Barrel expansion of Outer Membrane Protein G Nanopore through β-Hairpin Duplication

Abstract

Outer membrane β-barrel proteins (OMPs) are channels found in the outer membranes of Gram-negative bacteria characterized by a stable and diverse barrel architecture, which has made them attractive for nanopore sensing applications. Here, we systematically investigated the feasibility of expanding outer membrane protein G (OmpG) from its native fourteen-stranded β-barrel to an enhanced conductance variant by independently duplicating each of its seven hairpin units and inserting them downstream of their endogenous positions. Most combinations did not increase pore diameter, but duplication of the terminal seventh hairpin exhibited a rare population of pores with enhanced conductance, suggesting barrel enlargement. Further engineering efforts to optimize the terminal β-turn sequence have resulted in up to fifty percent of pores with increased conductance. Importantly, the enlarged pores retained the sensing functionality of the original scaffold, highlighting the potential of this approach for developing β-barrel OMP sensors with tunable dimensions.

Summary Statement:

Outer membrane β-barrel proteins (OMPs) are valuable for nanopore sensing due to their stable architecture but currently lack reliable methods to alter channel dimensions. We demonstrate that by duplicating the terminal hairpin of OmpG, a new population of enhanced conductance pores was observed that exhibited a three-fold conductance increase while retaining sensing functionality. These findings offer insights into β-barrel architecture and present a promising strategy for developing tunable nanopore sensors.

Introduction

Outer membrane proteins (OMPs) are a diverse category of proteins located at the outer membranes of Gram-negative bacteria and the endosymbiotic organelles (mitochondria & chloroplast) (Chaturvedi and Mahalakshmi, 2017). OMPs generally adopt a β-barrel structure except for Wza, a putative polysaccharide transporter (Dong et al., 2006). Among various β-barrel arrangements, monomeric β-barrel OMPs typically consist of eight to fourteen β-strands forming a hollow luminal channel (Hermansen et al., 2022) while sixteen to thirty-six β-strand OMPs often exist as homo-multimers with an extended internal loop folding into the luminal channel, providing structural support and resulting in a narrowed channel diameter (Tamm et al., 2004). Oligomeric β-barrels with shear-angle controlled stoichiometry are also observed (Hermansen et al., 2022). The diversity in OMP form is mirrored by their associated functions, such as nutrient uptake, homeostasis, efflux and secretion, and membrane-biogenesis (Hermansen et al., 2022; Mirus et al., 2010; Vergalli et al., 2020). Owing to their highly stable structures, OMPs have been an attractive repository of potential tools for use within the emerging nanopore sensing field (Crnković et al., 2021; Mayer et al., 2022).

Nanopore sensing employs a detection scheme in which a single nanometer-sized pore is inserted into an electrically insulating membrane separating two chambers containing a buffered electrolyte solution (Mayer et al., 2022). A potential is applied across the membrane to direct the bulk flow of ions through the pore resulting in a detectable current (Tang et al., 2022; Willems et al., 2020; Wilson et al., 2019). Analytes can be captured by externally affixed affinity reagents or held within the sensor pore itself depending on the sensing strategy. In either scenario, analyte capture transiently perturbs the measured current relative to the dimensions and surface chemistry of the analyte, enabling highly specific identification and quantitation (Tang et al., 2021; Wang et al., 2021). Using these general principles OMPs have now been extensively deployed for the detection and characterization of a wide range of biomolecules including small molecules/metabolites, nucleic acids, and peptides/proteins (Mayer et al., 2022). Nanopore sensing development is now often focused on expanding the analyte repertoire or improving the resolution of existing analytes (Tabatabaei et al., 2022). Achieving high resolution of a biomolecule requires close matching of the analyte and sensor pore’s physical and chemical properties, and as such the generation of pores with tunable attributes (such as size and charge) is highly desirable as a ‘one-size-fits-all’ sensor is unlikely to be sensitive to difference(s) between structurally similar analytes (Crnković et al., 2021; Zhou et al., 2020). While targeted mutagenesis can alter the luminal chemical properties of monomeric sensor pores with relative ease, a routine approach for adjusting pore size has not yet been achieved (Wang et al., 2018). Therefore, a generalizable method to manipulate the dimensions of OMP scaffolds is of significant interest in the field of nanopore sensor development.

Analysis of β-barrel OMPs has revealed patterns of sequence similarity between variably stranded members suggesting an evolutionary scenario that involves duplicative expansion and diversification from an ancestral β-hairpin, mirroring gene duplication observed in other systems (Franklin et al., 2018; Magadum et al., 2013; Remmert et al., 2010). β-barrel channels are formed by a repeating structural unit, the β-hairpin; an anti-parallel arrangement of amphipathic strands connected by a variable-length unstructured extracellular loop and adjoined to the following hairpin unit by a shorter periplasmic β-turn (Dhar et al., 2021; DuPai et al., 2021; Schulz, 2000). The hairpin units within each polypeptide chain form a substantial network of intra-chain backbone hydrogen bonds, closing the barrel structure through the self-association of the topologically first and last hairpin units (Jackups Jr and Liang, 2005). Based on the observed evolutionary relationship, many groups have reported enlarging OMP barrel diameter by duplicating the native hairpin modules (Korkmaz et al., 2015; Krewinkel et al., 2011; Liu et al., 2017; Tosaka and Kamiya, 2022). However, the incorporation of duplicated hairpin units into a stable barrel structure remains a non-trivial challenge as indicated by mixed reports of success using outer membrane protein G (OmpG) as a model system (Korkmaz et al., 2015; Tosaka and Kamiya, 2022). OmpG from Escherichia coli (E. coli) is a fourteen-stranded monomeric OMP consisting of seven hairpin units (Figure. 1) (Subbarao and van den Berg, 2006). OmpG presents a particularly intriguing case for two key reasons: (1) Previous attempts to enlarge the barrel targeted different hairpin regions yielding inconsistent results, suggesting a general positional dependence and/or the need for additional engineering considerations to refine duplicative expansion as a generalizable method for engineering enlarged barrels and (2) OmpG has been demonstrated to be a useful monomeric nanopore scaffold for highly selective detection of large biomolecules (Fahie et al., 2015; Foster et al., 2023; Kim et al., 2023; Pham et al., 2019). The general strategy for OmpG nanopore sensing involves capturing analytes via affinity reagents displayed on flexible external loops, which perturb OmpG’s basal gating behavior in analyte-specific ways through unique interactions with the analyte’s molecular surface. This approach allows for selective identification and quantification of target molecules, even among closely related homologs, with multiplexed sensing capability achieved by incorporating distinct affinity reagents onto different loops (Fahie et al., 2021; Foster et al., 2023). Therefore, engineering a larger OmpG pore with additional external loops would enhance multiplex sensing potential by increasing the number of available sites for hosting affinity reagents. Additionally, the anticipated increase in pore conductance could further improve the signal-to-noise ratio, thus enhancing overall detection sensitivity.

1.

Naïve duplicative expansion strategy. a) The amino acid sequence of OmpG, excluding the signal peptide. Hairpin units, each consisting of two β-strands, an unstructured loop, and an adjoining β-turn, are color-coded along a maroon-to-blue gradient. Solid arrows indicate β-strands (1–14), while dashed lines represent the connecting unstructured loops (1–7). The FLAG-tag reporter (-GDYKDDDDKG-) was inserted at the positions marked by orange triangles. b) Side view (left) of the OmpG barrel and top view (right) of OmpG with the loop sequences shown. c) Schematic cartoon of OmpG duplicative expansion strategy. β-strand residues are depicted as squares and loop/turn residues as circles. Each hairpin unit (HP1–7), containing a positionally-indicated FLAG-tag, was inserted downstream of the endogenous sequence as the sites marked by green stars, resulting in constructs delineated as OmpG^ExpHPX, with X denoting the duplicated hairpin unit.

Here, we report engineering OmpG from its native fourteen-stranded barrel structure into an enhanced conductance ‘expanded’ species via independent duplication of each of the seven hairpins. While most hairpin duplications failed to produce constructs with enhanced conductance, the duplication of hairpin 7, combined with optimization of the connecting β-turn sequence, resulted in a pore population exhibiting a three-fold increase in conductance, suggesting successful incorporation of the duplicated hairpin into an enlarged barrel structure. Notably, the expanded OmpG retained the sensing functionality of the original scaffold, although the enhanced conductance pore could not be enriched to homogeneity for structural determination.

Results

Naïve duplicative expansion

Prior reports of duplicative hairpin expansion have targeted both terminal and internal hairpin units in their expansion schemes, suggesting the β-barrel scaffold is relatively tolerant to this manner of sequence insertion (Arnold et al., 2007; Korkmaz et al., 2015; Krewinkel et al., 2011; Tosaka and Kamiya, 2022). Accordingly, to assess OmpG’s tolerance to duplicative expansion, we decided to pursue a “naïve” strategy in which each of the scaffold’s seven hairpin units (Figure 1a–b, termed HP1–7) was independently inserted directly downstream of its endogenous position (Figure 1c), resulting in a total of seven constructs delineated as OmpG^ExpHPX, with X indicating the duplicated hairpin unit (Figure S1). This systematic exploration of duplicative expansion is an attempt to bridge disparate reports of barrel expansion and establish a robust framework for the evaluation of this engineering strategy. Because the native HP7 lacks a β-turn at the C-terminus, a consensus β-turn sequence generated from Gram-negative bacterial outer membrane proteins (-NKNM-, Table S1) was appended to the N-terminus of the duplicated HP7 to facilitate the incorporation of the duplicated hairpin (Vorobieva et al., 2021). Each duplicated hairpin unit also contained a FLAG-tag motif reporter (-GDYKDDDDKG-) within the unstructured loop, with the boundaries of each hairpin unit and the associated insertion site manually determined from PDB ID 5MWV by using PyMOL rendered secondary structural features across all poses.

The incorporation of the duplicated hairpin into a stable β-barrel structure was assessed via single-channel current recording to obtain the channel conductance, a proxy measurement correlated with pore diameter, with the expectation that successful incorporation of duplicated sequences into the β-barrel will result in correspondingly increased conductance (Willems et al., 2020). Seven equivalent fourteen-stranded wild-type-like proteins harboring FLAG-tag at the corresponding loops, OmpG^L1-L7FLAG were previously generated (Foster et al., 2023) and used here as reference proteins for each OmpG^ExpHPX construct, as studies have shown that loop or luminal mutations may also influence pore conductance (Baldelli et al., 2024; Bredin et al., 2002; Orlik et al., 2002; Thakur et al., 2017; Zhuang and Tamm, 2014). All fifteen constructs (OmpG^WT, OmpG^L1-L7FLAG, OmpG^ExpHP1−7) were purified from E. coli inclusion bodies and refolded successfully in n-octyl-β-glucopyranoside (OG) detergent-containing solutions (Figure S2) (Chen et al., 2008b; Noinaj et al., 2015). The unitary conductance of all refolded proteins was assessed via single- channel recordings using 1 M potassium chloride (KCl) under an applied potential of ± 50 mV. Representative examples of different OmpG^WT behaviors are shown in Figure 2A, as this single-molecule technique can assess the population heterogeneity. Nominal OmpG^WT channel behavior (Figure 2a, left) is exemplified by an open pore conductance of 1.16 ± 0.08 nS (mean ± SD) and exhibits slightly asymmetric conductance and gating profiles (Chen et al., 2008b), dependent on the polarity of the applied transmembrane potential (Figure 2a, S3). Rapid gating events are attributed to conformational fluctuations of the pore’s seven loops, predominantly loop 6, which transiently occlude the lumen (Perez-Rathke et al., 2021; Yildiz et al., 2006; Zhuang et al., 2013) through a network of pH-modulated electrostatic interactions between loops and the barrel lumen. For comparative purposes, the predominant population of pores exhibiting asymmetric gating and conductive profile were termed α-pores for each OmpG construct. For simplicity, we standardize the presentation of all pores at the +50 mV within the main text, with the data acquired at the −50 mV presented in supplementary information (Figure S3) (Chen et al., 2008b). α-pores typically exhibited a conductance of 1.18 ± 0.08 ns with a gating frequency of 59.6 ± 7.6 /s. In all OmpG^WT samples, we also observe minor populations of pores that exhibited disparate conductance and gating profiles from the nominal α-pores. We categorized these minor population based on their conductance relative to α-pores. For instance, δ-pores exhibited approximately 80% of the α-pore conductance, whereas γ-pores displayed conductance at 50% or lower. The gating frequency of δ and γ were calculated as 183.1 ± 112.3 /s and 91.3 ± 2.8 /s respectively. These minor, aberrant δ and γ subpopulations likely represent non-native β-barrel conformations. While the gating mechanism of the 14-stranded wild-type OmpG (α-pore) has already been extensively characterized in our previous work (Fahie et al., 2022; Perez-Rathke et al., 2018; Pham et al., 2021a), an in-depth assessment of gating activity differences across these variant forms is not feasible due to the absence of structural information for δ- and γ-pores. Given their low abundance, these abdominal pores will not be further investigated. Instead, analyses will focus on pore populations with significant representation (>20%).

Figure 2.

Single-channel assessment of OmpG constructs. a) Representative current recordings of OmpG^WT pores, categorized into ‘α’ (nominal) pores and ‘δ/’γ’ pores, which are considered aberrant. b) Violin plots depicting the conductance distribution of OmpG^WT, fourteen-stranded FLAG-tagged controls, and OmpG^ExpHP1−7 constructs. Each dot represents the average conductance (between positive and negative applied potentials) of a unique pore (n = 20 per construct). c) Representative current traces of a-pore behavior for OmpG^LX-FLAG and OmpG^ExpHPX, with X indicating the FLAG-tag display loop or duplicated hairpin unit respectively. d) Proposed hairpin incorporation scenarios using duplication of HP6 as a representative example. All recordings were conducted in 50 mM Tris-HCl 1 M KCl pH 7.4 under an applied potential of ± 50 mV. For clarity, as OmpG can insert bidirectionally, all depicted pores are in the cis orientation with the loops experiencing a + 50 mV applied potential.

The conductance of the fourteen-stranded OmpG^L1-L7FLAG constructs exhibited a minor difference, up to 6% from that of OmpG^WT, depending on which loop contained the FLAG-motif (Figure 2b, S4). In contrast, the gating profile was significantly influenced by the FLAG display loop (Foster et al., 2023). For example, OmpG^L6-FLAG had drastically reduced gating activity due to the addition of the negatively charged FLAG sequence within the gating loop 6, in agreement with prior reports (Figure 2c, S3) (Foster et al., 2023; Pham et al., 2021b). The majority of Omp^ExpHPX constructs exhibited similar, or even reduced, conductance relative to OmpG^WT (Figure S4) except for OmpG^ExpHP7 which exhibited a rarely observed population (1/20) with unstable but significantly enhanced (~2x) conductance, termed a β-pore. Unfortunately, due to the low enrichment of this population structural determination was not feasible. Two other constructs, OmpG^ExpHP3 (1.20 ± 0.10 nS) and OmpG^ExpHP4 (1.27 ± 0.06 nS, P-value ≤ 0.01) exhibited modest conductance increases (≤ 10%) more in-line with prior reports of barrel enlargement (Krewinkel et al., 2011; Liu et al., 2017; Tosaka and Kamiya, 2022). Given that the enhanced conductance populations of these constructs were more enriched and homogenous relative to the rare OmpG^ExpHP7 β-pore, and loop 3 had demonstrated prior utility as a motif display site, we decided to determine the structure of OmpG^ExpHP3 by X-ray crystallography to validate barrel enlargement (Fahie et al., 2021; Foster et al., 2023). Unexpectantly, the crystal structure of Omp^ExpHP3 revealed a fourteen-stranded barrel structure (Figure S5), despite demonstrating enhanced conductance. Consequently, we did not pursue the crystallization of other OmpG^ExpHPX constructs as they were unlikely to adopt an expanded 16-barral structure.

Interestingly, the gating behavior of all observed OmpG^ExpHP6 pores resembled that of OmpG^L6-FLAG rather than OmpG^WT, exhibiting significantly reduced gating activity. The gating signals of OmpG predominantly arise from conformational changes in loop 6, which intrudes into and occludes the pore lumen (Chen et al., 2008a; Chen et al., 2008b; Yildiz et al., 2006), which invades into the lumen and occludes the pore. The electrostatic repulsion between negatively charged loop 6 and the negatively charged lumen regions constitutes the primary energy barrier that restricts loop 6 from entering the lumen (Perez-Rathke et al., 2021). Consequently, introducing additional negative charges into loop 6 can suppress gating activity, either by decreasing gating frequency or reducing the intensity of gating events (Perez-Rathke et al., 2021; Pham et al., 2021b). In L6-FLAG, the insertion of the FLAG-tag (-GDYKDDDDKG) into loop 6 introduced additional negative charges to loop 6 thereby greatly reducing gating behavior (Foster et al., 2023). The expanded constructs potentially fold into β-barrels via four distinct scenarios: (1) folding into the desired expanded 16-stranded barrel; (2) folding with the inserted HP6’ forming an extended periplasmic loop; (3) incorporation of HP6’ into the barrel structure with the original HP forming the periplasmic loop; or (4) barrel formation involving the original strand 11 and the newly introduced strand 12’, leaving the original strand 12 and new strand 11’ as part of an elongated loop 6. The observed decreased gating behavior in OmpG^ExpHP6 strongly suggests that these constructs likely fold according to scenarios 3 or 4, positioning the FLAG-tag within the extracellular loop 6 (Figure 2D). However, scenario 3 would result in a pore containing an extended periplasmic loop, potentially increasing gating activity, which contradicts the observed quiet gating behavior of OmpG^ExpHP6. Therefore, we propose that OmpG^ExpHP6 most likely folds as described in scenario 4.

2.2 |. Refined Engineering with β-turn Sequences

Because the OmpG^ExpHP7 construct showed a rare population of large conductance pores, we decided to optimize the grafted β-turn sequence attached to the duplicated HP7 to enhance the relative proportion of enhanced conductance pores. We tested several β-turn sequences (Figure 3a) including T4b, another consensus sequence from endogenously sixteen-stranded barrels (Table S1); T4c and T6, which were rationally designed based on the general positional frequency of amino acids in turns of the respective length (DuPai et al., 2021); and T5, a native non-terminal OmpG turn sequence. All constructs were able to refold in detergents but with reduced efficiency relative to OmpG^WT (Figure S6).

Figure 3.

β-turn engineering of OmpG^ExpHP7. a) Cartoon schematic of OmpG structure showing HP1 and HP7 closing the tertiary β-barrel. Engineering efforts targeted the region between the native and duplicated hairpin 7 for the introduction of a novel β-turn. b) Violin plots of population-level conductance assessments for the table delineated constructs compared to OmpG^WT and OmpG^L7-FLAG controls. Each dot represents a unique pore’s average (+/− potentials) conductance (n = 20 per construct). The dotted line marks the average conductance level of the controls. c) Representative single-channel current recordings of α- and β-pore behavior for OmpG^{ExpHP7-T5 (KLTDD)} and OmpG^{ExpHP7-T6 (GNDGKG)}. All recordings were conducted in 1 M KCl 50 mM Tris-HCl pH 7.4 under an applied transmembrane potential of ±50 mV.

Single-channel recordings of the turn-sequence engineered variants revealed that several constructs had increased the population of pores exhibiting two-to-three-fold higher conductance than the nominal α-pores (Figure 3b). These high conductance β-pores exhibited varied conductance enhancement and gating behavior (Figure S7–8). OmpG^{ExpHP7-T5 (KLTDD)} and OmpG^{ExpHP7-T6 (GNDGKG)} are particularly promising due to their stable open pore current (Figure S8). The β-pores of OmpG^ExpHP7-T5 exhibited two-fold increase in conductance compared to α- pores. However, the gating profiles of the two pores obtained for OmpG^ExpHP7-T5 were not consistent, exhibiting disparate occupancy of the gated state(s) (Figure S9), suggesting that they represent different folds/conformations of the protein and thus this construct is less suitable as a sensor due to potential uncontrolled variability. OmpG^ExpHP7-T6 exhibited a substantial population of pores with greatly increased conductance (3.01 ± 0.08 nS, n = 10), approximately three-fold relative to OmpG^WT, with half of the tested pores belonging to the β-pore population. The gating profile of these constructs was consistent among all β-pores (Figure S9) which exhibited rapid gating fluctuations at a frequency of 317.8 ± 76.3/s predominantly occupying around 0.5 I_res, in comparison to the gating frequency of 146.6 ± 33.6/s for α-pores. To exclude the possibility that the enhanced conductance reflected the stable association of the monomeric protein pore into a higher oligomeric state, OmpG^ExpHP7-T6 was assessed via SDS- and native-PAGE, revealing no evidence of higher oligomeric species (Figures S6 & S10). Most importantly, OmpG^ExpHP7-T6 exhibited single-step full current occlusion gating (Figure 3c), in stark contrast to the three-step closure behavior of trimeric pores or multi-level sub conductance states (Figure S11) indicative of multi-pore insertion (Danelon et al., 2003; Dargent et al., 1986; Delcour, 2009). Therefore, we concluded that β-pores of OmpG^ExpHP7-T6 likely adopt a monomeric β-barrel structures with increased diameter.

We also attempted to increase the β-pore proportion by testing the refolding of OmpG^ExpHP7-T6 in a panel of detergents with variable structural and micellular features (Figure S12). Unfortunately, no detergent outperformed the OG condition as assessed by improvement of refolding efficiency or generation of novel refolded species.

2.3 |. In vivo Folding and β-signal Mutagenesis

Thus far characterization of OmpG^ExpX constructs has been performed using an in vitro detergent-assisted protein refolding system. To test if chaperone-assisted direct folding to the outer membrane (OM) of E. coli may enhance the relative proportion of β-pores, we appended the endogenous signal peptide sequence, followed by a hexa-His tag, onto OmpG^WT, OmpG^L7-FLAG, OmpG^ExpHP7-T5, and OmpG^ExpHP7-T6 to facilitate the expression of these constructs in the OM (Figure 4a) (Chaturvedi and Mahalakshmi, 2017; Rollauer et al., 2015). The ability of these constructs to be folded in the E. coli OM was then assessed via flow cytometry using the FLAG tag as a reporter (Figure 4a) as well as gel-electrophoresis and immunoblot analysis of detergent-extracted bacterial membrane proteins (Figure 4b). Our control construct, OmpG^L7-FLAG was easily detectable on the bacterial surface as indicated by the enhanced anti-FLAG antibody fluorescence signal relative to OmpG^WT (Figure 4a, Top). In agreement, membrane extraction of OmpG^L7-FLAG showed strong enrichment of successfully folded protein as indicated by the high-intensity of the lower molecular weight band in the SDS-PAGE and the western blot (Figure 4b, Top). As the expanded constructs possess the FLAG-tag reporter only in the duplicated hairpin unit, we expected that only successful incorporation of the duplicated sequence into a stable barrel would result in detectable fluorescence due to the proposed sequential hairpin assembly by the Bam machinery and enforced topology of the bacterial membrane (Dhar et al., 2021; Rollauer et al., 2015; Tomasek and Kahne, 2021). Evaluation of OmpG^ExpHP7-T5 along the same metrics suggested inefficient membrane display, as indicated by a minor high fluorescence sub-population with most of the cells resembling the OmpG^WT profile (Figure 4a, Middle). This corresponds well with our observation of a low percentage of higher conductance pores in single-channel studies. Interestingly, the membrane-extraction results of this construct indicated substantial amounts of membrane-associated protein (Figure 4b, Middle), even higher than was seen for OmpG^L7-FLAG, despite the relatively low fluorescence signal. However, the mobility of this membrane-associated protein indicates it is not folded properly, suggesting that although the protein was correctly trafficked to the OM, terminal incorporation was inhibited by the additional sequence. Similar assessment of OmpG^ExpHP7-T6 revealed that this construct showed no detectable fluorescence signal above the wild-type control in flow cytometry (Figure 4a, Bottom) but exhibited minor enrichment of lower molecular weight membrane-associated OmpG^ExpHP7-T6 (Figure 4b, Bottom) signaling the proper folding of a barrel structure. The result can be explained as OmpG^ExpHP7-T6 has folded into a β-barrel with the FLAG-sequence facing inside of the cell, such that OmpG^HP7-T6 has adopted a ‘wild-type’ structure with the additional HP7 as a C-terminal overhang located at the periplasmic space (Figure 2d). Note that 50% of OmpG^ExpHP7-T6 proteins can refold into enhanced conductance pores in vitro, in disagreement with in vivo results.

Figure 4.

In vivo folding and membrane expression of expanded OmpG variants. a) Schematic cartoon of tag-location within each construct and the associated fluorescent profiles of bacterial membrane display flow cytometry for the indicated constructs. Successful construct display is indicated by rightward fluorescence shift relative to untagged OmpG^WT. Profiles are the average of three biological replicates with deviation as a shaded overlay. b) SDS-PAGE and anti-FLAG immunoblots of Ni-NTA IMAC purified membrane-extracted OmpG constructs. c) Cartoon schematic indicating the targeted endogenous β-signal replaced by the table of rationally designed mutation sequences. The fluorescence profiles of bacterial membrane display flow cytometry for the indicated β-signal mutant constructs are shown below. Data is presented as kernel density estimation of pooled data from three biological replicates. d) Representative single-channel current recordings of OmpG^{ExpHP7-KLTDD-B5} α-pores either prepared in vitro or in vivo. All recordings were conducted in 1 M KCl 50 mM Tris-HCl pH 7.4 under an applied transmembrane potential of ±50 mV.

Based on the observed accumulation of membrane-associated protein for OmpG^ExpHP7-T5, despite minimal fluorescence signal, we explored an alternative engineering approach that aimed to increase the in vivo population of enlarged diameter pores. Note, our current strategy duplicates HP7 that hosts a semi-degenerate regulatory motif the β-signal (Germany et al., 2024; Rollauer et al., 2015). Frequently located within the terminal β-strand, this motif sequence is important for the recognition of OMPs by periplasmic chaperones and their interaction with the folding machinery Bam complex (Germany et al., 2024; Leo and Linke, 2018; Robert et al., 2006; Sklar et al., 2007). Therefore, with OmpG^ExpHP7-T5 now containing two β-signals the second copy may interfere with the recognition of the terminal β-strand by the folding machinery. To facilitate the correct terminal recognition, we decided to mutate the β-signal motif of strand-14, while maintaining the endogenous sequence in the duplicated hairpin.

We assessed the positional amino acid frequency in known E. coli β-signals (Table S2) and generated a panel of five signal variants predominantly targeting the terminal three residues, as these are reported to be particularly important for chaperone interactions (Robert et al., 2006; Struyvé et al., 1991). The panel of variants attempted to maintain the chemical nature (i.e. polar, hydrophobic etc…) of the residue at each position but selected for the least frequently occurring amino acid in β-signals (Figure 4c). Using OmpG^ExpHP7-T5 as the template, constructs incorporating each of the β-signal mutated variants were generated and assessed using fluorescence activated cell sorting (Figure 4c, Bottom). We observed no fluorescence signal above the WT reference for four mutants, but the construct B5 matched the fluorescence intensity and distribution of OmpG^L7-FLAG suggesting the successful display of the flag-tag on HP7’ to the bacterial surface. However, single-channel assessment of membrane-extracted OmpG^{ExpHP7-KLTDD-B5} protein, as well as an in vitro refolded sample, revealed no evidence of enhanced pore conductance (Figure 4d). Notably, membrane-extracted proteins exhibited a higher gating frequency than their equivalent in vitro folded pores. We attribute this difference, at least in part, to the presence of the n-terminal his-tag at the periplasmic side on the in vivo folded protein. The his-tag my introduce noise due to interaction with the pore lumen. Furthermore, although the detergent-extracted proteins were purified from a Ni-NTA column, no effort was made to remove native lipids or other molecules from the solution. The potential presence of these compounds in the recording chamber could influence the bilayer composition and consequently impact OmpG gating levels. However, we do not believe this observation affects our interpretation of the data, as the gating noise was not used to assess sequence incorporation into the barrel in this context. In conclusion, our attempt at knocking out the endogenous β-motif did not enhance the in vivo folding and membrane expression of enlarged-barrel OmpG.

2.4 |. Evaluating the Impact of β-Barrel Expansion on Nanopore Sensing Performance

Although in vitro detergent-mediated refolding of OmpG^{ExpHP7-T6 (GNDGKG)} only achieved ~50% of β-pores, we still sought to assess if increasing the pore size would impact the nanopore sensor’s detection capability. Our group has previously reported on the detection of monoclonal anti-FLAG antibody FG4R using OmpG sensors that harbor a loop-displayed FLAG tag (Foster et al., 2023; Kim et al., 2023). Binding of FG4R to the FLAG tag on OmpG triggered a reversible change in the current and/or gating pattern allowing for detection of FG4R (Figure 5a). Here we compared the ability of fourteen-stranded OmpG^L7-FLAG and the putative expanded β-pore of OmpG^ExpHP7-T6 to detect FG4R. Since OmpG detection is mediated by loop-displayed affinity reagents, the observation of a specific and reversible signal elicited by FG4R upon addition to the chamber accessible to the flexible loops of the OmpG^{ExpHP7-GNDGKG} β-pore suggests that the duplicated hairpin carrying the FLAG motif was successfully incorporated into the barrel structure. For the purpose of comparison, binding of FG4R to OmpG^L7-FLAG elicits a significant reduction of gating frequency from 34.9 ± 6.7 Hz to 17.2 ± 6.0 Hz (N_Pore = 3, n_Event = 18) leading to an overall ‘quieter’ signal as visualized in the blue-colored regions of Figure 5b. Comparison of all-points histograms of OmpG^L7-FLAG under unbound and FG4R-bound conditions shows no significant difference in pore conductance, with a subtle increase in the FG4R-bound peak height arising from the quieter signal. In comparison, OmpG^ExpHP7-T6 showed less stability under this salt condition (300 mM KCl) than OmpG^L7-FLAG undergoing frequent channel closure to zero current. However, nominal behavior could be resumed by transient application of the opposite voltage, resulting in the thin vertical lines observed throughout the recording (Figure 5c). Binding of FG4R to the OmpG^ExpHP7-T6 β-pore exhibited an approximately ten percent decrease in the open-pore current (N_Pore = 3, n_Event = 18, P < 0.0001) relative to the unbound state (Figure 5c). However, for the OmpG^ExpHP7-T6/FG4R interaction no perturbation of gating frequency was observed, with the unbound pore gating at a frequency of 48.6 ± 4.6 Hz and the FG4R-bound regions gating at 47.6 ± 2.7 Hz (N_Pore = 3, n_Event = 18). Despite differences in signal appearance and reduced stability under lower salt conditions, the expanded pore retained its ability to detect FG4R effectively. These findings suggest that β-barrel enlargement through hairpin duplication is a viable strategy for tuning pore dimensions without compromising sensor functionality.

Figure 5.

Single-channel detection of FG4R. a) Cartoon schematic of nanopore detection set-up depicting OmpG^L7-FLAG for reference. b) Representative single-channel traces showing detection of FG4R (75 nM) by OmpG^L7-FLAG over 30 minutes (top) and 30 seconds (bottom). 100Hz filtered data (black) is overlaid on raw data (gray), with the FG4R-binding signal highlighted (blue). A representative all-points histogram of unbound and bound sections is depicted below. c) Equivalent presentation of OmpG^{ExpHP7-GNDGKG} β-pore detection of FG4R. All recordings were conducted in 300 mM KCl 50 mM Tris-HCl pH 7.4 under an applied transmembrane potential of ±50 mV.

Discussion

Our results have shown that while biophysical techniques (SDS-PAGE, circular dichroism, Fourier transform infrared spectroscopy) can effectively report on folding failures in engineered constructs, they lack the resolution to confirm stable incorporation of appended sequences into the tertiary β-barrel structure. Often, an increase in the unitary conductance has been used to verify the success of pore expansion (Liu et al., 2017; Tosaka and Kamiya, 2022). Our work demonstrated that a 10% conductance increase can be attributed to alterations in loop sequences rather than changes in barrel diameter. This finding is consistent with previous studies showing that luminal point mutations or tag presentations can affect conductance without altering the overall channel structure (Baldelli et al., 2024; Bredin et al., 2002; Orlik et al., 2002; Thakur et al., 2017). These observations highlight the importance of using adequate controls, such as sequence-randomized variants, in conductance measurements to attribute changes to the correct structural mechanism. As a result, we concluded that the minor conductance increases observed in some constructs do not provide strong evidence of barrel enlargement. Moreover, while duplicative expansion has been proposed as a broadly generalizable strategy to engineer larger β-barrel channels, our findings suggest that OmpG is less tolerant of β-hairpin insertions than previously reported (Krewinkel et al., 2011; Liu et al., 2017; Tosaka and Kamiya, 2022).

The evolutionary assessment of β-barrel enlargement through hairpin duplication suggests that accretion of new sequence(s) occurs primarily at the N-terminus (Franklin et al., 2018), consistent with our finding that no internal hairpin duplication constructs (OmpG^ExpHP2-HP6) exhibited conductance increases indicative of barrel incorporation. However, the construct exhibiting enhanced conductance, OmpG^{ExpHP7-GNDGKG}, was achieved through C-terminal rather than N-terminal hairpin duplication. Our data suggests that the duplication of hairpin 7 yielded more favorable interactions with adjacent strands than the comparable duplication of hairpin 1, indicating that strand complementarity at the barrel closure interface may be a sufficient determinant of successful hairpin incorporation in our in vitro system. This observation is consistent with the conclusions drawn by Arnold et al. from conceptually similar OmpX barrel recombination studies (Arnold, et al. 2000). While evolutionary trends suggest a preference for adding new sequences at the N-terminus during barrel expansion, they also indicate limited conservation of the newly acquired N-terminal sequences (Franklin et al., 2018). Our results further support this view, suggesting that in detergent-mediated in vitro refolding system governed by the thermodynamic stability of the refolded β-barrel structure, either N- or C-terminal duplications can be successfully accommodated, provided sufficient strand complementarity exists. In contrast, during in vivo chaperone-assisted folding, C-terminal duplication may be disfavored due to the duplication of the regulatory β-signal, thus favoring N-terminal duplication as the evolution route.

Interestingly, the OmpG^ExpHP7-T6 β-pores exhibit an approximately three-fold enhancement in conductance, despite a modest predicted increase in pore diameter from 28 Å to 32 Å (Figure 6), estimated from an AlphaFold2 OmpG-16 model (Mirdita et al., 2022). Theoretical conductance models, such as those by Hall (Hall, 1975) and Kowalzyck et al. (Kowalczyk et al., 2011), suggest this increase should result in only a 15–30% conductance enhancement, depending on ionic conductivity and model length. We propose three potential explanations for this discrepancy: 1) The AlphaFold model may fail to accurately predict certain physical properties of the β-barrel, such as shear or tilt, which can significantly impact conductance. During our use of AlphaFold tools, several limitations became evident. Predictions from AlphaFold2 and AlphaFold3 differed substantially, with AlphaFold3 consistently predicting sixteen-stranded barrels for all duplicated hairpin constructs—an outcome inconsistent with our crystallization data and single-channel measurements. This underscored how algorithm choice can profoundly impact model output. Even within a single algorithm, such as AlphaFold2, discrepancies persisted: constructs predicted to generate expanded-diameter OmpG models in silico often exhibited few or no enhanced-conductance pores in vitro. Conversely, constructs like OmpG^{ExpHP7-GDNGKG}, which empirically exhibited the highest proportion of enhanced-conductance pores, were predicted to perform poorly. Therefore, while AlphaFold and similar tools can potentially help identify regions of structural frustration, they currently lack the precision needed for reliable mutagenesis-based predictions in protein engineering. 2) Modifications to the terminal hairpin may have disrupted proper barrel closure, leading to the formation of a single barrel comprising 16 to 32 strands assembled from hairpin units contributed by two protein chains. However, the proposed “two-chain-one-barrel” model is inconsistent with results obtained from SDS and native gel analyses, which did not indicate the presence of pores formed by two protein chains. 3) Widening of the OmpG luminal constriction, combined with enhanced charge presentation due to the duplication of hairpin seven (introducing two glutamates likely positioned within the barrel lumen), synergistically enhanced conductance (Baldelli et al., 2024).

Figure 6.

Comparison of OmpG^WT structures to AlphaFold prediction of OmpG^ExpHP7-KLTDD sixteen-stranded pore. a) Top and side views of experimentally determined OmpG^WT structures (solid-state NMR [5MWV] left, and X-ray crystallography [2IWV] middle) in comparison to the top AlphaFold model predicted for OmpG^ExpHP7-KLTDD. The duplicated hairpin sequence is highlighted in yellow for the AlphaFold model. b) Plot showing the average per residue pLDDT score for the AlphaFold models of OmpG^ExpHP7-KLTDD above the corresponding top model. Plot and model are colored according to pLDDT score from blue (high) to red (low) according to the relative confidence of the mode, the region surrounded by the dashed line shows the duplicated hairpin sequence. Four of five models were predicted to be sixteen-stranded for this construct.

Overall, we do not have a definitive explanation for the substantially enhanced conductance observed for OmpG^{ExpHP7-GNDGKG} and were not able to achieve sample homogeneity to facilitate structural determination, thus future engineering efforts will focus on increasing the population of the β-pore to enable structural determination. However, we were able to generate a sensor with enhanced conductance, suggestive of barrel expansion, which maintained sensor attributes relative to the scaffold pore, suggesting the general viability of this strategy as an engineering direction in the future.

Conclusion

Here we demonstrated that OmpG can tolerate duplicative barrel expansion, with feasibility varying by position, and that rational protein engineering improves incorporation efficiency. However, optimizations in in vitro detergent-refolding may not correlate with in vivo outer membrane protein folding, indicating a need to prioritize one system, particularly in vitro for nanopore sensor development. In addition, the enhanced conductance pores retained the sensing functionality of the original scaffold, supporting duplicative expansion as a viable strategy to develop β-barrel OMP sensors with tunable dimensions.

Materials and Methods

Materials

DH10β E. coli strain was purchased from Thermo Scientific. BL21ΔABCF E. coli strain was purchased from Addgene. C43 E. coli strain was purchased from Lucigen. Phusion High-Fidelity DNA Polymerase kit, SapI and DpnI restriction enzymes, T4 DNA ligase, T4 polynucleotide kinase, Monarch Miniprep kit, and unstained broad-range protein standard were purchased from New England BioLabs. DNA clean & concentrator kit was purchased from Zymo Research. Q-Sepharose fast-flow resin was purchased from Cytiva. n-octyl-β-D-glucopyranoside (OG) was obtained from Chem Impex and octaethylene glycol monododecyl ether (C12E8) from Cayman Chemical. n-Dodecyl-β-D-maltopyranoside (DDM) and L-arabinose were purchased from GoldBio. Kanamycin and 3-[(3-cholamidopropyl)-dimethylammonio]-1-propane sulfonate] (CHAPS) were purchased from Sigma-Aldrich. Lauryldimethylamine-N-oxide (LDAO), n-nonyl-β-D-glucopyranoside (NG), n-decyl-β-D-maltopyranoside (DM), and (2,2-didecylpropane-1,3-bis-β-D-maltopyranoside) LMNG were purchased from Anatrace. Assorted labware, potassium chloride (KCl), hexadecane, pentane, and methanol were purchased from Fisher Scientific. 1,2-diphytanoyl-sn-glycerol-3-phosphocholine (DPhPC) lipid was purchased from Avanti Polar Lipids. 2,2,2-trichloroethanol (TCE), Pierce BCA protein assay kit, Pierce ECL western blotting substrate, BL21 A1E. coli strain, monoclonal mouse anti-FLAG antibody FG4R [RRID AB_1957945], FG4R conjugated to Dylight488 fluorophore [RRID AB_2537621], FG4R conjugated to horseradish peroxidase (HRP) [RRID AB2537626], and HisPur^™ Ni-NTA resin were purchased from Thermo Fisher Scientific. Immun-Blot PVDF membrane was purchased from Bio-Rad. Bovine serum albumin (BSA), fraction V was sourced from Caisson Labs. Powdered milk was purchased from Big Y. All other reagents were obtained from Research Products International unless otherwise stated.

Cloning of OmpG Constructs

OmpG constructs carrying the FLAG-tag motif (-GDYKDDDDKG-) in loops 1–7 were generated in prior work (Foster et al., 2023). All subsequent constructs were generated using oligonucleotides (Eurofins MWG operon) listed in Table S3. Before initiating construct cloning, internal SapI sites were removed from the desired inclusion body expression (pT7) and membrane expression (pET28b) vector backbones using oligo pairs ‘pT7_SapI_Heal’ and ‘28b_SapI_Heal’, respectively. For pT7/OmpG^ExpHPX constructs, a two-fragment Golden Gate cloning strategy was pursued. Fragment 1 containing the vector backbone was generated by PCR using sap-healed pT7/OmpG^WT as the template, with fragment 2 using pT7/OmpG^LX-FLAG as the template to amplify the desired FLAG-tagged hairpin unit. All fragments were subjected to DpnI digestion for 3+ hours at 37°C to remove template plasmid DNA. The digested PCR mixture was cleaned using the Zymo Clean & Concentrator kit. For each construct, the corresponding fragments 1 and 2 were then mixed at equimolar concentrations and subjected to a one-pot SapI-digestion/T4 ligation reaction. Ligated DNA was then transformed into electrocompetent DH10β E. coli under dual ampicillin (150 μg/mL) and streptomycin (50 μg/mL) selection. Colonies containing the desired mutant were confirmed via Sanger sequencing (Eurofins Genomics). Generation of pT7/OmpG^ExpHP7 constructs with novel β-turn sequences was achieved following a similar two-fragment Golden Gate strategy. Following sequence confirmation of the desired mutation, all constructs intended for in vitro refolding and single-channel characterization underwent a final round of PCR to remove the endogenous signal sequence using oligonucleotide set ‘ssdel’. The PCR mixture product was treated as before and then subjected to a one-pot T4 polynucleotide kinase and ligase reaction to phosphorylate and circularize the DNA for transformation. The ligation mixture was transformed into house-made electrocompetent DH10β as before. For OmpG constructs intended for outer membrane expression, an initial sub-cloning using oligo pair ‘Sbcln’ was performed to amplify the corresponding OmpG^ExpHPX sequence for restriction/ligation insertion (NcoI/NotI sites) into pET28b. Subsequently, an N-terminal hexaHis tag following the signal sequence was added to pET28b/OmpG^ExpHPX to facilitate purification following membrane expression.

Finally, for constructs designed to study the influence of β-signal variants, a new template, OmpG^ExpHP7alt, with a non-native nucleotide sequence of the duplicated hairpin seven, was assembled iteratively from template pET28b/OmpG^WT using oligo pairs ‘ExpHP7alt 1–3’. This new construct was used as a template to introduce the desired β-signal variants via mutagenic PCR using the corresponding oligos indicated in Table S3. Cloning proceeded under kanamycin (50 μg/mL) selection. All constructs containing the desired mutant were confirmed via Sanger sequencing.

Preparation and in vitro refolding of OmpG Constructs

OmpG^ExpHPX variants without the signal sequence were expressed in electrocompetent C43 E. coli cells. Cells were grown in a 2xYT medium containing 200 μg/mL ampicillin at 30°C until an OD₆₀₀ of 0.5–0.6 was reached. Protein expression was induced by adding 500 μM isopropyl ß-D-1-thiogalactopyranoside (IPTG), and the culture was allowed to grow at 30°C for an additional 3–4 hours. Cells were harvested by centrifugation, and the cell pellets were lysed in buffer (50 mM Tris-HCl pH 8.0, 1 mM EDTA) via sonication. The lysate was centrifuged at 20,000 rcf at 4 °C, and the supernatant was removed. The inclusion body pellet was washed once with wash buffer (1.5 M urea, 50 mM Tris-HCl pH 8.0) and then solubilized in binding buffer (8 M urea, 50 mM Tris-HCl pH 8.0) with constant stirring for 30 minutes. The solubilized pellet was centrifuged as before, and the OmpG protein was purified from the resultant supernatant using anion-exchange chromatography under gravity with a 3 mL Q-Sepharose bead volume (Cytiva #17051010). The column-bound protein was washed with buffers (8 M urea, 50 mM Tris-HCl pH 8.0, 75 mM NaCl) and (8 M urea, 50 mM Tris-HCl pH 8.0, 200 mM NaCl), and eluted with buffer (8 M urea, 50 mM Tris-HCl pH 8.0, 500 mM NaCl). Purity was assessed via 12% SDS-PAGE gel visualized with TCE. Protein concentration was determined using both a Pierce^™ BCA kit and A280 measurements via NanoDrop 2000, with extinction coefficient and predicted molecular weight calculated using the Benchling web GUI. Denatured OmpG was then diluted with refolding buffer (50 mM Tris-HCl pH 9.0 supplemented with the indicated detergent) at a 3:5 volume ratio (denatured OmpG to refolding buffer, with final detergent concentrations indicated in Figure S12). The protein was incubated at 37°C for approximately 24 hours, and refolding efficiency was assessed via SDS-PAGE gel shift assay.

Clear Native-PAGE

Purified proteins in a denatured or detergent-refolded state were mixed with 6X sample buffer (600 mM Tris-HCl pH 6.8, 300 μM Bromophenol Blue, 50% (v/v) glycerol) and loaded onto lab-cast fixed-percentage 12% polyacrylamide gels. Electrophoresis was carried out at 200V for 120min using cathode buffer (50 mM tricine, 7.5 mM imidazole, pH 7.0, 200 μM DDM, 1.15 mM deoxycholic acid (DOC)) and anode buffer (25 mM imidazole pH 7.0) recipes adapted from Wittig et al (Wittig and Schägger, 2008). Protein samples were visualized via TCE staining.

Crystallization

Refolded OmpG^ExpHP3 underwent crystallization screening using the sitting-drop vapor-diffusion method at 293 K using the sparse-matrix MemGold 2 screen (Molecular Dimensions). The protein was mixed with the precipitant in a 1:1 ratio using a Mosquito LCP robot (SPT Labtech). Needle crystal clusters appeared after 3 days in the following condition: 0.4 M Ammonium Sulfate, 0.1 M MES pH 6.5 and 12% PEG300. Single OmpG^ExpHP3 crystals were obtained in 0.4 M Ammonium Sulfate, 0.1 M MES pH 6.5 and 14% PEG300. Crystals were cryoprotected in a mixture of well solution supplemented with 30% PEG 300.

Data Collection

Diffraction data were collected at the I03 beamline at Diamond Light Source (DLS), Didcot, United Kingdom using an EIGER2 XE 16M detector. The crystals belonged to the space group P2₁2₁2₁. Diffraction frames were indexed and integrated using the autoproc (https://doi.org/10.1107/S0907444911007773). The data were scaled using AIMLESS in the CCP4 suite (https://doi.org/10.1107/S0907444913000061 and https://doi.org/10.1107/S2059798323003595).

Structure Solution

The structure of OmpG^ExpHP3 was solved by molecular replacement with the wild type OmpG structure (PDB ID: 2X9K) (https://doi.org/10.1016/j.jmb.2010.06.015) using Phenix (https://doi.org/10.1107/S2059798319011471). Electron density inspection was done in Coot (https://doi.org/10.1107/S0907444910007493). No density for the additional strands could be observed after refinement in Phenix (https://onlinelibrary.wiley.com/doi/pdf/10.1107/S0907444902016657).

The data collection and refinement statistics are summarized in Table S4.

Expression of OmpG constructs for assessment of in vivo folding

Plasmids pET28b/OmpG^ExpHPX containing OmpG variants with the signal sequence were transformed into electrocompetent BL21ΔABCF (DE3) E. coli cells. Cells were grown in a 300 mL 2xYT medium containing 50 μg/mL kanamycin at 30°C until an OD₆₀₀ of 0.5–0.6 was reached. Protein expression was induced by adding 500 μM IPTG, and the culture was allowed to grow at 23°C / 250 rpm for approximately 16 hours. Cells were harvested by centrifugation, and the cell pellets were lysed in buffer (1x PBS [137 mM NaCl, 2.7 mM KCl, 10 mM Na₂HPO₄, 1.8 mM KH₂PO₄] pH 7.4, 10 mM imidazole, 1 mM EDTA, 100 μM phenylmethylsulfonyl fluoride (PMSF)) via sonication using a Misonix instrument with the following parameters: 1/2-inch probe, 40% amplitude, 2-second pulse, 4-second rest, 2min process time, for ≥ 2 cycles on ice. The lysate was centrifuged at 75,000 rcf / 10 °C for 30 min, and the supernatant was removed. The membrane-fraction-containing pellet was resuspended in extraction buffer (1x PBS pH 7.4, 10 mM imidazole, 1% (w/v) DDM, 100 μM PMSF) and stirred for 30 minutes at 23°C. After the incubation, the extraction solution was centrifuged again at 75,000 rcf / 10 °C, and the resultant supernatant was applied to a 3 mL HisPur^™ Ni-NTA bead column. The column was washed with buffers (1x PBS pH 7.4, 100 μM PMSF, with 20-, 35-, 50-, and 75- mM imidazole) and eluted with buffer (1x PBS pH 7.4, 100 μM PMSF, 150 mM imidazole). Purity was assessed via 12% SDS-PAGE gel visualized with TCE.

For western blot, the protein samples on the gel were transferred to a 0.2 μm pore PVDF membrane using Tris-glycine running buffer (25 mM Tris-HCl, 192mM glycine, pH 7.5, 20% (v/v) methanol) by applying a constant 300 mA for 2 hours at 4 ⁰C. The membrane was blocked with 5% (w/v) milk in TBST (25 mM Tris-HCl, pH 7.5, 150 mM NaCl, 0.1% (v/v) Tween20) for at least 1 hour at 23 °C, washed twice with 20mL TBST, and then incubated with visualization antibody (1:2000 dilution of FG4R-HRP in TBST + 3% BSA) for 1 hour at 23°C. The membrane was washed three more times with 20 mL TBST and then briefly incubated with ECL substrate for chemiluminescence visualization.

Single-channel recording of OmpG proteins

Single-channel recordings of OmpG were performed similarly to prior studies (Foster et al., 2023). Briefly, experiments were conducted using a custom chip apparatus containing two chambers separated by a 25 μm thick Teflon film. An aperture of approximately 100 μm in diameter was created in the film using a briefly applied electric arc. The aperture was pretreated with hexadecane in pentane (10% v/v) before the chambers were filled with buffer. Ag/AgCl electrodes were immersed in each chamber, with the cis chamber grounded. DPhPC dissolved in pentane (10 mg/mL) was added to the surface of the buffer in both chambers, and monolayers were formed by pipetting the solution up and below the aperture multiple times. After successfully generating a planar lipid bilayer, refolded OmpG proteins (final concentration ~500 pM) were added to the cis chamber. A voltage potential of +250 mV was briefly applied to facilitate OmpG insertion. Following single-pore insertion, the applied voltage was reduced to ± 50 mV for recording. Conductance assessments were performed in buffer 50 mM Tris-HCl, pH 7.4, 1 M KCl. single-channel detection of FG4R was conducted in buffer 50 mM Tris-HCl, pH 7.4, 300 mM KCl. The current was amplified using an Axopatch 200B integrated patch clamp amplifier (Molecular Devices), with digitized acquisition at 10 kHz via a Digidata 1440A board (Molecular Devices) with 2kHz Bessel filter.

Analysis was performed using Clampfit 11.1 (Molecular Devices) and Origin 2020+ (OriginLab Corporation). OmpG pore orientation was determined based on previously reported voltage polarity-influenced gating behavior (Chen et al., 2008b). For analyte detection assessment, FG4R was added to the cis or trans chamber depending on pore orientation, and the solution was mixed by pipetting a volume representing one-tenth of the chamber volume ~ 10+ times. Data was acquired at a ‘noisy’ voltage bias (−50 mV for a cis loop-facing pore, +50 mV for a trans loop-facing pore).

Flow Cytometry assay

Plasmids containing OmpG variants were transformed into BL21 A1 E. coli for protein expression. Cells were grown at 30°C / 280 rpm until an OD₆₀₀ of ~0.6. 500 μM (final concentration) L-arabinose was added to cultures, which were then incubated at 30°C for an additional 2 hours. The OD₆₀₀ of the induced culture was measured again to normalize the number of cells for labeling. A volume representing 4 × 10⁸ cells was transferred to a sterile 2 mL Eppendorf tube, based on the conversion factor of an OD₆₀₀ of 1.0 = 8 × 10⁸ cells / mL. All samples were pelleted by centrifugation at 3,095 rcf at 4 °C for 3 minutes. The media was decanted, and the pellet was washed with 1 mL flow buffer (1x PBS pH 7.4, 1 mM EDTA, and 0.5% w/v BSA fraction V). Cells were pelleted again, resuspended in 200 μL of flow buffer, and then incubated with 16 nM FG4R-Dylight 488 at 4°C with gentle agitation at 15 rpm for 1 hour. Labeled cells were washed with 1 mL flow buffer as before and then resuspended in flow buffer at ~3 × 10⁷ cells / mL for flow cytometry analysis by using a BD Dual LSRFortessa 5-laser cytometer (BD Biosciences) with a 488 nm laser, a 505 nm long pass filter, and a 530/30 nm band pass filter.

Supplementary Material

Supplementary Material Description: amino acid sequence of protein variants, SDS- & Native PAGE images, additional current recording traces, electron density map for OmpG^ExpHP3, model of OmpG^ExpHP7-KLTDD b-pore, table of b-turn sequences, table of b-signal sequences, table of cloning oligonucleotides (PDF).

Additional supporting information can be found online in the Supporting Information section at the end of this article.