Permeability changes measured in cell plasma membranes

Assuming that viroporin-induced permeability changes reproduce a pattern akin to that observed in cultured mammalian cells upon viral infection, a series of expression-coupled cell permeability tests were devised to prove the viroporin activity of the cloned candidates. A common procedure involves the use of the antibiotic hygromycin B (HB), a protein translation inhibitor that cannot efficiently penetrate the cell membrane. Reduction of ³⁵S-labeled protein expression, followed by autoradiography after SDS-PAGE separation, can be observed in cells harvested from cultures previously induced to express the cloned viroporin gene. Inhibition of protein expression by HB has been reported in animal cells expressing among others PV 2B (Aldabe et al., Reference Aldabe, Barco and Carrasco1996; Madan et al., Reference Madan, Redondo and Carrasco2010a), Paramyxovirus SH (Masante et al., Reference Masante, El Najjar, Chang, Jones, Moncman and Dutch2014), E protein from murine hepatitis virus (Madan et al., Reference Madan, Garcia Mde, Sanz and Carrasco2005), NS3 protein of bluetongue virus (Han and Harty, Reference Han and Harty2004), human polyoma JC virus agnoprotein (Suzuki et al., Reference Suzuki, Orba, Okada, Sunden, Kimura, Tanaka, Nagashima, Hall and Sawa2010), or the Bovine ephemeral fever rhabdovirus α1 (Joubert et al., Reference Joubert, Blasdell, Audsley, Trinidad, Monaghan, Dave, Lieu, Amos-Ritchie, Jans, Moseley, Gorman and Walker2014). However, these measurements can neglect viroporin activity targeted at the cell endomembrane system and/or be limited by the cytotoxicity of the overexpressed protein. An alternative, more flexible method is to measure the increase in membrane permeability to small solutes induced in bacteria by the expression of the cloned genes. In the Escherichia coli BL21(DE3) system, the expression of potentially cytotoxic viroporins is tightly regulated by the inducible expression of the T7 RNA polymerase and the viroporin product, the latter cloned into a vector of the pET series after a T7lac promotor (Dubendorff and Studier, Reference Dubendorff and Studier1991; Lama and Carrasco, Reference Lama and Carrasco1992). Induction of viroporin expression can arrest cell growth, or even promote lysis in the case of BL21(DE3) pLysS cells expressing T7 lysozime constitutively, a process that can be followed by the changes in the optical density of the culture (Lama and Carrasco, Reference Lama and Carrasco1992; Sanz et al., Reference Sanz, Perez and Carrasco1994; Madan et al., Reference Madan, Garcia Mde, Sanz and Carrasco2005; Hyser et al., Reference Hyser, Collinson-Pautz, Utama and Estes2010; Strtak et al., Reference Strtak, Perry, Sharp, Chang-Graham, Farkas and Hyser2019). Expression can also facilitate entry of HB (Madan et al., Reference Madan, Garcia Mde, Sanz and Carrasco2005) or induce the release to the medium of radioactivity from E. coli cells preloaded with [³H]uridine (Perez et al., Reference Perez, Garcia-Barreno, Melero, Carrasco and Guinea1997; Madan et al., Reference Madan, Garcia Mde, Sanz and Carrasco2005) or [³H]choline (Sanz et al., Reference Sanz, Perez and Carrasco1994).

These bacteria-based methods have become quite popular as a primary test for the presence of viroporin activity (Guo et al., Reference Guo, Sun, Sun, Wei, Xu, Huang, Liu, Liu, Luo, Yin and Liu2013; Joubert et al., Reference Joubert, Blasdell, Audsley, Trinidad, Monaghan, Dave, Lieu, Amos-Ritchie, Jans, Moseley, Gorman and Walker2014; Ao et al., Reference Ao, Guo, Sun, Sun, Fung, Wei, Han, Yao, Cao, Liu and Liu2015). Used to discern among the distinct virally encoded products those bearing viroporin activity (Lama and Carrasco, Reference Lama and Carrasco1992), or to map crucial residues for function (Perez et al., Reference Perez, Garcia-Barreno, Melero, Carrasco and Guinea1997), these assays are not, however, free in all instances from spurious effects arising from the massive incorporation of the expressed products to the bacterial membrane. To surpass this limitation, a more sophisticated procedure, called ‘positive assay’, has been developed (Taube et al., Reference Taube, Alhadeff, Assa, Krugliak and Arkin2014). The assay takes advantage of the widespread, poor specificity of the viroporin ion-channel activity (i.e., its general implication in nonspecific homeostasis regulation; see section ‘Mechanisms of viroporin-induced membrane permeabilization’) and makes use of a K⁺-uptake deficient bacteria strain requiring high concentration of this cation in the medium to grow. Upon expression of a channel-forming viroporin, the bacteria are able to thrive in low K⁺ media. Thus, membrane permeability induced by viroporin activity impacts growth positively, avoiding potential toxicity issues. This approach has been recently applied to identification of new viroporins (Tomar et al., Reference Tomar, Oren, Krugliak and Arkin2019) and to high-throughput screening of inhibitors (Lahiri and Arkin, Reference Lahiri and Arkin2022).

Despite the correlation existing between expression and permeabilization, it is still necessary to distinguish the intrinsic ion-channel activity of the virally encoded protein from that possibly performed by the host cell proteins. Thus, alternative, complementary approaches have made use of purified specimens, GST- or MBP- fusion proteins expressed in bacteria, or proteins and peptides produced through chemical synthesis, which allow for performing quantitative measurements. For instance, the addition to the culture medium of P3, a peptide representing the amphipathic TMD of the PV 2B protein, reproduced in BHK-21 cells the membrane permeabilization phenomenon induced by the single expression of 2B from an alphavirus replicon (Madan et al., Reference Madan, Sanchez-Martinez, Vedovato, Rispoli, Carrasco and Nieva2007). Similarly, the external addition to COS cells of SV40 GST-VP2 and GST-VP3 induced the formation of pores with inner diameters in the 3–6 nm range (Giorda et al., Reference Giorda, Raghava, Zhang and Hebert2013). Note that these pores are considerably larger than the canonical voltage-dependent channels of neurons that display radius quite below the nanometer (Sato et al., Reference Sato, Sato, Iwasaki, Doi and Engel1998; Moldenhauer et al., Reference Moldenhauer, Díaz-Franulic, González-Nilo and Naranjo2016).

Viroporins also display ion-channel activity in patch-clamped X. laevis oocytes (Pinto et al., Reference Pinto, Holsinger and Lamb1992) and whole cells (Chizhmakov et al., Reference Chizhmakov, Geraghty, Ogden, Hayhurst, Antoniou and Hay1996), provided that expressed recombinant forms are properly transported to the plasma membrane (Cabrera-Garcia et al., Reference Cabrera-Garcia, Bekdash, Abbott, Yazawa and Harrison2021; Harrison et al., Reference Harrison, Abbott, Gentzsch, Aleksandrov, Moroni, Thiel, Grant, Nichols, Lester, Hartel, Shepard, Garcia and Yazawa2022; Breitinger et al., Reference Breitinger, Sedky, Sticht and Breitinger2023). Generalization of these electrophysiological approaches in the field followed the publication of the earliest works by Pinto et al. on the channel activity of IAV M2 protein (Pinto et al., Reference Pinto, Holsinger and Lamb1992; Wang et al., Reference Wang, Lamb and Pinto1994; Pinto and Lamb, Reference Pinto and Lamb2006). Tetramers of this small protein make the virion envelope and the Golgi membrane permeable to protons upon its activation at low pH (Pinto and Lamb, Reference Pinto and Lamb2006). During entry through the endocytic route, M2 activity allows the acidification of the virion core, facilitating the disassembly of the ribonucleoprotein complexes (Pinto and Lamb, Reference Pinto and Lamb2006; Stauffer et al., Reference Stauffer, Feng, Nebioglu, Heilig, Picotti and Helenius2014; Lamb, Reference Lamb2020). When located at the Golgi membrane, it prevents the acidification of this organelle, thereby protecting low-pH-activated spike hemagglutinin from premature conformational changes during transit to the plasma membrane (Pinto and Lamb, Reference Pinto and Lamb2006; Lamb, Reference Lamb2020). Voltage clamp measurements of X. laevis oocytes and whole cells induced pH-dependent cation channel currents upon expression of M2 and homologous products BM2 and CM2 derived from influenza viruses B and C, respectively (Mould et al., Reference Mould, Paterson, Takeda, Ohigashi, Venkataraman, Lamb and Pinto2003; Pinto and Lamb, Reference Pinto and Lamb2006). Subsequently reported evidence supported specificity for proton conductance (Chizhmakov et al., Reference Chizhmakov, Geraghty, Ogden, Hayhurst, Antoniou and Hay1996; Pinto et al., Reference Pinto, Dieckmann, Gandhi, Papworth, Braman, Shaughnessy, Lear, Lamb and Degrado1997), although the M2 channel showed capability to permeate NH₄⁺ cations as well (Mould et al., Reference Mould, Drury, Frings, Kaupp, Pekosz, Lamb and Pinto2000).

Similar standard patch clamp approaches have been followed to test the activity of other potential viroporins including HCV p7 (Breitinger et al., Reference Breitinger, Farag, Ali and Breitinger2016), Paramyxovirus SH (Gan et al., Reference Gan, Tan, Lin, Yu, Wang, Tan, Vararattanavech, Yeo, Soon, Soong, Pervushin and Torres2012), SARS-CoV E and 3a (Lu et al., Reference Lu, Zheng, Xu, Schwarz, Du, Wong, Chen, Duan, Deubel and Sun2006; Nieto-Torres et al., Reference Nieto-Torres, Dediego, Alvarez, Jimenez-Guardeno, Regla-Nava, Llorente, Kremer, Shuo and Enjuanes2011; Toft-Bertelsen et al., Reference Toft-Bertelsen, Jeppesen, Tzortzini, Xue, Giller, Becker, Mujezinovic, Bentzen, Kolocouris, Kledal and Rosenkilde2021; Breitinger et al., Reference Breitinger, Sedky, Sticht and Breitinger2023; Ewart et al., Reference Ewart, Bobardt, Bentzen, Yan, Thomson, Klumpp, Becker, Rosenkilde, Miller and Gallay2023), HIV-1 Vpu (Coady et al., Reference Coady, Daniel, Tiganos, Allain, Friborg, Lapointe and Cohen1998; Greiner et al., Reference Greiner, Bolduan, Hertel, Gross, Hamacher, Schubert, Moroni and Thiel2016), or Alphavirus 6K (Antoine et al., Reference Antoine, Montpellier, Cailliau, Browaeys-Poly, Vilain and Dubuisson2007), just to mention some examples. However, channel recordings in oocytes and whole cells requires performing a series of controls before interpreting these sole results as enough evidence of bona fide viroporin activity (Harrison et al., Reference Harrison, Abbott, Gentzsch, Aleksandrov, Moroni, Thiel, Grant, Nichols, Lester, Hartel, Shepard, Garcia and Yazawa2022; Breitinger et al., Reference Breitinger, Sedky, Sticht and Breitinger2023; Ewart et al., Reference Ewart, Bobardt, Bentzen, Yan, Thomson, Klumpp, Becker, Rosenkilde, Miller and Gallay2023). Expression of HIV-1 Vpu in oocytes (Coady et al., Reference Coady, Daniel, Tiganos, Allain, Friborg, Lapointe and Cohen1998) or SARS-CoV E in whole cells (Nieto-Torres et al., Reference Nieto-Torres, Dediego, Alvarez, Jimenez-Guardeno, Regla-Nava, Llorente, Kremer, Shuo and Enjuanes2011) can diminish membrane conductance, and in both instances this effect has been ascribed to interferences with plasma membrane expression of cell K⁺ channels. Overexpression of not only viroporins (Antoine et al., Reference Antoine, Montpellier, Cailliau, Browaeys-Poly, Vilain and Dubuisson2007), but also other small integral membrane proteins (Shimbo et al., Reference Shimbo, Brassard, Lamb and Pinto1995), can evoke endogenous currents carried out by Ca²⁺-activated Cl⁻¹ channels. Even though pure specimens of SARS-CoV1 3a were previously shown to display ion-channel activity in planar bilayers (Castaño-Rodriguez et al., Reference Castaño-Rodriguez, Honrubia, Gutiérrez-Álvarez, Dediego, Nieto-Torres, Jimenez-Guardeño, Regla-Nava, Fernandez-Delgado, Verdia-Báguena, Queralt-Martín, Kochan, Perlman, Aguilella, Sola and Enjuanes2018), a recent series of published works support or oppose the viroporin-like role of SARS-CoV2 3a protein (Toft-Bertelsen et al., Reference Toft-Bertelsen, Jeppesen, Tzortzini, Xue, Giller, Becker, Mujezinovic, Bentzen, Kolocouris, Kledal and Rosenkilde2021; Harrison et al., Reference Harrison, Abbott, Gentzsch, Aleksandrov, Moroni, Thiel, Grant, Nichols, Lester, Hartel, Shepard, Garcia and Yazawa2022; Miller et al., Reference Miller, Houlihan, Matamala, Cabezas-Bratesco, Lee, Cristofori-Armstrong, Dilan, Sanchez-Martinez, Matthies, Yan, Yu, Ren, Brauchi and Clapham2023). The discrepancy in the reported electrophysiology data was suggested to arise from technical issues and/or the contribution of endogenous channels.

Permeability changes measured in the eukaryotic endomembrane system

Besides modifications of plasma membrane permeability, viroporins may alter permeability of cell organelles to ions and solutes upon synthesis. Most conspicuously, the activity of several viroporins can subvert interorganellar Ca²⁺ homeostasis, which plays important roles in pathophysiology (see Chami et al., Reference Chami, Oules and Paterlini-Brechot2006; Chen et al., Reference Chen, Cao and Zhong2019; Mehregan et al., Reference Mehregan, Perez-Conesa, Zhuang, Elbahnsi, Pasini, Lindahl, Howard, Ulens and Delemotte2022, for a discussion on this issue). Ca²⁺ stored at high concentrations in the ER and Golgi can be released into the cytosol on physiological stimuli that activate the InsP3R channels (Raffaello et al., Reference Raffaello, Mammucari, Gherardi and Rizzuto2016). This complex, multi-ionic process that also depends on the electric potential (Campbell et al., Reference Campbell, Abushawish, Valdez, Bell, Haryono, Rangamani and Bloodgood2023) is followed by fast Ca²⁺ influx from the external medium through channels residing at the plasma membrane. The rapid and sustained increase in the cytosolic concentration of the free cation ([Ca²⁺]_cyt) regulates a number of Ca²⁺-dependent processes. Moreover, a local increase at ER contacts can also lead to Ca²⁺ overload of mitochondria through the uptake via VDAC-MCU channels (Singh and Mabalirajan, Reference Singh and Mabalirajan2021). Ca²⁺ overload can result in turn in the long-term activation of the mPTP channel, which is accompanied by the release of pro-apoptotic factors and activation of cell death through the intrinsic pathway of apoptosis. Ca²⁺ signals directed to the mitochondria can also activate autophagy. Furthermore, if cells are properly primed by cytokines and PAMP ligands, these oscillations in the concentrations of mitochondrial and cytosolic Ca²⁺ can lead to inflammasome activation (Swanson et al., Reference Swanson, Deng and Ting2019).

Activation of cell death through apoptotic/necrotic pathways, cell content recycling through autophagy or activation of inflammatory responses can help the cell and tissues to cope with replicating viruses (Chami et al., Reference Chami, Oules and Paterlini-Brechot2006). Viroporin channels assembled at the ER membrane can prevent the sudden and sustained oscillation in intracellular calcium concentration and block activation of these cellular responses (Aldabe et al., Reference Aldabe, Irurzun and Carrasco1997; Chami et al., Reference Chami, Ferrari, Nicotera, Paterlini-Brechot and Rizzuto2003; de Jong et al., Reference De Jong, De Mattia, Van Dommelen, Lanke, Melchers, Willems and Van Kuppeveld2008; Hyser et al., Reference Hyser, Collinson-Pautz, Utama and Estes2010; Luganini et al., Reference Luganini, Di Nardo, Munaron, Gilardi, Fiorio Pla and Gribaudo2018). Thus, ER-targeting viroporins could suspend the infection-induced cell responses, providing the virus with sufficient time to replicate (Brisac et al., Reference Brisac, Teoule, Autret, Pelletier, Colbere-Garapin, Brenner, Lemaire and Blondel2010).

Contrasting this assumption, other experimental results support that viroporin-induced increase in [Ca²⁺]_cyt can actually promote apoptosis or autophagy after virus infection (Berkova et al., Reference Berkova, Morris and Estes2003; Hyser et al., Reference Hyser, Collinson-Pautz, Utama and Estes2010; Crawford et al., Reference Crawford, Hyser, Utama and Estes2012; Hyser and Estes, Reference Hyser and Estes2015). Interestingly, the increase in the cytosolic content of this cation may play a direct role in the assembly of new viral particles (Zhou et al., Reference Zhou, Frey and Yang2009; Amarasinghe and Dutch, Reference Amarasinghe and Dutch2014; Chen et al., Reference Chen, Cao and Zhong2019; Rahman et al., Reference Rahman, Kerviel, Mohl, He, Zhou and Roy2020). One well-studied example of a virus featuring calcium dependence for replication and assembly is RV (Hyser and Estes, Reference Hyser and Estes2015). In this case, the NSP4 viroporin activity appears to activate STIM1 in the ER, a process required to maintain the activation of calcium influx through channels in the plasma membrane (Hyser et al., Reference Hyser, Utama, Crawford, Broughman and Estes2013).

Estimating changes in ER and Golgi permeability coupled to expression of potential viroporins involves a distinct set of methods. One of those methods employs the sarco/endoplasmatic reticulum calcium ATPase (SERCA) inhibitor thapsigargin to measure the amount of calcium releasable from stores. Upon addition of this compound to cells, calcium uptake by the ER, but not its release into the cytosol, is blocked. Thereafter, sudden elevations of [Ca²⁺]_cyt occur due to the entry of extracellular Ca²⁺ via the Ca²⁺ channels in the plasma membrane, which are activated by the cation released from the organelle. Hence, the response to thapsigargin depends on the amount of calcium previously stored in the ER at high concentration, which might be altered upon expression and ER localization of several viroporins (Campanella et al., Reference Campanella, De Jong, Lanke, Melchers, Willems, Pinton, Rizzuto and Van Kuppeveld2004; Luganini et al., Reference Luganini, Di Nardo, Munaron, Gilardi, Fiorio Pla and Gribaudo2018; Gladue et al., Reference Gladue, Largo, De La Arada, Aguilella, Alcaraz, Arrondo, Holinka, Brocchi, Ramirez-Medina, Vuono, Berggren, Carrillo, Nieva and Borca2018a).

In addition, the fluorescent Ca²⁺ indicator Fura-2 has been used to measure changes of [Ca²⁺]_cyt that demonstrated an increase in individual cells expressing RV NSP4 protein (Berkova et al., Reference Berkova, Morris and Estes2003), but a reduction in the amount of releasable calcium in cells expressing the HCMV US21 protein (Luganini et al., Reference Luganini, Di Nardo, Munaron, Gilardi, Fiorio Pla and Gribaudo2018) or the Coxsackievirus 2B protein (Campanella et al., Reference Campanella, De Jong, Lanke, Melchers, Willems, Pinton, Rizzuto and Van Kuppeveld2004). In the latter case, ER, Golgi, and cytosol were also selectively loaded with different versions of the Ca²⁺ indicator aequorin and its substrate coelenterazine to demonstrate that 2B decreases the cation content of both the ER and the Golgi (Campanella et al., Reference Campanella, De Jong, Lanke, Melchers, Willems, Pinton, Rizzuto and Van Kuppeveld2004). Other Ca²⁺ indicators as Indo-1 (Hyser et al., Reference Hyser, Collinson-Pautz, Utama and Estes2010), GCaMP5G (Hyser et al., Reference Hyser, Utama, Crawford, Broughman and Estes2013; Strtak et al., Reference Strtak, Perry, Sharp, Chang-Graham, Farkas and Hyser2019), and rhod-2/AM (Gladue et al., Reference Gladue, Largo, De La Arada, Aguilella, Alcaraz, Arrondo, Holinka, Brocchi, Ramirez-Medina, Vuono, Berggren, Carrillo, Nieva and Borca2018a; Gladue et al., Reference Gladue, Largo, Holinka, Ramirez-Medina, Vuono, Berggren, Risatti, Nieva and Borca2018b) have been described in the literature to measure [Ca²⁺]_cyt changes in cells expressing viroporins.

Dissipation of proton gradients at Golgi apparatus and ERGIC has been proposed as an additional viroporin function (Nieva et al., Reference Nieva, Madan and Carrasco2012), presumably to avoid premature low-pH activation of spike proteins in transit to the plasma membrane (Sakaguchi et al., Reference Sakaguchi, Leser and Lamb1996). Both HCV infection and expression of the p7 product increased the pH of acidic compartments, detected with LysoSensor Yellow/Blue DND-160 or LysoTracker Red DND-99, two pH fluorescent probes that exhibit a pH-dependent emission (Wozniak et al., Reference Wozniak, Griffin, Rowlands, Harris, Yi, Lemon and Weinman2010; Breitinger et al., Reference Breitinger, Farag, Ali and Breitinger2016). Overexpression of infectious bronchitis CoV E also alters Golgi pH as evidenced from changes in the co-expressed pH-sensitive GFP mutant pHluorin-TGN38 (Westerbeck and Machamer, Reference Westerbeck and Machamer2019). Similarly, SARS-CoV-2 E protein localized at the ERGIC and provoked an increase of pH in live NIH-3T3 cell internal organelles, as detected by changes in emission of Lysosensor Green DND-189 fluorescent probe (Cabrera-Garcia et al., Reference Cabrera-Garcia, Bekdash, Abbott, Yazawa and Harrison2021; Bekdash et al., Reference Bekdash, Yoshida, Nair, Qiu, Ahdout, Tsai, Uryu, Soni, Huang, Ho and Yazawa2024).

Notably, Wozniak et al. (Reference Wozniak, Griffin, Rowlands, Harris, Yi, Lemon and Weinman2010) further isolated a light membrane vesicle fraction containing lysosomes and ER from cells expressing the HCV viroporin p7. These vesicles were loaded with the pH-dependent fluorophore 8-hydroxypyrene-1,3,6-trisulfonic acid and subsequently used to measure proton permeability. Following this ex vivo strategy, these authors demonstrated that H⁺ efflux was promoted in vesicles bearing functional forms of p7, but not in those isolated from cells expressing non-functional mutants.

ER and Golgi compartments are also subject to profound remodeling by RNA viruses that replicate their genomes in the cytoplasm of the host cell. The involvement picornavirus 2B/2BC viroporin in the formation of new cytoplasmic vesicles or ‘viroplasm’, where genome replication may take place to protect double-stranded RNA intermediates from innate immune recognition, and affecting overall secretory pathways was well documented (Barco and Carrasco, Reference Barco and Carrasco1995; Suhy et al., Reference Suhy, Giddings and Kirkegaard2000; Gonzalez and Carrasco, Reference Gonzalez and Carrasco2003; de Jong et al., Reference De Jong, Melchers, Glaudemans, Willems and Van Kuppeveld2004; Hsu et al., Reference Hsu, Ilnytska, Belov, Santiana, Chen, Takvorian, Pau, Van Der Schaar, Kaushik-Basu, Balla, Cameron, Ehrenfeld, Van Kuppeveld and Altan-Bonnet2010; Nieva et al., Reference Nieva, Madan and Carrasco2012). It has not been until recently, however, that a molecular mechanism coupling viroporin-induced permeability to the activation of functional vesicular compartments (spherules) at the ER has been put forward (Nishikiori and Ahlquist, Reference Nishikiori and Ahlquist2018; Nishikiori and Ahlquist, Reference Nishikiori and Ahlquist2021). Genome replication of the alphavirus BMV requires the activity of the membrane-associated RNA replication protein 1a. This protein also makes yeast ER membranes permeable to phenoxyl radicals that are generated by an engineered ascorbate peroxidase enzyme confined within the lumen of the organelle, and has therefore been defined as a viroporin (Nishikiori and Ahlquist, Reference Nishikiori and Ahlquist2018). According to the proposed model (Nishikiori and Ahlquist, Reference Nishikiori and Ahlquist2021), ER permeabilization through 1a viroporin activity releases oxidizing potential from the lumen of the organelle to the cytoplasm. Oxidizing radicals are in turn required for the generation of disulfide bridges and the stabilization of the 1a multimers that activate late RNA replication functions at the neck-like connection between the interior of the spherules and the cytosol.

By comparison, the amount of evidence proving that viroporin activity can directly permeabilize mitochondrial membranes is very limited (Madan et al., Reference Madan, Castello and Carrasco2008; Madan et al., Reference Madan, Sanchez-Martinez, Carrasco and Nieva2010b; Nieva et al., Reference Nieva, Madan and Carrasco2012; Lee et al., Reference Lee, Cho, Lee, You, Yoo and Kim2018; You et al., Reference You, Cho, Lee, Lee, Yu, Yoon, Yoo, Kim and Lee2019; Cedillo-Barron et al., Reference Cedillo-Barron, Garcia-Cordero, Visoso-Carvajal and Leon-Juarez2024). However, the recent demonstration that mitochondrial membrane permeabilization by the NoV NS3 viroporin activates cell death processes for the induction of virus egress, constitutes a breakthrough in the field (Wang et al., Reference Wang, Zhang, Orchard, Hancks and Reese2023). It appears that permeabilization of mitochondria by the N-terminal domain of NS3 alters membrane potential and releases reactive oxygen species (ROS) and Cytochrome c to trigger programmed cell death and activation of cell plasma membrane lysis by Ninjurin-1, the latter process required for the release of the NoV particles to the medium. Depolarization of mitochondria can be followed by cytometry upon induction of viroporin expression in cells using the dye tetramethylrhodamine methyl ester, which is sequestered by active mitochondria (Lee et al., Reference Lee, Cho, Lee, You, Yoo and Kim2018; Wang et al., Reference Wang, Zhang, Orchard, Hancks and Reese2023). Mitochondrial ROS can be detected following a similar cytometry approach, after incubation of the viroporin-expressing cells with MitoSoX Red (Wang et al., Reference Wang, Zhang, Orchard, Hancks and Reese2023). In addition, mitochondria can be isolated from mice liver or cultured cells for incubation in vitro with purified forms of viroporins (You et al., Reference You, Cho, Lee, Lee, Yu, Yoon, Yoo, Kim and Lee2019; Wang et al., Reference Wang, Zhang, Orchard, Hancks and Reese2023). Following this approach, it has been shown that, upon incubation of the isolated organelles with hepatitis B virus (HBV) X protein (You et al., Reference You, Cho, Lee, Lee, Yu, Yoon, Yoo, Kim and Lee2019), membrane potential of mitochondria, which were stained with cationic dyes safranine O or JC-1, decreased significantly with respect to a carbonyl cyanide 3-chlorophenylhydrazone positive control. Using an alternative method, direct permeabilization of the mitochondrial membrane by NoV NS3 was assayed by determining Cytochrome c release into the medium (Wang et al., Reference Wang, Zhang, Orchard, Hancks and Reese2023.

Lipid vesicles: bulk measurements

A variety of minimal systems have been used to characterize viroporin function under more defined and controlled experimental conditions. These systems consist of model membranes of defined lipid composition incubated with pure protein/peptide specimens. Pore-forming activity can be assayed in liposomes through fluorescence-based methods both, in bulk and at the single-vesicle level. Bulk determinations of permeability changes should be preferably performed using large unilamellar vesicles (LUVs) with mean diameters ≥100 nm, which are devoid of the curvature stress and inter-layer surface asymmetry characteristic of small unilamellar vesicles produced by sonication. LUVs loaded with 5(6)-carboxyfluorescein at self-quenching concentrations, or high concentrations of fluorescent compounds together with quencher molecules (e.g., Tb³⁺/DPA or ANTS/DPX pairs), can be readily prepared through extrusion of freeze-thawed, water-dispersed lipid multilayers followed by gel-filtration or repeated centrifugation to eliminate non-encapsulated probe (Agirre et al., Reference Agirre, Barco, Carrasco and Nieva2002; StGelais et al., Reference Stgelais, Tuthill, Clarke, Rowlands, Harris and Griffin2007; Foster et al., Reference Foster, Verow, Wozniak, Bentham, Thompson, Atkins, Weinman, Fishwick, Foster, Harris and Griffin2011; Gervais et al., Reference Gervais, Do, Cantin, Kukolj, White, Gauthier and Vaillancourt2011; Gladue et al., Reference Gladue, Holinka, Largo, Fernandez Sainza, Carrillo, O’donnell, Baker-Branstetter, Lu, Ambroggio, Risatti, Nieva and Borca2012; Largo et al., Reference Largo, Gladue, Huarte, Borca and Nieva2014; Largo et al., Reference Largo, Verdia-Baguena, Aguilella, Nieva and Alcaraz2016; You et al., Reference You, Cho, Lee, Lee, Yu, Yoon, Yoo, Kim and Lee2019). Release to the medium after viroporin-induced membrane permeabilization results in the dilution of the molecule pairs and concomitant increase in fluorescence intensity of the probe component, which can be monitored in a fluorimeter as a function of time. This procedure substantiated the capacity for permeabilizing membranes of a wide range of potential viroporins, including PV 2B (Agirre et al., Reference Agirre, Barco, Carrasco and Nieva2002), HCV p7 (StGelais et al., Reference Stgelais, Tuthill, Clarke, Rowlands, Harris and Griffin2007; Foster et al., Reference Foster, Verow, Wozniak, Bentham, Thompson, Atkins, Weinman, Fishwick, Foster, Harris and Griffin2011; Gervais et al., Reference Gervais, Do, Cantin, Kukolj, White, Gauthier and Vaillancourt2011), CSFV p7 (Gladue et al., Reference Gladue, Holinka, Largo, Fernandez Sainza, Carrillo, O’donnell, Baker-Branstetter, Lu, Ambroggio, Risatti, Nieva and Borca2012; Largo et al., Reference Largo, Gladue, Huarte, Borca and Nieva2014; Largo et al., Reference Largo, Verdia-Baguena, Aguilella, Nieva and Alcaraz2016), HBV X protein (Lee et al., Reference Lee, Cho, Lee, You, Yoo and Kim2018; You et al., Reference You, Cho, Lee, Lee, Yu, Yoon, Yoo, Kim and Lee2019), NoV NS3 (Wang et al., Reference Wang, Zhang, Orchard, Hancks and Reese2023), and alphavirus 6K (Dey et al., Reference Dey, Siddiqui, Mamidi, Ghosh, Kumar, Chattopadhyay, Ghosh and Banerjee2019), among others. In addition, more sophisticated experimental setups have been used to generate electrochemical and/or pH gradients across liposomal membranes and test the capacity for translocating different cations (Na⁺, K⁺, Ca²⁺, and H⁺) by viroporins as IAV M2 (Lin et al., Reference Lin, Heider and Schroeder1997; Lin and Schroeder, Reference Lin and Schroeder2001; Leiding et al., Reference Leiding, Wang, Martinsson, Degrado and Arskold2010; Peterson et al., Reference Peterson, Ryser, Funk, Inouye, Sharma, Qin, Cross and Busath2011), HCV p7 (Wozniak et al., Reference Wozniak, Griffin, Rowlands, Harris, Yi, Lemon and Weinman2010; Gan et al., Reference Gan, Surya, Vararattanavech and Torres2014), or SARS-CoV 3a (Kern et al., Reference Kern, Sorum, Mali, Hoel, Sridharan, Remis, Toso, Kotecha, Bautista and Brohawn2021; Miller et al., Reference Miller, Houlihan, Matamala, Cabezas-Bratesco, Lee, Cristofori-Armstrong, Dilan, Sanchez-Martinez, Matthies, Yan, Yu, Ren, Brauchi and Clapham2023).

Since vesicles are made of pure lipids, this type of experiments also allows straightforward determination of the dependence of pore-forming activity on membrane lipid composition, which in turn can provide clues about the organelle targeted by a given viroporin (van Meer et al., Reference Van Meer, Voelker and Feigenson2008). For instance, PV 2B activity appears to depend on the anionic nature of the phospholipid headgroup and the length and degree of unsaturation of its acyl chains (Agirre et al., Reference Agirre, Barco, Carrasco and Nieva2002). Dependence of solute leakage on single lipids such as phosphatidylinositol or cardiolipin has been reported for viroporins that localize in the ER (Gladue et al., Reference Gladue, Holinka, Largo, Fernandez Sainza, Carrillo, O’donnell, Baker-Branstetter, Lu, Ambroggio, Risatti, Nieva and Borca2012; Largo et al., Reference Largo, Gladue, Huarte, Borca and Nieva2014; Largo et al., Reference Largo, Verdia-Baguena, Aguilella, Nieva and Alcaraz2016) or mitochondria (You et al., Reference You, Cho, Lee, Lee, Yu, Yoon, Yoo, Kim and Lee2019; Wang et al., Reference Wang, Zhang, Orchard, Hancks and Reese2023), respectively. Functional efficacy of M2 reconstituted in LUVs seems to depend on cholesterol (Leiding et al., Reference Leiding, Wang, Martinsson, Degrado and Arskold2010), a conspicuous component of Golgi and cell plasma membranes (van Meer et al., Reference Van Meer, Voelker and Feigenson2008).

LUVs can be loaded with compounds of different molecular weight sizes to prove solute-permeability through size-limited pores. Using this approach, molecules with a maximal size in the range of 1–10 kD have been described to leak out through pores established by potential viroporins (Agirre et al., Reference Agirre, Barco, Carrasco and Nieva2002; Shukla et al., Reference Shukla, Dey, Banerjee, Nain and Banerjee2015; Wang et al., Reference Wang, Zhang, Orchard, Hancks and Reese2023). In the LUV system, efficient solute release occurring at protein-to-lipid mole ratios lower than 1:100 together with the existence of an MW cutoff for the permeating solute are usually good indicators of the measured permeability increase resulting from bona fide viroporin activity, and not from spurious processes as the bilayer disintegration that may occur coupled to the massive incorporation of hydrophobic protein moieties (detergent-like effect; Shai, Reference Shai1999).

Fluorescence microscopy of single vesicles

In comparison, quantitative fluorescence microscopy performed in single vesicles provides more solid evidence on the capacity of viroporins to alter lipid bilayer permeability without compromising its overall integrity. For the purpose of lipid permeability testing, micrometer-sized giant unilamellar vesicles are produced by electroformation and labeled with a fluorescent lipid probe (e.g., N-(lissamine Rhodamine B sulfonyl) phosphatidylethanolamine) that allows their detection in a confocal microscope (see, e.g., Gladue et al., Reference Gladue, Largo, De La Arada, Aguilella, Alcaraz, Arrondo, Holinka, Brocchi, Ramirez-Medina, Vuono, Berggren, Carrillo, Nieva and Borca2018a). Changes in permeability induced by potential viroporins are determined in the intact vesicles after addition to the external medium of a soluble fluorescent probe as Alexa Fluor 488 (MW: 0.6 kDa). Access of this fluorescent marker to the internal volume of vesicles after viroporin treatment denotes the occurrence of efficient membrane permeabilization. Application of this approach may offer useful information on the virpoporin-induced permeabilization mechanism. First, as mentioned before, it allows us confirming that the detected changes in permeability occur while preserving the integrity of the lipid bilayer architecture. Second, the degree of filling, which can be quantitatively determined from the ratio of fluorescence intensities measured in the lumen of the vesicle with respect to that measured in the external medium (Largo et al., Reference Largo, Gladue, Torralba, Aguilella, Alcaraz, Borca and Nieva2018), may reveal if the process follows an all-or-none versus a graded mechanism of membrane permeabilization (Apellaniz et al., Reference Apellaniz, Nieva, Schwille and Garcia-Saez2010; Gladue et al., Reference Gladue, Largo, De La Arada, Aguilella, Alcaraz, Arrondo, Holinka, Brocchi, Ramirez-Medina, Vuono, Berggren, Carrillo, Nieva and Borca2018a). Stable pores assembled in lipid bilayers would tend to permeabilize membranes following the former mechanism, whereas transient, less defined structures are responsible for the latter. Third, the stability of the pores can be further tested after permeation to the Alexa Fluor 488 marker reaches equilibrium by incubating the vesicles with a second marker, for example, Alexa Fluor 647 (Gladue et al., Reference Gladue, Largo, De La Arada, Aguilella, Alcaraz, Arrondo, Holinka, Brocchi, Ramirez-Medina, Vuono, Berggren, Carrillo, Nieva and Borca2018a). Incorporation of similar amounts of this second marker to the vesicle internal volume supports a pore-forming activity stable in time. Finally, the approximate size of the pore can be estimated using in the medium FITC-dextrans of different MW-s as permeant/impermeant solutes (Largo et al., Reference Largo, Gladue, Torralba, Aguilella, Alcaraz, Borca and Nieva2018; You et al., Reference You, Cho, Lee, Lee, Yu, Yoon, Yoo, Kim and Lee2019).

Electrophysiology in planar membranes

Ion-channel activity of viroporins can be also studied in vitro by means of electrophysiology in planar lipid bilayers (PLBs). This technique commonly employs a cell with two compartments separated by an inert film such as Teflon in which a small hole is made (typically 100–200 μm diameter orifice). Membranes can be formed at each side by different two main methods. The first one is the so-called solvent-containing membrane. Lipids dissolved in organic solvents are spread at both sides of the orifice. The formation of the bilayer can be controlled by with a microscope so that when monolayers formed in each side come into contact, a gray–black spot starts to spread over the film, explaining why bilayers formed following this procedure are called black lipid membranes (Mueller et al., Reference Mueller, Rudin, Tien and Wescott1963; Van Gelder et al., Reference Van Gelder, Dumas and Winterhalter2000; Winterhalter, Reference Winterhalter2000). Alternatively, solvent-free planar lipid membranes can be prepared by using the Montal–Mueller technique (Montal and Mueller, Reference Montal and Mueller1972) or slight modifications of it (Bezrukov and Vodyanoy, Reference Bezrukov and Vodyanoy1993). Aliquots of lipid in organic solvent are added onto the salt solution subphases at both compartments (so-called cis and trans) of a Teflon chamber. Since Teflon is lipophobic, the orifices must be pre-treated with some kind of solution to make it lipophilic. After evaporation of the organic solvents used, the level of solutions in each compartment is raised above the orifice, so the planar bilayer is formed by apposition (or opposition) of the two monolayers. Within this technique, capacitance measurements are used to monitor correct bilayer formation. The two methods described to form PLBs yield comparable results for ion-channel reconstitution in terms of measured conductance and selectivity. However, solvent-containing membranes appear to be slightly more flexible than the solvent free ones, which may have an impact in facilitating the insertion of extremely hydrophobic proteins (Winterhalter, Reference Winterhalter2000), as it is the case of the pore-forming domains of class I and II viroporins. Overall, PLBs provide the significant advantages of working in well-controlled artificial environment using only tiny amounts of material. Also, unlike liposomes, one has direct access to both sides of the membrane and therefore it is possible to form asymmetric membranes (Van Gelder et al., Reference Van Gelder, Dumas and Winterhalter2000).

Electrophysiological studies in PLBs with conventional ion channels usually show random transitions between open and closed states that allow to identify a canonical unitary conductance (Gutsmann et al., Reference Gutsmann, Heimburg, Keyser, Mahendran and Winterhalter2015). Quite in contrast, most studies with viroporins have encountered a large heterogeneity of results in terms of diversity of conductive levels and in their dynamic behavior (Mehnert et al., Reference Mehnert, Routh, Judge, Lam, Fischer, Watts and Fischer2008; Montserret et al., Reference Montserret, Saint, Vanbelle, Salvay, Simorre, Ebel, Sapay, Renisio, Bockmann, Steinmann, Pietschmann, Dubuisson, Chipot and Penin2010; Hyser, Reference Hyser and Delcour2015; Largo et al., Reference Largo, Queralt-Martin, Carravilla, Nieva and Alcaraz2021). As shown in Figure 2a,b for CSFV p7 and SARS-CoV2 E, respectively, even when performing experiments with the same protocol and the same protein sample, it is possible to find traces with well-defined ‘opening’ and ‘closing’ rapid events, and others with larger current levels with longer lifetimes without almost any small flickering (Verdia-Baguena et al., Reference Verdia-Baguena, Nieto-Torres, Alcaraz, Dediego, Enjuanes and Aguilella2013; Largo et al., Reference Largo, Verdia-Baguena, Aguilella, Nieva and Alcaraz2016). Although such heterogeneity raised questions about reproducibility and/or the presence of detergent-like mechanisms (Hyser, Reference Hyser and Delcour2015; Mehnert et al., Reference Mehnert, Routh, Judge, Lam, Fischer, Watts and Fischer2008; Shai, Reference Shai1999), electrophysiological characterizations of viroporins have proved to be solid when data collected are large enough to generate histograms that are statistically significant (see Figure 2a,b). Interestingly, pore size estimates arising from PLB can be satisfactorily compared to permeability assays with solutes of different molecular weight stablishing a certain cutoff (Largo et al., Reference Largo, Gladue, Torralba, Aguilella, Alcaraz, Borca and Nieva2018). In the same line, data obtained in PLB become more trustworthy when regulatory factors of ion-channel activity are found in parallel in different techniques. For instance, pH or membrane composition that are critical for CSFV p7 permeabilization activity are found in PLB, leakage assay and even in atomic force microscopic imaging (see Figure 2c,d) (Largo et al., Reference Largo, Verdia-Baguena, Aguilella, Nieva and Alcaraz2016; Largo et al., Reference Largo, Queralt-Martin, Carravilla, Nieva and Alcaraz2021). The relevance and interpretation of PLB experiments with viroporins will be discussed in detail below in section ‘Mechanisms of viroporin-induced membrane permeabilization’.

Figure 2.

Electrophysiology and complementary approaches to study viroporin activity in vitro. (a) CSFV p7 ion channel activity in endoplasmic reticulum (ER)-like planar bilayers. Representative traces with current recordings without any protein addition (control) and after the addition of p7 protein show the different magnitudes of current jumps seen in experiments. Histogram of the current jump amplitude fitted to two Gaussian peaks. Current was recorded in 150 mM KCl, pH 5.0 at a potential of −50 mV. Reprinted with permission from Largo et al. (Reference Largo, Verdia-Baguena, Aguilella, Nieva and Alcaraz2016). (b) Representative current recordings of SARS-CoV2 E in ERGIC-mimetic lipid bilayers at 100 mM CaCl₂ show channel opening events with variable durations and conductance. Histogram of conductance jumps at +100 mV. Solid line indicates Gaussian fitting of the histogram. Reprinted with permission from Dregni et al. (Reference Dregni, Mckay, Surya, Queralt-Martin, Medeiros-Silva, Wang, Aguilella, Torres and Hong2023). (c) Comparison between vesicle leakage and channel formation induced by CSFV p7. Left: Percentage of planar bilayers displaying IC activity (light gray) is compared to the leakage percentage induced by CSFV-p7 addition to LUV (protein-to-lipid ratio, 1:250) (dark gray). Right: Representative conductance recordings for the same conditions. Reprinted with permission from Largo et al. (Reference Largo, Verdia-Baguena, Aguilella, Nieva and Alcaraz2016). (d) Atomic force microscopic images in ER-like lipid bilayers at pH 7.4 in control conditions (left) and in the presence of CSFV p7 at pH 7.4 (center) and pH 5.0 (right). CFSV p7:lipid ratio was 1:800. The color bar indicates the height in the z dimension, being white the highest and black the lowest area. Reprinted with permission from Largo et al. (Reference Largo, Queralt-Martin, Carravilla, Nieva and Alcaraz2021).

IAV M2 viroporin

IC activity of the bona fide IAV M2 channel and its blocking by amantadine/rimantadine are probably the viroporin activity and function inhibition processes that are best characterized on structural grounds (Schnell and Chou, Reference Schnell and Chou2008; Stouffer et al., Reference Stouffer, Acharya, Salom, Levine, Di Costanzo, Soto, Tereshko, Nanda, Stayrook and Degrado2008; Acharya et al., Reference Acharya, Carnevale, Fiorin, Levine, Polishchuk, Balannik, Samish, Lamb, Pinto, Degrado and Klein2010; Sharma et al., Reference Sharma, Yi, Dong, Qin, Peterson, Busath, Zhou and Cross2010; Nieva et al., Reference Nieva, Madan and Carrasco2012; Scott and Griffin, Reference Scott and Griffin2015; Jalily et al., Reference Jalily, Duncan, Fedida, Wang and Tietjen2020; Lamb, Reference Lamb2020). A line of inquiry into the mechanism of proton conduction was followed by DeGrado et al. utilizing an optimized synthetic peptide that represented the tetramerization, pore-forming domain of M2 (Stouffer et al., Reference Stouffer, Acharya, Salom, Levine, Di Costanzo, Soto, Tereshko, Nanda, Stayrook and Degrado2008). An X-ray structure obtained at pH 6.5 with a 1.65 Å resolution provided insights into a potential M2 ‘intermediate’ state (Acharya et al., Reference Acharya, Carnevale, Fiorin, Levine, Polishchuk, Balannik, Samish, Lamb, Pinto, Degrado and Klein2010) (Figure 3a). In the lumen of this M2 pore structure, up to five side-chain layers can be defined, namely, the Val27 valve, Ser31, the His-box (the selectivity filter), the Trp-basket, and Asp44-Arg45; plus three clusters of immobilized water molecules H-bonded to the protein, designated as ‘entry’, ‘bridging’, and ‘exit’. The comparison with structures previously solved at pHs close to 7.5 (neutral) (Schnell and Chou, Reference Schnell and Chou2008) and 5.0 (low pH) (Stouffer et al., Reference Stouffer, Acharya, Salom, Levine, Di Costanzo, Soto, Tereshko, Nanda, Stayrook and Degrado2008) suggested that tilting and bending of the helix N-terminus with respect to the pore axis can constrict the Val27 valve at low pH, whereas helix bundle opening, together with the flipping of Trp side chains in the basket, may result in water access at the C-terminus. The conformational oscillations between ‘Open-out’, ‘Intermediate’, and ‘Open-in’ linked to changes in the protonation state would facilitate H₃O⁺ diffusion down the concentration gradient, the overall process being regulated by the rate of deprotonation of the His-box (Acharya et al., Reference Acharya, Carnevale, Fiorin, Levine, Polishchuk, Balannik, Samish, Lamb, Pinto, Degrado and Klein2010).

Figure 3.

Structural features of IAV M2 and SARS-CoV2 E viroporins. (a) Structure of the proton-selective IAV M2. Left: tetrameric bundle structure solved by solid-state NMR spectroscopy in a lipid bilayer (PDB ID: 2L0J). Each monomer includes the transmembrane pore-forming domain (residues 22–46) and an interfacial amphipathic helix (residues 48–58); Right: internal anatomy of the pore-forming domain based on a crystal structure obtained with a resolution of 1.65 Å at the ‘intermediate’ pH 6.5 (PDB ID: 3LBW). Positions of the side-chain layers and water clusters are indicated. (b) Structures of the pore-forming transmembrane domain from SARS-CoV2 E solved by solid-state NMR spectroscopy in ERGIC-like bilayers. Top: a ‘closed’ state is favored at high pH and low Ca²⁺ concentration (PDB ID: 7K3G). Bottom: ‘open’ state adopted at low pH and high Ca²⁺ concentration (PDB ID: 8SUZ). Side chains of Leu18 are depicted to illustrate aperture of the pore. Side chains of Phe20, Phe23, and Phe26 undergo conformational changes coupled to the transition. Structure models rendered with Chimera (Pettersen et al., Reference Pettersen, Goddard, Huang, Couch, Greenblatt, Meng and Ferrin2004).

Availability of crystal structures of M2 subsequently obtained under different experimental conditions enabled the refinement of this mechanism (Thomaston et al., Reference Thomaston, Alfonso-Prieto, Woldeyes, Fraser, Klein, Fiorin and Degrado2015; Thomaston et al., Reference Thomaston, Woldeyes, Nakane, Yamashita, Tanaka, Koiwai, Brewster, Barad, Chen, Lemmin, Uervirojnangkoorn, Arima, Kobayashi, Masuda, Suzuki, Sugahara, Sauter, Tanaka, Nureki, Tono, Joti, Nango, Iwata, Yumoto, Fraser and Degrado2017; Thomaston et al., Reference Thomaston, Wu, Polizzi, Liu, Wang and Degrado2019). ‘Open-in’ structures (also designated as ‘Inward-open’) were obtained at high (8.0) and low (5.0) pHs in lipid cubic phases with 1.10 Å resolution (Thomaston et al., Reference Thomaston, Alfonso-Prieto, Woldeyes, Fraser, Klein, Fiorin and Degrado2015). These structures revealed that at the low pH, rather than organized in three defined clusters, hydrogen-bonded water molecules formed a network along the pore from its entrance until the His-box selectivity filter. In combination with in silico simulations, these structures suggested that in the ‘Inward-open’ state H₃O⁺ would channel protons through hydrogen bonds, which would orient in the water network as a function of pH to stabilize the protonation state of gating His37 residues. This mechanism of stabilization of the protonated His37 state received further support from diffraction studies performed at room temperature (Thomaston et al., Reference Thomaston, Woldeyes, Nakane, Yamashita, Tanaka, Koiwai, Brewster, Barad, Chen, Lemmin, Uervirojnangkoorn, Arima, Kobayashi, Masuda, Suzuki, Sugahara, Sauter, Tanaka, Nureki, Tono, Joti, Nango, Iwata, Yumoto, Fraser and Degrado2017). In addition, M2 structures solved in the presence of bound adamantanes emphasized the role of interfering with this internal water network in the mechanism of action of these inhibitors (Thomaston et al., Reference Thomaston, Wu, Polizzi, Liu, Wang and Degrado2019).

Only few viroporins display an ionic specificity comparable to that of M2, and even in this paradigmatic case, there is some controversy about the actual nature of the ion conduction in the channel. Thus, some experiments of M2 have shown a large ionic specificity, since H⁺ are transported in a ratio 10⁵ to 10⁶ with respect to of Na⁺ and K⁺ (Lin and Schroeder, Reference Lin and Schroeder2001), while other investigations reported mild selectivity to alkali cations with no clear specificity (Tosteson et al., Reference Tosteson, Pinto, Holsinger and Lamb1994; Stauffer et al., Reference Stauffer, Feng, Nebioglu, Heilig, Picotti and Helenius2014) or even permeability to small molecules such as anionic carboxyfluorescecin in liposomes or hygromicyin B in bacteria (Guinea and Carrasco, Reference Guinea and Carrasco1994; Scott et al., Reference Scott, Kankanala, Foster, Goldhill, Bao, Simmons, Pingen, Bentham, Atkins, Loundras, Elderfield, Claridge, Thompson, Stilwell, Tathineni, Mckimmie, Targett-Adams, Schnell, Cook, Evans, Barclay, Foster and Griffin2020; Devantier et al., Reference Devantier, Kjaer, Griffin, Kragelund and Rosenkilde2024). Interestingly, such differences have been mechanistically interpreted in terms of a dual function of M2 in which the channel could mediate the efflux of alkali cations necessary to maintain a charge balance across the membrane once protons flow into the virion along their concentration gradient (Tosteson et al., Reference Tosteson, Pinto, Holsinger and Lamb1994; Stauffer et al., Reference Stauffer, Feng, Nebioglu, Heilig, Picotti and Helenius2014).

SARS-CoV E viroporin

Functional complexity has been associated with SARS-CoV E viroporin too (Surya et al., Reference Surya, Tavares-Neto, Sanchis, Queralt-Martín, Alcaraz, Torres and Aguilella2023). The preliminary evidence that the TMD of SARS-CoV1 E protein may form pentamers in gels (Parthasarathy et al., Reference Parthasarathy, Ng, Lin, Liu, Pervushin, Gong and Torres2008) was subsequently supported by 3D structures derived from NMR spectroscopy of peptides solubilized in lipid micelles (Pervushin et al., Reference Pervushin, Tan, Parthasarathy, Lin, Jiang, Yu, Vararattanavech, Soong, Liu and Torres2009; Surya et al., Reference Surya, Li and Torres2018). The COVID-19 pandemic intensified notably research efforts on the structure and function of E as a potential target for antiviral drugs (To et al., Reference To, Surya and Torres2016; Devantier et al., Reference Devantier, Kjaer, Griffin, Kragelund and Rosenkilde2024). The result of this increased research activity was the publication of two high-resolution structures of the SARS-CoV-2 E TMD pore domain in ERGIC-like membranes determined by solid state NMR spectroscopy (Mandala et al., Reference Mandala, Mckay, Shcherbakov, Dregni, Kolocouris and Hong2020; Medeiros-Silva et al., Reference Medeiros-Silva, Dregni, Somberg, Duan and Hong2023) (Figure 3b). In the initially reported structure (Mandala et al., Reference Mandala, Mckay, Shcherbakov, Dregni, Kolocouris and Hong2020), the SARS-CoV2 E TMD was modeled as a pentamer, whose structure featured a closed pore displaying Val and Leu interdigitation and an aromatic belt composed of Phe20, Phe23, and Phe26 (Figure 3b, top). Since SARS-CoV-1 E and SARS-CoV-2 E are very similar (they share 95% of the sequence and have comparable TMDs, only differing by three residue substitutions and one deletion localized at the C-terminus; Surya et al., Reference Surya, Tavares-Neto, Sanchis, Queralt-Martín, Alcaraz, Torres and Aguilella2023), the structural models for both proteins agree in showing quite rigid channels that include very narrow, almost impermeable, sections (approximately 2–3 Å) with no significant lipid involvement in the overall structure. Nonetheless, there are some differences such as minor changes in the N-terminal side and in the tilting of TMD helices and more importantly, the hydrophobic Phe26 is facing the pore lumen in SARS-CoV-1 E (Surya et al., Reference Surya, Li and Torres2018), whereas it is facing the lipid acyl chains in SARS-CoV-2 E (Mandala et al., Reference Mandala, Mckay, Shcherbakov, Dregni, Kolocouris and Hong2020). For the sake of clarity, it should be mentioned that these structures of SARS-CoV-1 and SARS-CoV-2 E were obtained using truncated protein versions that differ in several charged residues outside the TM (Duart et al., Reference Duart, García-Murria and Mingarro2021). These differences could slightly influence membrane topology and/or channel structure via electrostatic interactions. Anyway, detailed computational analysis confirmed that both structures represent a hydrophobically occluded pore in which the transition from the closed to the open states is not observed throughout the simulations (Yang et al., Reference Yang, Wu, Wang, Rubino, Nickels and Cheng2022).

Subsequent work suggested that in response to lower pH or higher Ca²⁺ concentration, the Phe20 and Phe26 residues could reorient their side chains causing the aperture of the pore (Medeiros-Silva et al., Reference Medeiros-Silva, Somberg, Wang, Mckay, Mandala, Dregni and Hong2022; Medeiros-Silva et al., Reference Medeiros-Silva, Dregni, Somberg, Duan and Hong2023) (see Figure 3b, bottom). The ubiquitous presence of both, protons and calcium cations, is a common theme underlying SARS-CoV E research for two reasons. On the one hand, expression of the E protein increases intracellular-Golgi pH (Cabrera-Garcia et al., Reference Cabrera-Garcia, Bekdash, Abbott, Yazawa and Harrison2021) and alters Ca²⁺ homeostasis triggering the activation of the NLRP3 inflammasome leading to IL-1β overproduction (Nieto-Torres et al., Reference Nieto-Torres, Verdia-Baguena, Castano-Rodriguez, Aguilella and Enjuanes2015a). On the other hand, E-induced pentameric structures delineate narrow aqueous pores with a precise architecture that could be potentially suited to yield H⁺- and/or Ca²⁺-activated channels that operate in resemblance to canonical ion channels in neurons (Hille, Reference Hille2001; Delcour, Reference Delcour2015). Although both pieces of information seem the perfect match for each other, this line of reasoning could be an oversimplification of the actual situation if additional insights are considered.

Thus, experimental evidence obtained for both SARS-CoV-1 E and SARS-CoV-2 E supports that fully functional open channels can be obtained without acidification of the medium or the presence of calcium (Wilson et al., Reference Wilson, Mckinlay, Gage and Ewart2004; Parthasarathy et al., Reference Parthasarathy, Ng, Lin, Liu, Pervushin, Gong and Torres2008; Verdia-Baguena et al., Reference Verdia-Baguena, Aguilella, Queralt-Martin and Alcaraz2021; Surya et al., Reference Surya, Tavares-Neto, Sanchis, Queralt-Martín, Alcaraz, Torres and Aguilella2023). Furthermore, although experiments performed with SARS-CoV2 E in (Xia et al. Reference Xia, Shen, He, Pan, Liu, Wang, Yang, Fang, Wu, Duan, Zuo, Xie, Jiang, Xu, Chi, Li, Meng, Zhou, Zhou, Cheng, Xin, Jin, Zhang, Yu, Li, Feng, Chen, Jiang, Xiao, Zheng, Zhang, Shen, Li and Gao2021) were consistent with an amplitude and open probability that gradually increased when pH decreased, larger permeabilities were also found for monovalent cations than for divalent ones. In contrast, systematic experiments reported in Verdia-Baguena et al. (Reference Verdia-Baguena, Nieto-Torres, Alcaraz, Dediego, Enjuanes and Aguilella2013, Reference Verdia-Baguena, Aguilella, Queralt-Martin and Alcaraz2021)) indicate that both low pH and addition of calcium over salts of monovalent cations substantially decreased the channel conductance of SARS-CoV1 E. Thus, exhaustive electrophysiological characterization of SARS-CoV1 E and SARS-CoV2 E seems to indicate that these channels can transport different types of ions (Na⁺, K⁺, Ca²⁺, H⁺, and Cl⁻) with only mild selectivity and no ion specificity (Wilson et al., Reference Wilson, Mckinlay, Gage and Ewart2004; Verdia-Baguena et al., Reference Verdia-Baguena, Nieto-Torres, Alcaraz, Dediego, Enjuanes and Aguilella2013; Verdia-Baguena et al., Reference Verdia-Baguena, Aguilella, Queralt-Martin and Alcaraz2021). In this scenario, we may speculate that E-induced channels would allow multi-ionic transport so that their ion conduction depended on the overall composition of the cellular compartments that they connect (Appenzeller-Herzog and Hauri, Reference Appenzeller-Herzog and Hauri2006) and not on proton and calcium concentration gradients alone. Interestingly, recent studies suggest that SARS-CoV2 E could perturb calcium homeostasis not directly connecting diffusively compartments, but indirectly acting as an exoregulin affecting the SERCA (Berta et al., Reference Berta, Tordai, Lukacs, Papp, Enyedi, Padanyi and Hegedus2024).

The existence of a complex behavior regarding the mechanism of ion permeation and the role of Ca²⁺ in the SARS-CoV2 E-induced channels is also evident in the computational field. Thus, in contrast to recent models based on NMR spectroscopy data (Medeiros-Silva et al., Reference Medeiros-Silva, Somberg, Wang, Mckay, Mandala, Dregni and Hong2022; Medeiros-Silva et al., Reference Medeiros-Silva, Dregni, Somberg, Duan and Hong2023), some computational studies identify Leu10 and Phe19 as the hydrophobic gates of the SARS-CoV2 E channel (Cao et al., Reference Cao, Yang, Wang, Lee, Zhang, Zhang, Sun, Xu and Meng2020), showing also that the channel should be impermeable to divalent cations and relatively low permeable to monovalent ones. Interestingly, the collapse of the pentameric NMR structure into a closed configuration, expelling water from interior, has been reproduced by subsequent computational approaches using different membrane compositions, enhanced sampling methods, or extended equilibration with constraints (Mehregan et al., Reference Mehregan, Perez-Conesa, Zhuang, Elbahnsi, Pasini, Lindahl, Howard, Ulens and Delemotte2022; Cubisino et al., Reference Cubisino, Milenkovic, Conti-Nibali, Musso, Bonacci, De Pinto, Ceccarelli and Reina2024). However, recent molecular dynamics simulations suggest a new arrangement for the SARS-CoV2 E TMD monomers in the pore to effectively conduct ions (Cubisino et al., Reference Cubisino, Milenkovic, Conti-Nibali, Musso, Bonacci, De Pinto, Ceccarelli and Reina2024). Following a similar route to previous studies in which collapsed pore structures in α-helical barrels are turned into conductive ones (Scott et al., Reference Scott, Niitsu, Kratochvil, Lang, Sengel, Dawson, Mahendran, Mravic, Thomson, Brady, Liu, Mulholland, Bayley, Degrado, Wallace and Woolfson2021), each E TMD monomer was rotated 180^o so that the short amphipathic helices faced the interior of the pore and the long hydrophobic helix stretches oriented toward the exterior, in direct contact with the membrane (Cubisino et al., Reference Cubisino, Milenkovic, Conti-Nibali, Musso, Bonacci, De Pinto, Ceccarelli and Reina2024). Despite this new arrangement involved a U-turn in monomers forming the pore, calcium was decisive again acting as an amplifier of the ionic current, being the calculated in silico conductance almost two orders of magnitude higher when calcium was added over 50 mM KCl than in pure 150 mM KCl (Cubisino et al., Reference Cubisino, Milenkovic, Conti-Nibali, Musso, Bonacci, De Pinto, Ceccarelli and Reina2024). Again, these computational predictions are at odds with experimental results showing that Ca²⁺ added over KCl acts as a partial blocker and not as an enhancer (Verdia-Baguena et al., Reference Verdia-Baguena, Aguilella, Queralt-Martin and Alcaraz2021), mimicking the anomalous mole fraction effect observed in calcium selective channels and other nonspecific pores (Gillespie et al., Reference Gillespie, Boda, He, Apel and Siwy2008).

Up to this point, the discussion of the physiological role of the E protein usually has mainly included pentameric structures. However, a growing number of studies involving different techniques suggest that E-induced oligomerization may be heterogenous and dynamic, including pentamers but also other structures (from monomers or dimers to even decamers; Surya et al., Reference Surya, Tavares-Neto, Sanchis, Queralt-Martín, Alcaraz, Torres and Aguilella2023) that are crucially regulated by membrane composition in terms of membrane charge, cholesterol and other constituents (Somberg et al., Reference Somberg, Sucec, Medeiros-Silva, Jo, Beresis, Syed, Doudna and Hong2024; Volovik et al., Reference Volovik, Denieva, Gifer, Rakitina and Batishchev2024). In particular, electrophysiological recordings of both SARS-CoV1 E with SARS-CoV2 E show a large variability of conducting states (average conductance values are usually obtained via histograms of currents involving large error bars; Verdia-Baguena et al., Reference Verdia-Baguena, Nieto-Torres, Alcaraz, Dediego, Torres, Aguilella and Enjuanes2012; Verdia-Baguena et al., Reference Verdia-Baguena, Nieto-Torres, Alcaraz, Dediego, Enjuanes and Aguilella2013; Verdia-Baguena et al., Reference Verdia-Baguena, Aguilella, Queralt-Martin and Alcaraz2021; Surya et al., Reference Surya, Tavares-Neto, Sanchis, Queralt-Martín, Alcaraz, Torres and Aguilella2023) whose properties are strongly lipid-dependent (Aguilella et al., Reference Aguilella, Verdia-Baguena and Alcaraz2014). Interestingly, SARS-CoV1 E channels are more conductive in neutral lipids than in charged ones in concentrated solutions (Aguilella et al., Reference Aguilella, Verdia-Baguena and Alcaraz2014), what clashes with purely electrostatic arguments suggesting that charged lipids should induce ion accumulation in the channel mouths and hence increase channel conductance (Queralt-Martin et al., Reference Queralt-Martin, Lopez, Aguilella-Arzo, Aguilella and Alcaraz2018). This emphasizes the importance of protein–lipid interactions and particularly the role of the hydrophobic mismatch in pore formation (Grau-Campistany et al., Reference Grau-Campistany, Strandberg, Wadhwani, Reichert, Burck, Rabanal and Ulrich2015; Grau-Campistany et al., Reference Grau-Campistany, Strandberg, Wadhwani, Rabanal and Ulrich2016).

Although the critical influence of lipids on the conductive properties of SARS-CoV1 E with SARS-CoV2 E channels is beyond doubt, and it has been suggested that these pores have proteolipidic structure (Aguilella et al., Reference Aguilella, Verdia-Baguena and Alcaraz2014; Verdia-Baguena et al., Reference Verdia-Baguena, Nieto-Torres, Alcaraz, Dediego, Torres, Aguilella and Enjuanes2012), it is unclear what this term implies structurally. In the literature of pore-forming proteins, the term ‘proteolipidic’ usually refers to toroidal pores where the walls of the channel are formed by both protein monomers and lipid molecules in some sort of intercalated fashion (Gilbert et al., Reference Gilbert, Dalla Serra, Froelich, Wallace and Anderluh2014; Cosentino et al., Reference Cosentino, Ros and Garcia-Saez2016; Vandenabeele et al., Reference Vandenabeele, Bultynck and Savvides2023). Since none of the resolved pentameric structures of the E protein show significant lipid presence, we may speculate that proteolipidic pores correspond to arrangements involving a low number of monomers (Surya et al., Reference Surya, Tavares-Neto, Sanchis, Queralt-Martín, Alcaraz, Torres and Aguilella2023; Volovik et al., Reference Volovik, Denieva, Gifer, Rakitina and Batishchev2024), or alternatively, lipid molecules interact strongly with some specific residues (e.g., Phe19, Phe20, or Phe 26) that act as hydrophobic gates of the pore (Cao et al., Reference Cao, Yang, Wang, Lee, Zhang, Zhang, Sun, Xu and Meng2020).

Overall, it is unknown why high-resolution structural studies mainly report very tight pentameric E channel structures in closed configurations that are allegedly activated by particular stimuli (Ca²⁺ and H⁺), whereas most available functional data suggest a variety of E-triggered mildly selective pores in which lipid molecules may exert a tight modulation. We hypothesize that the existence of contrasting evidence could be a manifestation of the protein versatility and its capacity to play different functional roles (the so-called dualism). Such flexibility seems to be obvious in the case of SARS-CoV E protein, but, as discussed previously, appears subtly even in the case of IAV M2.

Indeed, in the general context of viroporin-mediated pathophysiological effects, the existence of dual mechanisms can be understood considering that the successive stages of the virus cycle may involve permeabilization mechanisms of entirely different complexity. Thus, hijacking of the host cell biosynthetic machinery and hindering immune responses during the early stages of viral infection may probably involve highly selective pores (usually narrow and poorly conductive (Aguilella et al., Reference Aguilella, Queralt-Martin, Aguilella-Arzo and Alcaraz2011) with specificity for certain ions (H⁺ and Ca²⁺) (Nieva et al., Reference Nieva, Madan and Carrasco2012), whereas more conductive and nonselective pores may be more suitable for later steps where indiscriminate membrane disruption and cell death occur (Nieto-Torres et al., Reference Nieto-Torres, Dediego, Verdia-Baguena, Jimenez-Guardeno, Regla-Nava, Fernandez-Delgado, Castano-Rodriguez, Alcaraz, Torres, Aguilella and Enjuanes2014; Largo et al., Reference Largo, Queralt-Martin, Carravilla, Nieva and Alcaraz2021). However, the dualism may also express the general concern in the field of pore-forming proteins about the physiological relevance of some results obtained either in cell culture experiments (where the protein under scrutiny may inadvertently coexist with other native pore-forming proteins) or after applying in vitro techniques that may not reproduce the actual membrane environment favoring the pore assembly (Cosentino et al., Reference Cosentino, Ros and Garcia-Saez2016). It has been also argued that the use of unrealistic high mole fractions of proteins in experiments may lead to artifactual membrane defects that allow polar molecules to cross the membrane nonspecifically (Delcour, Reference Delcour2015; Perera et al., Reference Perera, Ganesan, Siskind, Szulc, Bielawska, Bittman and Colombini2016).

Functional classification of viroporins: conventional versus unconventional ion channels

In an effort to overcome the confusion created when a given viroporin shows both traits corresponding to canonical bona fide ion channels and other characteristics suggesting unclear regulation, we have attempted a classification of viroporins taking into account the functional data available in the literature (Figure 4 and Tables 1 and 2). We adopted the terminology conventional/unconventional ion channels suggested by A. Delcour, remarking that the notion of ion channel is too restrictive when refers only to the conventional voltage-dependent channels of neurons (and similar ones), showing well-defined single channel conductance, a highly selectivity for particular ions, tightly gated by certain stimuli (ligands and/or voltage) and typically constructed as oligomers of α-helical segments (‘barrel staves’) forming a narrow aqueous pore (Delcour, Reference Delcour2015) (notably, these characteristics match the requirements commented above in Harrison et al., Reference Harrison, Abbott, Gentzsch, Aleksandrov, Moroni, Thiel, Grant, Nichols, Lester, Hartel, Shepard, Garcia and Yazawa2022, discussing whether certain viroporins are really ion channels). As we will discuss later on this section, many well-known pore-forming entities (i.e., bacterial and mitochondrial porins, toxins, and connexins) that share common traits with most viroporins do not meet all these criteria, so that all them can be grouped under the notion of unconventional ion channels (Delcour, Reference Delcour2015; Hyser and Estes, Reference Hyser and Estes2015; Syrjanen et al., Reference Syrjanen, Michalski, Kawate and Furukawa2021) rather than being in a gray zone of non-reproducible pores like those created by detergent-like mechanisms (Shai, Reference Shai1999).

Figure 4.

Mechanisms of membrane permeabilization by viroporins. Class IA IAV M2 exemplifies the case of a conventional channel, operated by the pH gradient and selective for protons. The structure of the conducting pore is stabilized through interactions of the helix bundle with the surrounding membrane lipids (model based on structure with PDB ID: 2L0J). Class IIB PV 2B, or more generally the 2B protein of enteroviruses, portraits features of a conventional channel that conducts Ca²⁺, but also behaves as a pore allowing free diffusion of solutes below approximately 1,000 Da, whose aperture seems to depend on anionic phospholipids bearing long, unsaturated acyl chains. 3D structure of PV 2B hairpin transmembrane domain derived from Alpha-Fold (Senior et al., Reference Senior, Evans, Jumper, Kirkpatrick, Sifre, Green, Qin, Zidek, Nelson, Bridgland, Penedones, Petersen, Simonyan, Crossan, Kohli, Jones, Silver, Kavukcuoglu and Hassabis2020). The more complex pore-forming domain of NoV NS3 assemble pores in the outer mitochondrial membrane that allow leakage of Cytochrome c to the cytosol (monomers based on the PDB ID: 4BTF structure). In analogy with the mitochondrial apoptosis-induced channel, the model proposes the release of the protein (depicted in magenta) through toroidal proteolipidic megapores that depend on cardiolipin (Vandenabeele et al., Reference Vandenabeele, Bultynck and Savvides2023).

Figure 4 depicts a gradient that goes from conventional channels to unconventional ones as regards gating, selectivity, and lipid implication in the pore structure of viroporins. Our figure suggests that because of their dual character, most viroporins occupy a continuous portion of the spectrum, rather than a well-defined position in it that would allow a more precise classification. At one end of the spectrum, we have placed IAV M2, the paradigm of bona fide ion channel regarding the specificity of the transported ion, the operation of a gating mechanism, and the adoption of robust, functional structures irrespective of the lipid matrix. At the other end, we propose NoV NS3 as an example of viroporin displaying the capacity to generate macropores (toroidal pores), in this case within the outer membrane of mitochondria, which, following a mechanism akin to that displayed by proteins of the BAX/BAK/BOK family (Vandenabeele et al., Reference Vandenabeele, Bultynck and Savvides2023), includes lipids of high curvature. Somewhere in the middle, we have placed PV 2B or, more generally, non-structural 2B proteins from picornaviruses, which are simultaneously described as ion channels displaying specificity for the transported ion (i.e., Ca²⁺), or pores with a size cutoff of approximately 1,000 Da for the permeant solutes.

Tables 1 and 2 show the characteristic features of conventional and unconventional channels, respectively, with some viroporins that match one or the other category. The characterization of these tables aims to provide relevant examples to understand the concepts rather than attempting an exhaustive classification, in this sense some excellent reviews are available (Hyser and Estes, Reference Hyser and Estes2015; To et al., Reference To, Surya and Torres2016; Devantier et al., Reference Devantier, Kjaer, Griffin, Kragelund and Rosenkilde2024). From the point of view of transport phenomena, differences between highly selective and specific channels like M2 and other weakly selective viroporins such as SARS-CoV2 E lie on the compromise between two complementary properties, permeability (how many molecules flow through the system and how fast they do it) and selectivity (how a desired molecule is separated from the rest) (Hille, Reference Hille2001; Aguilella et al., Reference Aguilella, Queralt-Martin, Aguilella-Arzo and Alcaraz2011). However, this balance cannot be understood only in terms of pore size: the ability to display specific selectivity for a particular specie requires highly sophisticated structure-based physicochemical mechanisms. For example, aquaporins that allow flux of water molecules in cells while being impermeable to protons (Agre et al., Reference Agre, Preston, Smith, Jung, Raina, Moon, Guggino and Nielsen1993) operate by a combination of hydrophobic effects and precise steric restraints (Hub and De Groot, Reference Hub and De Groot2008). Also, in potassium channels, coordination of the protein channel carbonyl group with unsolvated permeant ions is essential to achieve 1,000-fold K⁺/Na⁺ selectivity (Noskov et al., Reference Noskov, Berneche and Roux2004). Since the majority of viroporins are relatively small proteins with almost no conserved motifs characteristic of specific canonical channels (with the exception of the chlorovirus Kcv potassium channel; Gazzarrini et al., Reference Gazzarrini, Severino, Lombardi, Morandi, Difrancesco, Van Etten, Thiel and Moroni2003), one could expect a very limited amount of specific viroporin ion channels. As seen in Table 1, only few examples are available apart from Kcv, all of them specific to protons such as IAV M2 and HPV16 E5, configuring pH-gated channels. These sophisticated structures that allow specificity are well-defined oligomeric protein structures where lipid molecules have no significant presence (Devantier et al., Reference Devantier, Kjaer, Griffin, Kragelund and Rosenkilde2024; Hyser, Reference Hyser and Delcour2015). Accordingly, membrane composition has limited impact on viroporin performance as ion channel (see Table 1). Also, the existence of critical conserved motifs in the amino acid sequence allows to identify both mutations inhibiting the channel function and a specific pharmacology in terms of channel blocker (see Table 1). However, even in these conventional channels, contradictory information appears, such as IAV M2 conducting alkali cations (Tosteson et al., Reference Tosteson, Pinto, Holsinger and Lamb1994; Stauffer et al., Reference Stauffer, Feng, Nebioglu, Heilig, Picotti and Helenius2014), or the alleged behavior of HCV p7 as a proton channel (given the presence of a highly conserved motif similar to the M2 selectivity filter; Hyser, Reference Hyser and Delcour2015), whereas conduction of monovalent and divalent cations is observed (Premkumar et al., Reference Premkumar, Wilson, Ewart and Gage2004; Montserret et al., Reference Montserret, Saint, Vanbelle, Salvay, Simorre, Ebel, Sapay, Renisio, Bockmann, Steinmann, Pietschmann, Dubuisson, Chipot and Penin2010) (see Table 2).

Accordingly, having a narrow pore is a necessary but not sufficient condition to observe specific ion selectivity, extremely precise structural arrangements are necessary. This idea is crucial to understand why tireless structure-based and computational studies trying to turn viroporins into calcium-specific ion channels find extreme difficulties: viroporins described up to date can conduct calcium together with other ions but not while excluding any other ions as canonical calcium channels do (Wu et al., Reference Wu, Yan, Li, Qian, Lu, Dong, Zhou and Yan2016; see Table 2) (see also Hyser and Estes, Reference Hyser and Estes2015; Verdia-Baguena et al., Reference Verdia-Baguena, Aguilella, Queralt-Martin and Alcaraz2021). As mentioned above, a relevant example in this sense are molecular dynamic simulations trying to turn closed structures of SARS-CoV E into calcium-activated channels (Cubisino et al., Reference Cubisino, Milenkovic, Conti-Nibali, Musso, Bonacci, De Pinto, Ceccarelli and Reina2024), where it is acknowledged that calcium conducting mechanisms do not present any regularity compared to the Ca²⁺ specific selectivity of calcium channels (Liu et al., Reference Liu, Zhang, Yan and Song2021a).

In contrast to specialized channels such as IAV M2, chlorovirus Kcv, and HPV16 E5, other viroporins can display high conductive levels, weak ionic selectivity, permeability to small solutes and ohmic (not rectifying) conduction (see Table 2). Such traits usually involve pore diameters that can be significantly larger than for conventional channels so that they have been named ‘large-pore channels’ (Syrjanen et al., Reference Syrjanen, Michalski, Kawate and Furukawa2021) in a dramatic twist of the channel/pore dichotomy. Their function can be described embracing the notion of unconventional ion channels, meaning that their regulatory mechanisms can be unusually complex compared to the conventional channels, involving multiple bioelectrochemical stimuli (diverse salt ions, membrane constituents, small solutes, solution pH, metabolites, drugs, etc.) that operate together in a concerted manner coupled by electrochemical gradients and electroneutrality requirements. Within this line of reasoning, the large heterogeneity in the conductive levels and weak selectivity found in most studies of viroporins (see Table 2) may not arise from failed attempts to achieve a ‘canonical’ homo-oligomeric conformation, but are probably linked to the capacity of viroporins to form a variety of hetero-oligomeric structures with flexible architectures in which even membrane lipids may be involved (Verdia-Baguena et al., Reference Verdia-Baguena, Nieto-Torres, Alcaraz, Dediego, Enjuanes and Aguilella2013; Cosentino et al., Reference Cosentino, Ros and Garcia-Saez2016). Well-characterized examples of viroporins support this structural heterogeneity, for example, HCV p7, which has been observed to form hexamers and heptamers (Clarke et al., Reference Clarke, Griffin, Beales, Gelais, Burgess, Harris and Rowlands2006; Ouyang et al., Reference Ouyang, Xie, Berardi, Zhao, Dev, Yu, Sun and Chou2013).

Likewise, the absence of a marked ion discrimination capacity in relatively highly conductive structures could be reasonably associated with the physiological role of facilitating the exchange of small solutes, including metabolites, between different cellular compartments. Relevant examples of such unconventional behavior can be found in maltoporin (LamB), a maltodextrin transport-channel in which the static (different levels) and dynamic (variations with time) disorder found in ion conductance values does not affect sugar binding specificity (Bezrukov et al., Reference Bezrukov, Kullman and Winterhalter2000; Kullman et al., Reference Kullman, Gurnev, Winterhalter and Bezrukov2006). Also, bacterial porins, mitochondrial channels, and connexin gap junctions that show weak ionic selectivity may display sophisticated specificity for antibiotics and small metabolites like ATP (Rostovtseva et al., Reference Rostovtseva, Komarov, Bezrukov and Colombini2002; Danelon et al., Reference Danelon, Nestorovich, Winterhalter, Ceccarelli and Bezrukov2006; Alcaraz et al., Reference Alcaraz, Nestorovich, Lopez, Garcia-Gimenez, Bezrukov and Aguilella2009; Syrjanen et al., Reference Syrjanen, Michalski, Kawate and Furukawa2021).

One of the factors that blurs the distinction between conventional and unconventional channels is understanding how selected mutations abrogate channel function. In canonical viroporins, non-functional mutants arise from the existence of a well-defined structure and a detailed knowledge of the mechanism of pore function. For instance, in IAV M2, the identification of H37xxxW41 motif as responsible for its gating and proton selectivity is used to engineer non-functional mutants (Balannik et al., Reference Balannik, Carnevale, Fiorin, Levine, Lamb, Klein, Degrado and Pinto2010). Non-functional mutants have also been found in the case of SARS-CoV1 E and SARS-CoV2 E (Surya et al., Reference Surya, Tavares-Neto, Sanchis, Queralt-Martín, Alcaraz, Torres and Aguilella2023; Verdia-Baguena et al., Reference Verdia-Baguena, Nieto-Torres, Alcaraz, Dediego, Torres, Aguilella and Enjuanes2012). However, quite in contrast to IAV M2, it is intriguing how profound is the impact that single mutations such as N15A and V25F exert on the activity of these viroporins, particularly when having in mind the versatility of the CoV E protein (Verdia-Baguena et al., Reference Verdia-Baguena, Aguilella, Queralt-Martin and Alcaraz2021; Surya et al., Reference Surya, Tavares-Neto, Sanchis, Queralt-Martín, Alcaraz, Torres and Aguilella2023) and the ability of lipid molecules to stabilize different proteolipidic conformations in the bilayer (Grau-Campistany et al., Reference Grau-Campistany, Strandberg, Wadhwani, Rabanal and Ulrich2016; Largo et al., Reference Largo, Verdia-Baguena, Aguilella, Nieva and Alcaraz2016; Perini et al., Reference Perini, Aguilella-Arzo, Alcaraz, Peralvarez-Marin and Queralt-Martin2022). Of note, it must be pointed out that problems in the rationalization of functional experiments with mutants involving amino acid replacements may have an origin in effects induced by collateral impacts on protein expression and/or localization and membrane insertion. Last but not least, Tables 1 and 2 also reflect the sharp contrast between the well-defined specific pharmacology (see Devantier et al., Reference Devantier, Kjaer, Griffin, Kragelund and Rosenkilde2024, for extensive details) of conventional channels like IAV M2 (Cady et al., Reference Cady, Luo, Hu and Hong2009) and the use of broad-spectrum channel inhibitors like in unconventional ones SARS-CoV1 E and SARS-CoV 3a (Nieto-Torres et al., Reference Nieto-Torres, Verdia-Baguena, Jimenez-Guardeno, Regla-Nava, Castano-Rodriguez, Fernandez-Delgado, Torres, Aguilella and Enjuanes2015b). To date, the pharmaceutical approach to viroporins has targeted the direct blocking of channel activity or the disruption of protein oligomerization. In most cases, this has been achieved with generic inhibitors that show low potency and a variety of unwanted (or unknown) adverse effects. In this sense, remarkable efforts to develop viroporin-targeting drugs, are described in a recent review (Devantier et al., Reference Devantier, Kjaer, Griffin, Kragelund and Rosenkilde2024).