Distinguishing functional from structural roles of conserved pore residues during formate translocation by the FocA anion channel

Abstract The formate‐specific anion channel FocA of Escherichia coli belongs to the superfamily of homopentameric formate‐nitrite transporters (FNT). Minimally nine amino acid residues are conserved in the formate translocation pore of each protomer of the pentamer, including a histidine (H209) and a threonine (T91), both of which are crucial for bidirectional formate translocation through the pore. Information regarding in vivo functional or structural roles for the other seven conserved residues is limited, or nonexistent. Here, we conducted an amino acid‐exchange analysis of these seven conserved residues. Using an established formate‐responsive lacZ‐based assay to monitor changes in intracellular formate levels and anaerobic growth rate due to the inhibitory formate analog hypophosphite, we identified five of the seven residues analyzed to be important for the structural integrity of the pentamer, in particular, two highly conserved asparagine residues, N213 and N262. The remaining two conserved residues, K156 and N172, were essential for formate/hypophosphite translocation. K156 is located on the periplasmic fringe of the pore and aids the attraction of formate to the channel. Here, we show that this residue is also important for formate efflux from the cytoplasm to the periplasm, suggesting a role in formate release from the pore. N172 could be replaced by alanine with retention of low‐level bidirectional anion translocation function; however, exchange for threonine abolished anion translocation. N172 is, therefore, crucial for bidirectional formate translocation, possibly through its interaction with the conserved pore residue, T91.

F I G U R E 1 Conserved residues lining the FocA pore. (a) Structural overview of the Escherichia coli FocA protomer (Protein Data Bank structure 3KCU; Wang et al., 2009). The peptide backbone of a monomer (gray) is displayed in the cartoon representation and its pore (light orange) was modeled using MOLE 2.5 (see Section 2). The amino acid residues analyzed in this study (b), as well as the mechanistically relevant residues T91 and H209, are shown in the stick representation. (b) Frequency plot of the occurrence of amino acid residues chosen for exchange and subsequent analysis of in vivo anion translocation ability was created using WebLogo 3 (Crooks et al., 2004). A total of 165 annotated formatenitrite transporters channels were aligned with native FocA from E. coli and assessed with the online tool. "Bits" indicates the relative frequency of occurrence of the residues. The relative percentage conservation of the E. coli FocA residues among the 165 formate-nitrite transporter sequences is as follows: D88 20%, K156 81%, N172 89%, E208 78%, S210 47%, N213 87%, and N262 94%. Amino acid residues with an apolar side chain (A, F, I, L, M, P, V, W) are shown in gray, while polar side chains are displayed in green (C, G, N, Q, T, S, Y). Acidic amino acids are shown in red (D, E) and basic residues are indicated in blue (H, K, R). The coloring of the residues correlates with that used in the structural representation shown in (a).

| INTRODUCTION
The homopentameric formate-nitrite transporter (FNT) superfamily of membrane channel proteins includes several thousand members (Mukherjee et al., 2017). These channels have roles in the translocation of monovalent anions such as formate (represented by FocA), hydrosulfide (HSC), lactate (PfFNT), and nitrite (NirC) (Czyzewski & Wang, 2012;Golldack et al., 2017;Jia & Cole, 2005;Wang et al., 2009). FNTs share strong structural similarities and each protomer of the pentamer exhibits structural homology to the fold of aquaporin channels (Agre et al., 2002;Stroud et al., 2003;Sui et al., 2001). The FNT protomers are tightly packed giving the homopentamer a rigid structure within the membrane (Lü et al., 2013;Waight et al., 2013). Each protomer has a narrow, hydrophobic pore through which the monovalent anionic cargo can pass. Access to the hydrophobic core of the pore is achieved through two funnel-like vestibules, one on the periplasmic side of the membrane and the other on the cytoplasmic side ( Figure 1a). The structural organization of the cytoplasmic vestibule of FocA from Escherichia coli is incomplete because the structure of the flexible N-terminal domain of the protein could not be adequately resolved (Wang et al., 2009).
All FNTs have a common set of conserved amino acid residues that line the pore (Figure 1a): these include a centrally located threonine residue (T91 in the numbering of E. coli FocA; Figure 1a), which is located on the tip of a flexible Ω-loop and which links two portions of a broken transmembrane helix (TM2a and b) (Waight et al., 2010) and a histidine residue (H209 in Figure 1a), located on the end of a so-called S-loop, which acts as a linker between a second broken transmembrane helix (TM5a and b; Wang et al., 2009). T91 and H209 represent two of the most conserved residues in all FNT channels (Lü et al., 2013;Mukherjee et al., 2017;Waight et al., 2013). An in vivo assay using a formatesensitive fdhF P ::lacZ-based reporter was established to study FocA from E. coli (Beyer et al., 2013;Suppmann & Sawers, 1994).
The reporter assay allows monitoring of changes in intracellular formate concentration and has shown that H209 is crucial for pH-dependent uptake of formate by FocA (Kammel, Trebbin, Pinske, et al., 2022), while T91 has an essential function in formate efflux out of the cell Kammel & Sawers, 2022a, 2022b). X-ray crystal structure data have revealed that T91 can adopt two positions relative to the fixed location of H209; T91 either forms a hydrogen bond with H209, or it moves roughly 5 Å away from it, toward the cytoplasmic vestibule Waight et al., 2010). When T91 is separated from H209 it can hydrogen-bond with asparagine N172, or it can interact with N172 indirectly via a bridging water molecule . Together, these structural observations suggest that N172 might be important for, and involved in, the anion translocation process. N172 belongs to a set of seven other conserved residues lining the pore and for which little or no in vivo information is available regarding any potential functional or structural role each might adopt (Figure 1a,b). The other conserved residues include two further asparagine residues N213 and N262, together with E208, S210, D88, and K156 ( Figure 1). Preliminary studies carried out in our lab using the formate-responsive reporter system showed impaired formate translocation for FocA K156E, E208A, and N213D variants . Moreover, heterologous expression studies performed in yeast with K156 amino acid-exchange variants of FocA from E. coli have suggested that it has an important role in the electrostatic attraction of formate to the periplasmic side of the pore, before its pH-dependent translocation into the cell .
Together, the conservation of T91, H209 plus the seven other residues in the pore strongly suggests that FNTs share a common mechanism to translocate their respective anion cargo. While isolated FNTs are indeed able to translocate several different cargo molecules , in vivo data obtained for E. coli FocA (Suppmann & Sawers, 1994) indicate that it exhibits specificity for formate, suggesting that other members of the superfamily also might show anion specificity in vivo. In the case of FocA, this anion specificity has been proposed to be achieved through interaction between the cytoplasmic enzyme PflB (pyruvate formate-lyase) and the Nterminal, cytoplasmically oriented domain of FocA Kammel et al., 2021). The PflB interaction is suggested to determine the position of T91 (located at the tip of the Ω-loop) within the pore to control channel gating (Kammel et al., 2021).
The aim of the current study, therefore, is to examine what influence substitutions in the seven conserved residues shown in Figure 1b have on formate translocation out of the cytoplasm. As well as using our formate-responsive reporter system, we also adopt a hypophosphite-based growth-sensitivity assay as a proxy to examine the potential effects of the residue exchanges on formate uptake. Hypophosphite is a chemical analog of formate and, when taken up into the E. coli cell by FocA, severely reduces the growth of anaerobically growing E. coli cells by irreversibly inhibiting PflB enzyme activity (Suppmann & Sawers, 1994); the reduction in growth rate is due to the prevention of formate and acetyl-CoA generation from pyruvate (Plaga et al., 1988). Mutants lacking focA exhibit resistance to hypophosphite and, consequently, their growth rate is unaffected by low concentrations of the inhibitory anion (Kammel et al., 2021;Suppmann & Sawers, 1994). Using these approaches, we show that the conserved residues fall into two basic categories: those that are required for the structural integrity of the pentamer and consequently the translocation pore; and those that, when exchanged, have a direct impact on anion translocation.

| Bacterial strains, plasmids, and general cultivation conditions
The strains and plasmids used in this study are listed in Table 1.
The cells were grown anaerobically in an M9 minimal medium, pH 7.0 (Sambrook et al., 1989), at 37°C, with 0.8% (w/v) glucose as a carbon source (Kammel et al., 2021). The DH701 (focA) strain was transformed with plasmids carrying genes encoding different FocA variants and the cells were cultivated in 15 ml Hungate tubes for studies aimed at determining changes in the intracellular formate concentration using the in vivo fdhF P ::lacZ reporter system. To analyze whether the different FocA variants were stably synthesized, growth was carried out in 500 ml serum bottles filled to the top with the medium. In both types of growth, the cells were cultivated statically until they reached the late exponential growth phase (OD 600~0 .7−0.9) whereupon they were harvested by centrifugation and analyzed (Kammel et al., 2021).
The assay to determine the sensitivity of strains toward the toxic formate analog hypophosphite (0.5 mM final concentration) was carried out in microtiter plates, exactly as described (Kammel et al., 2021;Kammel, Trebbin, Pinske, et al., 2022). When required, antibiotics were added to a final concentration of 50 μg/ml for kanamycin and 100 μg/ml for ampicillin.

| Construction of plasmids
The expression vector pfocA (Table 1) was used as a template for mutagenesis of the focA gene, which in this vector was not fused to codons encoding a C-terminal Strep II-tag (Kammel et al., 2021).
When plasmids carrying genes encoding FocA variants with the C-terminal Strep II-tag (e.g., pfocA3-K156E and pfocA3-N213D; Table 1) were already available, they were used as a template for the introduction of a stop codon at the end of the focA gene (using oligonucleotides focA_stop_fw and focA_stop_re, Table A1). Plasmids in which the focA gene was mutated at codon 172 (Asn to either Ala or Thr) were generated using nonoverlapping oligonucleotides and the mutagenesis procedure was performed according to the New England Biolabs KLD (Kinase, Ligase, DpnI) mutagenesis protocol (Kammel & Sawers, 2022a). Replacements of the other codons were carried out by site-directed mutagenesis following the Agilent QuikChange procedure (Kammel, Trebbin, Pinske, et al., 2022).

| Hypophosphite-sensitivity test and β-galactosidase enzyme activity assay
Reduction in the rate of anaerobic cell growth due to supplementation of 0.5 mM sodium hypophosphite to the growth medium was determined as described (Kammel et al., 2021). The intracellular formate level was measured indirectly using the β-galactosidase enzyme activity assay (Kammel et al., 2021;Miller, 1972).
All analyses were performed in triplicate with minimally three biological replicates and the values obtained are presented with the standard deviation of the mean.

| Computational tools
The degree of conservation of the amino acids chosen for substitution was analyzed using the WebLogo tool (online version 2.8.2) (Crooks et al., 2004). The frequency of occurrence of the residues was assessed by analysis of an alignment of 165 annotated FNTs.
A representation of the location of these amino acid residues within the FocA protomer for the translocation pore was visualized with PyMOL (The PyMOL Molecular Graphics System, version 2.5, Schrodinger, LLC). To do this, the pore of either E. coli FocA (Protein Data Bank [PDB] structure 3KCU; Wang et al., 2009) or Vibrio cholerae FocA (PDB structure 3KLY; Waight et al., 2010) was modeled using the MOLE 2.5 software (Sehnal et al., 2013). The parameters for the calculation of the pore were probe radius 5 Å; interior threshold 1.1 Å; surface cover radius 10 Å; origin radius 5 Å; bottleneck radius 1.2 Å; bottleneck length 3 Å; and cut-off ratio 0.7.
The PyMOL software was also applied for the determination of  however, this has never been verified in vivo for these residues.
A further asparagine residue (N172) is located near T91 ( Figure 1a) and they form a hydrogen bond when T91 is in the "down" orientation (Wang et al., 2009), or they interact indirectly by coordinating a water molecule between them, which is bound to T91 when in the "up" position in the NirC structure . The findings of these structural studies suggest that N172 might have a key function in anion translocation. The residue is also conserved in almost 90% of FNT channels, with the only naturally occurring variant in some FNTs being a glycine residue ( Figure 1b).
At the edge of the periplasmic vestibule, there is a positively charged lysine, K156, which is conserved in approximately 80% of the analyzed FNT proteins (see Figure 1b) and it has been proposed to be important for the uptake of formate by guiding it into the channel through electrostatic attraction .
Combined experimental evidence , its conservation in numerous FNTs, coupled with the fact that various residues (see Figure 1b) appear to be able to replace it when it is not essential, nonetheless support a function for this residue in most, but not all, channels. 3.2 | K156 is essential for formate efflux out of fermenting E. coli cells Of the seven conserved residues (excepting T91 and H209) lining the FocA pore, exchanges in K156 and E208 have been preliminarily analyzed in vivo using homologous or heterologous hosts . While the exchange of K156 for a negatively charged glutamate residue inactivated formate translocation by FocA in an E. coli background , the exchange of the same residue for either C, A, or M reduced pHdependent formate-uptake activity in a heterologous yeast expression system . In contrast, exchanges in E208 for A or Q had no significant effect on either formate efflux  or formate uptake . Here, using our recently optimized in vivo reporter systems to detect changes in formate efflux or hypophosphite uptake (Kammel et al., 2021;Kammel, Trebbin, Pinske, et al., 2022), we re-examined some of these exchanges as controls for the impact exchanges in the other five conserved residues had on formate translocation and hypophosphite uptake (Figure 2). Remarkably, the growth of DH701 synthesizing FocA K156A was still highly sensitive to sodium hypophosphite, where an approximate 70% reduction in growth rate was measured compared with when the same strain synthesized native FocA (Figure 2b). This result indicates that hypophosphite uptake by FocA was still functional when an apolar alanine residue replaced the positively charged lysine residue. In contrast, when the K156E exchange in FocA was analyzed, the results revealed a strong reduction in sensitivity to hypophosphite ( Figure 2b). This result indicates that the negatively charged glutamate side-chain likely repelled uptake of the formate analog, and agrees with a previous formate import study using strain DH201 (focA pflB) carrying pfocA3-K156E, where no uptake of formate was detected . Together, these results suggest that the lysine residue appears to be required for formate efflux into the periplasm by FocA but it is not essential for hypophosphite import by FocA.

| Evidence that the conserved glutamine residue N172 has a role in formate translocation
Three conserved glutamine residues line the FocA pore, with N213 and N262 being located in the central regions of the pore. On the other hand, N172 is located in proximity to the essential T91 residue ( Figure 1a). Crystal structure data suggest that N172 might be important for anion translocation (Lü et al., 2011;Wang et al., 2009).
Consequently, N172 was substituted with either a nonhydrogenbonding alanine residue or with threonine, which has a similar size to asparagine and retains hydrogen-bonding potential. Plasmids carrying the correspondingly modified focA genes encoding these protein variants were introduced into DH701 (focA) and after anaerobic growth, the changes in the levels of intracellular formate, as reflected Indeed, when a suitably mutated focA gene was introduced on pfocA-N213A or pfocA-N262A into DH701 (focA) to test its influence on intracellular formate levels, both variants failed to show any capability of translocating formate out of the cell (Figure 2a).
Similarly, the synthesis of FocA N262A even caused significant intracellular accumulation of formate, possibly also suggesting a role for this residue in the formate translocation process. Western blot analysis of membrane fractions derived from strains synthesizing these variants confirmed that FocA N262A was synthesized, but at significantly lower levels than the native protein (Figure 3b), while FocA N213A could not be detected. The latter finding explains the lack of formate export by the corresponding strain (Figure 2a). Both strains also showed a significant reduction in their capacity to take up hypophosphite (Figure 2b).
To confirm the hypothesis that the asparagine residue at position 213 is important for hydrogen-bond network stabilization, we also analyzed a variant with an aspartate residue replacing asparagine; the carboxylate group on the side chain of aspartic acid can readily participate in hydrogen-bond formation. The fact that when strain

| DISCUSSION
The conservation of several key amino acids that line the pores of the FNT superfamily members strongly suggests that a common mechanism is used for the export and import of anions or their conjugate acids (Mukherjee et al., 2017). While slight variations in the residues present especially in the vestibules of different FNT subfamily members might influence which anion/neutral acid can pass through the respective pore, it has been clearly shown that purified FNT proteins can passage a variety of anions . However, the overall architecture of the pores restricts cargo molecules to small monovalent anions (Lü et al., 2013;Waight et al., 2013). Anion specificity is, therefore, likely determined by specific partner protein(s) that interact with the FNT, as has been proposed for enterobacterial FocA Kammel et al., 2021). Together, this feature of FNTs makes it all the more important to determine what functions the conserved residues lining the pore have, particularly concerning distinguishing whether they have a structural role or whether they are involved in the translocation process. The functional importance of the conserved KAMMEL AND SAWERS | 7 of 11 histidine (H209) and threonine (T91) residues for formate translocation by FocA of E. coli has been established (Kammel, Trebbin, Pinske, et al., 2022;Kammel & Sawers, 2022b). Here, we have examined the roles of seven further conserved residues that line the pore of E. coli FocA and other FNT members. Our findings reveal that two of these residues (K156 and N172) are important in facilitating formate translocation by FocA, while the remaining five residues (D88, E208, S210, N213, and N262) have more or less important functions in maintaining the integrity of the pore's structure, although it cannot at this stage be excluded that N262 might also be important during anion translocation.
The structurally conserved residues fall into two classes, whereby the two glutamine residues, N262 and N213, have the strongest impact on formate translocation because when they are exchanged for the small apolar residue alanine, they cause severe loss of stability of the respective FocA variants in the membrane. This is in accord with their involvement in stabilizing hydrogen-bonding networks, as predicted from the crystal structure data (Waight et al., 2010;Wang et al., 2009). They appear to be crucial in maintaining rigidity within the pentamer, particularly in the upper, periplasmically oriented part of the protein (see Figure 1a). In the case of N262, the side chain is located within the hydrogen-bonding distance of T91-N172 in the E.
coli FocA structure when T91 is separated from H209 (Wang et al., 2009). However, in the V. cholerae FocA structure, the equivalent Asn residue is at an approximate distance of 2.9 Å from the histidine residue when it forms a hydrogen bond with the threonine (Waight et al., 2010). Thus, N262 might also directly or indirectly influence the ability of H209-T91 to translocate formate through FocA.
The side-chains of the other three less well-conserved residues (D88, E208, and S210) appear to play less important roles in stabilizing the protein because exchanges with residues that have small, compact side-chains were readily tolerated. It is likely, therefore, that backbone contacts (hydrogen bonds and electrostatic interactions) are more important for these residues (Wang et al., 2009 Saccharomyces cerevisiae indicated that exchange of the K156 for either alanine, cysteine, or methionine impaired proton-coupled uptake of formate into yeast cells . The requirement for low pH was not observed in our study, suggesting that in the natural host, the mechanism of hypophosphite/formate uptake is different than in the heterologous yeast system. The hypophosphite anion has a lower pK a (1.1) than that of formate (pK a = 3.7), and thus it would be expected that FocA K156A should also take up formate. Indeed, we could also show that when strain DH601, which lacks active PflA and FocA and cannot make formate intracellularly (Kammel et al., 2021), synthesized FocA K156A , it was still able to import formate, but FocA K156E was strongly impaired in this regard (data not shown). As has been previously proposed (Kammel, Trebbin, Pinske, et al., 2022), it is likely that additional regulatory features, such as control of formate translocation through interaction with PflB (Kammel et al., 2021), account for the different results determined using the homologous and heterologous systems; yeast cells do not have PflB.
These new results also suggest that K156 appears to be more important for the release of formate into the periplasm than for its reimport into the cell, although the blockage of hypophosphite uptake by generating a K156E exchange variant of FocA indicates that K156 does at least partially contribute to electrostatic attraction of the anion in the uptake direction . Whether K156's involvement in the release of formate is somehow mediated via a long-range effect induced by the PflB interaction on the cytoplasmic side of the membrane, or via an interaction with a further, as yet unidentified, protein at the periplasmic surface of the membrane, is currently unclear.
Providing an answer to this question will require further extensive experimental study.
The asparagine at residue position 172 is well conserved and the only naturally occurring alternative is glycine, which is found in certain fungi and bacterial FNT proteins of unknown function. Our observation that an exchange of the residue for alanine with its compact methyl side-chain allowed FocA to retain bidirectional anion translocation function is in accord with a glycine substitution being functional (Figure 1b). The unexpected finding that introduction of a threonine residue prevented both efflux of formate and significantly reduced import of hypophosphite suggests that the γ-methyl group on its side-chain sterically blocks the pore (see   observed in the NirC protein, suggests a role for N172 in cargo passage, possibly in the proton-relay mechanism proposed by Lü et al. (2013). In this proposal, during the uptake of formate, the anion is transiently protonated by H209 to allow passage of the neutral acid through the hydrophobic core of the pore (Atkovska & Hub, 2017). Lü and coworkers (Lü et al., 2013) have hypothesized that after abstraction of a proton from T91 by the imidazolium cation of H209, the now nucleophilic T91 on the tip of the Ω-loop moves away from the "fixed" H209 residue down toward the cytoplasmic vestibule where recapture of the proton from formic acid occurs. It is proposed that N172 facilitates this process by helping position the nucleophilic T91 residue to recapture the proton. Replacing N172 with alanine resulted in some reduction in hypophosphite translocation capacity, suggesting that the residue exchange reduced the efficiency of translocation, which would be in accord with the proposal above. Further detailed studies will be required to prove or refute this hypothesis, particularly in light of the recent demonstration that uptake of exogenously supplied formate (and possibly hypophosphite) by E. coli cells differs mechanistically from the reimport of formate that originated in the cytoplasm and was exported to the periplasm by FocA during exponential growth (Metcalfe et al., 2022).

| CONCLUSIONS
Here, we have been able to characterize an additional seven conserved amino acid residues common to the pores of FNT channels into two categories: those having a structural role; and those with a direct role in formate translocation by FocA. While asparagine residues N213 and N262 likely maintain the integrity of the pore, particularly around the more rigidly structured S-loop region, three other structurally conserved residues (D88, E208, and S210) tolerate exchange for residues with small or compact side-chains. This suggests backbone interactions are probably important to aid the structural integrity of the pore in the case of these residues. The stable pentameric structural assembly is likely important for channel function and is similar to what has been observed for the tetrameric aquaporins (Stroud et al., 2003).
A further asparagine residue, N172, was shown to be important in formate translocation through the pore. N172 likely aids correct positioning of, or interaction between, the essential H209 and T91 residues during bidirectional formate/formic acid translocation.
Finally, a lysine residue located at the entry to the periplasmic vestibule of FNTs, when exchanged for an apolar alanine residue, proved to be impaired in formate efflux by FocA. This suggests that K156 is required for the release of the anion into the periplasm; the partial retention of hypophosphite uptake by the FocA K156A variant F I G U R E 4 The inactive formate translocation in FocA N172T is due to sterical hindrance within the pore. Structural depiction of the central part of the translocation pore of a Vibrio cholerae FocA protomer (Protein Data Bank structure 3KLY; Waight et al., 2010). The peptide backbone of the monomer (chain A) is displayed in cartoon representation in gray and its pore (darker gray) was modeled using MOLE 2.5 (see Section 2). The residues H208 (in gray, corresponding to H209 in Escherichia coli FocA), T90 (residue and the Ω-loop in blue, corresponding to T91 in E. coli FocA), and N/A/T171 (in green, corresponding to N172 in E. coli FocA) are shown with their respective side chain in stick representation. The distances between the indicated atoms of the side chains were determined using PyMOL (a). PyMOL was also used for the virtual mutagenesis of asparagine N171 to alanine (b) and threonine (c-e). For the threonine exchange, three potential rotamers (T171 1, 2, and 3) are shown and each of them had a predicted frequency of occurrence in this orientation of between 17% and 22%.
verifies that efflux and uptake of anions by FocA involve different mechanisms.

ACKNOWLEDGMENTS
This work was supported by the Martin-Luther University of Halle-Wittenberg and the federal state of Saxony-Anhalt. Open Access funding enabled and organized by Projekt DEAL.

CONFLICT OF INTEREST
None declared.

DATA AVAILABILITY STATEMENT
All data generated or analyzed during this study are included in this published article.

ETHICS STATEMENT
None required.