In silico analysis of zosurabalpin-LptB2FG binding in Acinetobacter spp., Klebsiella pneumoniae, and Shigella flexneri: mechanisms underlying its differential efficacy

Abstract

Zosurabalpin, a novel tethered macrocyclic peptide antibiotic, exhibits potent activity against Acinetobacter spp., particularly carbapenem-resistant Acinetobacter baumannii (CRAB). Zosurabalpin inhibits lipopolysaccharide (LPS) transport by targeting the LptB2FG protein complex, resulting in toxic LPS accumulation and bacterialdeath. This study investigates zosurabalpin’s molecular specificity against Acinetobacter spp., its ineffectiveness against Klebsiella pneumoniae, and its potential efficacy against Shigella flexneri. Comparative analysis of LptB2FG sequences and structures, revealed significant differences in LptB2FG protein conformations, pocket geometry and electrostatic surface surrounding the binding pocket among the three species, which may influence zosurabalpin binding. Docking results for zosurabalpin showed lower binding affinities for K. pneumoniae and S. flexneri compared to Acinetobacter baylyi. Additionally, other zosurabalpin derivatives were tested showing improved binding affinities for K. pneumoniae but not for S. flexneri. These findings underscore the need for tailored zosurabalpin derivatives to enhance efficacy against a broader spectrum of Gram-negative bacteria.

Supplementary Information

The online version contains supplementary material available at 10.1007/s40203-025-00343-3.

Introduction

Antibiotic resistance in Gram-negative bacteria poses a significant threat to global health, as their asymmetrical outer membrane (OM) serves as a potent defense mechanism against antibiotics (Fair and Tor 2014; Silhavy et al. 2010; Magri et al. 2023). These drug-resistant pathogens exhibit a unique bilayer membrane structure. The inner membrane (IM) corresponds to the typical phospholipid bilayer structure, whereas the outer leaflet of the OM consists essentially of glycolipids, mainly lipopolysaccharide (LPS) (Raetz and Whitfield 2002).

LPS is a glycolipid composed of a lipid moiety called lipid A (contains four to seven acyl chains attached), a core of oligosaccharides and O-antigenic polysaccharides (Raetz and Whitfield 2002). LPS contribute to the high impermeability against small hydrophobic molecules that might easily cross phospholipid bilayers, increasing the resistance of Gram-negative bacteria to many antimicrobial compounds. LPS can also play a critical role in the virulence of gram-negative bacteria by enhancing their ability to trigger pathologies (Bertani and Ruiz 2018).

The transport of lipopolysaccharides (Lpt) is mediated by a system composed of seven essential proteins (LptB2FGCADE), which form a bridge between the IM and OM by crossing the periplasm. These proteins are linked through jellyroll domains presenting in LptF, LptG, LptC, LptA, and LptD (Sperandeo et al. 2017). At the inner membrane, the LptB2FG transporter forms an ATP-binding cassette (ABC) that couples ATP hydrolysis to LPS extraction (Li et al. 2019). Indeed, the LptB2 dimer, representing the nucleotide-binding domain, is responsible for ATP hydrolysis, while LptF/G interacts with LPS and transports it to LptC and LptA, located in the periplasm, which then transfer LPS to LptDE located in OM (Li et al. 2019).

Targeting the seven Lpt proteins that form the LPS transporter may offer an effective therapeutic strategy to disrupt OMintegrity by preventing the transport of LPS. This strategy can increase susceptibility to other antibiotics by making membrane more permeable (Pahil et al. 2024). Furthermore, it causes LPS to build up to toxic levels inside the cell, ultimately leading to cell death Li et al. (2019).

In early 2024, a group of researchers reported a new macrocyclic peptide (MCP) called zosurabalpin, which selectively targets LptB2FG in A. baumannii after screening a set of around 45,000 MCPs. Zosurabalpin has demonstrated significant efficacy in inhibiting the proliferation of diverse carbapenem-resistant Acinetobacter baumannii (CRAB) strains, including multidrug-resistant strains, in both in vitro and in vivo studies. Animal models have confirmed its therapeutic potential, showing a substantial decrease in bacterial load without toxicity (Zampaloni et al. 2024). Furthermore, in Phase 1 clinical trials, zosurabalpin was well tolerated and safe at doses up to 2000 mg, with no significant adverse effects, reinforcing its clinical promise (Guenther et al. 2023).

The analysis of Cryo-EM 3D structures of LptB2FG revealed that zosurabalpin engages LptB2FGC only when the complex is bound to LPS, highlighting the importance of the presence of LPS for its mode of action (Zampaloni et al. 2024). Zosurabalpin was tested and found to be inactive against a wide range of Gram-positive and Gram-negative bacterial species, including E. coli ATCC 25922, K. pneumoniae ATCC 700603, P. aeruginosa ATCC 27853 and S. aureus ATCC 29213, as well as the fungal species such as C. albicans ATCC 90028. These findings indicate that zosurabalpin is highly specific and selective for Acinetobacter species, particularly A. baumannii (Zampaloni et al. 2024).

In this study, we aimed to elucidate the molecular basis underlying zosurabalpin’s specificity for Acinetobacter spp. and its lack of activity against Klebsiella pneumoniae. We further explored its potential efficacy against Shigella flexneri. To achieve this, we conducted a comparative analysis of the LptB2FG protein sequences and structures across these three bacterial species using multiple sequence alignment, molecular modeling, and molecular docking approaches.

Materials and methods

Sequence alignment of LptF, LptB, and LptG proteins

Sequence alignments for LptF, LptB, and LptG from A. baylyi, K. pneumoniae, S. flexneri were performed using the Clustal Omega tool available on the EMBL-EBI web server (Madeira et al. 2019). The alignment was conducted with default parameters, including the output format set to ClustalW with character counts, MBED-like clustering enabled for guide tree generation and iteration, and the output order set to ‘aligned.’ De-alignment of input sequences, distance matrix generation, and combined iterations (set to 0) were disabled. Guide tree and HMM iterations were kept at their default values. All protein sequences were retrieved from UniProt (Table 1).

Table 1.

Uniprot identifiers for LptB, LptF, and LptG proteins of A. baylyi, K. pneumoniae, and S. flexneri

Protein	Uniprot identifiers for each species
	A. baylyi	K. pneumoniae	S. flexneri
LptB	Q6FC66	A0A2X3II39	P0A9V4
LptF	Q6FFD7	A0A1Y0Q3P9	P0AFA1
LptG	Q6FFD6	A0A1Y0Q3C0	A0A0H2V3J7

Structural and electrostatic analysis of LptB2FG complexes

To provide insights into the structural and electrostatic differences among the LptB2FG complexes of the three species, the Cryo-EM 3D structures of LptB2FG from A. baylyi (PDB ID: 8FRN, resolution: 3.30 Å) in complex with zosurabalpin and LPS, as well as those from K. pneumoniae (PDB ID: 7EFO, resolution: 3.85 Å) and S. flexneri (PDB ID: 6S8H, resolution: 3.70 Å) in complex with LPS only, were obtained from the Protein Data Bank (PDB). These structures were superimposed and visualized using UCSF ChimeraX 1.4 (Pettersen et al. 2021), enabling a detailed comparison of their conformations. The electrostatic surfaces of these structures were then assessed using Chimera X 1.4 (Pettersen et al. 2021).

Ligand preparation for docking

The zosurabalpin structure was extracted from PDB entry 8FRN and processed using AutoDockTools (Morris et al. 2009). During preparation, hydrogen atoms were added, Gasteiger charges were computed and assigned to the ligand, the torsion tree root was defined, and rotatable bonds were identified. Finally, the prepared ligand was saved in PDBQT format, ready for subsequent docking studies. Additionally, the structures of newly designed zosurabalpin derivatives were sketched using the 2D sketcher in Maestro software by replacing the two main positively charged amino acids (ornithine and lysine) with negatively charged amino acids (aspartic acid and glutamic acid). The resulting compounds were saved as 2D SDF files, then converted to 3D and saved in PDBQT format using Open Babel (O’Boyle et al. 2011). These derivatives were further prepared using AutoDock Tools with the same steps as zosurabalpin.

Preparation of LptB2FG structures and molecular docking

The Cryo-EM 3D structures of LptB2FG from A. baylyi (PDB ID: 8FRN), K. pneumoniae (PDB ID: 7EFO), and S. flexneri (PDB ID: 6S8H) were prepared for docking studies using AutoDock Tools (Morris et al. 2009). The LPS molecules were retained in each structure to maintain the native binding pocket. Receptor preparation involved removing water molecules, adding polar hydrogens, computing Kollman charges, assigning missing atoms, and assigning atom types. Additionally, to ensure a unified grid box for docking, the three structures were superimposed. The grid box was then optimized to cover the zosurabalpin binding pocket, with the grid box centered at coordinates (111.93, 588.6, 119.51) and a size of 25Å × 25Å × 25Å.

Molecular docking for zosurabalpin and the newly designed derivatives against the prepared structures was conducted using AutoDock Vina 1.2.0 (Eberhardt et al. 2021), with an exhaustiveness of 300.

ADMET prediction

ADMET (Absorption, Distribution, Metabolism, Excretion, and Toxicity) properties of the compounds were predicted using the ADMETLab 2.0 tool (Xiong et al. 2021).

Results and discussion

Zosurabalpin, a novel tethered macrocyclic peptide antibiotic, demonstrates potent and selective activity against Acinetobacter spp., particularly carbapenem-resistant Acinetobacter baumannii strains (Zampaloni et al. 2024). This new antibiotic, currently undergoing Phase I clinical trials (Guenther et al. 2023), works by inhibiting the transport of LPS in the bacteria. It targets a protein complex called LptB2FG, which is essential for transporting LPS to the bacterial surface and building the OM, a hallmark of Gram-negative bacteria (Pahil et al. 2024). By blocking this transport, zosurabalpin causes LPS to accumulate to toxic levels within the cell, ultimately leading to cell death (Gugger and Hergenrother 2024).

In this study, we aimed to decipher the molecular basis, to understand why zosurabalpin is effective against Acinetobacter spp., but not against Klebsiella pneumoniae, according to experimental studies. We extended this analysis to also explore the zosurabalpin's potential efficacy against Shigella flexneri, a bacterium not previously investigated. To achieve this, we conducted a comparative analysis of the LptB2FG protein sequences and structures across all three bacterial species. This employs a multi-faceted approach involving multiple alignment techniques, molecular modeling, and molecular docking simulations.

Given that the 3D structure of LptB2FG of A. baumannii is still unresolved, in this study, we carried out a comparison using that of Acinetobacter baylyi. These homologous bacterial proteins showed a high degree of sequence conservation compared to those of A. baumannii, sharing 81.7% identity for LptF and 82% identity for LptG, with almost all residues (13 of 16) that interact with LPS or zosurabalpin conserved, except for two residues in LptF (arginine 320 and arginine 325, both replaced by lysine) and one in LptG (serine 40 replaced by glycine) (Zampaloni et al. 2024).

Assessing sequence conservation among LptB2FG proteins

To assess the sequence conservation of LptB, LptF and LptG proteins among A. baylyi, K. pneumoniae, and S. flexneri, we retrieved their sequences from UniProt and performed multiple sequence alignment using the Clustal Omega (Madeira et al. 2019; Magri and Abdel-Mawgoud 2022).

The alignment of sequences from S. flexneri and K. pneumoniae against A. baylyi revealed a lower degree of similarity for LptF and LptG, with medium similarity observed for LptB (Table 2). Indeed, aligning LptF of K. pneumoniae with A. baylyi showed 98.40% coverage and 24% identity. LptG showed 100% coverage with 29.70% identity, while LptB showed 96.8% coverage and 66% identity (Table 2; Fig. 1). For S. flexneri, LptF aligned with A. baylyi exhibited 98.40% coverage and 24.50% identity; LptG reached 100% coverage and 29.70% identity, and LptB showed 96.80% coverage and 64.70% identity (Table 2; Fig. 1). Focusing on the 16 residues implicated in zosurabalpin interaction 3,K. pneumoniaeK. pneumoniae revealed 3 conserved residues in LptF (58, 249 and 271) and 1 conserved residue in LptG (39) compared to A. baylyiA. baylyi. Similarly, S. flexneriS. flexneri showed 2 conserved residues in LptF (85 and 271) and 1 in LptG (39) when compared to A. baylyiA. baylyi (Table 3; 1).It should be noted that previous mutagenesis studies on the key interacting residues showed that point mutations in these residues can reduce zosurabalpin sensitivity by increasing the Minimum Inhibitory Concentration (MIC), indicating that these residues play a crucial role in maintaining zosurabalpin binding and activity (Pahil et al. 2024). Notably, the LptF-E249K mutation resulted in a 283-fold increase in MIC, from 0.06 µM to 17 µM; LptF-I317N showed a 150-fold increase, from 0.06 µM to 9 µM; LptF-R320T led to a 28-fold increase, from 0.06 µM to 1.4 µM; and LptG-L36Q resulted in a 7.33-fold increase, from 0.06 µM to 0.44 µM (Pahil et al. 2024).

Table 2.

Comparison of percentage identity and coverage of LptB, LptF, and LptG proteins in K. pneumoniae and S. flexneri against A. baumannii and A. Baylyi

	A. Baylyi
	LptB		LptF		LptG
	Identity	Coverage	Identity	Coverage	Identity	Coverage
K. pneumoniae	66.00%	96.80%	24.00%	98.40%	29.70%	100%
S. flexneri	64.70%	96.80%	24.50%	98.40%	29.70%	100%

Fig. 1.

Multiple sequence alignment of LptB (A), LptF (B), and LptG (C) sequences from A. baylyi, K. pneumoniae, and S. flexneri Using Clustal Omega. Asterisks indicate key residues involved in the interaction with zosurabalpin; blue asterisks denote conserved residues, and red asterisks indicate non-conserved residues

Fig. 3.

Docking poses, 2-D diagram interaction and binding affinity of zosurabalpin in complex with LptB2FG of A. baylyi

Table 3.

Alignment of active site residues of LptG and LptF interacting with Zosurabalpin across the three species

	LptF												LptG				Matched amino acids
Positions	55	58	125	249	271	314	317	318	320	321	322	325	36	37	39	40
A. baylyi	R	E	L	E	W	V	I	A	R	T	R	R	L	G	L	S	–
K. pneumoniae	G	E	S	E	W	L	T	S	K	S	N	K	V	D	L	K	4
S. flexneri	G	E	S	D	W	L	T	S	K	S	N	K	V	D	L	K	3

On the other hand, sequences comparison between K. pneumoniae and S. flexneri showed a high degree of conservation, with LptF and LptG sharing identities of 85.48% and 92.50%, respectively, and complete coverage (100%), distinguishing them from the limited similarity observed when compared to that of A. baylyi. Furthermore, the key amino acids involved in zosurabalpin interaction in A. baylyi were similarly conserved between K. pneumoniae and S. flexneri. In LptF, 11 conserved residues were identified (55, 58, 125, 271, 314, 317, 318, 320, 321, 322, and 325), while LptG displayed conservation in 4 residues (36, 37, 39, and 40) (Table 3; Fig. 1).

Taken together, the high conservation between K. pneumoniae and S. flexneri, and the poor similarity of both bacteria to A. baylyi, may cause significant differences in protein structure conformation, helix orientation as well as differences in some protein surface properties, such as electrostatic surface properties. These differences could affect the binding mode of LPS and zosurabalpin within the V-shaped cavity.

Assessing structural variations among LptB2FG proteins

To explore how sequence variations might affect the protein structure, we retrieved Cryo-EM structures of LptB2FG from A. baylyi in complex with zosurabalpin and LPS, as well as structures from K. pneumoniae and S. flexneri in complex with LPS only (Fig. 2). A comparison of these three structures revealed distinct positioning of LPS, particularly for the LPS inner core, within the central V-shaped cavity formed by the transmembrane domains (TMDs) of LptFG (Fig. 2F, G and H). This variation stems from the orientation of the transmembrane helices of LptFG across the three complexes (Fig. 2D and E). The differences in the orientation of these helices significantly impact the overall orientation of LPS within the cavity, as numerous hydrophobic residues from α1, α2, and α5 of both LptF and LptG envelop the six acyl chains of Lipid A within the V-shaped cavity (Luo et al. 2021).

Fig. 2.

A–C, Overall structure of thequaternary LPS transporter LptB2FG from A. baylyi (A), K.pneumoniae (B) and S. flexneri (C). Two copies of LptBconstitute the nucleotide-binding domains (NBDs), and the transmembrane (TMD)domains consist of transmembrane helices of LptF and LptG. D, E,Superimposition of the 3D structures of A. baylyi (in gold) and K.pneumoniae (in light blue) and S. flexneri (in light green) LptB₂FG.Panel D superimposes the structures to underscore differences in their3D conformations and helix orientations, with Panel E focusing on theV-shaped cavity formed by the transmembrane helices of LptF and LptG domains. F–H,I–K, surface and cartoon representations of the different pocketsformed upon the binding of LPS to within the LptB2FG V-shaped cavity of A.baylyi (F, I), K. pneumoniae (G, J) andS. flexneri (H, K). The zosurabalpin binding pocket observedin A. baylyi, formed between LptGF and LPS, is absent in K.pneumoniae and S. flexneri. Conversely, two other pockets (pocket 1and pocket 2) formed between LPS and LptF are observed in K. pneumoniae andS. flexneri. The red grid illustrates the grid box used for docking. L–N,the electrostatic surface of A. baylyi (L), K. pneumoniae(M) and S. flexneri (N) LptB₂FG, with negativesurfaces shown in red and positive surfaces in blue.

Indeed, in both K. pneumoniae and S. flexneri, the inner core is oriented towards helices 5 and 6 of LptF, resulting in the formation of two distinct pockets between LPS and LptF (Fig. 2G, H, J and K). The first, larger pocket (pocket 1) forms between LPS and helices 1, 2, 3, and 4 of LptF, while the second, smaller pocket (pocket 2), forms between LPS and helix 5 of LptG along with helix 1 of LptF. In contrast, in A. baylyi, the inner core of LPS is oriented towards helices 1 and 2, creating a larger pocket surrounded by LPS and helices 1, 2, and 3 of LptF, as well as helix 1 of LptG (Fig. 2F and I). It is noteworthy that zosurabalpin in A. baylyi engages LptFG only when the complex is bound to LPS, specifically at the pocket formed between helices 4 and 5 of LptF, helix 1 of LptG, and the acyl chains of LPS (Gugger and Hergenrother 2024), indicating the importance of this specific pocket in zosurabalpin recognition and fixation. Additionally, the hydroxybenzoic acid moiety of zosurabalpin is accommodated within a small channel situated between helix 1 of LptF and helix 5 of LptG at the binding site. This channel is present in pocket 2 of both K. pneumoniae and S. flexneri. However, due to the limited volume of this pocket, steric occlusion may occur, potentially obstructing the binding of zosurabalpin in this region.

Beyond the differences in LPS localization and the geometry of pockets formed in each species, we conducted a detailed analysis of the electrostatic surfaces surrounding these pockets to determine whether they share similar electrostatic properties. Our findings revealed notable differences in the electrostatic surfaces, particularly the presence of a significantly negatively charged region within the pocket in A. baylyi (Fig. 2L). In contrast, pockets 1 and 2 of K. pneumoniae were characterized by a predominance of positively charged or neutral residues, with a small surface area exhibiting a slight negative charge in pocket 1 (Fig. 2M). Similarly, pockets 1 and 2 of S. flexneri were generally neutral or positively charged (Fig. 2N). It is noteworthy that the primary amino acids (lysine and ornithine) of zosurabalpin lodge into a negative pocket in LptFG of A. baylyi (Pahil et al. 2024) that does not exist in LptFG pockets of K. pneumoniae and S. flexneri (Fig. 2L). This may result in weaker electrostatic interactions between zosurobalpin and the pockets (Bitencourt-Ferreira et al. 2019).

To verify whether zosurabalpin can bind to these pockets despite the differences in their geometry and electrostatic nature, we performed molecular docking with a grid box covering both LptFG pockets of K. pneumoniae and S. flexneri using AutoDock Vina (Eberhardt et al. 2021). The results showed that zosurabalpin binds to pocket 1 of K. pneumoniae with an affinity score of − 7.4 kcal/mol and pocket 2 with − 6.0 kcal/mol (Fig. 4A), while in S. flexneri, zosurabalpin binds only to pocket 1 with an affinity score of − 5.25 kcal/mol (Fig. 4B). These scores remain lower compared to A. baylyi, which exhibited an affinity score of − 11.1 kcal/mol (Fig. 3), suggesting that zosurabalpin may have lower binding efficacy or specificity towards the pockets of K. pneumoniae and S. flexneri compared to A. baylyi. Taken together, these results could explain the lack of efficacy of zosurabalpin against K. pneumoniae and suggest its potential ineffectiveness against S. flexneri.

Fig. 4.

A, B, Docking poses, 2-D diagram interaction and binding affinity of zosurabalpin in complex with LptB2FG of K. pneumoniae (A) and S. flexneri (B)

In silico assessment of zosurabalpin derivatives with enhanced binding potential

To improve the binding affinity, we tested a series of zosurabalpin derivatives in which the two main positively charged amino acids (ornithine and lysine) were replaced by the negatively charged amino acids, aspartic acid and glutamic acid (Table 4). This modification aimed to enhance the electrostatic interactions within the binding pockets. The ADMET properties of these molecules were evaluated using ADMETLab 2.0. The ADMET predictions indicated that the majority of the properties of the designed derivatives were similar to those of zosurabalpin, as presented in Table S1.

Table 4.

Structures and Docking scores of newly designed compounds

Compound	Chemical structure	Substitution pattern		Docking score (kcal/mol)
		R1	R2	A. baylyi	K. pneumoniae	S. flexneri
Zosurabalpin		Lysine	Ornithine	− 11.1	− 7.4	− 5.25
C1		Glutamic Acid	Ornithine	− 9.20	− 8.95	− 7.00
C2		Aspartic Acid	Ornithine	− 9.61	− 9.33	− 7.44
C3		Aspartic Acid	Aspartic Acid	− 9.80	− 10.04	− 7.91
C4		Glutamic Acid	Glutamic Acid	− 9.8	− 9.66	− 7.43
C5		Glutamic Acid	Aspartic Acid	− 9.48	− 9.46	− 6.66
C6		Aspartic Acid	Glutamic Acid	− 9.56	− 9.35	− 7.14
C7		Lysine	Glutamic Acid	− 9.15	− 9.28	− 6.73
C8		Lysine	Aspartic Acid	− 9.08	− 9.11	− 6.96

The value in bold indicates the molecule with the highest docking score

These molecules were then subjected to molecular docking against A. baylyi, K. pneumoniae and S. flexneri. The results showed that the designed molecules exhibited docking affinity scores ranging from − 8.95 to − 10.04 kcal/mol for K. pneumoniae, − 5.25 to − 7.91 kcal/mol for S. flexneri, and − 9.08 to − 9.80 kcal/mol for A. baylyi (Table 4). Notably, for K. pneumoniae, the high docking scores (compared to zosurabalpin) indicate that these macrocyclic peptide derivatives of zosurabalpin have the potential to act as effective inhibitors. However, for S. flexneri, even though some molecules achieved scores around − 7 kcal/mol, this is still significantly lower compared to the other species. These results suggest that this class of molecules may not be suitable for specifically targeting LptB2FG in S. flexneri. Interestingly, the docking scores for A. baylyi were also lower than that of zosurabalpin (− 11.1 kcal/mol), with the highest scoring molecule reaching only − 9.80 kcal/mol.

To gain deeper insights and understand why zosurabalpin derivatives exhibited high affinity in K. pneumoniae and reduced affinity in A. baylyi, we further analyzed the intermolecular interactions of the top-scoring derivative, C3, which featured substitutions of both lysine and ornithine with aspartic acid (Fig. 5). For A. baylyi, zosurabalpin formed hydrogen bonds and salt bridge interactions through the NH₃ group of its lysine side chain with Glu58, hydrogen bonds with Thr321, π–π stacking with Trp271, a salt bridge with Arg325, and interacted with LPS through two hydrogen bonds and a salt bridge via the NH₃ group of the ornithine side chain (Fig. 3).

Fig. 5.

Docking poses, 2-D diagram interaction, and binding affinity of C3 in complex with LptB2FG of A. baylyi (A) and K. pneumoniae (B). The electrostatic surfaces surrounding the C3 binding pocket are illustrated to emphasize the charge distribution in the interaction sites. Negative surfaces shown in red and positive surfaces in blue

In contrast, C3 interacts less extensively with A. baylyi (Fig. 5A). Indeed, C3 interacts with LPS through only one hydrogen bond via the amino function of aspartic acid in R2 and forms a salt bridge with Arg55 (Fig. 5A). This change in interaction mode explains the predicted decrease in binding affinity for C3 compared to zosurabalpin and highlights the importance of positively charged residues in zosurabalpin for interactions, as it interacts with negatively charged residues like Glu58. The importance of this interaction was further demonstrated in a previous mutagenesis study, where the E58V mutation in LptF of A. baylyi led to a 57-fold increase in MIC, rising from 0.06 µM to 3.6 µM (Pahil et al. 2024). This underscores the critical role of the salt bridge formed between the positively charged lysine of zosurabalpin and the negatively charged Glu58 in LptF. The electrostatic attraction between these opposite charges stabilizes the interaction, while additional hydrogen bonds enhance binding affinity and specificity. In contrast, C3, which lacks such extensive interactions, binds less effectively to A. baylyi.

In K. pneumoniae, while zosurabalpin interacts with Lys30 via a π-cation interaction and forms hydrogen bonds with LPS through the carboxyl group of ornithine (Fig. 4A), C3 exhibits a distinct binding profile. Specifically, C3 establishes hydrogen bonds with Ser125 through its R2 aspartic acid residue and forms a salt bridge through the R1 aspartic acid with Lys322, a charged positive amino acid (Fig. 5B). Additionally, C3 interacts with LPS via hydrogen bonds formed by the carboxylic functions of tryptophan (Fig. 5B). The formation of these specific interactions in C3 likely explains the predicted increase in binding affinity.

Conclusion

In summary, our study reveals the molecular basis for zosurabalpin’s selective efficacy against Acinetobacter spp. and its lack of effectiveness against K. pneumoniae and S. flexneri. Comparative sequence and structural analyses of the LptB2FG protein complex highlighted significant differences in the conformation and electrostatic properties of the binding pockets among these species. Zosurabalpin binds effectively to a unique pocket in A. baylyi, characterized by specific negatively charged regions that interact strongly with zosurabalpin through its primary amino acids (lysine and ornithine). This pocket is absent in K. pneumoniae and S. flexneri, which display different pocket architectures with more neutral or positively charged surfaces, leading to lower binding affinities. The tested derivatives of zosurabalpin were predicted to show improved binding to alternative pockets in K. pneumoniae, but not in S. flexneri. These findings emphasize the importance of binding pocket characteristics in drug efficacy and suggest that optimizing zosurabalpin to target specific pocket features can enhance its antibacterial activity against a broader range of Gram-negative pathogens. To further validate these findings, future works could involve molecular dynamics simulations followed by experimental assays, such as MIC measurements and binding affinity estimations for zosurabalpin derivatives.

Electronic supplementary material

Below is the link to the electronic supplementary material.

Author contributions

M.M. Conceptualization, Methodology, Software, Formal analysis, Investigation, Visualization and Writing—Original Draft. E.B. Writing—review & editing, Writing—original draft, Visualization, Validation, Supervision, Software, Resources, Methodology, Investigation, Formal analysis, Conceptualization.A.I. Writing—review & editing, Validation, Supervision, Project administration, Resources, Methodology, Investigation, Conceptualization.R.E. Writing—review & editing, Validation, Supervision, Resources.L.B. Resources, Validation, Supervision.

Funding

This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.

Data availability

No datasets were generated or analysed during the current study.

Declarations

Competing interests

The authors declare no competing interests.

Ethics approval

This study did not involve human or animal subjects, and therefore, ethical approval was not necessary.

Informed consent

This study did not involve human participants, so informed consent was not required.