Proton transfer pathway in anion channelrhodopsin-1

  1. Masaki Tsujimura
  2. Keiichi Kojima
  3. Shiho Kawanishi
  4. Yuki Sudo  Is a corresponding author
  5. Hiroshi Ishikita  Is a corresponding author
  1. Department of Applied Chemistry, The University of Tokyo, Japan
  2. Graduate School of Medicine, Dentistry and Pharmaceutical Sciences, Okayama University, Japan
  3. Research Center for Advanced Science and Technology, The University of Tokyo, Japan

Abstract

Anion channelrhodopsin from Guillardia theta (GtACR1) has Asp234 (3.2 Å) and Glu68 (5.3 Å) near the protonated Schiff base. Here, we investigate mutant GtACR1s (e.g., E68Q/D234N) expressed in HEK293 cells. The influence of the acidic residues on the absorption wavelengths was also analyzed using a quantum mechanical/molecular mechanical approach. The calculated protonation pattern indicates that Asp234 is deprotonated and Glu68 is protonated in the original crystal structures. The D234E mutation and the E68Q/D234N mutation shorten and lengthen the measured and calculated absorption wavelengths, respectively, which suggests that Asp234 is deprotonated in the wild-type GtACR1. Molecular dynamics simulations show that upon mutation of deprotonated Asp234 to asparagine, deprotonated Glu68 reorients toward the Schiff base and the calculated absorption wavelength remains unchanged. The formation of the proton transfer pathway via Asp234 toward Glu68 and the disconnection of the anion conducting channel are likely a basis of the gating mechanism.

Introduction

Anion channelrhodopsins (ACRs) are light-gated anion channels that undergo photoisomerization at the retinal chromophore, which is covalently attached to a conserved lysine residue via the protonated Schiff base, from all-trans to 13-cis. Natural ACRs were identified in the cryptophyte Guillardia theta (GtACR1 and GtACR2) (Govorunova et al., 2015). ACRs hyperpolarize the membrane through anion import and can widely be used as neural silencing tools in optogenetics (Wiegert et al., 2017; Miyazaki et al., 2019). Microbial rhodopsins have acidic residues or Cl at the Schiff base moiety to stabilize the protonated Schiff base as counterions. Counterions play a major role in determining the absorption wavelength and the function of the protein (Tsujimura and Ishikita, 2020). The X-ray crystal structures of GtACR1 show that two acidic residues, Glu68 and Asp234, exist at the corresponding positions (Figure 1; Kim et al., 2018; Li et al., 2019).

GtACR1.

(a) Overview of the GtACR1 structure. (b) Residues near the Schiff base in GtACR1 (PDB code: 6EDQ; Li et al., 2019). Dotted lines indicate H-bonds.

It was proposed that both Glu68 and Asp234 were protonated in GtACR1 (Kim et al., 2018; Sineshchekov et al., 2016; Yi et al., 2016; Kandori, 2020) in contrast to other microbial rhodopsins because the absorption wavelengths remain unchanged upon the E68Q and D234N mutations (Kim et al., 2018; Sineshchekov et al., 2016; Yi et al., 2016). Indeed, the C=C stretching frequency of the retinal is not significantly affected upon the E68Q and D234N mutations in resonance Raman spectroscopy, which implies that the electrostatic interaction between the retinal and protein environment remains unchanged (Yi et al., 2016). In addition, the C=O stretching frequency for a protonated carboxylate, which is observed in the wild-type GtACR1, disappears in the E68Q (Yi et al., 2017; Dreier et al., 2021) and D234N (Kim et al., 2018) GtACR1s according to Fourier transform infrared (FTIR) spectroscopy analysis.

Nevertheless, it is an open question whether Asp234 is protonated. Kim et al. pointed out that the loss of photocurrent in the D234N GtACR1 cannot be easily understood if Asp234 is protonated as the influence of the mutation of protonated aspartate to asparagine on the protein function is often small (Kim et al., 2018). GtACR1 crystal structures show that the residues at the Schiff base moiety are highly conserved between GtACR1 and bacteriorhodopsin (BR). Tyr57, Arg82, Tyr185, and Lys216, which are responsible for the low pKa of –2 for the counterion Asp212 in BR (Saito et al., 2012), are fully conserved as Tyr72, Arg94, Tyr207, and Lys238 in GtACR1. Note that the counterion Asp85 in BR, which increases pKa(Asp212) by 6 (Saito et al., 2012), is replaced with Ser97 in GtACR1. This suggests that the pKa of Asp234 in GtACR1 is even lower than the low pKa of –2 for Asp212 in BR. According to Li et al., Tyr72 and Tyr207 donate H-bonds to Asp234 in GtACR1 (Li et al., 2019): this suggests that deprotonated Asp234 is stabilized, decreasing pKa(Asp234), as observed for deprotonated Asp212 in BR. In addition, resonance Raman spectroscopy analysis indicates that the Schiff base Lys238 is also protonated (Yi et al., 2016) as observed in other microbial rhodopsins. The presence of the positively charged Schiff base needs to have an adjacent negative charge (e.g., deprotonated acidic residue) to effectively decrease the energy in GtACR1. To the best of our knowledge, microbial rhodopsins have more than one deprotonated acidic residue adjacent to the retinal Schiff base (e.g., Tsujimura and Ishikita, 2020). This also holds true for channelrhodopsin from Chlamydomonas noctigama (Chrimson) and rhodopsin phosphodiesterase (Rh-PDE), which have both deprotonated and protonated acidic residues near the Schiff base (Vierock et al., 2017; Watari et al., 2019). That is, either Glu68 or Asp234 may be deprotonated in GtACR1. Deprotonation of Asp234 is energetically more favorable than deprotonation of Glu68 in the presence of protonated Schiff base as the electrostatic interaction with Asp234 (3.2 Å) is larger than with Glu68 (5.3 Å). Alternatively, Cl may exist and act as a counterion, as observed in Cl pumping rhodopsins (Kim et al., 2016; Kolbe et al., 2000). However, the corresponding electron density is not observed in GtACR1 (Kim et al., 2018; Li et al., 2019). In addition, no spectral changes are reported upon deionization of the sample or exchange from Cl to SO42– buffer (Sineshchekov et al., 2016). So far, the counterion of GtACR1 remains unknown.

Recently, Dreier et al. proposed that Asp234 is deprotonated in the dark and acts as a counterion according to FTIR measurements and molecular dynamics (MD) simulations (Dreier et al., 2021). The C=O stretching frequencies of 1740 (–)/1732 (+) cm–1 for protonated Asp234 at 77 K observed by Kim et al., 2018 were not observed at 293 K by Dreier et al., 2021. In addition, MD simulations indicated that the H-bond network that involves Tyr72, Tyr207, and Asp234 was stable with deprotonated Asp234 but unstable with protonated Asp234 (Dreier et al., 2021). Indeed, the presence of deprotonated Asp234 was already suggested based on the homology modeling of GtACR2 (Kojima et al., 2018) before the crystal structures of GtACR1 were reported.

GtACR1 undergoes a photocycle including K, L, M, N, and O intermediates (Figure 2; Sineshchekov et al., 2016). The L-state represents the anion conducting state. The L- to M-state transition involves the deprotonation of the Schiff base and the photocurrent decay. The fast photocurrent decay (fast channel closing) corresponds to the M-state formation (i.e., proton release from the Schiff base), and the slow photocurrent decay corresponds to the M-state decay (Sineshchekov et al., 2016; Sineshchekov et al., 2015; Figure 2). Glu68 is likely to accept a proton from the Schiff base upon the M-state formation as a decrease in the accumulation of the M-state was observed in the E68Q GtACR1 (Sineshchekov et al., 2016). However, it remains unclear whether Glu68 is the initial proton acceptor in the wild-type GtACR1. The GtACR1 crystal structures show that Asp234 is closer to the Schiff base (Figure 1; Kim et al., 2018; Li et al., 2019; Li et al., 2021). In addition, the E68Q mutation did not completely inhibit the M-state formation, which indicates that an alternative proton accepter exists (Sineshchekov et al., 2016).

Figure 2 with 1 supplement see all
Photocycle of GtACR1 proposed in Sineshchekov et al., 2016.

The L- to M-state and M- to N/O-state transitions involve deprotonation and reprotonation of the Schiff base, respectively.

To identify the counterion and clarify the proton-mediated gating mechanism of GtACR1, we investigate the Glu68 and Asp234 mutant proteins (E68Q, E68D, D234N, D234E, and E68Q/D234N) expressed in HEK293 cells. The protonation states are calculated by solving the Poisson–Boltzmann equation and evaluated by conducting MD simulations. Using a quantum mechanical/molecular mechanical (QM/MM) approach, the absorption wavelengths are calculated and the microscopic origin of the wavelength shifts upon the mutations is analyzed.

Results

Protonation states of Glu68 and Asp234

The protonation pattern (Table 1) and pKa values (Table 2 and Table 3) calculated solving the linear Poisson-Boltzmann equation show that Asp234 is deprotonated, whereas Glu68 is protonated in the GtACR1 crystal structures (Kim et al., 2018; Li et al., 2019; Table 1; Supplementary file 1A). The calculated protonation pattern shows that Asp234 is deprotonated in the wild-type GtACR1 even using the MD-generated conformations with protonated Glu68/protonated Asp234 or deprotonated Glu68/protonated Asp234 (Table 1, Supplementary file 1B, Figure 2—figure supplement 1). These results suggest that deprotonation of Asp234, ‘the only residue directly interacting with the protonated Schiff base (Li et al., 2019)’, is a prerequisite to stabilize the protonated Schiff base, as suggested by Dreier et al., 2021.

Table 1
Protonation states of Glu68 and Asp234.

Changes in the protonation states are in bold. – indicates not applicable.

Fixed state during MD*Calculated state after MD
Wild-type (1)E68ProtonatedProtonated
D234DeprotonatedDeprotonated
Wild-type (2)E68ProtonatedProtonated
D234ProtonatedDeprotonated
Wild-type (3)E68DeprotonatedProtonated
D234ProtonatedDeprotonated
E68D (1)D68ProtonatedProtonated
D234DeprotonatedDeprotonated
E68D (2)D68DeprotonatedProtonated
D234DeprotonatedDeprotonated
E68Q (1)Q68
D234DeprotonatedDeprotonated
E68Q (2)Q68
D234ProtonatedDeprotonated
D234E (1)E68ProtonatedProtonated
E234DeprotonatedDeprotonated
D234E (2)E68ProtonatedProtonated
E234ProtonatedProtonated
D234N (1)E68DeprotonatedDeprotonated
N234
D234N (2)E68ProtonatedProtonated §
N234
E68Q/D234NQ68
N234
  1. *

    The system was equilibrated for 5 ns. 10 conformations were sampled at 0.1 ns intervals during the 1 ns production run.

  2. Protonation patterns obtained using the MD-generated conformations.

  3. Although we were able to obtain the MD-generated conformation with protonated Glu68 and protonated Glu234, which was confirmed in the calculated protonation pattern, the conformation cannot reproduce the experimentally measured absorption wavelength (Supplementary file 1D) and is unlikely relevant to the D234E GtACR1.

  4. §

    Although we were able to obtain the MD-generated conformation with protonated Glu68, which was confirmed in the calculated protonation pattern, the protonation state is not consistent with deprotonated Glu68 suggested in FTIR studies by Dreier et al., 2021.

Table 2
pKa for Asp234 in GtACR1 and Asp212 in BR.
GtACR1BR
pKaAsp234−4.9Asp212−2.0*
Contribution
Schiff base−9.7Schiff base−8.7*
Arg94−3.3Arg82−6.3*
Tyr72−3.3Tyr57−3.4*
Tyr207−3.4Tyr185−3.3*
Ser970.5Asp856.0*
  1. *
Table 3
pKa for Glu68 in GtACR1.

The corresponding sites of BR are shown in the same line. – indicates not applicable.

GtACR1BR
pKa*Glu6812Ala53
Contribution
Schiff base−6.6Schiff base
Arg94−1.3Arg82
Asn2392.5Val217
Asp2344.9Asp212
  1. *

pKa(Asp234) = –5 (Table 2) is significantly low and even lower than pKa(Asp212) = –2 in BR (Saito et al., 2012). The crystal structures show that the residues at the Schiff base moiety are highly conserved between GtACR1 and BR. Tyr72 and Tyr207 donate H-bonds to each carboxyl O site of Asp234 in GtACR1 (Li et al., 2019; Dreier et al., 2021), whereas Tyr57 and Tyr185 donate H-bonds to each carboxyl O site of Asp212 in BR (Saito et al., 2012). Thus, each tyrosine residue stabilizes the deprotonated state of Asp234, decreasing pKa(Asp234) in GtACR1 by ~3 (Table 2), as observed in BR (Saito et al., 2012). The tendency is also observed for the conserved residue pairs, Arg94/Arg82 and Lys238/Lys216, in GtACR1/BR (Table 2). Asp85, which increases pKa(Asp212) in BR by 6, is replaced with Ser97, which has no influence on pKa(Asp234) in GtACR1 (Table 2). This discrepancy contributes to the low pKa(Asp234) in GtACR1, which is lower than pKa(Asp212) in BR. As far as the original geometry of the GtACR1 crystal structure is analyzed, no residue that increases pKa(Asp234) significantly is identified (Table 2).

Recently, Li et al. reported the GtACR1 conformation (pre-activating state), where Arg94 forms a salt-bridge with Asp234 (Li et al., 2021). The influence of Arg94 on pKa(Asp234) (~3) indicates that the electrostatic link between Arg94 and Asp234 exists even in the ground state (Table 2). It seems possible that the electrostatic interaction between deprotonated Asp234 and channel-gating Arg94 (Table 2) is absent in the D234N GtACR1, leading to the loss of the photocurrent (Kim et al., 2018).

In contrast, pKa(Glu68) is high, 12 (Table 3), which is consistent with the reported protonation state of Glu68 (Yi et al., 2016; Yi et al., 2017; Dreier et al., 2021). The high pKa(Glu68) value can be primarily due to the presence of anionic Asp234 whose deprotonated state is stabilized by the protonated Schiff base (Table 3). Charge neutral Ala53 exists at the corresponding position in BR, which is also consistent with the protonation of Glu68 (Table 3).

Exceptionally, Glu68 is deprotonated only in the D234N GtACR1. The influence of the protonated Schiff base is weaker on Glu68 than on Asp234 (Table 2 and Table 3), which allows to stabilize the putative protonated Glu68 conformation in MD simulations (Table 1). However, the experimentally measured absorption wavelength cannot be reproduced unless Glu68 is deprotonated in the D234N GtACR1 (Table 4, Supplementary file 1C). This is consistent with the absence of the 1708 cm–1 band in the D234N GtACR1, which is assigned to protonated Glu68 in FTIR measurements (Dreier et al., 2021). These results confirm that the presence of a negative charge at the protonated Schiff base moiety is a prerequisite to stabilize the protonated Schiff base, as observed in other microbial rhodopsins (Tsujimura and Ishikita, 2020), including Chrimson and Rh-PDE (Vierock et al., 2017; Watari et al., 2019).

Table 4
Calculated and experimentally measured absorption wavelengths λmax (nm).

Δλ (nm) denotes the shift with respect to the wild-type GtACR1.

Calculated*Measured
λmaxΔλλmaxΔλ
Wild-type505(493)0(0)5130
E68D499(483)−6(−10)493−20
E68Q505(493)0(0)5152
D234E499(481)−6(−12)503−10
D234N(deprotonated E68)507(487)2(−6)510−3
D234N(protonated E68)534(517)29(24)
E68Q/D234N538(522)33(29)5207
E68Q/D234N(with Cl)509(491)4(−2)
  1. *

    Average values of 10 MD-generated structures, which are finally QM/MM-optimized. Absorption wavelengths were calculated using Equation 1 with ΔEHOMO-LUMO. Absorption wavelengths calculated using Equation 2 with ETD-DFT are shown in parenthesis.

  2. With Cl binding at the Schiff base moiety (see below).

Absorption wavelengths of the Glu68 and Asp234 mutant GtACR1s

The experimentally measured absorption wavelengths for the Glu68 and Asp234 mutant proteins (E68Q, E68D, D234N, D234E, and E68Q/D234N) expressed in HEK293 cells are shown in Figure 3.

Figure 3 with 2 supplements see all
Absorption spectrum of (a) wild-type, (b) E68Q, (c) E68D, (d) D234N, (e) D234E, and (f) E68Q/D234N GtACR1s in DDM micelles.

Spectra are normalized at peak absorbance. The spectra for the wild-type (dotted curves in panels b–f) are shown for comparison. For each absorption spectrum, the results of four-traces were averaged to improve the signal-to-noise ratio. Absorption wavelengths were measured from the averaged spectra.

In the present study, the experimentally measured absorption wavelengths are the same for the wild-type and D234N GtACR1s (Table 4), which is consistent with results reported previously (Kim et al., 2018; Sineshchekov et al., 2016; Yi et al., 2016). Notably, the calculated absorption wavelengths are also the same for the wild-type and D234N GtACR1s irrespective of deprotonated Asp234 in the wild-type GtACR1 (Table 4, Supplementary file 1C, Figure 3—figure supplements 1 and 2). This can be explained as, in the D234N GtACR1, deprotonated Glu68 moves toward the positively charged Schiff base, which fully substitutes a role of deprotonated Asp234 in stabilizing the positively charged Schiff base (Figure 4a and b, Table 5). A similar conformation of Glu68, which orients toward the Schiff base, was previously reported for the corresponding residues of GtACR2 (Glu64) (Kojima et al., 2018) and channelrhodopsin from Chlamydomonas reinhardtii (Glu90) (Volkov et al., 2017). Thus, the absence of the change in the absorption wavelength upon D234N mutation (Kim et al., 2018) does not necessarily indicate that Asp234 is protonated in the wild-type GtACR1.

Figure 4 with 2 supplements see all
The H-bond network of the Schiff base of (a) wild-type, (b) D234N, and (c) D234E GtACR1s.

Representative MDgenerated conformations, whose absorption wavelengths are closest to the average value for the 10 MD-generated structures, are shown. Dotted lines indicate H-bonds.

Table 5
Components that contribute to the absorption wavelength in the wild-type and D234N GtACR1s (nm).

The contributions were analyzed using a single MD-generated structure whose absorption wavelength is closest to the average value of the 10 MD-generated structures.

Wild-typeD234N
Glu680−20
Asp234−28
Asn2340
H2O0−7
Total−28−27

The existence of deprotonated Asp234 in the wild-type GtACR1 can also be understood from the absorption wavelength in the D234E GtACR1. The distance between Glu234 and the Schiff base (2.9 Å in MD-generated conformations) in the D234E GtACR1 is shorter than that between Asp234 and the Schiff base (3.4 Å in MD-generated conformations) in the wild-type GtACR1 because glutamate is longer than aspartate (Figure 4a and c, Figure 4—figure supplement 1). The absorption wavelength is short as the electrostatic interaction between the deprotonated counterion and the protonated Schiff base is strong (Tsujimura and Ishikita, 2020). Remarkably, the D234E mutation leads to a decrease in the absorption wavelength (Table 4, Figure 3, Figure 3—figure supplements 1 and 2), which suggests that Asp234 is deprotonated in the wild-type GtACR1. The decrease in the measured absorption wavelength of 10 nm could not be reproduced when we forced Asp234 in wild-type and Glu234 in D234E GtACR1s to protonate (Supplementary file 1D). The electrostatic contributions of charge-neutral protonated Asp234 in wild-type and protonated Glu234 in D234E GtACR1s to the absorption wavelengths are small (Supplementary file 1D).

As far as we are aware, the absorption wavelength of the isolated E68Q/D234N GtACR1 is not reported (e.g., Sineshchekov et al., 2016; Dreier et al., 2021). We successfully isolated a photoactive form of E68Q/D234N GtACR1 using the HEK293 cell expression system, which has been widely used for the functional expression in animal rhodopsins (Kojima et al., 2017; Yamashita et al., 2010). The experimentally measured absorption wavelength in the isolated E68Q/D234N protein is 7 nm longer than that in the wild-type protein (Figure 3), which indicates that Glu68 or Asp234 must be deprotonated in the wild-type GtACR1. As Asp234 is closer to the Schiff base (3.2 Å) than Glu68 (5.3 Å) (Li et al., 2019), it seems more likely that Asp234 is deprotonated in the wild-type GtACR1.

Microbial rhodopsins, including Chrimson and Rh-PDE (Vierock et al., 2017; Watari et al., 2019), have more than one deprotonated acidic residue adjacent to the Schiff base (Tsujimura and Ishikita, 2020). The loss of two acidic residues upon the E68Q/D234N mutation requires an additional negative charge as far as the Schiff base remains protonated. Thus, it seems possible that Cl exist to stabilize the protonated Schiff base specifically in the E68Q/D234N GtACR1 because the next closest acidic residue, Glu60, is 10 Å away from the Schiff base. The presence of Cl in the E68Q/D234N GtACR1 is not reported. To investigate the existence of Cl, isolated E68Q/D234N samples were solubilized in Cl-free buffer. However, denaturation of the samples did not allow us to conclude the existence of Cl. In QM/MM calculations and MD simulations, the binding of Cl at Thr71/Asn234 or Ser97/Lys238 is more stable in the E68Q/D234N GtACR1 than in the wild-type protein (Figure 4—figure supplement 2, Supplementary file 1E). The increase in the calculated absorption wavelength upon the E68Q/D234N mutation (33 nm) is overestimated in the absence of Cl, whereas the corresponding increase (4 nm) is at the same level as that measured experimentally in the presence of Cl (Table 4). Thus, Cl is likely to exist near the protonated Schiff base to compensate for the loss of two acidic residues in the E68Q/D234N GtACR1.

The E68Q mutation does not alter the absorption wavelength (Table 4) as reported previously (Sineshchekov et al., 2016; Yi et al., 2016), thereby suggesting that Glu68 is protonated in the presence of deprotonated Asp234 (e.g., wild-type GtACR1) (Table 1).

In general, blue light-sensitive microbial rhodopsins (e.g., Sensory rhodopsin II and Chlamydomonas channelrhodopsins) show the main absorbance peak with spectral shoulder at shorter wavelength region (e.g., Takahashi et al., 1990). Based on these, it seems likely that the wide band of E68D is due to the existence of the spectral shoulder of this blue-shifted mutant.

Discussion

Our finding of deprotonated Asp234 in the ground state of GtACR1 can explain the following observations: loss of photocurrent upon the D234N mutation (Kim et al., 2018) can be due to loss of Asp234, which is deprotonated in the wild-type GtACR1. It seems possible that the electrostatic interaction between deprotonated Asp234 and channel-gating Arg94 (Li et al., 2021; Table 2) is absent in the D234N GtACR1, leading to loss of the photocurrent (Kim et al., 2018). Intriguingly, MD simulations show that upon the D234N mutation, deprotonated Glu68 reorients toward and interferes with the channel bottle neck (Figure 5). It seems likely that Glu68 acts as a proton acceptor, forming the M-state (i.e., fast channel closing), in the D234N GtACR1 (Sineshchekov et al., 2016), as deprotonated Glu68 is sufficiently close to the protonated Schiff base (Figure 4b). The reorientation of deprotonated negatively charged Glu68 toward the protonated Schiff base and the interference with the channel bottle neck may also explain why the photocurrent owing to the anion conduction is abolished (Kim et al., 2018) irrespective of the accumulation of the M-state (Sineshchekov et al., 2016) (i.e., with deprotonated Schiff base) in the D234N GtACR1. It seems likely that the anion conduction is inhibited in the anion conducting L-state in the D234N GtACR1 because Glu68 already interferes with the channel bottle neck in the ground state. Thus, not only the gating (Arg94) but also the conduction (Glu68) can be inhibited in the D234N GtACR1.

Figure 5 with 1 supplement see all
Channel space in GtACR1.

(a) Channel space and side-chain orientations in the wild-type (gray sticks, PDB code 6EDQ; Li et al., 2019) and D234N (green sticks, a representative MD-generated conformation) GtACR1s. The yellow mesh indicates the channel space in the wild-type GtACR1 analyzed using the CAVER program (Chovancova et al., 2012). Note that the channel space is consistent with that reported by Li et al., 2019. Channel space and side-chain orientations in the representative MD-generated structures of (b) wild-type and (c) D234N GtACR1s. Chovancova et al., 2012 The red arrow indicates the decrease in the channel space (radius) owing to the approach of Glu68. (d) Channel radii along the channel in wild-type (dotted black line) and D234N (solid red line) GtACR1s. The red open circle indicates the constriction created by reoriented Glu68 in D234N GtACR1.

The mechanism is also likely to hold true for the M-state in the wild-type GtACR1, although the MD simulations were conducted based on the dark state structures. As far as we are aware, no intermediate structures of GtACR1 have been reported. The recent time-resolved X-ray free electron laser (XFEL) structures of cation channelrhodopsin C1C2 show that the distance between the Schiff base and Glu129 (Glu68 in GtACR1) remains unaffected during the early part of the photocycle irrespective of the isomerization of the retinal (Figure 5—figure supplement 1; Oda et al., 2021). In C1C2, not only the isomerization of the retinal but also a protein conformational change is required for the conducting-channel formation during the photocycle, as no continuous channel exists in the ground state (Kato et al., 2012). In contrast, a continuous channel spanning through the protein already exists in the ground state of GtACR1 (Figure 5; Li et al., 2019). From the analogy, it seems plausible that the Schiff base interacts electrostatically with Glu68 in the M-state, inhibiting the anion conduction.

The formation of an H-bond between Asn234 and deprotonated Glu68 in the D234N GtACR1 (Figure 5) also suggests that Glu68 accepts the proton from protonated Asp234 in the M-state of the wild-type GtACR1 (Figure 6). Based on the observation of the ground state structure, it seems possible that the proton transfer pathway that proceeds from the protonated Schiff base via deprotonated Asp234 toward protonated Glu68 can also form in the M-state. Protonated Glu68 can accept the proton from transiently protonated Asp234 and simultaneously donates the proton to the adjacent acceptor group in the H-bond network e.g., Asp-L213 in the bacterial photosynthetic reaction center (Sugo et al., 2021).

Formation of the proton transfer pathway and disconnection of the anion conduction channel during the L- to M-state transition.

The channel is open (red open arrow) in the anion conducting L-state. The reorientation of the Glu68 side chain (black and green curved arrow) leads to the formation of the H-bond network that proceeds from the Schiff base to Glu68 via Asp234 (blue dotted lines) and the closing of the channel bottle neck (fast channel closing). Thus, the release of the proton from the Schiff base toward Glu68 occurs in the M-state (blue open arrow).

Then, the absence of Glu68 as a proton acceptor of Asp234 may affect the release of the proton from the Schiff base in the E68Q GtACR1. Indeed, it has been reported that the E68Q mutation affects the channel-gating mechanism (Sineshchekov et al., 2015), leading to a decrease in the M-state (i.e., deprotonated Schiff base) accumulation (Sineshchekov et al., 2016). The conformation of Glu68 as a proton acceptor of Asp234 interferes with the channel bottle neck (Figure 5). This may explain why the M-state formation (i.e., release of the proton from the Schiff base via Asp234 to Glu68) corresponds to the fast channel closing (Figures 2 and 6). This may also explain why the mutation of Glu68 to glutamine leads to a suppression of the fast channel closing at a physiological pH (Sineshchekov et al., 2015).

Conclusions

It was proposed that Asp234 was protonated in the wild-type GtACR1 (Kim et al., 2018; Sineshchekov et al., 2016; Yi et al., 2016; Kandori, 2020) from the following results: (i) the absorption wavelength remains unchanged upon the D234N mutation (Kim et al., 2018; Sineshchekov et al., 2016; Yi et al., 2016); (ii) the C=C stretching frequency of the retinal is not significantly affected upon the D234N mutation in resonance Raman spectroscopy, which implies that the electrostatic interaction between the retinal and the protein environment remains unchanged (Yi et al., 2016); and (iii) the C=O stretching frequencies of 1740 (–)/1732 (+) cm–1 for a protonated carboxylate, which is observed in the wild-type GtACR1, disappear in the D234N GtACR1 at 77 K (Kim et al., 2018). However, the C=O stretching frequencies of 1740 (–)/1732 (+) cm–1 for protonated Asp234 at 77 K were not observed at 293 K by Dreier et al., 2021.

In contrast, the present results show that Asp234 is deprotonated in the wild-type GtACR1, as indicated by the following findings. (i) The E68Q/D234N mutation leads to an increase in the absorption wavelength (Table 4), which indicates that Glu68 or Asp234 is deprotonated in the wild-type GtACR1 (Table 1). (ii) The absorption wavelength in the D234E GtACR1 is shorter than in the wild-type protein (Table 4), which can be explained only by the presence of a deprotonated acidic residue (Table 1, Supplementary file 1D). (iii) The calculated pKa value of –5 for Asp234 is lower than that of –2 for the corresponding residue Asp212 in BR (Saito et al., 2012; Table 2). The significantly low pKa value can be understood as Asp85 in BR, which increases pKa(Asp212) by 6 (Saito et al., 2012), being replaced with Ser97 in GtACR1 (Table 2). The calculated protonation pattern shows that Asp234 is deprotonated in the wild-type GtACR1 even using the MD-generated conformations with protonated Asp234 (Table 1). (iv) Glu68, which is protonated in the wild-type GtACR1, is deprotonated in the D234N GtACR1 (Table 1). If Glu68 remained protonated in the D234N GtACR1, the absorption wavelength would be significantly longer as compared with the wild-type GtACR1 (Table 4). This is consistent with the FTIR measurements, which show that Glu68 is deprotonated in the D234N GtACR1 (Dreier et al., 2021). In any GtACR1, a negative charge needs to be present as far as the Schiff base is protonated. (v) Mutation of deprotonated Asp234 to uncharged asparagine does not alter the calculated absorption wavelength because deprotonated Glu68 reorients and interacts with the Schiff base in the D234N GtACR1, compensating for the change in the charge at the 234 site (Table 4, Table 5, Figure 4). Thus, the absence of changes in the absorption wavelength upon the D234N mutation (Kim et al., 2018; Sineshchekov et al., 2016; Yi et al., 2016) does not serve as a basis of the presence of protonated Asp234 in the wild-type GtACR1. The charge compensation by Glu68 can also explain why the C=C stretching frequency of the retinal, which reflects the electrostatic interaction between the retinal and the protein environment, does not significantly change upon the D234N mutation in resonance Raman spectroscopy (Yi et al., 2016).

The following mechanism can be deduced from the present findings: in D234N GtACR1, anionic Glu68 reorients toward the Schiff base to interact electrostatically. If Asp234 accepts a proton from the Schiff base in the M-state of the wild-type GtACR1, Glu68 is likely to reorient toward the channel, decreasing in the channel radius and inhibiting the anion conduction structurally. Simultaneously, the approach of anionic Glu68 toward the channel pore inhibits anion conduction electrostatically (Figure 5). The mechanism presented here explains why (i) the loss of photocurrent occurs upon the D234N mutation (Kim et al., 2018), (ii) the M-state formation corresponds to the fast channel closing (Sineshchekov et al., 2016), and (iii) the Glu68 to Gln mutation leads to a suppression of the fast channel closing at a physiological pH (Sineshchekov et al., 2015). The formation of the proton transfer pathway in the M-state, which proceeds from the Schiff base via Asp234 and Glu68 toward the protein bulk surface (Figure 6), can explain (iv) the accumulation of the M-state (i.e., deprotonation of the Schiff base) in the wild-type, D234N, and E68Q GtACR1s.

When the properties of a protein (e.g., absorption wavelength) remain unchanged upon the mutation of aspartate to asparagine, one may assume that the aspartate is protonated. However, this does not hold true for the following cases: (i) when the aspartate is adjacent to the focusing site (e.g., forming an H-bond), because the H-bond character (e.g., polarity and pattern) of asparagine is not identical to that of protonated aspartate irrespective of the same net charge; (ii) when another titratable residue exists near the aspartate (e.g., Glu68 in GtACR1) and the protonation states of the two residues are linked. The present example shows that asparagine mutation is not always equivalent to protonated aspartate especially when it is directly involved in the H-bond with the focusing site.

Materials and methods

Coordinates and atomic partial charges

Request a detailed protocol

The atomic coordinates were taken from the X-ray structure of GtACR1 monomer unit ‘A’ (PDB code 6EDQ; Li et al., 2019). All crystal water molecules were included explicitly in calculations if not otherwise specified. During the optimization of hydrogen atom positions with CHARMM (Brooks et al., 1983), the positions of all heavy atoms were fixed, and all titratable groups (e.g., acidic and basic groups) were ionized. The Schiff base was considered protonated. Atomic partial charges of the amino acids and retinal were obtained from the CHARMM22 (MacKerell et al., 1998) parameter set.

Protonation pattern

Request a detailed protocol

The computation was based on the electrostatic continuum model, solving the linear Poisson–Boltzmann equation with the MEAD program (Bashford and Karplus, 1990). The difference in electrostatic energy between the two protonation states, protonated and deprotonated, in a reference model system was calculated using a known experimentally measured pKa value (e.g., 4.0 for Asp; Nozaki and Tanford, 1967). The difference in the pKa value of the protein relative to the reference system was added to the known reference pKa value. The experimentally measured pKa values employed as references were 12.0 for Arg, 4.0 for Asp, 9.5 for Cys, 4.4 for Glu, 10.4 for Lys, 9.6 for Tyr, (Nozaki and Tanford, 1967), and 7.0 and 6.6 for the Nε and Nδ atoms of His, respectively (Tanokura, 1983a; Tanokura, 1983b; Tanokura, 1983c). All other titratable sites were fully equilibrated to the protonation state of the target site during titration. The dielectric constants were set to 4 inside the protein and 80 for water. All water molecules were considered implicitly. All computations were performed at 300 K, pH 7.0, and with an ionic strength of 100 mM. The linear Poisson–Boltzmann equation was solved using a three-step grid-focusing procedure at resolutions of 2.5, 1.0, and 0.3 Å. The ensemble of the protonation patterns was sampled by the Monte Carlo (MC) method with the Karlsberg program (Rabenstein and Knapp, 2001). The MC sampling yielded the probabilities [protonated] and [deprotonated] of the two protonation states of the molecule.

MD simulations

Request a detailed protocol

The GtACR1 assembly was embedded in a lipid bilayer consisting of 258 1-palmitoyl-2-oleyl-sn-glycero-3-phosphocholine (POPC) molecules using CHARMM-GUI (Jo et al., 2008), and soaked in 29070–29072 TIP3P water models, and 5–7 chloride ions were added to neutralize the system using the VMD plug-ins (Humphrey et al., 1996). After structural optimization with position restraints on heavy atoms of the GtACR1 assembly, the system was heated from 0.1 to 300 K over 5.5 ps with time step of 0.01 fs, equilibrated at 300 K for 1 ns with time step of 0.5 fs, and annealed from 300 to 0 K over 5.5 ps with time step of 0.01 fs. The heating and annealing processes to energetically relax the positions of POPC and TIP3 water molecules were performed with time step of 0.01 fs, as done in previous studies (Kurisaki et al., 2015a; Kurisaki et al., 2015b). To avoid the influence of changes in the retinal Schiff base structure on the excitation energy, the position restraints on heavy atoms of side chains were released and MD simulations were performed; the system was heated from 0.1 K to 300 K over 5.5 ps with time step of 0.01 fs and equilibrated at 300 K for 1 ns with time step of 0.5 fs. The system was equilibrated at 300 K for 5 ns with time step of 1.0 fs, and a production run was conducted over 1 ns with 1.0 fs step for sampling of side-chain orientations. 10 conformations were sampled at 0.1 ns intervals during the 1 ns production run. All MD simulations were conducted with the CHARMM22 (MacKerell et al., 1998) force field parameter set using the MD engine NAMD version 2.11 (Phillips et al., 2005). For MD simulations with time step of 1.0 fs, the SHAKE algorithm for hydrogen constraints was employed (Ryckaert et al., 1977). For temperature and pressure control, the Langevin thermostat and piston were used (Feller et al., 1995; Kubo et al., 1991).

POPC molecules have little effect on the calculated pKa values of Asp234 and Glu68 nor the absorption wavelengths (Supplementary file 1F and G). Using the resulting coordinates, the protonation state of the titratable residues was finally determined with the MEAD (Bashford and Karplus, 1990) and Karlsberg (Rabenstein and Knapp, 2001) programs in the absence of POPC molecules.

QM/MM calculations

Request a detailed protocol

Using 10 MD-generated protein conformations, the geometry was optimized using a QM/MM approach in the absence of POPC molecules. The restricted density functional theory (DFT) method was employed with the B3LYP functional and LACVP* basis sets using the QSite (Schrödinger, 2012) program. The QM region was defined as the retinal and Schiff base (Lys238). All atomic coordinates were fully relaxed in the QM region, and the protonation pattern of titratable residues was implemented in the atomic partial charges of the corresponding MM region. In the MM region, the positions of H atoms were optimized using the OPLS2005 force field (Jorgensen et al., 1996), while the positions of the heavy atoms were fixed.

The absorption energy of microbial rhodopsins is highly correlated with the energy difference between highest occupied molecular orbital (HOMO) and lowest unoccupied molecular orbital (LUMO) (ΔEHOMO-LUMO) or the lowest excitation energy calculated using time-dependent (TD) DFT (ETD-DFT) of the retinal Schiff base (Tsujimura and Ishikita, 2020; Tsujimura et al., 2021). To calculate absorption energies and corresponding wavelengths, the energy levels of HOMO and LUMO and the lowest excitation energies were calculated in the absence of POPC molecules. The absorption energy (Eabs in eV) was calculated using the following equations, which are obtained for wild-type and six mutant GtACR1s (coefficients of determination R2 = 0.93 for Equation 1 and 0.73 for Equation 2; Tsujimura et al., 2021):

(1) Eabs=0.842ΔEHOMOLUMO+0.408
(2) Eabs=1.455ETDDFT1.056

The HOMO-LUMO energy gap and the lowest excitation energy were calculated based on 10 MD-generated/QM/MM-optimized protein conformations (see Source data 1 for the atomic coordinates). Empirically, the correlation between the calculated and experimentally measured absorption energies is higher in ΔEHOMO-LUMO than in ETD-DFT among 13 microbial rhodopsin crystal structures (Tsujimura and Ishikita, 2020) and GtACR1 mutants (Tsujimura et al., 2021). In the present study, we analyze the absorption wavelengths of microbial rhodopsin proteins using Equation 1 based on the empirically corrected Eabs (Zhang and Musgrave, 2007) (in eV) or the corresponding wavelength (in nm).

A QM/MM approach utilizing the polarizable continuum model (PCM) method with a dielectric constant of 78 for the bulk region, in which electrostatic and steric effects created by a protein environment were explicitly considered in the presence of bulk water, was employed. In the PCM method, the polarization points were placed on the spheres with a radius of 2.8 Å from the center of each atom to describe possible water molecules in the cavity. The radii of 2.8–3.0 Å from each atom center and the dielectric constant values of ~80 are likely to be optimal to reproduce the excitation energetics, as evaluated for the polarizable QM/MM/PCM approach (Tamura et al., 2020). The TD-DFT method with the B3LYP functional and 6-31G* basis sets was employed using the GAMESS program (Schmidt et al., 1993). The trends in the shifts of absorption wavelength with respect to wild-type GtACR1 remain unchanged when the functional/basis set is replaced (e.g., the CAM-B3LYP functional; Yanai et al., 2004; Figure 3—figure supplement 2).

The electrostatic contribution of the side chain in the MM region to the absorption wavelength of the retinal Schiff base was obtained from the shift in the HOMO-LUMO energy gap upon the removal of the atomic charges of the focusing side chain.

Gene preparation

Request a detailed protocol

The cDNA of GtACR1 (GenBank accession number: KP171708) was optimized for human codon usage and fused to a C-terminal sequence encoding a hexahistidines-tag. The fusion product was inserted into the pCAGGS mammalian expression vector, as previously described (Kojima et al., 2017; Kojima et al., 2020). GtACR1 cDNAs containing mutations were constructed using the In-Fusion Cloning Kit according to the manufacturer’s instructions (Kojima et al., 2017; Kojima et al., 2020).

Protein expression and purification of GtACR1

Request a detailed protocol

The expression plasmids were transfected into HEK293T cells using the calcium-phosphate method (Kojima et al., 2017; Kojima et al., 2020). HEK 293T cells were a gift from Dr. Satoshi Koike (Tokyo Metropolitan Organization for Medical Research). We have confirmed that the identity has been authenticated by STR profiling and the cell lines tested negative for mycoplasma contamination. We have not used any cell lines from the list of commonly misidentified cell lines maintained by the International Cell Line Authentication Committee. After 1-day incubation, all-trans-retinal (final concentration = 5 μM) was added to transfected cells. After another day incubation, the cells were collected by centrifugation (7510 × g for 10 min) at 4°C and suspended in Buffer-A (50 mM HEPES [pH 7.0] and 140 mM NaCl). All-trans-retinal (final concentration = 0.31 μM) was added to the cell suspension to reconstitute the photoactive pigments by shaking rotatory for more than 12 hr at 4°C. Then, the cells were collected by centrifugation (12900 × g for 30 min) at 4°C and suspended in Buffer-A and solubilized in Buffer-B (20 mM HEPES [pH 7.4], 300 mM NaCl, 5% glycerol, and 1% dodecyl maltoside [DDM]). The solubilized fraction was collected by ultracentrifugation (169,800 × g for 20 min) at 4°C, and the supernatant was applied to a Ni2+ affinity column to purify the pigments. After the column was washed with Buffer-C (20 mM HEPES [pH 7.4], 300 mM NaCl, 5% glycerol, 0.02% DDM, and 20 mM imidazole), the pigment was eluted with a linear gradient of imidazole by Buffer-D (20 mM HEPES [pH 7.4], 300 mM NaCl, 5% glycerol, 0.02% DDM, and 1 M imidazole). Purified samples were concentrated by centrifugation using an Amicon Ultra filter (30,000 Mw cut-off; Millipore, USA) and the buffer was exchanged using PD-10 column (GE Healthcare, USA) to Buffer-E (20 mM HEPES [pH 7.4], 300 mM NaCl, 5% glycerol, and 0.02% DDM).

Spectroscopic analysis

Request a detailed protocol

Absorption spectra of the purified proteins were recorded with a UV–visible spectrophotometer (Shimadzu, UV-2450, UV-2600) in Buffer-E. The samples were kept at 15°C using a thermostat.

Data availability

All data generated or analysed during this study are included in the manuscript and supporting files.

The following previously published data sets were used

References

  1. Book
    1. Molecular Operating Environment (MOE)
    (2021)
    2019.01; Chemical Computing Group ULC, 1010 Sherbooke St. West, Suite #910, Montreal, QC, Canada, H3A 2R7
    Chemical Computing Group ULC.

Decision letter

  1. Qiang Cui
    Reviewing Editor; Boston University, United States
  2. José D Faraldo-Gómez
    Senior Editor; National Heart, Lung and Blood Institute, National Institutes of Health, United States
  3. Qiang Cui
    Reviewer; Boston University, United States

Our editorial process produces two outputs: i) public reviews designed to be posted alongside the preprint for the benefit of readers; ii) feedback on the manuscript for the authors, including requests for revisions, shown below. We also include an acceptance summary that explains what the editors found interesting or important about the work.

Decision letter after peer review:

Thank you for submitting your article "Proton-mediated gating mechanism of anion channelrhodopsin-1" for consideration by eLife. Your article has been reviewed by 3 peer reviewers, including Qiang Cui as the Reviewing Editor and Reviewer #1, and the evaluation has been overseen by José Faraldo-Gómez as the Senior Editor.

The reviewers have discussed their reviews with one another, and the Reviewing Editor has drafted this to help you prepare a revised submission.

Essential revisions:

1) Several reviewers raised concern in using the HOMO-LUMO gap to model the absorption maximum instead of state-of-the-art QM/MM methods based on, for example, TD-DFT or ab initio methods. This should be possible for the current system considering the modest size of the QM region.

2) Essentially all reviewers are concerned about the statistical errors associated with the reported simulation data. These should be analyzed carefully and reported explicitly.

3) One reviewer highlighted the importance of examining an alternative protonation state of the system [E68(-)/D234(H)].

4) As one reviewer noted, the title of the manuscript implies insight into the "gating mechanism" of GtACR1, but it was actually never simulated like, for example, in the work from Ardevol and Hummer https://www.pnas.org/content/115/14/3557. Instead the conclusions about the gating mechanism are based on simulations of the resting state of the protein. You explained that the actual channel function is correlated with later intermediates of the photocycle (Figure 2 in the manuscript). However, already in the first intermediate the retinal changes the conformation thus the protein conformation might change, which could impact the distance to E68. Therefore, it is important to discuss the mechanistic implication of using the ground state structure in the computational analysis.

Reviewer #1 (Recommendations for the authors):

Overall, the authors did an excellent job in combining different computational methodologies to address the issue of residue titration states in proteins.

Reviewer #2 (Recommendations for the authors):

1) One of the major concerns is that the authors use HOMO-LUMO energy difference to compare them to experimental absorption maxima. A HOMO-LUMO energy gap is not an excitation energy and not the absorption maximum. The main conclusions in this paper are based on these calculations. Therefore, I am concerned about the reliability of the results. To improve it, there is a wide range of methods that can be used instead. I don't mind if the authors use TD-DFT or wave function methods as long as excitation energies are computed.

2) The authors report that 10 geometries were used for HOMO-LUMO gap computation and then averaged. However, the numbers are not shown anywhere. I would like to ask to list it in the SI. It would be for example interesting to see the fluctuation of the values because the differences between the different mutants are rather small.

3) The discussion of the gating mechanism is done on the basis of the dark state structure (Figure 5 and 6), i.e. the retinal is not isomerized and the protein conformation didn't change. So there is some uncertainty in the prediction of the gating mechanism which should be mentioned and discussed.

4) There is very little or no data from the MD trajectories which are used to draw conclusions about the protonation states. For example the RMSD from each trajectory as well as critical distances. This can be also added to the SI.

5) The main conclusion is that Asp234 is deprotonated because the D234E mutant is blue shifted. But this can also be explained by the ability of the protonated D234 to form hydrogen bonds with the Schiff base. Then the closer proximity of E234 would also result in a blue shift.

6) Also the fact that D234E/E68D double mutants is red shifted by only 7 nm can explained by the fact that in D234E the residue E68 is closer to the Schiff base but in the double mutant both are neutral. In addition, the shift is of 7 nm is rather small.

7) In several places the authors mention the similarity between the electrostatic potential in bR and GtACR1. However, bR has a second counter ion (D85) and GtACR1 has a neutral residue in this position. This is close to the Schiff bas region and is therefore important. So I recommend to rethink this phrasing.

8) How can E68 accept a proton in the wild type if it is protonated?

Reviewer #3 (Recommendations for the authors):

This is interesting work and I support its publication following revision. My concerns and suggestions for improvement are:

1) It is clear that one of the acidic residues, E68 or D234, must be deprotonated in order for the Schiff base to be protonated. However, the paper would be improved by a more thorough comparison of the interplay or exchangeability of E68 and D234. In many of the analyses, the authors only compare E68(H)/D234(-) to E68(H)/D234(H). They do not test E68(-)/D234(H)-e.g., MD simulations in Table 1, pKa values in tables 2 and 3, and the calculated spectroscopic shifts in table 4. The mutant results for D234N already suggest that E68 can take D234's role in stabilizing the protonated Schiff base. However, this mutant should not take the place of the wildtype system with E68(-)/D234(H). As emphasized in their concluding paragraph, Asp and Asn can behave differently.

2) For example, focusing on the measured shifts in absorption wavelengths (Table 4), mutating either residue to the opposite acidic residue (i.e., E68D and D234E) results in a decrease in wavelength. Similarly, mutating either to neutral (E68Q and D234N) results in very little shift (+2 and -3, respectively). Is this difference significant or within error of being equivalent? If the latter, why is it that the authors conclude only E68(H)/D234(-) is present in the room temperature wild type system? Couldn't E68(-)/D234(H) be a contributing species? Similarly, the authors could have calculated the shifts for E68Q in both the protonated and deprotonated forms of D234 (Table 4).

3) Focusing on pKa calculations, which will clearly hold their trends for the crystal structure, the pKa of E68 should also be measured post simulation and relaxation of the E68(-)/D234(H) state.

4) An additional concern is the lack of reported error bars. The magnitude of error should be reported for each quantitative calculated and measured result. This is essential to discern which conclusions are statistically supported and outside of their measured or calculated error bars.

5) It would help the reader to clearly delineate what is new in this work and what has been previously reported. It seems only the absorption of E68Q/D234N is new? What conclusions are new?

6) The final concluding paragraph about the influence of Asp to Asn mutations does not seem well supported by these findings. In this presented case there is a functional change due to the D234N mutation, even though there is little change in absorption due to compensating D68 deprotonation. This conclusion should perhaps be reconsidered/reframed. If kept as is, the motivation based on the reported findings needs to be more clearly explained.

7) The DFT used in the QM/MM was fairly basic (B3LYP with 6-31G* basis sets). How was this choice justified for the given system? I.e., to what higher level theory and larger basis sets was this compared to in order to verify it works well enough for these calculations?

8) Line 311 on page 17 replace ', which are caused by…asparagine,.…' with 'that are caused by.… asparagine.…' removing the commas since the clause specifies the subject-it is not additional information about the subject.

[Editors' note: further revisions were suggested prior to acceptance, as described below.]

Thank you for resubmitting your work entitled "Proton transfer pathway in anion channelrhodopsin-1" for further consideration by eLife. Your revised article has been reviewed by 3 peer reviewers, one of whom is a member of our Board of Reviewing Editors, and the evaluation has been overseen by José Faraldo-Gómez as the Senior Editor.

Editors and reviewers concur that the manuscript has been improved; however, there are some remaining issues that must be addressed or clarified before the manuscript can be accepted for publication, especially about the TD-DFT calculations and MD simulations for the pKa evaluations. We urge you to provide a compelling resolution to these issues.

Reviewer #1:

The authors included additional TD-DFT calculations, and also sampled more configurations as well as protonation patterns. They have also clarified a few questions raised in the last round of review. In my opinion, the revision is acceptable for publication in eLife.

Reviewer #2:

1) The authors have provided TD-DFT results, following the suggestion from all three reviewers. The results are qualitatively similar to the HOMO-LUMO energy gaps. But there is one important difference. The structure with Cl- is blue shifted with respect to the parent state at the TD-DFT level, while it is red shifted when the HOMO-LUMO energy gap is used for the calculation of excitation energies.

A comparison of the differences in orbital energies to absorption maxima is physically incorrect (also pointed out by other reviewers). Therefore, I recommend the authors to add the TD-DFT results to the table 4 in the manuscript.

2) I appreciate adding the excitation energies from the 10 snapshots of each GtACR1 variant. It shows that the wide distribution for E68D agrees with the wide band from experiment. Maybe this is due to some heterogeneity in the structure. It would be interesting to check this.

3) It is not clear how the authors know that the M intermediate is the conducting intermediate in GtACR1. As far as I know the structure of the M intermediate is also not known for C1C2, the artificial chimaera of channelrhodopsin 1 and 2. More importantly, it is not just the isomerization but also a conformational change in the protein which allows the pore formation. It is not clear how this is taken into account in the simulation. I'm not asking the author to redo the simulations, but simply inform the reader about this important differences.

4) The proton transfer is outside of my expertise. I thought the Grotthuss mechanism is proposed for a network of water molecules.

In addition to these replies, the use of a 0.01 fs time step for the numerical integration of the equations of motions in MD makes no sense. This is at least 50 times too small.

Reviewer #3:

My one remaining concern is that the length of the MD simulations in each protonation state should be stated and the method for selecting conformations from those simulations should be described. The reason for this is that the pKa calculations will be most dependent on conformation. Their conclusions hinge on (1) having run simulations long enough to stabilize each charge state, and (2) selecting a reasonable number of conformations that were representative of the stablilized ensemble. Ideally they would report the number of conformations they chose, how probable those structures were in the simulations (cluster and count based on positions of the important residues), and show a representative figure of the structures in the SI.

Other than this, I am happy with the author's revisions.

https://doi.org/10.7554/eLife.72264.sa1

Author response

Essential revisions:

1) Several reviewers raised concern in using the HOMO-LUMO gap to model the absorption maximum instead of state-of-the-art QM/MM methods based on, for example, TD-DFT or ab initio methods. This should be possible for the current system considering the modest size of the QM region.

We have calculated in all cases using (i) TD-DFT and (ii) different basis set and functional accordingly (Supplementary files 1C, D, and G, Figure 3—figure supplement 1 and 2).

2) Essentially all reviewers are concerned about the statistical errors associated with the reported simulation data. These should be analyzed carefully and reported explicitly.

We have shown the statistical error in Figure 3—figure supplement 1 and 2.

3) One reviewer highlighted the importance of examining an alternative protonation state of the system [E68(-)/D234(H)].

According to the suggestion from reviewer 3 (points 1, 2, and 3), we have conducted MD simulations assuming the E68(-)/D234(H) state and repeated electrostatic calculations for the protonation pattern, solving the Poisson-Boltzmann equation. The results have been added mainly to Table 1.

4) As one reviewer noted, the title of the manuscript implies insight into the "gating mechanism" of GtACR1, but it was actually never simulated like, for example, in the work from Ardevol and Hummer https://www.pnas.org/content/115/14/3557. Instead the conclusions about the gating mechanism are based on simulations of the resting state of the protein. You explained that the actual channel function is correlated with later intermediates of the photocycle (Figure 2 in the manuscript). However, already in the first intermediate the retinal changes the conformation thus the protein conformation might change, which could impact the distance to E68. Therefore, it is important to discuss the mechanistic implication of using the ground state structure in the computational analysis.

As the editor pointed out, we discussed only based on the ground state structure. We realized that the main focus is on the protonation states and proton transfer, rather than gating mechanism.

Thus, we have modified the title of the manuscript as follows: “Proton transfer pathway in anion channelrhodopsin-1”.

Reviewer #2 (Recommendations for the authors):

1) One of the major concerns is that the authors use HOMO-LUMO energy difference to compare them to experimental absorption maxima. A HOMO-LUMO energy gap is not an excitation energy and not the absorption maximum. The main conclusions in this paper are based on these calculations. Therefore, I am concerned about the reliability of the results. To improve it, there is a wide range of methods that can be used instead. I don't mind if the authors use TD-DFT or wave function methods as long as excitation energies are computed.

The comment is correct. We have also shown the results of TD-DFT. The two results are essentially the same (Figure 3—figure supplement 1 and 2 and Supplementary file 1C). Just empirically, the correlation between the experimentally measured absorption and calculated wavelengths is higher in ∆EHOMO-LUMO than in ETD-DFT among 13 microbial rhodopsin crystal structures (J. Phys. Chem. B 124 (2020) 11819) and GtACR1 mutants (Biochim. Biophys. Acta 1862 (2021) 148349), although the reason is unclear.

2) The authors report that 10 geometries were used for HOMO-LUMO gap computation and then averaged. However, the numbers are not shown anywhere. I would like to ask to list it in the SI. It would be for example interesting to see the fluctuation of the values because the differences between the different mutants are rather small.

We have shown all values for the 10 geometries of all mutant GtACR1s in Supplementary file 1C. To show the fluctuation of each value with error bar, we have also made Figure 3—figure supplement 1.

3) The discussion of the gating mechanism is done on the basis of the dark state structure (Figure 5 and 6), i.e. the retinal is not isomerized and the protein conformation didn't change. So there is some uncertainty in the prediction of the gating mechanism which should be mentioned and discussed.

The comment is correct. To evaluate whether the present mechanism deduced from the ground state structure is applicable to the M-state, we have analysed the structural difference between the all-trans and 13-cis conformations in the reported structures of C1C2. The most crucial distance, the Schiff base…counterion distance, is identical for the two conformations in C1C2 (7.6 Å and 7.2 Å, respectively, Figure 5—figure supplement 1). The overall geometry at the Schiff base/counterion moiety remains unchanged. From the analogy, we concluded that the present mechanism is likely to hold true for the M-state in GtACR1.

We have illustrated the structural difference in C1C2 (Figure 5—figure supplement 1) and discussed the mechanism for GtACR1 from the analogy (near Figure 5).

4) There is very little or no data from the MD trajectories which are used to draw conclusions about the protonation states. For example the RMSD from each trajectory as well as critical distances. This can be also added to the SI.

We consider N(Schiff Base)…O(Asp/Glu234) as the most critical distance in the present study, because the protonation state of Asp/Glu234 is the central focus and it affects the absorption wavelength most critically.

We have shown the changes in N…O distance in wild type and D234E GtACR1s during MD simulations (Figure 4—figure supplement 1).

5) The main conclusion is that Asp234 is deprotonated because the D234E mutant is blue shifted. But this can also be explained by the ability of the protonated D234 to form hydrogen bonds with the Schiff base. Then the closer proximity of E234 would also result in a blue shift.

As a careful check suggested by the reviewer, we have conducted MD simulations with protonated Asp/Glu234. The calculated absorption wavelength with protonated Glu234 in the D234E mutant remains unchanged with respect to wild type, which cannot explain the experimentally measured blue-shift of 10 nm. This is because when Asp/Glu234 are protonated, the electrostatic influence of charge neutral residues on the Schiff base are significantly small (with respect to deprotonated anionic Asp/Glu234), being almost the same (Supplementary file 1D).

We have summarized the calculated absorption wavelengths with protonated Asp/Glu234 in Supplementary file 1D.

6) Also the fact that D234E/E68D double mutants is red shifted by only 7 nm can explained by the fact that in D234E the residue E68 is closer to the Schiff base but in the double mutant both are neutral. In addition, the shift is of 7 nm is rather small.

The D234E/E68D mutant was not analysed in the present study. From “red shifted by only 7 nm”, we guess that the reviewer probably referred to E68Q/D234N mutant. In this case, however, the suggestion is unclear to us.

7) In several places the authors mention the similarity between the electrostatic potential in bR and GtACR1. However, bR has a second counter ion (D85) and GtACR1 has a neutral residue in this position. This is close to the Schiff bas region and is therefore important. So I recommend to rethink this phrasing.

The comment is correct. Asp85 in BR is replaced with Ser97 in GtACR1. We have rephrased the corresponding sentence in page 4.

8) How can E68 accept a proton in the wild type if it is protonated?

As far as protonated acidic residues are involved in a Grotthuss-like conduit, it can mediate proton transfer, accepting the proton from the donor site and simultaneously donating the proton to the acceptor site (e.g., protonated AspL213 in the proton transfer pathway toward quinone in the photosynthetic reaction center. It has long been suggested that Asp-L213 is protonated and accepts a proton from Asp-M17 in the upstream and donates the proton to Ser-L213, the direct proton donor to the quinone). We have added an explanation to Discussion.

Reviewer #3 (Recommendations for the authors):

This is interesting work and I support its publication following revision. My concerns and suggestions for improvement are:

1) It is clear that one of the acidic residues, E68 or D234, must be deprotonated in order for the Schiff base to be protonated. However, the paper would be improved by a more thorough comparison of the interplay or exchangeability of E68 and D234. In many of the analyses, the authors only compare E68(H)/D234(-) to E68(H)/D234(H). They do not test E68(-)/D234(H)-e.g., MD simulations in Table 1, pKa values in tables 2 and 3, and the calculated spectroscopic shifts in table 4. The mutant results for D234N already suggest that E68 can take D234's role in stabilizing the protonated Schiff base. However, this mutant should not take the place of the wildtype system with E68(-)/D234(H). As emphasized in their concluding paragraph, Asp and Asn can behave differently.

We have performed MD simulations with E68(-)/D234(H). Although the protonation state was fixed to E68(-)/D234(H) during MD simulations, the protonation pattern calculated using the MD-generated geometry indicates that Glu68 is protonated and Asp234 is deprotonated. That is, the protein electrostatic environment is already preorganized to stabilize protonated Glu68 and deprotonated Asp234, eliminating the possibility of E68(-)/D234(H). The result has been added to Table 1.

Using the 10 MD-generated conformations, we have calculated the pKa values. All conformations show that Glu68 is protonated and Asp234 is deprotonated. The result has been summarized in Supplementary file 1B.

These results suggest that the protein electrostatic environment does not allow to have E68(-)/D234(H) at least in the original geometry of the ground-state structure.

Because the E68(-)/D234(H) state cannot be obtained in all MD-generated conformations, we were unable to calculate the corresponding absorption wavelength.

2) For example, focusing on the measured shifts in absorption wavelengths (Table 4), mutating either residue to the opposite acidic residue (i.e., E68D and D234E) results in a decrease in wavelength. Similarly, mutating either to neutral (E68Q and D234N) results in very little shift (+2 and -3, respectively). Is this difference significant or within error of being equivalent? If the latter, why is it that the authors conclude only E68(H)/D234(-) is present in the room temperature wild type system? Couldn't E68(-)/D234(H) be a contributing species? Similarly, the authors could have calculated the shifts for E68Q in both the protonated and deprotonated forms of D234 (Table 4).

E68(-)/D234(H) is not stable in the original geometry of the crystal structure (i.e., without performing MD simulations, Tables 2 and 3).

Even if we intentionally force to stabilize the E68(-)/D234(H) state at a room temperature (i.e., MD simulations, Table 1), the resulting conformation is far from having E68(-)/D234(H) (Table 1, wild type (3)). The energetically unstable E68(-)/D234(H) in any conformation indicates that electrostatic interactions between E68/D234 with the protein environment originating from the residues listed in Tables 2 and 3 cannot allow them to have E68(-)/D234(H). Since QM/MM calculations need energetically optimized geometry optimization before calculating absorption wavelength (otherwise it makes no sense), no further calculation for the energetically unstable E68(-)/D234(H) conformation can be performed.

We have added the result of E68(-)/D234(H) to Table 1.

3) Focusing on pKa calculations, which will clearly hold their trends for the crystal structure, the pKa of E68 should also be measured post simulation and relaxation of the E68(-)/D234(H) state.

We have shown the result in Table 1, wild type (3), and in Supplementary file 1B.

4) An additional concern is the lack of reported error bars. The magnitude of error should be reported for each quantitative calculated and measured result. This is essential to discern which conclusions are statistically supported and outside of their measured or calculated error bars.

We have summarized the statistical error of the reported data for calculations in Figure 3—figure supplement 1.

For each measured absorption spectrum, the results of 4-traces were averaged to improve the signal-to-noise ratio. Absorption wavelengths were measured from the averaged spectra. This was not mentioned in the original version. We have added the explanation to the caption, Figure 3.

5) It would help the reader to clearly delineate what is new in this work and what has been previously reported. It seems only the absorption of E68Q/D234N is new? What conclusions are new?

We have explained the present new findings more clearly, making the “Conclusion” chapter and described the points therein.

6) The final concluding paragraph about the influence of Asp to Asn mutations does not seem well supported by these findings. In this presented case there is a functional change due to the D234N mutation, even though there is little change in absorption due to compensating D68 deprotonation. This conclusion should perhaps be reconsidered/reframed. If kept as is, the motivation based on the reported findings needs to be more clearly explained.

The original sentence was unclear to readers. We have rephrased the paragraph entirely, as we consider that this message is the most important outcome of the present study (that is, “Asp234 is deprotonated even though the absorption wavelength remained unchanged upon the D234N mutation”) and should be presented appropriately.

7) The DFT used in the QM/MM was fairly basic (B3LYP with 6-31G* basis sets). How was this choice justified for the given system? I.e., to what higher level theory and larger basis sets was this compared to in order to verify it works well enough for these calculations?

We have also calculated with (i) 6-31G** (basis set) and (ii) CAM-B3LYP (functional). The trends in the shifts of absorption wavelength with respect to wild type GtACR1 remain unchanged. The results have been shown in Figure 3—figure supplement 2.

8) Line 311 on page 17 replace ', which are caused by…asparagine,.…' with 'that are caused by.… asparagine.…' removing the commas since the clause specifies the subject-it is not additional information about the subject.

Thank you very much for pointing out our grammatical error. The corresponding sentence has been removed to address the point 6.

[Editors' note: further revisions were suggested prior to acceptance, as described below.]

Editors and reviewers concur that the manuscript has been improved; however, there are some remaining issues that must be addressed or clarified before the manuscript can be accepted for publication, especially about the TD-DFT calculations and MD simulations for the pKa evaluations. We urge you to provide a compelling resolution to these issues.

Reviewer #2:

1) The authors have provided TD-DFT results, following the suggestion from all three reviewers. The results are qualitatively similar to the HOMO-LUMO energy gaps. But there is one important difference. The structure with Cl- is blue shifted with respect to the parent state at the TD-DFT level, while it is red shifted when the HOMO-LUMO energy gap is used for the calculation of excitation energies.

A comparison of the differences in orbital energies to absorption maxima is physically incorrect (also pointed out by other reviewers). Therefore, I recommend the authors to add the TD-DFT results to the table 4 in the manuscript.

Thank you very much for your helpful advice. We have added the TD-DFT results to Table 4.

2) I appreciate adding the excitation energies from the 10 snapshots of each GtACR1 variant. It shows that the wide distribution for E68D agrees with the wide band from experiment. Maybe this is due to some heterogeneity in the structure. It would be interesting to check this.

After long discussion with the members, we have concluded that this is more likely due to “spectral shoulder” than “some heterogeneity in the structure”.

We have stated in Results as follows:

“In general, blue light-sensitive microbial rhodopsins (e.g., Sensory rhodopsin II and Chlamydomonas channelrhodopsins) show the main absorbance peak with spectral shoulder at shorter wavelength region (e.g. Takahashi, T. et al. Biochemistry 1990, 29, 8467-8474). Based on these, it seems likely that the wide band of E68D is due to the existence of the spectral shoulder of this blue-shifted mutant.”

3) It is not clear how the authors know that the M intermediate is the conducting intermediate in GtACR1.

The M intermediate state is not the conducting intermediate. Instead, the L intermediate is the anion conducting intermediate and the fast channel closing occurs during the L to M transition, as shown in Figure 2.

As far as I know the structure of the M intermediate is also not known for C1C2, the artificial chimaera of channelrhodopsin 1 and 2.

The comment is correct. The structure shown in Figure 5—figure supplement 1 (PDB code 7E6X) is just a “putative” M-state structure (the structure at the time point of ∆t = 4 ms, where the M-state is accumulated in the C1C2 crystals (Oda, K. et al. eLife 2021, 10, e62389)).

Accordingly, removing “the M-state” from C1C2, we have corrected the main text and the legend of Figure 5—figure supplement 1 to avoid confusion.

More importantly, it is not just the isomerization but also a conformational change in the protein which allows the pore formation. It is not clear how this is taken into account in the simulation. I'm not asking the author to redo the simulations, but simply inform the reader about this important differences.

We have stated that not only the isomerization but also a conformation change is required for the conducting-channel formation during the photoycle (in Discussion).

4) The proton transfer is outside of my expertise. I thought the Grotthuss mechanism is proposed for a network of water molecules.

The comment is correct. We have replaced “Grotthuss-like proton conduit” with “H-bond network”.

In addition to these replies, the use of a 0.01 fs time step for the numerical integration of the equations of motions in MD makes no sense. This is at least 50 times too small.

The comment is indeed correct. As the reviewer suggested, the time step of 0.01 fs is too short. Although the short time step was used only for heating and annealing processes, it makes no sense. We will not use a too-short time step next time. Thank you very much for the important comment.

Reviewer #3:

My one remaining concern is that the length of the MD simulations in each protonation state should be stated and the method for selecting conformations from those simulations should be described. The reason for this is that the pKa calculations will be most dependent on conformation. Their conclusions hinge on (1) having run simulations long enough to stabilize each charge state, and (2) selecting a reasonable number of conformations that were representative of the stablilized ensemble.

We have shown the length of the MD simulations and the method for selecting conformations in the Method section and the footnote of Table 1.

Ideally they would report the number of conformations they chose, how probable those structures were in the simulations (cluster and count based on positions of the important residues), and show a representative figure of the structures in the SI.

In each protonation state, deprotonated/protonated Asp234 forms H-bonds with the adjacent residues through the 1 ns product run.

We have shown the representative conformations and the distances of the H-bonds of Asp234 in Figure 2—figure supplement 1.

https://doi.org/10.7554/eLife.72264.sa2

Article and author information

Author details

  1. Masaki Tsujimura

    Department of Applied Chemistry, The University of Tokyo, Tokyo, Japan
    Contribution
    Investigation, Writing – original draft
    Contributed equally with
    Keiichi Kojima
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-7554-7372
  2. Keiichi Kojima

    Graduate School of Medicine, Dentistry and Pharmaceutical Sciences, Okayama University, Okayama, Japan
    Contribution
    Investigation
    Contributed equally with
    Masaki Tsujimura
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-4729-0511
  3. Shiho Kawanishi

    Graduate School of Medicine, Dentistry and Pharmaceutical Sciences, Okayama University, Okayama, Japan
    Contribution
    Investigation
    Competing interests
    No competing interests declared
  4. Yuki Sudo

    Graduate School of Medicine, Dentistry and Pharmaceutical Sciences, Okayama University, Okayama, Japan
    Contribution
    Investigation
    For correspondence
    sudo@okayama-u.ac.jp
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-8155-9356
  5. Hiroshi Ishikita

    1. Department of Applied Chemistry, The University of Tokyo, Tokyo, Japan
    2. Research Center for Advanced Science and Technology, The University of Tokyo, Tokyo, Japan
    Contribution
    Investigation, Supervision, Writing – original draft, Writing - review and editing
    For correspondence
    hiro@appchem.t.u-tokyo.ac.jp
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-5849-8150

Funding

Japan Agency for Medical Research and Development (20dm0207060h0004)

  • Yuki Sudo

Core Research for Evolutional Science and Technology (JPMJCR1656)

  • Yuki Sudo
  • Hiroshi Ishikita

Japan Society for the Promotion of Science (JP21K15054)

  • Keiichi Kojima

Japan Society for the Promotion of Science (JP20K21482)

  • Yuki Sudo

Japan Society for the Promotion of Science (JP21H0040413)

  • Yuki Sudo

Japan Society for the Promotion of Science (JP21H0244613)

  • Yuki Sudo

Japan Society for the Promotion of Science (JP18H05155)

  • Hiroshi Ishikita

Japan Society for the Promotion of Science (JP18H01937)

  • Hiroshi Ishikita

Japan Society for the Promotion of Science (JP20H03217)

  • Hiroshi Ishikita

Japan Society for the Promotion of Science (JP20H05090)

  • Hiroshi Ishikita

University of Tsukuba (Interdisciplinary Computational Science Program in CCS)

  • Hiroshi Ishikita

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

This research was supported by AMED (20dm0207060h0004) to YS and JST CREST (JPMJCR1656 to YS and HI), JSPS KAKENHI (JP21K15054 to KK, JP20K21482, JP21H0040413, and JP21H0244613 to YS, and JP18H05155, JP18H01937, JP20H03217, and JP20H05090 to HI), and the Interdisciplinary Computational Science Program in CCS, University of Tsukuba.

Senior Editor

  1. José D Faraldo-Gómez, National Heart, Lung and Blood Institute, National Institutes of Health, United States

Reviewing Editor

  1. Qiang Cui, Boston University, United States

Reviewer

  1. Qiang Cui, Boston University, United States

Publication history

  1. Received: July 16, 2021
  2. Preprint posted: July 22, 2021 (view preprint)
  3. Accepted: November 25, 2021
  4. Version of Record published: December 21, 2021 (version 1)

Copyright

© 2021, Tsujimura et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 533
    Page views
  • 78
    Downloads
  • 1
    Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Open citations (links to open the citations from this article in various online reference manager services)

Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)

  1. Masaki Tsujimura
  2. Keiichi Kojima
  3. Shiho Kawanishi
  4. Yuki Sudo
  5. Hiroshi Ishikita
(2021)
Proton transfer pathway in anion channelrhodopsin-1
eLife 10:e72264.
https://doi.org/10.7554/eLife.72264

Further reading

    1. Biochemistry and Chemical Biology
    Sudipta Mondal et al.
    Research Article

    Chain-length specific subsets of diacylglycerol (DAG) lipids are proposed to regulate differential physiological responses ranging from signal transduction to modulation of the membrane properties. However, the mechanism or molecular players regulating the subsets of DAG species remains unknown. Here, we uncover the role of a conserved eukaryotic protein family, DISCO-interacting protein 2 (DIP2) as a homeostatic regulator of a chemically distinct subset of DAGs using yeast, fly and mouse models. Genetic and chemical screens along with lipidomics analysis in yeast reveal that DIP2 prevents the toxic accumulation of specific DAGs in the logarithmic growth phase, which otherwise leads to endoplasmic reticulum stress. We also show that the fatty acyl-AMP ligase-like domains of DIP2 are essential for the redirection of the flux of DAG subspecies to storage lipid, triacylglycerols. DIP2 is associated with vacuoles through mitochondria-vacuole contact sites and such modulation of selective DAG abundance by DIP2 is found to be crucial for optimal vacuole membrane fusion and consequently osmoadaptation in yeast. Thus, the study illuminates an unprecedented DAG metabolism route and provides new insights on how cell fine-tunes DAG subspecies for cellular homeostasis and environmental adaptation.

    1. Biochemistry and Chemical Biology
    2. Microbiology and Infectious Disease
    Atanas Radkov et al.
    Research Article

    Members of the bacterial T6SS amidase effector (Tae) superfamily of toxins are delivered between competing bacteria to degrade cell wall peptidoglycan. Although Taes share a common substrate, they exhibit distinct antimicrobial potency across different competitor species. To investigate the molecular basis governing these differences, we quantitatively defined the functional determinants of Tae1 from Pseudomonas aeruginosa PAO1 using a combination of nuclear magnetic resonance (NMR) and a high-throughput in vivo genetic approach called deep mutational scanning (DMS). As expected, combined analyses confirmed the role of critical residues near the Tae1 catalytic center. Unexpectedly, DMS revealed substantial contributions to enzymatic activity from a much larger, ring-like functional hot spot extending around the entire circumference of the enzyme. Comparative DMS across distinct growth conditions highlighted how functional contribution of different surfaces is highly context-dependent, varying alongside composition of targeted cell walls. These observations suggest that Tae1 engages with the intact cell wall network through a more distributed three-dimensional interaction interface than previously appreciated, providing an explanation for observed differences in antimicrobial potency across divergent Gram-negative competitors. Further binding studies of several Tae1 variants with their cognate immunity protein demonstrate that requirements to maintain protection from Tae activity may be a significant constraint on the mutational landscape of tae1 toxicity in the wild. In total, our work reveals that Tae diversification has likely been shaped by multiple independent pressures to maintain interactions with binding partners that vary across bacterial species and conditions.