The SH protein of mumps virus is a druggable pentameric viroporin

Abstract

Viral infections are on the rise and drugs targeting viral proteins are needed. Viroporins constitute a growing group of virus-encoded transmembrane oligomeric proteins that allow passage of small molecules across the membrane. Despite sparsity in viroporin structures, recent work has revealed diversity in both the number of transmembrane helices and oligomeric states. Here, we provide evidence that the small hydrophobic protein (SH) from mumps virus is a pentameric viroporin. From extensive biophysical data, a HADDOCK model of full-length SH shows its intracellular C-terminal region to form an extended structure crucial to stabilization of the pentamer. Heterologous expression of wild-type SH and variants in Xenopus laevis oocytes reveals the viroporin as a chloride channel, with transport facilitated by conserved hydroxyl-carrying residues lining the pore. The channel function of SH is inhibited by the small-molecule BIT225, highlighting the potential for antiviral targeting through SH.

Interdisciplinary research on viral SH offers insight into its structure, channel function, and potential for antiviral targeting.

INTRODUCTION

Mumps virus (MuV) is a highly infectious virus known to cause inflammatory symptoms, commonly parotitis and orchitis (1). MuV is neurotropic and can cause central nervous system infections or long-term effects such as seizures, deafness, or infertility (1, 2). MuV belongs to the family Paramyxoviridae and is an enveloped, negative-sense RNA virus (1). It is one of three viruses targeted by the MMR (measles, mumps, and rubella) vaccine; a historically highly efficient vaccine used in vaccination programs since the 1960s (1). Despite this, recent sporadic outbreaks have been observed, even in majorly vaccinated populations (2–4), reviving a focus on MuV. To combat these outbreaks, new options for treatment after infection, as a complement to vaccination, could be considered.

The small hydrophobic protein (SH) is one of the nine proteins encoded by MuV. SH is a 57-residue single-pass type I membrane protein, with the C terminus residing in the cytosol (Fig. 1, A and B) (5). It is not fully known which membranes SH locates to, with some studies suggesting that SH is only found in infected host cells, such as the plasma membrane (5), and not on the surface of circulating MuV (1). The gene encoding SH is in one of the more variable parts of the MuV genome, and its sequence is used to distinguish between the 12 genotypes of MuV (fig. S1) (6, 7) . In recent years, the most prevalent genotype for recorded MuV cases has been genotype G (4).

Fig. 1. Oligomerization and structural characterization of SH_FL.

(A) Consensus amino acid sequence of SH from genotype G. (B) Membrane topology of SH. (C) Left: SDS-PAGE of SH_FL in 100 mM SDS. Right: SDS-PAGE of SH_FL in 1:120 POPC. (D) Native MS of SH_FL in DPC. (E) Far-UV CD spectrum of SH_FL in 1:500 DHPC. (F) ¹H-¹⁵N HSQC spectra of SH_FL in different membrane mimetics and detergents as indicated. (G) SCS analysis of SH_FL in 66% TFE:33% H₂O (v/v). (H) ¹H-¹⁵N HSQC spectra of the Trp⁵¹ indole peak from (F). Peaks colored by lipid or detergent head group charge. Anionic, red; Zwitterionic, blue.

The overall effects of SH in the viral life cycle in humans, its single natural host, are not well understood. Studies in cell cultures have shown that SH is not essential for viral replication (5) but can contribute to viral pathogenesis (8). MuV SH has been shown to interfere with tumor necrosis factor–α–mediated apoptosis (9) and be interchangeable with SH from the related paramyxoviruses human respiratory syncytial virus (hRSV), simian virus 5, and J paramyxovirus (JPV) despite low sequence similarity (10, 11). Of interest, hRSV SH has been described as a viroporin (12, 13). Viroporins are a family of virus-encoded membrane proteins that form pores in virions and host membranes, altering the permeability to certain ions, leading to successful propagation of viral infections (14, 15). Only a limited number of viroporins have been extensively characterized. Notable examples of these are M2 of influenza A virus (16, 17) and E protein of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) (18, 19). M2 and E display a diversity of structural states, both within the individual viroporin and compared to each other and other viroporin structures (15). Viroporins are known to have a diverse set of functions with actions in multiple steps of the viral life cycle, including entry to egress and immune evasion, and they are likely underused as targets for antiviral drugs (15, 20).

Given the resemblance of MuV SH to viroporins, we address here whether SH has viroporin functions. We use an interdisciplinary approach to reveal SH as an oligomeric chloride channel and provide an integrative structural model of the pentameric pore formed by full-length SH. Using mutagenesis, we identify key determinants for the channel function of SH and for stabilizing its pentameric state. In demonstrating the possible inhibition of SH by a small-molecule inhibitor, the potential use of SH as a drug target in antiviral defense is revealed. Collectively, our data establish SH as a viroporin.

RESULTS

SH forms oligomers

Many membrane proteins are known to oligomerize, including viroporins and the related viroporin hRSV SH (12, 21). We therefore assessed the oligomeric state of SH using several approaches. A consensus sequence from genotype G, the most prevalent circulating genotype (4), was selected as a relevant representative of the SH protein (Fig. 1A). First, we recombinantly produced full-length SH (SH_FL) allowing stable isotope enrichment for nuclear magnetic resonance (NMR) spectroscopy analyses. Analyzing pure SH_FL on SDS–polyacrylamide gel electrophoresis (SDS-PAGE), showed several bands distributed as a ladder with step sizes corresponding to monomeric SH_FL of an expected size of 6.8 kDa (Fig. 1C, left). This finding indicated that SH_FL forms oligomers, with the largest size corresponding to that of a decamer. To investigate which oligomeric state(s) of SH_FL were dominant and even stabilized in lipids, SH_FL reconstituted in 1:120 1-palmitoyl-2-oleoyl-glycero-3-phosphocholine (POPC) was investigated by SDS-PAGE, increasing the sample buffer SDS from 2 to 4% (w/v) and refraining from sample heating (Fig. 1C, right). Here, we observed that SH_FL primarily migrated to the apparent size of ~34 kDa, corresponding to a pentamer. From the pattern of consecutive peaks, native mass spectrometry (MS) of SH_FL in detergent also indicated a pentameric state (Fig. 1D). Thus, the dominant oligomeric state of SH_FL in multiple environments is that of a pentamer.

SH_FL is helical in reconstituted environments

To determine the structure of SH_FL, a far–ultraviolet (UV) circular dichroism (CD) spectrum of SH_FL in 1,2-dicaproyl-sn-glycero-3-phosphocholine (DHPC) at a 1:500 protein:lipid ratio were recorded at 25°C. The spectrum showed a characteristic signature of α-helical structure of SH_FL (Fig. 1E). The fractional helicity (f_H) based on the mean residue ellipticity at 222 nm (MRE₂₂₂) was 43% corresponding to 25 residues. In contrast, AlphaFold3 predictions of SH predict the protein to be mostly helical, ≥48 residues/84%, in both monomeric and pentameric models (fig. S2), although with predicted low local distance difference test (pLDDT) scores and an interface predicted template modeling (ipTM) of 0.26 for the pentameric model. SH_FL was tested in a variety of membrane mimetics and displayed helical properties in all conditions with f_H ~ 40% (fig. S3A). To obtain a higher level of structural information for SH, a detergent and lipid screen with commonly used detergents and lipids was performed by NMR using ¹⁵N-labeled SH_FL. For all detergents and lipids, the corresponding ¹H-¹⁵N heteronuclear single-quantum coherence (HSQC) spectra of SH_FL showed only few peaks compared to the expected number for SH_FL (Fig. 1F), suggesting a higher-order structure. The two spectra of SH_FL reconstituted in lipids, 1-myristoyl-2-hydroxy-sn-glycero-3-phospho-(1′-rac-glycerol) (LMPG) and POPC, had slightly more peaks compared to those in detergents; however, none of these conditions gave spectra that would allow assignment and further structural characterization. Close to the expected number of peaks were observed for SH_FL in 66% trifluoroethanol (TFE):33% H₂O (v/v), and assignments of the backbone and sidechain nuclei of SH_FL were possible using a ¹³C,¹⁵N-labeled sample (fig. S4). Secondary chemical shift (SCS) analysis, comparing the obtained chemical shifts to those of a predicted random coil of SH_FL, revealed that SH_FL was helical throughout most of the sequence, including residues 8 to 54, leaving only the N- and C-terminal 1 to 7 and 55 to 57 residues disordered (Fig. 1G). As TFE is known to induce helicity (22), the effect of varying concentrations of TFE was explored. A correlation between increasing TFE concentration and fractional helicity in SH_FL was observed (fig. S3B), suggesting that the helicity of SH in high TFE was artificially increased compared to the native protein structure. One of the few peaks visible in every spectrum was from a Trp side chain indole, which showed chemical shift changes depending on the solubilizing conditions (Fig. 1H). As SH_FL contains two Trp residues, Trp²⁷ and Trp⁵¹, with Trp²⁷ in the transmembrane (TM) region, the observable peak most likely corresponds to Trp⁵¹, indicating that it is close enough to the membrane-water interface to be affected by the type of membrane mimetics.

The helicity of SH is confined to the TM

To address the effect of the C terminus, an SH variant with a truncated C terminus (SH_1–34) was purified, investigated by CD and NMR, and compared to SH_FL. Here, the ¹H-¹⁵N HSQC spectrum of SH_1–34 in 1:500 DHPC displayed the expected number of peaks. Native MS of SH_1–34 revealed a change in oligomeric state to a tetrameric state, further supported by SDS-PAGE (fig. S5, A and C). Thus, the C terminus of SH is important for stabilizing the pentamer. The spectral quality made backbone assignment possible (Fig. 2A), allowing for SCS analysis and characterization of the dynamics of SH_1–34 (Fig. 2B). SCS analysis showed α-helical content from residue 8 to 31, without distinct secondary structure in either terminus. The recorded NMR relaxation experiments revealed that SH_1–34 displays low R₁, high R₂, and HetNOE values in the region where the protein is helical (Fig. 2B), corresponding to the region of the TM domain (TMD). By CD, the profile of SH_1–34 was α-helical (Fig. 2C) with an f_H of 75% compared to 43% for SH_FL. Converted to number of residues, both variants have 25 residues in the helical conformation. Thus, in conclusion, only the TM part of SH is helical, whereas the C-terminal tail has nonhelical structure that exerts a stabilizing effect on the pentamer and likely positioned in proximity to the membrane.

Fig. 2. SH is helical only in its TMD.

(A) ¹H-¹⁵N HSQC spectrum of SH_1–34 in 1:500 DHPC, with assignment. (B) SCS analysis of SH_1–34 in (A). R₁, R₂, and HetNOE values for SH_1–34 in 1:500 DHPC. (C) Far-UV CD spectra of SH_1–34 in 1:500 DHPC. (D) ¹H-¹⁵N HSQC spectrum of SH_2–12, with assignment. (E) SCS analysis of SH_2–12 in (D). (F) Far-UV CD spectra of SH_2–12. (G) ¹H-¹⁵N HSQC spectrum of SH_39–57, with assignment. (H) SCS analysis of SH_39–57 in (G). (I) Far-UV CD spectra of SH_35–57.

The extramembrane regions of SH show disordered and extended structural properties

Because the comparative analyses of the helical content suggested that SH_FL contains nonhelical regions, we sought to address how the N- and C-terminal regions would behave in isolation. An N-terminal peptide of the first 12 residues, SH_2–12, was characterized by NMR and CD spectroscopy (Fig. 2, D to F). From NMR SCS analysis (Fig. 2E) a lack of secondary structure within the SH_2–12 N terminus was evident, supported by a distinct minimum at 198 nm in the CD spectrum (Fig. 2F). Thus, the N terminus of SH is likely disordered up to the beginning of the TM region around residue 8. A similar analysis was performed for C-terminal peptides, SH_35–57 or SH_39–57 (Fig. 2, G to I). No indication of a defined secondary structure was observed for SH_39–57 (Fig. 2H). As opposed to SH_2–12, the CD spectrum of SH_39–57 had a minimum at 218 nm, indicating an extended structure (Fig. 2I). Neither POPC nor a combination of POPC/palmitoyl-oleoyl phosphatidylserine (POPS) altered the CD profile of SH_39–57, suggesting that the presence of a membrane surface does not induce secondary structure (fig. S3C). Collectively, these data show that the helicity of SH is restricted to the TM region, whereas the extra membranous regions are more disordered (N-terminal) or form extended conformations (C-terminal).

SH forms a pore with polar residues exposed toward the lumen

Using the multibody docking capabilities of HADDOCK (23–25), we next integrated the information obtained from the above experiments to generate a structural model of SH. First, a model of SH residues 7 to 34 (SH_7–34), corresponding to the TM region (Fig. 2B) was predicted with AlphaFold2 v.2.3.0 through ColabFold v1.5.2 (26, 27) showing pLDDT scores consistently above 70 (fig. S6A). This model was used as a starting point for the reconstruction of the multimer form of SH. Because native MS and SDS-PAGE indicated that SH primarily adopted a pentameric state (Fig. 1, C and D), five copies of SH_7–34 were included in the model, imposing C5 symmetry between chains, and with center-of-mass restraints and additional torsion angles restraints derived from the NMR data for SH_1–34. The top-ranking resulting structure clusters all generated a pore exposing most of the polar residues of SH facing inward toward the lumen of the pore, including Tyr¹⁹ and Thr²³ (Fig. 3, A to C, and fig. S7). SH_FL and SH_1–34 were also modeled as pentamers using AlphaFold3 (figs. S2B and S6B); however, all models had pLDDT scores below 70 throughout the sequence, ipTM scores below 0.5, and did not consistently position the polar residues toward the pore (fig. S6B). This disparity likely highlights the sparsity of viral proteins in the PDB. We thus continued with the data-supported model of SH_7–34. The cross section of the HADDOCK generated pore was analyzed using the HOLE program (Fig. 3, B and C) (28), showing the pore to be narrower toward the N-terminal end, with Phe¹² and Tyr¹⁹ creating the smallest cross sections of the pore. The interfaces between adjacent helices are dominated by extensive hydrophobic interactions, such as the interactions of methyl-rich side chains in the region of Ile¹⁶ of helix i and Leu¹⁵ and Leu¹⁸ on helix i + 1, as well as Thr²³ and Leu²⁴ of helix i and Val²⁶ on helix i + 1 (Fig. 3D). Aromatic side chains also partake in the stabilization of the helix-helix interface, e.g., through a CH-π interaction between Phe¹² from helix i and Thr11 i + 1. The interface-oriented Trp²⁷ of helix i form a cross-helix hydrogen bond with the hydroxyl of Thr³³ on helix i + 1, in addition to forming hydrophobic interactions with Ile²⁹ (Fig. 3D).

Fig. 3. The SH pentamer.

(A) Five helix bundle of the transmembrane region SH_7-34. (B) Simplified two helix view of pore-lining residues and HOLE-calculated pore diameter. (C) Pore radius of SH_7–34 along with pore-lining residues. (D) Examples of helix stabilizing residues in the helix-helix interface. (E) Clusters from MD simulation of SH_35–57. (F) Two pentameric models of SH_FL with extended C-termini. Insert: Zoom of Trp⁵¹ proximity to “membrane.”

The extramembrane regions of SH are extended

The N-terminal region of the SH_FL AlphaFold3 model (residues 1 to 6) displayed disorder (Fig. S2A). Disorder in the N terminus is in full accordance with the experiments (Fig. 2, B and D to F). Therefore, the six-residue N-terminal peptide was joined to residue seven of the SH_7–34 segments using PyMoL (29) and torsion angles optimized using Coot (30) (Fig. 3F). In contrast, the C-terminal region of the SH_FL AlphaFold3 models (fig. S2), SH_35–57, was not in agreement with experimental CD and NMR data (Fig. 1E and Fig. 2, G to I). The AlphaFold3 model showed continued helical structure with pLDDT scores above 70 (fig. S2A). The experimental data demonstrated a lack of helical secondary structure, and a 100-ns molecular dynamics (MD) simulation was run on the isolated SH_35–57 peptide. Based on root mean square deviations (RMSDs), conformations of the C-terminal peptide were clustered (Fig. 3E). Clusters were excluded if there was steric hindrance to residue 35, which needed to be available to form a peptide bond with residue 34 of the TM region. Five clusters of mostly extended structures, matching the observation from CD and NMR, were used for a second round of multibody docking in HADDOCK, docking five copies of SH_35–57 to the generated TM pentamers, imposing distance restraints between residue 34 on SH_7–34 and residue 35 on SH_35–57 to restore the peptide connectivity. A shape consisting of fake beads representing the phosphates in the head groups of a POPC lipid bilayer was added to the molecular docking as a guideline for the steric hindrance imposed by the bilayer. The resulting clusters showed diversity in the orientation of SH_35–57, with the C termini of some clusters colliding with the membrane, excluding these from further consideration (fig. S8). Because the NMR spectra of SH_FL suggested the indole group of Trp⁵¹ to be in proximity of the membrane surface, an ambiguous restraint was placed on Trp⁵¹ of SH_35–57 to be within 2 Å of any shape bead on the C-terminal side of the bilayer, further restricting the movement of SH_35–57. Two different clusters fulfilled the set criteria, in which most of the C-terminal regions are extended near the membrane surface (Fig. 3F), and where Trp⁵¹ is near the membrane (Fig. 3F inset). Thus, in conclusion, our structural model of SH_FL indicates a transmembrane helix with a disordered N terminus and an extended C terminus in proximity to the membrane surface, forming pentamers, stabilized by the presence of the C terminus. This contrasts the model predicted by AlphaFold.

SH is a viroporin that conducts Cl⁻ currents

From the properties of the pentameric structure, it was relevant to explore whether SH is a viroporin with ion channel properties. SH_FL with a C-terminal His6-tag was heterologously expressed in Xenopus laevis (l.) oocytes, a system with low background expression of other ion channels and characterized with conventional two-electrode voltage clamp (Fig. 4A). Uninjected oocytes served as controls to detect background current. SH_FL-expressing oocytes showed significantly higher ion conductance than uninjected oocytes (Fig. 4, B and C). The amplitude of the SH_FL-mediated current depended on the amount of RNA microinjected into the oocytes and thus the inferred expression level of the viroporin (Fig. 4, D and E). Thus, heterologously expressed SH_FL mediate ion conductance in a dose-dependent fashion.

Fig. 4. SH is a viroporin and conducts Cl⁻ currents in Xenopus l. oocytes.

(A) Illustration of the two-electrode voltage clamp setup employed. One microelectrode is used for current injection, and one is used for voltage sensing, measuring the membrane potential (V_m) compared to a command voltage, which is leveled by an amplifier. (B) Summarized and averaged current-voltage (I/V) relations in SH_FL-expressing oocytes compared to control (uninjected) oocytes. n = 11. (C) Current activity at −85 mV in SH_FL-expressing oocytes compared to uninjected oocytes from (B). Inset: Representative current traces. (D) Summarized I/V curves of uninjected or with 4 or 50 ng of SH_FL RNA/oocyte microinjected. n = 9. (E) Current activity at −85 mV in SH_FL-expressing oocytes compared to uninjected oocytes from (D). (F) Summarized I/V curves of SH_FL-induced currents in the presence or absence of Cl⁻. n = 10. (G) Normalized data at V_m = −85 mV from (F). (H) Summarized I/V curves of SH_FL-mediated currents in the presence or absence of Ca²⁺. n = 9. (I) Normalized data at V_m = −85 mV from (H). (J) Summarized I/V curves of SH_FL-induced currents in the presence or absence of Na⁺. n = 7. (K) Normalized data at V_m = −85 mV from (J). All summarized current traces represent means ± SEM. Each point in the bar charts indicates an individual oocyte. The magnitude of SH_FL-mediated currents (at V_m = −85 mV) was compared to those of uninjected oocytes using an unpaired t test (C), one-way analysis of variance (ANOVA) (E), or between the presence or absence of indicated ion with paired t test [(G), (I), and (K)]. *P < 0.033, **P < 0.002, and ***P < 0.001.

To characterize the current mediated by SH_FL, selected ions in the control recording solution were replaced. Replacement of Cl⁻ with equi-osmolar gluconate reduced the SH_FL-mediated current (Fig. 4F; compare −632 ± 103 nA in control solution to −165 ± 26 nA with gluconate substitution; n = 10, P < 0.001; Fig. 4G). The removal of Ca²⁺ from the control solution significantly increased the SH_FL-mediated current (Fig. 4H; compare −583 ± 116 nA in control solution to −1689 ± 186 nA without Ca²⁺; n = 9, P < 0.01; Fig. 4I), suggesting possible modulation of the pore. Replacement of Na⁺ with equi-osmolar choline left the SH_FL-mediated current unchanged, suggesting that the current was independent of the presence of the monovalent cation (Fig. 4J; −compare 847 ± 196 nA in control solution to −800 ± 183 nA with choline substitution; n = 7, P = 0.48; Fig. 4K). Combined, these data illustrate that the major contributor to SH_FL-mediated current in Xenopus l. oocytes is Cl⁻, and the viroporin is negatively modulated by Ca²⁺.

Oligomerization is essential for the ion channel function of SH

To resolve the identity of the SH conducting parts and validate the HADDOCK model, we constructed a set of gradually truncated SH constructs and an L24A, W27A (SH_LW), and a Y19A (SH_Y) variant, expected to affect the oligomer (Fig. 5). These residues are mostly conserved across the genotypes or, when not, as in the case of Y19, the observed substitutions maintain the polar properties of the observed residue (fig. S1). Four of the five variants—SH_Y, SH_LW, SH_13–57, and SH_1–34—were less conductive compared to SH_FL (Fig. 5, B and C). Because SH_10–57 was conducting, but SH_13–57 was not, the disordered N-terminal before the beginning of the TM helix plays no role in ion conductance, whereas residues 10 to 13 form part of the interhelical hydrophobic network stabilizing the oligomer (Fig. 3D). SDS-PAGE analysis and changes in the NMR spectra compared to SH_FL showed altered dynamics in the two variants SH_LW and SH_1–34, the spectrum of SH_LW being more heterogenous and the spectrum of SH_1–34 having more well-defined peaks compared to the spectrum of SH_FL. The observed changes indicate that these modifications affected the stability of oligomeric state, explaining the lack of ion conductance (fig. S5). The point mutation in SH_Y reduced the conductance compared to SH_FL, suggesting that the tyrosine residue could be involved in ion conductance. Because Y19A would make the pore wider in an otherwise unaltered pore (Fig. 5E), it is also possible that the mutation affects the overall organization of the pore and even leads to pore collapse. Together, these data strongly support the HADDOCK model of SH, highlighting the notion that either destabilizing SH within the TMD interfaces or removing the C-terminal tail is detrimental to its pore-conducting abilities.

Fig. 5. SH channel function is dependent on oligomerization and polar residues and inhibited by BIT225.

(A) Left: Schematic representation of SH variants. Right: Sequences of SH WT and variants. All electrophysiology experiments included a C-terminal His6-tag. SH_LW: L24A, W27A. SH_Y: Y19A. (B) Summarized I/V curves of SH variants. n = 9,22. (C) Data at V_m = −85 mV from (B). (D) Surface side view of SH_FL and SH_LW. (E) Surface N-terminal view of SH_FL and SH_Y. (F) Summarized I/V curves of SH_FL-expressing oocytes with 45-min preincubation with 10 μM BIT225 or matched negative controls. n = 9. (G) Data at V_m = −85 mV from (F). (H) Concentration dependent inhibition of SH_FL by BIT225. All summarized current traces represent means ± SEM. Each point in the bar charts indicates an individual oocyte. Statistical significance was determined with unpaired t test. *P < 0.033, **P < 0.002, and ***P < 0.001.

The amiloride analog BIT225 is a known inhibitor of HIV-1 Vpu viroporin (31) and SARS-CoV-2 E (32) but not TMEM16, an endogenously expressed Cl⁻ ion channel in Xenopus l. (32). To resolve whether BIT225 inhibits SH viroporin Cl⁻ conductance, we preincubated SH_FL-expressing oocytes for 45 min with 10 μM BIT225 before recordings. Preincubation and local perfusion of the oocytes during recordings with BIT225 resulted in a reduction in current activity (Fig. 5, F to G). By varying the BIT225 concentration, it could be seen that the inhibitory effect increased with increasing concentration (Fig. 5H). NMR spectra of SH_FL in the presence of an excess of BIT225 showed chemical shift perturbations of selected peaks indicating an interaction between BIT225 and the protein (fig. S9); however, it was not possible to assign the affected residues. Together, these experiments showcase that SH can be targeted by small-molecule inhibitors.

DISCUSSION

SH has previously been identified as a type I membrane protein with implications in the immune evasion of MuV (5, 9, 10). However, like many other viral membrane proteins, structural and functional insight was lacking. Here, we show that SH forms a helical pentamer with chloride conductance that is negatively modulated by calcium and inhibited by the small-molecule BIT225. Through a sectional and interdisciplinary approach, a structural model of the full-length SH is provided with a dominating pentameric state observed in multiple environments (Fig. 1, C and D). The resulting model shows a mostly hydrophobic TM pore (Fig. 3, C and D), with hydrophobic side chains stabilizing the helix-helix interfaces and where a few polar residues positioned in the pore lumen, Tyr¹⁹ and Thr²³, facilitate ion conduction (Fig. 5B). These data establish SH as a viroporin.

The model of SH_FL displays nonhelical extra-membranous regions. The six N-terminal residues show intrinsically disordered properties (33), whereas the C terminus have extended characteristics and are tethered to the membrane surface through its many hydrophobic residues, including Trp⁵¹ (Fig. 3F). Most of the current structures of viroporins have mainly focused on the transmembrane regions of the proteins, and the few structures with extra-membranous regions generally show distinct secondary structural elements, e.g., a second α-helix on M2 from influenza A virus perpendicular to the membrane (34). In the case of SH, no secondary structure was observed for the C terminus and helicity could only be induced in very high concentrations of TFE (Fig. 1G and fig. S3B). In some viroporins, short β sheets have been suggested in the juxtamembrane regions of the protein, such as for hRSV SH (12) and SARS-CoV-2 E (19), more alike the extended structure observed for SH. Although a lack of distinct secondary structure was observed in the C terminus, it is expected to remain close to the membrane, as it has a high content of hydrophobic residues and displays distinct head-group effects dependent on the lipid and detergents used (Fig. 1H). As many membrane proteins are inherently sensitive to their lipid environment (35), specific lipids might affect the conformation and activity of SH, such as affecting the organization of the C terminus.

We identified several important features of SH. The intracellular C terminus was directly linked to the pore function, likely through its stabilizing effect promoting the ability of SH to form oligomers. However, the exact structure and dynamics of the C terminus is not clear. It is possible that this stabilizing effect may act through interactions between the C-terminal residues of one helix with side chains in the TM region of helices positioned (i,i ± 1) or (i,i ± 2). However, our computational modeling did not capture any of these potential contacts. As we saw no signs of intermolecular interactions between chains of the C-terminal regions, there is no evidence that these would engage in an interchain structure rich in β sheets, as seen for other viroporins (12, 19). A possible explanation for the stabilization of the pentamer is through entropy optimization (36). Given that the C-terminal region is extended and in contact with the bilayer and given that the C termini do not interact with each other, optimizing the distances between them on the bilayer surface may provide an entropic force driving the pentamer together and thereby stabilizing it. The removal of the C-terminal region gave access to the structure of the TM-helix through disruption of the pentamer, into a state with favorable NMR dynamics. In contrast, the removal of the N terminus before the beginning of the TM-helix had no effect on conductance or oligomericity. Other, more specific residues were also shown to be important for the stability of the pore. The L24A and W27A mutations abolished the pore function, probably due to disruption of the network of hydrophobic residues lining the helix-helix interface, causing a similar destabilization of the pore. The Y19A mutation also showed reduced conductivity, highlighting its potential role in ion binding or in pore formation. It is interesting to note that Y19 and T23 are mostly or fully conserved, respectively, across genotypes (fig. S1), and the genotypes where Y19 is not conserved have polar residues (His, Ser, or Cys) at position 19. Thus, it is possible that the Cl⁻ conductance is preserved across genotypes. As pore dynamics likely play a role in ion conductance as shown for SARS-CoV-2 E (37), it is also possible that mutations that conversely would stabilize the structure could influence ion conductance.

The oligomerization and consequent pore function establish SH to belong to the viral protein family of viroporins. Most identified viroporins have shown a preference for cation conduction, e.g., protein 3a of SARS-CoV2 and p7 of hepatitis C virus have preferences for divalent (Ca²⁺) (38) or monovalent cations (Na⁺ and K⁺) (39), respectively. In contrast, SH displayed a preference for conducting Cl⁻ ions (Fig. 4F). This has also been observed for M2 from influenza C and D, which both contain a conserved luminal Tyr residue, where the proteins were proposed to enhance viral budding by promoting an efflux of Cl⁻ ions; however, the mechanism is unknown (40). Cl⁻ conduction was also suggested for hRSV SH at pH 4 because of His protonation (21). The described effect of SH on immune evasion (9) has not yet been connected to its channel function. It is therefore possible that SH is multifunctional. The conductance of SH was negatively affected by the presence of Ca²⁺, highlighting a potential allosteric effect. It was not elucidated if and where on SH the Ca²⁺ ions bind. A recent study on Ca²⁺ binding to disordered proteins found that binding depends on a significant amount of negatively charged residues (41). This is not a feature of either the C- or N-terminal regions of SH, although a pentameric state may jointly support Ca²⁺ binding through colocalization of up to five Asp⁵⁴. As many of the lipid head groups also bind Ca²⁺ (42), one mechanism could be to destabilize the headgroup interactions of the C-terminal regions enough to weaken the pore stability and, thus, ion conductance. The effect could also come from a combination of these effects or indirectly by affecting interactions with other proteins or it may act through binding to phosphoryl groups if phosphorylation of the conserved serine residues in the C-terminal tail occurs. More data are needed to elucidate how this effect is established; however, in silico predictions of phosphorylation (43, 44) suggest that only the most C-terminal serines, Ser⁵² and Ser⁵⁶, may be targets for kinases.

The conductivity of SH was shown to be inhibited by BIT225 (Fig. 5F), highlighting the possible druggability of SH and introducing SH as a target for antiviral drug design. Similar druggability has been explored for viroporins from coronaviruses (45, 46). Given the importance of structure-based drug design (47), this current SH model offers insights into the organization of the oligomer and pinpoints residues important for the pore function, which could be used for specific targeting. With the recombinant production and reconstitution of the full-length SH, there is a relevant potential for initiating future high-throughput screening campaigns of larger small-molecule libraries using, e.g., saturation-transfer-difference NMR (48), which can detect selective binding of compounds at low SH concentrations.

This combined work on the structure and function of SH presents a full-length structural model of a viroporin with chloride conductance function and critical dependence on the C-terminal region for its formation, not seen for other viroporins. The data highlight SH as a potential drug target, given its implications on immune evasion and verified druggability. Besides the evidence that SH can be blocked by a generic inhibitor, we also pinpoint key residues and features in the pentameric model, including the C terminus, that could be sites of interactions to be targeted by antiviral drugs.

MATERIALS AND METHODS

Protein purification

Expression and purification of SH (WT and variants) for structural characterization was performed as described in (49). In brief, a pGEX-4 T-1 vector coding for SH downstream of a glutathione S-transferase (GST) carrier protein and a thrombin cleavage site was acquired from GenScript (US). The plasmid was transformed into competent Escherichia coli BL21(DE3) cells, and protein was expressed unlabeled in LB media. Labeled ¹⁵N or ¹³C,¹⁵N protein was expressed in M9 media [KH₂PO₄ (3 g/liter), Na₂HPO₄·H₂O (7.5 g/liter), NaCl (5 g/liter), 1 mM MgSO₄, 1 ml of M2 trace solution, glucose (3 g/liter), (NH₄)₂SO₄ (1.5 g/liter) [if ¹³C-labeled: ¹³C-D-glucose (ISOTEC), if ¹⁵N-labeled: (¹⁵NH₄)₂SO₄, (ISOTEC)] added ampicillin (100 μg ml⁻¹; Fisher BioReagents). Cells were collected by centrifugation (15 min, 5000g, 4°C) and resuspended in cell culture lysis buffer [40 ml/liter; 1x phosphate-buffered saline buffer (pH 7.4), 25% (w/v) sucrose, 5 mM ethylenediaminetetraacetic acid (EDTA), 1 mM phenylmethylsulfonyl fluoride (PMSF)]. Inclusion bodies were harvested by sonicating the cells on ice (5× 30 s with 30-s rest between rounds at 80% amplitude) and then centrifugating (25 min, 20,000g, 4°C), and the supernatant was discarded. Resuspension, sonication, centrifugation, and discard of the supernatant was repeated twice, followed by resuspension of the pellet in 50 mM tris-HCl (pH 7.4) and centrifugation (20 min, 20,000g, 4°C).

The pellet was resuspended in 12 ml of 1.5% (w/v) sarkosyl, 10 mM dithiothreitol (DTT) and 20 mM tris-HCl (pH 7.4) and incubated 3 hours at room temperature with mild agitation, followed by centrifugation (20 min, 20,000g, 4°C). The supernatant containing GST-SH was dialyzed against 0.5% (w/v) sarkosyl, 10 mM NaCl, and 50 mM tris-HCl (pH 7.4) at 4°C to remove DTT and cleaved with thrombin (SERVA) to release GST, leaving two residues (GS) at the N terminus. Following cleavage, the solution was lyophilized and resuspended in 2 ml Milli-Q water. The solution was split into 100 μl of aliquots to which each was added to 750 μl of 1:2 chloroform:methanol solution and mixed well before addition of 300 μl of Milli-Q water, and the solution was mixed further and centrifuged (14,000g, 2 min, 4°C). Centrifugation resulted in three layers, the top aqueous layer was carefully removed, and 500 μl of methanol was added to the solution, followed by mixing. The solution was incubated on ice for 20 min, centrifuged (16,000g, 40 min, 4°C), and the supernatant containing the target protein was transferred to a glass vial where the organic solvent was evaporated with a stream of N₂. The peptide SH_39–57, Ac-RHAALYQRSFFHWSFDHSL, was acquired from TAG Copenhagen (DK) at 97% purity.

Peptide synthesis

The peptide SH_2–12, H-Pro-Ala-Ile-Gln-Pro-Pro-Leu-Tyr-Leu-Thr-Phe-NH₂, was synthesized using the Fmoc/t-Bu protection strategy on a Rink amide linker functionalized solid support, yielding the C-terminal amidated peptide after cleavage from the resin. Amino acids with acid-labile sidechain protecting groups were used, allowing the simultaneous removal of protecting groups and release of peptide from the resin. The synthesis was conducted on an Initiator+ AlstraTM Automated Microwave Peptide Synthesizer (Biotage). Each peptide coupling was performed twice at 75°C for 5 min using diisopropylcarbodiimide/Oxyma as the coupling reagent (5.00 eq. per coupling), while Fmoc deprotection was executed with 20% piperidine in dimethylmethanamide (DMF) for 10 min at room temperature. Swelling and washing of the resin were carried out using DMF at 70°C for 20 min. Upon completion of the automated peptide synthesis, the resin was washed successively with DMF, MeOH, and CH₂Cl₂ before left with suction for 1 hour. Side-chain deprotection and cleavage of the peptide from the resin were achieved by adding trifluoroacetic acid (TFA)/H₂O/ triisopropyl silane in a 95:2.5:2.5 ratio to the resin, followed by overnight shaking. The resulting suspension was filtered, and the filtrate was added cooled diethyl ether and TFA (1:9 TFA:diethyl ether). The precipitate was isolated by centrifugation and washed with diethyl ether (×3) and then air-dried overnight. The crude peptide was divided into portions of maximum 25 mg, dissolved in a minimum volume of DMF, and purified by preparative high-performance liquid chromatography (HPLC). The pure fractions were collected, concentrated by air, and lyophilized, yielding the final pure peptide product. ultra-performance LC-MS mass/charge ratio: 1274.2 [M + H⁺], calculated C₆₃H₉₇N₁₄O₁₄⁺: 1273.7. Automated solid-phase peptide synthesis was carried out using ChemMatrix resin beads as the solid support, with a loading of 0.3 mmol/g. The synthesis was performed in a flat-bottomed phycoerythrin syringe fitted with polypropylene element filter, Teflon tubing, and valve. Standard proteinogenic amino acids and the Rink amide linker were used as received from commercial sources. Preparative HPLC purification of the peptide was performed on a Waters auto-purification system consisting of a 2767 sample manager, 2545 gradient pump, and 2998 photodiode array detector. Column: XBridge Peptide BEH C18 OBD Prep Column, 130 Å, 5 μM, 19 mm × 100 mm. Solvent A1: 0.1% formic acid in water; solvent B1: 0.1% formic acid in acetonitrile. Flow rate: 20 ml/min.

Lipids and detergents

DHPC, DPC, LMPG, 1-palmitoyl-2-hydroxy-sn-glycero-3-phospho-(1′-rac-glycerol), POPC, and POPS were purchased from Avanti Polar Lipids. SDS was purchased from Avantor Performance Materials. N-lauroylsarcosine sodium salt (sarkosyl), chloroform with 0.5 to 1.0% ethanol as stabilizer, methanol (≥99.9%), and 2,2,2-trifluoroethanol (≥99.0%) was purchased from Sigma-Aldrich.

Reconstitution of protein in detergents and lipids

Lipids were first lyophilized and then re-solubilized in buffer by repeated cycles of sonication baths, vortexing, and resting on ice. Lipid vesicles were made homogenous in size using a 100 nm polycarbonate filter-membrane (Whatman) and in an Avanti Mini-Extruder (Avanti Polar Lipids) and checked by dynamic light scattering. Lyophilized protein in amounts required for the final concentration was resolubilized directly in detergent and homogeneous lipids bicelles of choice at the indicated protein:detergent or protein:lipid ratio in 20 mM Na₂HPO₄/NaH₂PO₄ (pH 7.2). After reconstitution, the sample was washed on a 3K spin filter in 20 mM Na₂HPO₄/NaH₂PO₄ to remove excess lipids. For the mixed lipid POPC/POPS composition, the lipids were dissolved in chloroform and mixed in a 3:1 ratio of POPC to POPS before evaporating the organic solvent with a stream of N₂ and resolubilizing the lipids in 20 mM Na₂HPO₄/NaH₂PO₄ (pH 7.2) as described above.

Mass spectrometry

SH_FL in DPC was buffer exchanged to 100 mM ammonium acetate and SH_1–34 in DPC was buffer exchanged to 200 mM ammonium acetate with 2× CMC (0.046%) lauryldimethylamine-N-oxide using Zeba biospin columns (7 kDa). The samples were further diluted to ~30 μM before being transferred to a nESI capillary (Thermo Fisher Scientific). Using an offline nanospray source, the samples were analyzed on a Waters Synapt G1 TWIMS MS modified for analysis of intact protein complexes (MS Vision, The Netherlands) in positive ionization mode. The capillary voltage was 1.5 kV, the trap voltage was 10 V, and the cone voltage was 10 V. The source temperature was 30°C. The source pressure was set to 8 mbar.

CD spectroscopy

CD spectra in the far (190 to 260 nm) UV ranges were recorded at 25°C, scan speed of 10 nm/min, 1-nm bandwidth, and 2-s response using a Jasco J-810 or J-815 spectropolarimeter equipped with a sample holder Peltier temperature control and a path length of 0.1 cm using a quartz SUPRASIL cell (Hellma). For each condition, 10 scans were accumulated, averaged, background spectrum–recorded at identical conditions, and subtracted. The spectra were smoothed with a convolution width of five data points. Concentrations of proteins and peptides varied between 5 and 30 μM in 20 mM Na₂HPO₄/NaH₂PO₄ (pH 7.2, or pH 6.5 for SH_34–57), with detergents and lipids as indicated in the figure or figure legend. Mean residue ellipticity ([Ɵ]_MRE) was calculated using the following equation, where l is the path length in cm, c is the molar concentration, and n is the number of peptide bonds (50)

$${\left[\mathrm{Ɵ}\right]}_{\text{MRE}}=\frac{\text{mdeg}}{10·l·c·n}$$

Fraction helicity was calculated based on the MRE at 222 nm, using the following equation

$$f_{\mathrm{H}}=\frac{{\left[\mathrm{\theta }\right]}_{\text{MRE},222}-3000}{-\mathrm{36,000}-3000}·100\%$$

NMR spectroscopy

NMR samples for the detergent screen typically contained 200 μM SH_FL in 20 mM Na₂HPO₄/NaH₂PO₄ (pH 7.2) with 10% D₂O (v/v) in the presence of a molar excess of detergents or lipids at an indicated ratio. Spectra for the assignment of backbone and side-chain nuclei were recorded on ¹³C,¹⁵N-labeled samples of 0.6 mM SH_FL in 66% TFE:23% H₂O:10% D₂O (v/v/v) and 0.8 mM SH_1–34 in 1:500 DHPC in 20 mM Na₂HPO₄/NaH₂PO₄ (pH 7.2) and 10% D₂O (v/v) or on unlabeled samples of 0.6 mM SH_2–12 in 20 mM Na₂HPO₄/NaH₂PO₄ (pH 7.2), 10% D₂O (v/v) and 1 mM SH_39–57 in Milli-Q, and 10% D₂O (v/v). NMR samples of SH_FL with BIT225 dissolved in dimethyl sulfoxide (DMSO) were recorded in a 1:10 molar excess of BIT225 and compared to a spectrum of SH_FL added an equivalent volume of DMSO.

All NMR samples were recorded in 5-mm Shigemi BMS tubes (Bruker), and all NMR spectra were recorded on Bruker Avance III HD 600-, 750-, or 800-MHz spectrometers equipped with cryogenic probes and Z-field gradients at 37°C, except for SH_2–12 recorded at 10°C. Free induction decays were transformed and visualized in NMRPipe (51) or TopSpin (Bruker BioSpin) and subsequently analyzed using the CCPNmr Analysis software (52). Proton chemical shifts were internally referenced to dextran sulfate sodium at 0.00 parts per million, with heteronuclei referenced by relative gyromagnetic ratios. A ¹H-¹⁵N HSQC spectrum was recorded for all samples. Further, for assignment of backbone and side-chain nuclei of ¹³C,¹⁵N-labeled samples manual backbone assignment was performed on the basis of the analysis of HNCO, HN(CA)CO, HNCA, HN(CO)CA, HNCACB, and HN(CO)CACB spectra recorded with nonuniform sampling and processed using qMDD (53). For backbone assignment of unlabeled samples, natural abundance ¹H-¹H total correlated spectroscopy and ¹H-¹H rotating frame overhauser effect spectroscopy spectra were recorded, and manual backbone assignment was performed. To acquire R₁ and R₂ relaxation rates for SH_1–34, triplicates of varying relaxation delays were recorded in a randomized order, and peak intensities were fitted to single-exponential decay. Relaxation rates for T1 were 20, 60, 100, 180, 300, 500, 800, and 1200 ms, and for T2 16.96, 32.92, 67.84, 101.76, 135.68, 169.6, 203.52, and 271.36 ms. Dihedral angles for SH_1–34 were calculated on the basis of the chemical shift assignment using TALOS+ program (54).

Random coil chemical shifts for secondary structure analysis were generated using (www1.bio.ku.dk/english/research/bms/sbinlab/randomchemicalshifts2/) (55), and SCS analysis was calculated using the following equation (56).

$$\text{SCS}=\mathrm{\Delta \delta }={\mathrm{\delta }}_{\text{observed}}–{\mathrm{\delta }}_{\text{random coil}}$$

Model reconstruction

The pentameric assembly of the SH TM region was determined using the multibody docking capabilities of the HADDOCK webserver v.2.4 (25). A model of SH_7–34 was generated using AlphaFold2. Five copies of SH_7–34, where the AlphaFold model had a pLDDT score of ≥70, center-of-mass restraints, and the dihedral angles obtained from NMR/TALOS imposed for the five monomers were used to obtain the TM pore. C5 symmetry restraints were imposed on the C-terminal peptides (chains B to F) maintaining equal distances between the chains and noncrystallographic restraints were imposed between sequential pairs of chains (BC-CD-DE-EF-FB), effectively preserving identical conformations of the five copies of the C-terminal peptide. Models (10,000) were generated at the rigid body docking stage of which the best 400 were refined and clustered using the fraction of native contacts (57) with a 0.6 cutoff. Clusters were ranked based on the default HADDOCK scoring function consisting of the intermolecular energies calculated using the OPLS force field (58) (100% van der Waals and 20% electrostatics) an empirical desolvation energy term (59) and the restraint energy. A 100-ns MD simulation of a peptide corresponding to the 23 C-terminal residues was performed using the CHARMM36m force field (60) for GROMACS (61). Prominent conformations were split into eight clusters, and three were removed due to steric hindrance for N terminus of the peptide. Five copies of SH_35–57 were added to multibody docking interface, restraining the distance of Lys35 from the peptides to Tyr³⁴ of the TM helixes. In addition, a shape consisting of fake beads was added to represent the phosphates in a POPC bilayer, and Trp⁵¹ was ambiguously restrained to be within a 2 Å proximity to any shape bead. Peptide bonds were created between Lys³⁵ in the C-terminal peptides and Tyr³⁴ in the TM using PYMOL v. 2.5.5, and torsion angles were optimized in COOT. Similarly, the first six N-terminal residues were added from the AlphaFold3 model onto the pentameric model using PYMOL, and torsion angles were optimized in COOT. A cross section of the pentameric pore was analyzed using the HOLE program (28). The SH models in figs. S2 and S6 were made using AlphaFold3 (62).

Multiple sequence alignment

Multiple sequence alignments of SH sequences from the 12 MuV genotypes were performed using two reference sequences for each genotype obtained from (6, 63). Multiple sequence alignment was performed with Clustal Omega (64) and visualized with Jalview (65).

Heterologous protein expression in Xenopus l. oocytes

Oocytes were purchased from Ecocyte (Germany) and kept in Kulori medium [90 mM NaCl, 1 mM KCl, 1 mM MgCl₂, 1 mM CaCl₂, and 5 mM Hepes adjusted to pH 7.4 with 2 M tris-base (HOCH₂)₃CNH₂)] at 18°C. For oocyte experiments, at least three preparations of oocytes from different frog donors were used. For the electrophysiological experiments, n is the number of oocytes. cDNA encoding C-terminally His6-tagged WT and truncated versions of SH (GenScript, Netherlands) in the pXOOM expression vector (66), linearized downstream from the polyadenylate segment and in vitro transcribed using T7 mMessage machine in accordance with the manufacturer’s instructions (Ambion, Austin, TX, USA). Complementary RNA was extracted with MEGAclear (Ambion) and stored at −80°C. For microinjection into defolliculated Xenopus l. oocytes, 50 ng of SH RNA/oocyte was used (4 ng of RNA was also tested in the dose-dependent experiment), and the oocytes were kept for 3 to 4 days in Kulori medium at 18°C before experiments. Approximately 50 to 70% of the oocytes died upon injection of SH RNA and before measurements, whereas <5% of uninjected oocytes died in this period.

Two-electrode voltage clamp

Conventional two-electrode voltage clamp studies were performed using a DAGAN CA-1B High-Performance oocyte clamp (DAGAN, Minneapolis, MN, USA) with a Digidata 1440A interface controlled by pCLAMP software version 10.5 (Molecular Devices, Burlingame, CA, USA). Electrodes were pulled (HEKA Elektronik, Lambrecht, Germany) from borosilicate glass capillaries to a resistance of 1.5 to 3 megohm when filled with 1 M KCl. The current traces were obtained by stepping the clamp potential from −20 mV to test potentials ranging from +50 to −130 mV (pulses of 200 ms) in increments of 15 mV. Recordings were low pass-filtered at 500 Hz, sampled at 1 kHz, and the steady-state current activity was analyzed at 160 to 180 ms after applying the test pulse. The standard solution used consisted of (100 mM NaCl, 2 mM KCl, 1 mM CaCl₂, 1 mM MgCl₂, and 10 mM Hepes). Ion substitution experiments were measured on the same oocytes with or without the indicated ion, the Na⁺-free solution was made with equi-osmolar replacement with choline chloride, Ca²⁺-free solution was made with equi-osmolar replacement with Na⁺, and the Cl⁻-free solution was made with equi-osmolar replacement with Na⁺-gluconate. All solutions were adjusted to a pH of 7.4 with 2 M tris base. To test inhibition by BIT225, oocytes expressing SH_FL were preincubated with BIT225 dissolved in DMSO (concentrations: 0.1, 3, 10, or 100 μM) with matched negative control oocytes for 45 min before measurements. During the recordings, the oocytes were continuously perfused with standard solution with BIT225 or DMSO.

Synthesis of BIT225

All reagents were of commercial grade. Silica gel 60 F254 precoated plates (Merck) were used for TLC. Flash chromatography of compounds was performed using silica gel 60 (40 to 64 μm). ¹H NMR spectra were obtained at 400 or 600 MHz, and ¹³C NMR spectra were obtained at 100 or 151 MHz and calibrated relative to residual solvent peaks. Analytical HPLC was performed on a Dionex UltiMate system using a Gemini-NX C18 column (4.6 mM by 250 mM, 3 μM, 110 Å) with mobile phase A: water:TFA (100:0.1, v/v) and mobile phase B: MeCN:water:TFA, (90:10:0.1, v/v/v). Data were acquired and processed using the Chromeleon Software v.6.80. The analysis was performed by a gradient of 0 to 100% mobile phase B in mobile phase A over 15 min.

Abbreviations: Acetyl (Ac), aqueous (aq.), broad (br), doublet (d), double doublet (dd), dichloromethane (DCM), dimethylformamide (DMF), dimethyl sulfoxide (DMSO), ethyl (Et), multiplet (m), methyl (Me), room temperature (rt), singlet (s), triplet (t), tetrahydrofuran (THF).

Reactions and conditions: (i) 1-methyl-1H-pyrazole-4-boronic acid pinacol ester, aq. NaOH (2.0 M), Na₂CO₃, MeCN, 90°C, 18 hours, 61%; (ii) (a) oxalyl chloride, DCM, DMF, rt (3.5 hours)–40°C (1 hour), (b) guanidine hydrochloride, aq. NaOH (2.0 M), THF, rt, 30 min, 65%.

5-(1-Methyl-1H-pyrazol-4-yl)-2-naphthoylguanidine (BIT225) (67) Step i: 5-(1-methyl-1H-pyrazol-4-yl)-2-naphthoic acid: A vial charged with 5-bromo-2-naphthoic acid (500 mg, 1.99 mmol), 1-methyl-1H-pyrazole-4-boronic acid pinacol ester (435.1 mg, 2.09 mmol), and Pd(PPh₃)₄ (115.1 mg, 99.6 μmol) was evacuated and backfilled with argon (3×). Then, aq. NaOH (2.0 M, 2.4 ml) and MeCN (9.5 ml) were added. The vial was sealed and stirred at 90°C for 18 hours. The reaction mixture was cooled down to rt and aq. HCl (1.0 M, 7.5 ml) was added, followed by water (5 ml) and extracted with EtOAc (3 × 15 ml). The combined organic phase was dried over anhydrous Na₂SO₄, filtered, and concentrated under reduced pressure. The crude product was recrystallized from DCM to afford a white solid (307 mg, 61%): R_f = 0.18 (MeOH:DCM); 5:95; ¹H NMR (400 MHz, DMSO-d₆): δ = 13.14 (br s, 1H), 8.63 (d, J = 1.8 Hz, 1H), 8.22 (d, J = 8.9 Hz, 1H), 8.10 (s, 1H), 8.08 to 8.03 (m, 1H), 7.99 (dd, J = 8.8, 1.8 Hz, 1H), 7.75 (d, J = 0.8 Hz, 1H), 7.65 to 7.56 (m, 2H), 3.96 (s, 3H); ¹³C NMR (151 MHz, DMSO–d₆): δ = 167.4, 138.6, 132.8, 132.7, 131.0, 130.7, 130.3, 128.5, 128.3, 128.2, 126.5, 125.6, 125.5, 119.4, 38.7. Step ii: A flame-dried vial, charged with 5-(1-methyl-1H-pyrazol-4-yl)-2-naphthoic acid (100 mg, 0.40 mmol), was evacuated and backfilled with argon (3×). Then, anhydrous DCM (17 ml) and DMF (two drops) were added followed by oxalyl chloride (100 μl). The mixture was stirred at rt. for 3.5 hours and then heated for 1 hour at 40°C. The cooled reaction mixture was concentrated under reduced pressure. The resulting crude acid chloride was suspended in anhydrous tetrahydrofuran (4 ml), and this mixture was added dropwise to a solution of guanidine hydrochloride (174.2 mg, 1.82 mmol) in aq. NaOH (2.0 M, 2.4 ml, 4.8 mmol), and the reaction mixture was then stirred for 30 min. Water (3 ml) was added and extracted with EtOAc (3 × 5 ml). The combined organic phase was washed sequentially with aq. NaOH (1 M, 10 ml) and water (10 ml), then dried over anhydrous Na₂SO₄, and filtered. The filtrate was evaporated onto Celite and purified by flash chromatography (3 to 10% EtOAc in heptane) to afford the title compound as a white solid (75 mg, 65%). R_f = 0.23 (MeOH:DCM; 1:9); ¹H NMR (600 MHz, CD₃OD): δ = 8.65 to 8.60 (m, 1H), 8.13 (d, J = 1.2 Hz, 2H), 7.94 to 7.88 (m, 2H), 7.73 (s, 1H), 7.54 to 7.49 (m, 2H), 4.02 (s, 3H); ¹³C NMR (151 MHz, CD₃OD): δ = 179.4, 165.0, 140.1, 137.2, 134.8, 134.3, 131.8, 130.8, 129.6, 129.2, 127.0, 126.9, 126.0, 122.3, 39.0; HPLC: t_R = 9.12 min, purity: 96.3%.

Statistical analysis

Statistical analyses were performed with Prism7 (GraphPad Software Inc., La Jolla, CA, USA), and the test was stated in the figure legends. An outlier was excluded from Fig. 4I, after Grubbs test. Data represent the means ± SEM, and P < 0.05 was considered statistically significant.

Supplementary Materials

This PDF file includes:

Figs. S1 to S9