. Photobiol. B 172, 70–76 (2017).

30. Tsunoda, S. P. et al. Functional characterization of sodium-pumping rhodopsins with different pumping properties. PLoS ONE (2017) doi:10.1371/journal.pone.0179232.

31. Bork, P. et al. Tara Oceans studies plankton at Planetary scale. Science (2015). doi:10.1126/science.aac5605.

32. Dimroth, P., Kornberg, H. L. & Henderson, P. J. F. Mechanisms of sodium transport in bacteria. Philos. Trans. R. Soc. Lond. B Biol. Sci. 326, 465–477 (1990).

33. Häse, C. C. & Barquera, B. Role of sodium bioenergetics in Vibrio cholerae. Biochim. Biophys. Acta BBA - Bioenerg. 1505, 169–178 (2001).

34. Mulkidjanian, A. Y., Galperin, M. Y., Makarova, K. S., Wolf, Y. I. & Koonin, E. V. Evolutionary primacy of sodium bioenergetics. Biol. Direct 3, 13 (2008).

35. Shalaeva, D. N., Galperin, M. Y. & Mulkidjanian, A. Y. Eukaryotic G protein- coupled receptors as descendants of prokaryotic sodium-translocating rhodopsins. Biol. Direct (2015) doi:10.1186/s13062-015-0091-4.

36. Shalaeva, D. N., Cherepanov, D. A., Galperin, M. Y., Vriend, G. & Mulkidjanian,

105

A. Y. G protein-coupled receptors of class A harness the energy of membrane potential to increase their sensitivity and selectivity. Biochim. Biophys. Acta BBA - Biomembr. 1861, 183051 (2019).

37. Mamedov, A. M. et al. Identification of the key determinant of the transport promiscuity in Na+-translocating rhodopsins. Biochem. Biophys. Res. Commun. 499, 600–604 (2018).

38. Kato, Y., Inoue, K. & Kandori, H. Kinetic Analysis of H+-Na+ Selectivity in a Light-Driven Na+-Pumping Rhodopsin. J. Phys. Chem. Lett. (2015) doi:10.1021/acs.jpclett.5b02371.

39. Gerwert, K., Freier, E. & Wolf, S. The role of protein-bound water molecules in microbial rhodopsins. Biochimica et Biophysica Acta - Bioenergetics (2014). doi:10.1016/j.bbabio.2013.09.006.

40. Nango, E. et al. A three-dimensional movie of structural changes in bacteriorhodopsin. Science 354, 1552–1557 (2016).

41. Luecke, H., Schobert, B., Richter, H. T., Cartailler, J. P. & Lanyi, J. K. Structure of bacteriorhodopsin at 1.55 \AA resolution. J. Mol. Biol. 291, 899–911 (1999).

42. Nishimura, N., Mizuno, M., Kandori, H. & Mizutani, Y. Distortion and a Strong Hydrogen Bond in the Retinal Chromophore Enable Sodium-Ion Transport by the Sodium-Ion Pump KR2. J. Phys. Chem. B (2019) doi:10.1021/acs.jpcb.9b00928.

43. Shigeta, A. et al. Solid-state nuclear magnetic resonance structural study of the retinal-binding pocket in sodium ion pump rhodopsin. Biochemistry (2017) doi:10.1021/acs.biochem.6b00999.

44. Gushchin, I. et al. Crystal structure of a light-driven sodium pump. Nat. Struct. Mol. Biol. 22, 390–396 (2015).

45. Kovalev, K. et al. Structure and mechanisms of sodium-pumping KR2 rhodopsin. Sci. Adv. 5, eaav2671 (2019).

46. FTIR Spectroscopy of a Light-Driven Compatible Sodium Ion-Proton Pumping Rhodopsin at 77 K | The Journal of Physical Chemistry B. https://pubs.acs.org/doi/10.1021/jp500756f.

47. Tomida, S., Ito, S., Inoue, K. & Kandori, H. Hydrogen-bonding network at the 106 cytoplasmic region of a light-driven sodium pump rhodopsin KR2. Biochim. Biophys. Acta BBA - Bioenerg. 1859, 684–691 (2018).

48. Kato, H. E. et al. Structural basis for Na + transport mechanism by a light-driven Na + pump. Nature 521, 48–53 (2015).

49. Inoue, K., Konno, M., Abe‐Yoshizumi, R. & Kandori, H. The Role of the NDQ Motif in Sodium-Pumping Rhodopsins. Angew. Chem. 127, 11698–11701 (2015).

50. Brown, L. S. & Ernst, O. P. Recent advances in biophysical studies of rhodopsins – Oligomerization, folding, and structure. Biochimica et Biophysica Acta - Proteins and Proteomics (2017). doi:10.1016/j.bbapap.2017.08.007.

51. Shibata, M. et al. Oligomeric states of microbial rhodopsins determined by high- speed atomic force microscopy and circular dichroic spectroscopy. Sci. Rep. 8, (2018).

52. Pescitelli, G. et al. Exciton Circular Dichroism in Channelrhodopsin. J. Phys. Chem. B 118, 11873–11885 (2014).

53. Bayburt, T. H., Grinkova, Y. V. & Sligar, S. G. Assembly of single bacteriorhodopsin trimers in bilayer nanodiscs. Arch. Biochem. Biophys. 450, 215–222 (2006).

54. Kaur, J. et al. Solid-state NMR analysis of the sodium pump Krokinobacter rhodopsin 2 and its H30A mutant. J. Struct. Biol. (2018) doi:10.1016/j.jsb.2018.06.001.

55. Morizumi, T. et al. X-ray Crystallographic Structure and Oligomerization of Gloeobacter Rhodopsin. Sci. Rep. (2019) doi:10.1038/s41598-019-47445-5.

56. Bratanov, D. et al. Unique structure and function of viral rhodopsins. Nat. Commun. 10, 4939 (2019).

57. Hunt, J. F., McCrea, P. D., Zaccaı̈ , G. & Engelman, D. M. Assessment of the aggregation state of integral membrane proteins in reconstituted phospholipid vesicles using small angle neutron scattering11Edited by M. F. Moody. J. Mol. Biol. 273, 1004–1019 (1997).

58. Kolbe, M., Besir, H., Essen, L. O. & Oesterhelt, D. Structure of the light-driven chloride pump halorhodopsin at 1.8 \AA Resolution. Science (2000) doi:10.1126/science.288.5470.1390.

59. Yun, J. H. et al. Non-cryogenic structure of a chloride pump provides crucial clues

107 to temperature-dependent channel transport efficiency. J. Biol. Chem. (2019) doi:10.1074/jbc.RA118.004038.

60. Volkov, O. et al. Structural insights into ion conduction by channelrhodopsin 2. Science 358, (2017).

61. Ran, T. et al. Cross-protomer interaction with the photoactive site in oligomeric proteorhodopsin complexes. Acta Crystallogr. D Biol. Crystallogr. (2013) doi:10.1107/S0907444913017575.

62. Needham, D. M. et al. A distinct lineage of giant viruses brings a rhodopsin photosystem to unicellular marine predators. Proc. Natl. Acad. Sci. (2019) doi:10.1073/pnas.1907517116.

63. Kovalev, K. et al. High-resolution structural insights into the heliorhodopsin family. Proc. Natl. Acad. Sci. U. S. A. (2020) doi:10.1073/pnas.1915888117.

64. Gushchin, I. et al. Structure of the light-driven sodium pump KR2 and its implications for optogenetics. FEBS J. 283, 1232–1238 (2016).

65. Otomo, A., Mizuno, M., Inoue, K., Kandori, H. & Mizutani, Y. Allosteric Communication with the Retinal Chromophore upon Ion Binding in a Light-Driven Sodium Ion- Pumping Rhodopsin. Biochemistry (2020) doi:10.1021/acs.biochem.9b01062.

66. Baldwin, J. M., Henderson, R., Beckman, E. & Zemlin, F. Images of purple membrane at 2.8 Å resolution obtained by cryo-electron microscopy. J. Mol. Biol. 202, 585–591 (1988).

67. Dencher, N. A. & Heyn, M. P. Bacteriorhodopsin monomers pump protons. FEBS Lett. (1979) doi:10.1016/0014-5793(79)80552-9.

68. Skopintsev, P. et al. Femtosecond-to-millisecond structural changes in a light- driven sodium pump. Nature 1–5 (2020) doi:10.1038/s41586-020-2307-8.

69. Kovalev, K. et al. Molecular mechanism of light-driven sodium pumping. Nat. Commun. 11, 2137 (2020).

70. Konno, M. et al. Mutant of a Light-Driven Sodium Ion Pump Can Transport Cesium Ions. J. Phys. Chem. Lett. (2016) doi:10.1021/acs.jpclett.5b02385.

108

71. Inoue, K., Nomura, Y. & Kandori, H. Asymmetric functional conversion of eubacterial light-driven ion pumps. J. Biol. Chem. 291, 9883–9893 (2016).

72. Kaneko, A., Inoue, K., Kojima, K., Kandori, H. & Sudo, Y. Conversion of microbial rhodopsins: insights into functionally essential elements and rational protein engineering. Biophysical Reviews (2017). doi:10.1007/s12551-017-0335-x.

73. Vogt, A. et al. Engineered Passive Potassium Conductance in the KR2 Sodium Pump. Biophys. J. (2019) doi:10.1016/j.bpj.2019.04.001.

74. Hoque, M. R. et al. A Chimera Na+-Pump Rhodopsin as an Effective Optogenetic Silencer. PLOS ONE 11, e0166820 (2016).

75. Grimm, C., Silapetere, A., Vogt, A., Bernal Sierra, Y. A. & Hegemann, P. Electrical properties, substrate specificity and optogenetic potential of the engineered light-driven sodium pump eKR2. Sci. Rep. 8, 9316 (2018).

76. Red-shifting mutation of light-driven sodium-pump rhodopsin | Nature Communications. https://www.nature.com/articles/s41467-019-10000-x.

77. Inoue, H., Nojima, H. & Okayama, H. High efficiency transformation of Escherichia coli with plasmids. Gene 96, 23–28 (1990).

78. Studier, F. W. Protein production by auto-induction in high-density shaking cultures. Protein Expr. Purif. (2005) doi:10.1016/j.pep.2005.01.016.

79. Landau, E. M. & Rosenbusch, J. P. Lipidic cubic phases: A novel concept for the crystallization of membrane proteins. Proc. Natl. Acad. Sci. 93, 14532–14535 (1996).

80. Caffrey, M. & Cherezov, V. Crystallizing membrane proteins using lipidic mesophases. Nat. Protoc. 4, 706–731 (2009).

81. Gordeliy, V. I. et al. Molecular basis of transmembrane signalling by sensory rhodopsin II-transducer complex. Nature 419, 484–487 (2002).

82. Von Stetten, D. et al. In crystallo optical spectroscopy (icOS) as a complementary tool on the macromolecular crystallography beamlines of the ESRF. Acta Crystallogr. D Biol. Crystallogr. (2015) doi:10.1107/S139900471401517X.

83. Kabsch, W. XDS. Acta Crystallogr Biol Crystallogr (2010)

109 doi:10.1107/S0907444909047337.

84. Winn, M. D. et al. Overview of the CCP4 suite and current developments. Acta Crystallogr. D Biol. Crystallogr. 67, 235–242 (2011).

85. Tickle, I. J. et al. STARANISO. Cambridge, United Kingdom: Global Phasing Ltd. (2018).

86. Weierstall, U. et al. Lipidic cubic phase injector facilitates membrane protein serial femtosecond crystallography. Nat. Commun. 5, 3309 (2014).

87. White, T. A. et al. CrystFEL : a software suite for snapshot serial crystallography. J. Appl. Crystallogr. 45, 335–341 (2012).

88. Duisenberg, A. J. M. Indexing in single-crystal diffractometry with an obstinate list of reflections. J. Appl. Crystallogr. 25, 92–96 (1992).

89. Ginn, H. M. et al. TakeTwo : an indexing algorithm suited to still images with known crystal parameters. Acta Crystallogr. Sect. Struct. Biol. 72, 956–965 (2016).

90. White, T. A. Processing serial crystallography data with CrystFEL : a step-by-step guide. Acta Crystallogr. Sect. Struct. Biol. 75, 219–233 (2019).

91. Vagin, A. & Teplyakov, A. Molecular replacement with MOLREP. Acta Crystallogr. D Biol. Crystallogr. (2010) doi:10.1107/S0907444909042589.

92. Murshudov, G. N. et al. REFMAC5 for the refinement of macromolecular crystal structures. Acta Crystallogr. D Biol. Crystallogr. (2011) doi:10.1107/S0907444911001314.

93. Adams, P. D. et al. PHENIX: A comprehensive Python-based system for macromolecular structure solution. Acta Crystallogr. D Biol. Crystallogr. (2010) doi:10.1107/S0907444909052925.

94. Emsley, P. & Cowtan, K. Coot: model-building tools for molecular graphics. Acta Crystallogr. D Biol. Crystallogr. 60, 2126–2132 (2004).

95. Jo, S., Kim, T., Iyer, V. G. & Im, W. CHARMM-GUI: A web-based graphical user interface for CHARMM. J. Comput. Chem. 29, 1859–1865 (2008).

96. Abraham, M. J. et al. GROMACS: High performance molecular simulations through multi-level parallelism from laptops to supercomputers. SoftwareX 1–2, 19–25 (2015). 110

97. Huang, J. et al. CHARMM36m: an improved force field for folded and intrinsically disordered proteins. Nat. Methods 14, 71–73 (2017).

98. Zhu, S., Brown, M. F. & Feller, S. E. Retinal Conformation Governs p K a of Protonated Schiff Base in Rhodopsin Activation. J. Am. Chem. Soc. 135, 9391–9398 (2013).

99. Laio, A. & Parrinello, M. Escaping free-energy minima. Proc. Natl. Acad. Sci. 99, 12562–12566 (2002).

100. Tribello, G. A., Bonomi, M., Branduardi, D., Camilloni, C. & Bussi, G. PLUMED 2: New feathers for an old bird. Comput. Phys. Commun. 185, 604–613 (2014).

101. Ho, B. K. & Gruswitz, F. HOLLOW: Generating accurate representations of channel and interior surfaces in molecular structures. BMC Struct. Biol. 8, (2008).

102. Borshchevskiy, V. I., Round, E. S., Popov, A. N., Büldt, G. & Gordeliy, V. I. X- ray-radiation-induced changes in bacteriorhodopsin structure. J. Mol. Biol. 409, 813–825 (2011).

103. Borshchevskiy, V. et al. Low-dose X-ray radiation induces structural alterations in proteins. Acta Crystallogr. D Biol. Crystallogr. 70, 2675–2685 (2014).

104. Moukhametzianov, R. et al. Development of the signal in sensory rhodopsin and its transfer to the cognate transducer. Nature 440, 115–119 (2006).

105. Kato, H. E. et al. Structural basis for Na + transport mechanism by a light-driven Na + pump. Nature 521, 48–53 (2015).

106. Kato, H. E. et al. Crystal structure of the channelrhodopsin light-gated cation channel. Nature (2012) doi:10.1038/nature10870.

107. Li, H. et al. Crystal structure of a natural light-gated anion channelrhodopsin. eLife (2019) doi:10.7554/eLife.41741.

108. Kim, Y. S. et al. Crystal structure of the natural anion-conducting channelrhodopsin GtACR1. Nature (2018) doi:10.1038/s41586-018-0511-6.

109. Fenalti, G. et al. Molecular control of δ-opioid receptor signalling. Nature (2014) doi:10.1038/nature12944.

110. Meier, T., Polzer, P., Diederichs, K., Welte, W. & Dimroth, P. Structure of the rotor ring of F-type Na+-ATPase from Ilyobacter tartaricus. Science (2005) 111 doi:10.1126/science.1111199.

111. Weinert, T. et al. Proton uptake mechanism in bacteriorhodopsin captured by serial synchrotron crystallography. Science (2019) doi:10.1126/science.aaw8634.

112. Vogt, A. et al. Engineered Passive Potassium Conductance in the KR2 Sodium Pump. Biophys. J. (2019) doi:10.1016/j.bpj.2019.04.001.

112

5. Appendix

Table 5.1. Data collection and refinement statistics for the ground state of the wild type KR2.

Pentameric, Pentameric, Pentameric Monomeric “dry” “wet” pH 8.0 6.0 5.0 6.0 8.0 8.9 8.0 8.0

Data collection

Space group C2221 C2221 C2221 I222 I222 I222 C2221 C2221

Cell

dimensions 131.36, 131.43, 130.24, 40.33, 40.41, 40.60, 128.18, 130.63, a, b, c (Å) 239.59, 240.79, 241.39, 81.25, 82.32, 82.25, 239.72, 240.47, 135.61 135.25 135.74 135.11 233.36 233.35 234.18 131.91 90, 90, 90, 90, 90, 90, 90, 90, 90, 90, α, β, γ (°) 90, 90, 90 90, 90, 90 90, 90, 90 90 90 90 90 90 Wavelength 1.000 1.000 0.978 1.000 0.978 1.000 0.976 0.976 (Å) 48.03- 48.21- 40.62- 40.60- 40.70- 48.08-2.6 2.2 2.7 2.3 1.8 2.5 47.28-3.0 48.13-2.8 Resolution (Å) (2.66- (2.24- (2.77- (2.39- (1.84- (2.60- (3.12-3.00) (2.89-2.80) 2.60) 2.20) 2.70) 2.30) 1.80) 2.50) 13.0 19.0 8.7 10.5 3.9 10.7 Rmerge (%) 30.4 (147.5) 12.9 (118.7) (143.5) (130.8) (174.6) (105.8) (42.8) (153.7) 1.5 9.1 4.9 1.9 5.3 Rpim (%) 4.1 (79.4) 12.7 (61.8) 10.0 (91.5) (16.7) (62.7) (50.6) (20.1) (75.0) 30.8 10.4 19.1 I/σI 7.9 (1.1) 9.7 (0.9) 9.4 (0.9) 5.7 (1.4) 5.6 (1.2) (4.4) (1.5) (3.3) 99.6 99.1 99.9 99.7 99.9 99.9 CC1/2 (%) 98.9 (66.2) 99.7 (78.2) (96.2) (63.2) (44.8) (68.3) (92.5) (56.0) Completeness 100.0 96.9 98.9 99.7 99.9 99.6 99.9 (100.0) 99.6 (99.4) (%) (100.0) (98.6) (99.8) (99.5) (99.9) (99.9) Unique 107966 57277 64789 17576 36723 14043 41018 (4562) 52539 (4497) reflections (5398) (4469) (4559) (1841) (2179) (1557)

Refinement

48.08- 48.26- 20.00- 20.00- 20.00- Resolution (Å) 48.13-2.6 47.32-3.0 48.17-2.8 2.2 2.7 2.3 1.8 2.6 No. reflections 102,502 54,451 61,464 16,711 35,277 11,878 38,937 49,912 14.9/17. 22.3/25. 14.2/18. 22.1/27. Rwork/Rfree (%) 19.1/24.5 20.2/25.8 21.7/25.2 18.8/22.3 2 0 1 1

No. atoms

Protein 10886 10872 10910 2125 2247 2129 10753 10872

Water 541 316 97 47 93 13 41 192

113

Lipid fragments 1210 899 417 275 185 54 201 895

Retinal 100 100 100 20 20 20 100 100

Sodium ions 5 5 5 1 1 - 5 5

B factors (Å2)

Protein 35.1 38.3 75.7 43.7 35.4 64.3 45.5 62.2

Water 44.5 35.2 72.1 50.3 52.3 58.3 36.1 55.7

Lipid fragments 76.6 69.0 93.7 78.3 62.2 73.6 56.0 95.1

Retinal 28.7 32.3 71.1 35.3 30.2 62.0 52.3 61.4

Sodium ions 25.6 33.1 69.9 58.9 44.1 - 34.9 50.3 R.m.s

deviations Protein bond 0.0041 0.0024 0.0023 0.0029 0.0035 0.0023 0.0030 0.0028 lengths (Å) Protein bond 0.831 0.560 0.541 0.584 0.767 0.504 0.590 0.600 angles (°)

114

Table 5.2. Data collection and refinement statistics for the K+-pumping and oligomerization interface mutant forms of KR2.

G263F, G263F, S254A, S254A, Y154F, H30K,

monomeric pentameric monomeric pentameric monomeric monomeric

pH 4.3 8.0 4.3 8.0 8.0 8.0 Data

collection

Space group I222 C2221 I222 C2221 I222 I222

Cell

dimensions 40.77, 81.96, 131.68, 40.80, 83.00, 131.40, 40.47, 81.85, 40.74, 84.18, a, b, c (Å) 232.93 239.68, 134.58 234.10 240.04, 135.14 233.57 234.54

α, β, γ (°) 90, 90, 90 90, 90, 90 90, 90, 90 90, 90, 90 90, 90, 90 90, 90, 90

Wavelength 0.976 1.003 0.968 1.000 0.972 0.968 (Å) Resolution 40.5-2.0 (2.05- 47.94-2.40 40.78-2.1 48.05-2.4 40.93-1.8 42.09-2.2 (Å) 2.00) (2.45-2.40) (2.16-2.10) (2.45-2.40) (1.84-1.80) (2.27-2.20)

Rmerge (%) 4.8 (40.5) 12.7 (167.4) 8.8 (167.4) 11.9 (148.1) 6.1 (87.0) 15.8 (170.2)

Rpim (%) 2.7 (22.6) 5.3 (68.0) 4.6 (86.6) 6.0 (74.2) 3.1 (42.8) 7.3 (79.6)

I/σI 14.8 (3.2) 9.8 (1.3) 10.9 (1.0) 9.1 (1.0) 13.2 (1.7) 7.4 (1.0)

CC1/2 (%) 99.9 (90.1) 99.8 (71.0) 99.9 (53.0) 99.8 (52.1) 99.8 (88.5) 99.7 (47.6)

Completeness 99.5 (100.0) 99.9 (99.9) 99.4 (100.0) 99.9 (99.9) 99.6 (99.6) 99.9 (100.0) (%) Unique 26,858 (1983) 83,145 (4490) 23679 (1936) 83365 (4530) 36488 (2168) 21097 (1825) reflections

Refinement

Resolution 20.00-2.0 47.98-2.4 20.00-2.1 44.91-2.4 20.00-1.8 20.00-2.2 (Å) No. 25,525 78,915 22,551 79,117 36,450 20,046 reflections

Rwork/Rfree (%) 17.7/22.7 17.2/20.5 22.6/27.9 18.1/21.2 16.3/20.3 20.9/23.8

No. atoms

Protein 2161 10877 2126 10867 2239 2111

Water 54 462 28 328 97 38 Lipid 337 1648 291 894 162 289 fragments Retinal 20 100 20 100 20 20

Sodium ions 1 5 1 5 1 1

115

B factors (Å2)

Protein 39.2 49.7 51.4 52.4 32.4 38.8

Water 47.0 56.3 48.7 55.4 50.4 39.8 Lipid 80.4 98.6 84.0 91.7 56.0 74.6 fragments Retinal 31.3 43.4 42.1 48.0 26.7 32.2

Sodium ions 42.4 41.4 69.6 42.7 37.3 60.9 R.m.s

deviations Protein bond 0.0043 0.0033 0.0028 0.0028 0.0016 0.0038 lengths (Å) Protein bond 0.769 0.704 0.570 0.628 0.614 0.730 angles (°)

116

Table 5.3. Data collection and refinement statistics for the ground and O-states of the wild type KR2.

2.50)

- #

2.7

293K - - 8.0 6YC0 C222 50 58,999 0.0011 1.0751 (serial) 19.1/22.0 WT 2.50 (2.54 11.3 11.3 (0.7) 90, 90, 90 -

99.7 (81.0) 15.2 15.2 (150.4) 100.0 100.0 (100.0) 77,967 77,967 (7691) 136656/38761 Illuminated state Illuminated 69.19 138.37 240.34, 135.21,

2.6

293K - - 8.0 3/3 2.59 2.59 (2.65 2.59) 6YC4 - C222 50 138.35 59,103 0.0012 1.0765 6.7 6.7 (1.1) 19.9/21.8 WT 90, 90, 90 99.4 99.4 (50.7) 99.6 (94.8) 23.7 23.7 (213.1) 69,594 69,594 (4259) (single crystal) (single 134.92, 239.78, 239.78, 134.92, Illuminated state Illuminated 48.81

2.50)

2.5

293K - - 8.0 3/3 6YC2 C222 50 74,363 0.0023 1.1192 17.4/19.8 WT 2.50 (2.55 10.1 (1.2) 90, 90, 90 Dark Dark state 99.8 99.8 (81.7) - 21.9 21.9 (236.5) 100.0 100.0 (100.0) 78,134 78,134 (4418) 48.84 135.15, 239.89, 138.35 239.89, 135.15,

2.10)

2.1

100K - state - 8.0 1/1 - C222 6XYT 50 0.0027 1.0479 O 101,547 17.6/20.0 WT 2.10 (2.14 20.1 (1.1) 90, 90, 90 7.8 7.8 (276.0) 99.8 (56.0) 99.8 (99.9) - 123,865 123,865 (6133) 48.06 131.16, 240.63, 135.04 240.63, 131.16,

2.00)

2.0

100K - - 8.0 1/1 6YC3 C222 50 0.0027 1.0671 135,234 17.7/20.3 WT 2.00 (2.03 13.2 13.2 (1.2) 90, 90, 90 7.6 7.6 (174.9) 99.9 (74.3) - 99.9 99.9 (100.0) Ground state Ground 144,508 144,508 (7135) 48.16 131.87, 240.32, 135.51 240.32, 131.87,

(%)

(°)

split (Å)

γ free

c σ , / (%) CC*) , R pH (Å) I # / β b or R or ( reflections

, crystals , PDB ID Number Number α a 1/2 work

Space Space group Refinement R merge Resolution (Å) Resolution (Å) Resolution No. indexed/merged indexed/merged R.m.s R.m.s deviations CC R Completeness (%) Completeness Unique reflections Unique Protein bond lengths lengths Protein bond Data Data collection Cell dimensions (°) angles Protein bond

117

Table 5.4. Data collection and refinement statistics for the D116N and H30A mutant forms of KR2.

D116N-100K D116N-100K H30A-100K

monomeric pentameric pentameric PDB ID 6YBY 6YBZ 6YC1 pH 4.6 8.0 8.0 Data collection

Space group I222 C2221 C2221 Number indexed/merged 1/1 1/1 1/1 crystals Cell dimensions a, b, c (Å) 40.89, 83.60, 233.83 131.34, 240.48, 135.41 131.09, 239.73, 135.13 α, β, γ (°) 90, 90, 90 90, 90, 90 90, 90, 90 Resolution (Å) 40.81-1.80 (1.84-1.80) 48.13-2.35 (2.39-2.35) 48.01-2.20 (2.24-2.20)

Rmerge or Rsplit 4.6 (100.0) 11.0 (192.5) 12.2 (209.9) (%) I/σI 16.2 (1.7) 11.3 (1.0) 11.5 (0.9) # CC1/2 ( CC*) 99.9 (90.5) 99.9 (78.7) 99.9 (48.9) (%) Completeness 99.8 (99.7) 99.9 (99.9) 98.9 (99.2) (%) Unique 37,771 (2191) 89,161 (4530) 106,450 (5261) reflections Refinement Resolution (Å) 20-1.8 50-2.35 50-2.20 No. reflections 35,822 74,883 91,343

Rwork/Rfree (%) 20.2/22.0 19.9/22.3 18.2/20.5 R.m.s deviations Protein bond 0.0054 0.0096 0.0019 lengths (Å) Protein bond 0.7962 1.0876 1.0502 angles (°)

118

Acknowledgments

I’m deeply thankful to my supervisor Prof. Dr. Valentin Gordeliy, who has been mentoring me over my undergraduate and Ph.D. work, for his support and valuable discussions. I would like to thank him for all the opportunities he gave me and for his invaluable advice.

I appreciate the help of my friends and colleagues in Forschungszentrum Jülich: Dmytro Volkov, Dmitry Bratanov, Dmitrii Zabelskii, Oleksandr Volkov, and Svetlana Vaganova. I would like to express special gratitude to Taras Balandin for the supervising of my work in the lab, mentoring of the protein production for the projects, and adventurous ideas. I also would like to thank Christian Baeken, who has produced KR2 during the last two years.

A large part of this work was done in the lab of membrane transporters at IBS, Grenoble. Crystallization trials were set up there and it would not be possible without help from my colleagues: Roman Astashkin, Vitaly Polovinkin, Maksim Rulev, Egor Zinovev, and Igor Melnikov.

The data collection was performed at ESRF in Grenoble and at PETRAIII in Hamburg, and I’m deeply thankful to ESRF and PETRAIII staff and, in particular, to Dr. Alexander Popov and Dr. Gleb Bourenkov for their help and useful recommendations. I’m also thankful to Dr. Antoine Royant for the help with the spectroscopy on KR2 crystals and for always being ready to provide everything he could for experiments.

I would like to thank Dr. Ivan Gushchin and Dr. Valentin Borshchevskiy at the Moscow Institute of Physics and Technology for their wisdom and advice during my Ph.D. and help with data analysis and planning of the experiments.

My special thanks go to my best top dear friend, Alexey Alekseev, who always was nearby even staying 3000 km far. Without him I would never been where I am now. Thank you, man!

I should say the same to my friend and colleague Vitaly Shevchenko, who mentored me since my first day in the lab.

I’m very thankful to my parents, who always supported me and believed in me. And I’m also enormously thankful to my wife, who has been with me through all the way of wins and failures, acceptances and rejections, who never stopped cheering me up and finding positive sides everywhere. She is the one who gave me the strength to get all that I have at this step. Thank you for always being there for me.

119

Abstract

Ion transport across the cell membrane is a fundamental biological process. It allows cells to maintain a proper inner composition, required for their survival. Moreover, it creates electrochemical potential at the membrane, used for many essential processes, such as ATP synthesis, nutrients uptake, etc. Ion translocation is performed by special proteins - ion transporters, which may drive active or passive transfer of different types of ions through the lipid membrane. One of the largest and most diverse families of such proteins is microbial rhodopsins (MRs) family. MRs are light-sensitive membrane proteins (MPs), which comprises a retinal chromophore as a prosthetic group. Most of MRs are light-driven ion transporters; however, they can perform other important biological functions, being photosensors, enzymes, etc.

The first MR found was a light-driven H+ pump bacteriorhodopsin from archaeon Halobacterium salinarum (HsBR). It is at the moment the most studied MR and serves as a model protein for all MRs, and MPs in general. The mechanism of light-driven H+ pumping is thus the most investigated and understood.

The discovery of channelrhodopsins (ChRs) in chlorophyte alga Chlamydomonas reinhardtii (CrChR1 and CrChR2, respectively), became a basement of the optogenetics – a technique for the control of living tissue by the light. Optogenetics utilizes MRs for optical control of cells and tissues with unprecedented temporal and spatial resolutions. The development of the optogenetics, together with the evolution of the functional metagenomics, resulted in numerous big findings in the rhodopsins field of research. In the 2010s, many novel MRs with unique properties were discovered and functionally and structurally characterized. Among them, there were light-driven Na+ pumps (NaRs). The existence of NaRs in nature was doubtful since all MRs contain a protonated positively charged Schiff base in the core of the molecule. This should have created a conflict of the simultaneous presence of two positive charges (Na+ ion and the protonated Schiff base) in close proximity. Despite that, in 2013, the first NaR, KR2, was found in marine bacteria Krokinobacter eikastus. The high ion selectivity of KR2 to Na+, but not to K+, Ca2+, and H+ makes it an outstanding potential silencing optogenetic tool. Moreover, the first light-driven K+ and Cs+ pumps were engineered based on KR2, expanding the toolkit of the optogenetics even further.

This work is dedicated to the structural and functional studies of KR2. This rhodopsin is a unique light-driven non-proton cation transporter. A fundamental difference between KR2 and other active transporters is that it cannot utilize Grotthuss or tunneling mechanisms for Na+

120 translocation, which are used by H+ pumps like, for instance, HsBR. Therefore, NaRs should have completely different mechanism of functioning from that described for HsBR. Also, KR2 does not bind the transported substrate, a Na+ ion, in its resting state. Instead, Na+ is uptaken from the cytoplasm and released to the extracellular space by the protein during its working cycle, called photocycle. As in other MRs, the photocycle of KR2 has several spectrally distinct metastable intermediate states. Namely, there are K, L, M, and O states. It was demonstrated that Na+ ion is transiently bound inside KR2 protomer only in the late O intermediate. This fact additionally stresses the existence of a notable difference between the mechanisms of the light-driven H+ and non-H+ cation pumping.

The importance of the results of this work is many-fold. First, we determined the structure of a biologically relevant pentameric form of KR2 under physiological conditions. It revealed the organization of Na+-pumping state of the NaR. Second, by additional structural and functional analysis of H30L/K and Y154F mutants of KR2, we showed that pentamerization of the NaR is vital for efficient Na+ pumping. Indeed, pentameric assembly is critical for both ion uptake and release, since it makes the inner cavities more accessible from the outer space. Oligomerization is also obligatory factor for the organization of the Na+-pumping state of KR2. Third, we obtained a 2.1 Å structure of the O-state of KR2 photocycle with transiently bound Na+ ion in the core of the protein protomer. The architecture of the Na+ binding site revealed the principles of ion selectivity in KR2. Fourth, we describe the Na+ translocation pathway and demonstrated its rearrangements during KR2 photocycle. Fifth, we reported structural switches corresponding to the Na+-to-H+ pump conversion upon pH decrease. Sixth, these results, together with additional functional and structural characterization of several critical KR2 mutants, allowed us to propose the molecular mechanism of light-driven Na+ pumping. We showed, that the mechanism includes principles of passive ion transport through the cavities of the protein, similar to that in channelrhodopsins. At the same time, in contrast to them, NaRs possess a Na+ binding site near the Schiff base, the existence of which provides unidirectional translocation of ions through the protein. Seventh, we solved the first structures of K+-pumping KR2 variants G263F and S254A. The results were published in 2 manuscripts in Science Advances (2019) and Nature Communications (2020).

Therefore, this work provides insights into the molecular mechanisms of NaRs functioning and deepens and expands the understanding of this fundamental biological process. The obtained structures open the way for the further rational design of enhanced KR2-based optogenetic tools including those with modified ion selectivity.

121

Abstrakt

Der Ionentransport durch die Zellmembran ist ein grundlegender biologischer Prozess. Es ermöglicht den Zellen, eine richtige innere Zusammensetzung aufrechtzuerhalten. Darüber hinaus erzeugt es ein elektrochemisches Potential an der Membran, das für viele wesentliche Prozesse wie ATP-Synthese, Nährstoffaufnahme usw. verwendet wird. Die Ionentranslokation wird von speziellen Proteinen durchgeführt - Ionentransportern, die den aktiven oder passiven Transfer verschiedener Arten von Ionen antreiben können die Lipidmembran. Eine der größten und vielfältigsten Familien solcher Proteine ist die Familie der mikrobiellen Rhodopsine (MRs). MRs sind lichtempfindliche Membranproteine (MPs), die ein retinales Chromophor als prothetische Gruppe umfassen. Die meisten MRs sind lichtgetriebene Ionentransporter; Sie können jedoch andere wichtige biologische Funktionen erfüllen, z. B. Photosensoren, Enzyme usw.

Der erste gefundene MR war ein lichtgetriebenes H+-Pump Bakteriorhodopsin aus dem Archäon Halobacterium salinarum (HsBR). Es ist derzeit das am meisten untersuchte MR und dient als Modellprotein für alle MRs und MPs im Allgemeinen. Die Entdeckung von Channelrhodopsinen (ChRs) in der Chlorophytenalge Chlamydomonas reinhardtii (CrChR1 und CrChR2) wurde zu einem Fundament der Optogenetik - einer Technik zur Kontrolle von lebendem Gewebe durch Licht. Optogenetik verwendet MRs zur optischen Kontrolle von Zellen und Geweben mit beispielloser zeitlicher und räumlicher Auflösung. Die Entwicklung der Optogenetik führte zusammen mit der Entwicklung der funktionellen Metagenomik zu zahlreichen großen Erkenntnissen auf dem Gebiet der Rhodopsine. In 2010 wurden viele neuartige MRs mit einzigartigen Eigenschaften entdeckt und funktionell und strukturell charakterisiert. Unter ihnen befanden sich lichtgetriebene Na+-Pumpen (NaRs). Die Existenz von NaRs in der Natur war zweifelhaft, da alle MRs eine protonierte positiv geladene Schiff-Base im Kern des Moleküls enthalten. Dies hätte zu einem Konflikt des gleichzeitigen Vorhandenseins von zwei positiven Ladungen (Na+-Ion und protonierte Schiff-Base) in unmittelbarer Nähe führen müssen. Trotzdem wurde 2013 das erste NaR, KR2, in marinen Bakterien Krokinobacter eikastus gefunden. Die hohe Ionenselektivität von KR2 gegenüber Na+, jedoch nicht gegenüber K+, Ca2+ und H+ macht es zu einem hervorragenden optogenetischen Werkzeug zur potenziellen Stummschaltung.

Diese Arbeit widmet sich den strukturellen und funktionellen Studien von KR2. Dieses Rhodopsin ist ein einzigartiger lichtgetriebener Nicht-Protonenkationentransporter. Ein grundlegender Unterschied zwischen KR2 und anderen aktiven Transportern besteht darin, dass es keine Grotthuss- oder Tunnelmechanismen für die Na+-Translokation verwenden kann, die von H+-Pumpen verwendet werden. Außerdem bindet KR2 das transportierte Substrat, ein Na+-Ion, im 122

Ruhezustand nicht. Stattdessen wird Na+ aus dem Zytoplasma aufgenommen und vom Protein während seines Arbeitszyklus, dem sogenannten Photozyklus, in den extrazellulären Raum freigesetzt. Wie bei anderen MRs weist der Photozyklus von KR2 mehrere spektral unterschiedliche metastabile Zwischenzustände auf. Es gibt nämlich K-, L-, M- und O-Zustände. Es wurde gezeigt, dass Na+-Ionen nur im späten O-Intermediat vorübergehend im KR2-Protomer gebunden sind. Diese Tatsache unterstreicht zusätzlich die Existenz eines bemerkenswerten Unterschieds zwischen den Mechanismen des lichtgetriebenen H+- und Nicht-H+- Kationenpumpens.

Die Bedeutung der Ergebnisse dieser Arbeit ist vielfältig. Zunächst haben wir die Struktur einer biologisch relevanten pentameren Form von KR2 unter physiologischen Bedingungen bestimmt. Es zeigte die Organisation des Na+-Pumpzustands des NaR. Zweitens haben wir gezeigt, dass die Pentamerisierung des NaR für ein effizientes Na+-Pumpen von entscheidender Bedeutung ist. In der Tat ist die pentamere Anordnung sowohl für die Ionenaufnahme als auch für die Ionenfreisetzung entscheidend, da sie die inneren Hohlräume vom Weltraum aus leichter zugänglich macht. Die Oligomerisierung ist auch ein obligatorischer Faktor für die Organisation des Na+-Pumpzustands von KR2. Drittens erhielten wir eine 2.1 Å-Struktur des O-Zustands des KR2-Photozyklus mit transient gebundenem Na+-Ion im Kern des Proteinprotomers. Viertens beschreiben wir den Na+-Translokationsweg und demonstrierten seine Umlagerungen während des KR2-Photozyklus. Fünftens berichteten wir über Strukturschalter, die der Umwandlung der Na+ - zu-H+-Pumpe bei pH-Abnahme entsprechen. Sechstens ermöglichten diese Ergebnisse den molekularen Mechanismus des lichtgetriebenen Na+-Pumpens vorzuschlagen. Wir haben gezeigt, dass der Mechanismus Prinzipien des passiven Ionentransports durch die Hohlräume des Proteins umfasst, ähnlich wie bei Channelrhodopsinen. Gleichzeitig besitzen NaRs im Gegensatz zu ihnen eine Na+-Bindungsstelle in der Nähe der Schiff Base, deren Existenz eine unidirektionale Translokation von Ionen durch das Protein ermöglicht. Siebtens haben wir die ersten Strukturen der K+-Pumping-KR2-Varianten G263F und S254A gelöst. Die Ergebnisse wurden in 2 Manuskripten in Science Advances (2019) und Nature Communications (2020) veröffentlicht.

Daher liefert diese Arbeit Einblicke in die molekularen Mechanismen der Funktion von NaRs und vertieft und erweitert das Verständnis dieses grundlegenden biologischen Prozesses. Die erhaltenen Strukturen eröffnen den Weg für das weitere rationale Design verbesserter optogenetischer Werkzeuge auf KR2-Basis, einschließlich solcher mit modifizierter Ionenselektivität.

123