Inhibitors of protein translocation across membranes of the secretory pathway: novel antimicrobial and anticancer agents
- 722 Downloads
Proteins routed to the secretory pathway start their journey by being transported across biological membranes, such as the endoplasmic reticulum. The essential nature of this protein translocation process has led to the evolution of several factors that specifically target the translocon and block translocation. In this review, various translocation pathways are discussed together with known inhibitors of translocation. Properties of signal peptide-specific systems are highlighted for the development of new therapeutic and antimicrobial applications, as compounds can target signal peptides from either host cells or pathogens and thereby selectively prevent translocation of those specific proteins. Broad inhibition of translocation is also an interesting target for the development of new anticancer drugs because cancer cells heavily depend on efficient protein translocation into the endoplasmic reticulum to support their fast growth.
KeywordsProtein translocation Signal peptide Translocon Endoplasmic reticulum Translocation inhibitor Sec61 SecY
Binding immunoglobulin protein
Corticotropin-releasing factor receptor 1
Endoplasmic reticulum-associated protein degradation
Guided entry of tail-anchored proteins
Human herpes virus-7
Human immunodeficiency virus
Heat shock cognate 70
Heat shock protein 70
Intercellular adhesion molecule 1
Proton motive force
Ribosome-nascent chain complex
Signal peptidase complex
Signal recognition particle
Small secreted protein
Tail-anchored membrane protein
Tumor necrosis factor alpha
Translocating chain-associated membrane protein
Transmembrane domain recognition complex
Unfolded protein response
Vascular cell adhesion molecule
Different signal peptide-dependent translocation pathways
More than 30% of all human genes encode proteins destined for the extracellular environment, cell membrane or components of the secretory pathway . Since protein synthesis occurs in the cytosol, translocation of proteins across biological membranes is essential for cellular function. Multiple pathways of protein translocation have been proposed. In general, translocation from the cytosol requires three key steps: (1) substrate recognition and targeting to the destination membrane, while maintaining the substrate in a translocation-competent state (2) translocation across or integration into that membrane, which usually requires energy expenditure in the form of GTP, ATP or proton motive force and (3) release, folding and maturation of the protein substrate.
Targeting signals contain the principal information that drives protein translocation. They direct newly synthesized proteins to their target membrane for translocation or membrane integration . Signals present at the N-terminus of the synthesized protein are termed signal peptides (SP) or signal sequences. SPs are cleaved from the mature protein after translocation by the signal peptidase complex (SPC) and are characterized by a short (8–12 amino acids) hydrophobic segment, but their length and amino acid composition are highly divergent [3, 4]. Alternatively, targeting is facilitated by uncleaved amino-terminal signals termed signal anchors (SA). Signal anchors can act as a transmembrane segment and usually contain about 20 (or more) hydrophobic residues, a length required to physically span the approximately 3 nm wide hydrophobic interior of the phospholipid bilayer in an α-helical fold . A third distinction is made for a class of proteins called tail-anchored membrane proteins (TA proteins), which have a single hydrophobic transmembrane region at their C-terminus that acts both as a targeting signal and membrane anchor .
The conserved Sec-dependent pathway is used for the translocation of most eukaryotic proteins . The central component of this system is the heterotrimeric Sec61 translocon complex, also known as the SecY complex in bacteria and archaea . It forms an aqueous channel in the endoplasmic reticulum (ER) membrane which allows protein transport across the membrane and it facilitates insertion of hydrophobic protein segments into the lipid bilayer. The Sec pathway operates in two major modes of translocation: (1) co-translational translocation couples the ribosomal protein synthesis directly to translocation through the channel, which efficiently uses the energy from mRNA translation in ribosomes to drive protein translocation across the membrane, (2) while post-translational translocation delivers completely synthesized polypeptide chains to the membrane which is best understood in fungi and bacteria. Mitochondrial, chloroplast and peroxisomal protein import, as well as specialized bacterial secretion systems (suggested reviews: [9, 10, 11, 12]) are not discussed here, but the general concepts of signal peptide-dependent translocation (use of targeting signals and specialized protein-conducting channels) are conserved in these systems.
The process of co-translational translocation is a multistep sequence that depends on dynamic interactions between many factors. For most of these steps, natural and synthetic inhibitors have been discovered that usually affect translocation of a broad range of co-translational substrates. However, some compounds are able to operate in a signal peptide-selective way.
The Sec translocon is a dynamic protein complex
The Sec61 complex consists of a central Sec61α subunit (referred to as SecY in bacteria and archaea) which forms the channel, and two smaller peripheral subunits Sec61β and Sec61γ . Sec61γ is homologous to SecE in bacteria and archaea, but Sec61β shows little homology to the bacterial SecG subunit. Sec61α contains ten transmembrane helices divided into two halves of the channel (TM 1-5 and TM 6-10) with a hinge point between TM helix 5 and 6, often referred to as a ‘clam shell’ design. The ‘lateral gate’ of Sec61α, formed by the interface of transmembrane helices 2 and 7, allows opening of the channel towards the lipid bilayer for lipid insertion of transmembrane domains . It also serves as the recognition site of signal peptides  and allows hydrophobic peptide region access to the lipid layer . The inside of the channel is hourglass-shaped, with a ring at the center consisting of six bulky hydrophobic amino acid residues (the pore ring) which position their side chains to the center of the pore. This ring prevents leakage of ions through the inactive channel and during translocation of a protein substrate. The lumenal side of the closed channel is occupied by a short helix (TM2a) called the plug domain.
The translocon provides a dynamic interface between the water filled inside of the channel and the lipid environment. Hence, most eukaryotic membrane proteins with a (trans)membrane domain (such as a hydrophobic α-helix) are inserted into the ER membrane during co-translational translocation. In the current understanding of membrane integration, individual TM segments insert sequentially in the membrane layer through the lateral gate of Sec61 . Furthermore, the channel can accommodate several TM helices at the same time, and facilitates early folding of these segments before release into the membrane . Multiple accessory factors are dynamically recruited to the translocon and assist the Sec channel (Fig. 1), e.g., through the chaperoning functions of translocon-associated protein (TRAP) and translocating chain-associated membrane protein (TRAM). Other factors associate with the translocon to perform post-translational modification of the peptide substrates: oligosaccharyl–transferase (OST) complex facilitates N-linked glycosylation, while the SPC cleaves signal peptides from the mature protein.
Non-selective inhibition of RNC transfer to the translocon
Inhibitors of translocon gating
The polyketide macrolide mycolactone (Fig. 2) is a virulence factor produced by the human pathogen Mycobacterium ulcerans and causes necrotizing lesions of the skin without acute inflammation . Hall et al. showed that mycolactone is a non-selective inhibitor of Sec61-dependent translocation across the ER . Additionally, the inhibitory effect on cells was irreversible, indicative of a high affinity binding. The lack of immune response to this molecule is thus due to its suppression of inflammatory cytokine and receptor production in immune cells, and due to an indirect inhibition of antigen cross-presentation . The eukaryotic Sec61 channel was recently identified as the target of mycolactone. Chemical crosslinking data suggest that the compound induces a conformational change of the channel that significantly disturbs co-translational translocation efficiency, but has less impact on post-translational translocation substrates .
The Pseudomonas aeruginosa protein exotoxin A is a cytotoxic ADP-ribosyltransferase that enters the eukaryotic cytosol trough retrograde transport and inhibits retrograde export of immunogenic peptides from the ER towards the cytosol. It binds to Sec61α and prevents both co- and post-translational translocation [32, 33]. Exotoxin A also competes with cytosolic protein calmodulin (CaM) for binding to an N-terminal IQ motif on Sec61α and prevents Ca2+ leakage through the channel in human cells . These observations suggest that the protein keeps the Sec61 channel in a closed state.
A group of cyclic heptadepsipeptides are derived from the fungal macrocycle HUN-7293. The latter inhibits expression of three endothelial cell adhesion molecules: intercellular adhesion molecule 1 (ICAM-1), vascular cell adhesion molecule (VCAM-1) and E-selectin . One derivative called cotransin (Fig. 2) was shown to inhibit the co-translational translocation of several proteins into the ER, in a signal peptide-selective way . These initial studies reported inhibition of VCAM-1, P-selectin, angiotensinogen, β-lactamase, and corticotropin-releasing factor receptor 1 (CRF-R-1). Later studies also identified endothelin B receptor , human epidermal growth factor receptor 3  and tumor necrosis factor alpha (TNFα) , a type II integral membrane protein with uncleaved signal anchor, as targets of cotransin.
Cotransin does not affect SRP recognition or targeting, but prevents access of NCs to the ER lumen, suggesting that the compound inhibits signal peptide-dependent gating of the Sec61 channel (Fig. 1). Accessory translocon factors such as TRAP, TRAM, Sec62/63 and binding immunoglobulin protein (BiP) are not required for cotransin activity, as the compound was able to selectively prevent translocation of VCAM-1 in minimal proteoliposomes (containing only Sec61 and SR) . Garrison et al. suggested that cotransin either stabilizes the channel in a closed conformation or that it allosterically alters the signal peptide binding site of Sec61. These hypotheses, respectively, restrict productive interaction of low-affinity SPs or decrease the SP binding site flexibility, which both result in substrate selection at the translocon.
It must be noted that the reported compound concentrations used in the different translocation assays varies widely, which is important for the interpretation of the selectivity concept. For example, cotransin operates selectively at low nanomolar concentrations . In contrast, Klein et al. have recently shown that a saturating concentration of cotransin (30 µM) actually inhibits translocation of a broad range of secreted proteins, while integral membrane proteins are mostly unaffected .
Decatransin is a fungal cyclic decadepsipeptide (Fig. 2) that prevents growth of human carcinoma cells . It is synthesized by a non-ribosomal peptide synthetase. Such very large modular enzymes are often used by microorganisms to produce complex secondary metabolites . Decatransin prevents Sec61/SecY-dependent co- and post-translational translocation into the ER lumen but does not affect SRP recognition or SR targeting .
Apratoxins are natural secondary metabolites isolated from a marine cyanobacterium. They are also produced by a non-ribosomal peptide synthetase . The cyclic depsipeptide apratoxin A (Fig. 2) was discovered as a cytotoxic antitumor drug and prevents growth of various cancer cell lines by inducing G1 cell cycle arrest and apoptosis [43, 44]. Proteomic data showed that expression of a subset of secreted and membrane proteins are downregulated by apratoxin A, and this effect was due to an inhibition of their co-translational translocation .
Gating inhibitors likely operate through a common mechanism
Mutations in Sec61 provide cross-resistance to gating inhibitors
Accordingly, resistance mutations for apratoxin A were all located near the Sec61 plug domain and several of these mutations confer cross-resistance against cotransin . Apratoxin A also competes with cotransin for translocon binding, suggesting a mutually exclusive binding site near the lumenal plug region. Compared to cotransin, apratoxin A does not arrest the TNFα TMD in a preferred orientation on the cytosolic side of Sec61. This suggests that the compound blocks TNFα translocation in an earlier step, before TMD insertion occurs .
Decatransin might act similar to cotransin, as several (but not all) mutations in Sec61 that provide resistance to cotransin also confer decatransin resistance. Additionally, both decatransin and CP2, a compound very similar to HUN-7293/cotransin, inhibit translocation in yeast Sec61α and bacterial SecY , suggesting that the translocon might contain a universal binding site for natural translocation inhibitors. Junne et al. hypothesized that these hydrophobic, peptide-like molecules mimic signal peptides and can bind to the Sec translocon, where they block incoming peptides and stabilize the closed translocon conformation .
Mycolactone showed a broad effect not only on co-translational translocation but also dose-dependently competed with a cotransin variant (CT7) for translocon binding suggesting that they likely share a binding site . When tested for cross-resistance in the cotransin-resistant cell lines, several mutations near the lumenal plug indeed provided mycolactone resistance (Fig. 3).
Gating inhibitors stabilize the translocon in a closed state
The prl (protein localization) phenotype of bacteria and yeast, characterized by the ability to translocate defective signal peptides, has historically been associated with mutations in the translocon plug and pore ring, as these mutations destabilize the closed translocon conformation [50, 51]. Interestingly, the observed resistance mutations for HUN-7293 derivatives, decatransin, apratoxin and mycolactone also map to residues located on the lumenal side of the translocon lateral gate, near the plug domain and thus resemble the prl phenotype. The resistance mutations are believed to increase the flexibility of the translocon, which overrules the action of the inhibitors, i.e., keeping the translocon in a closed state . Paatero et al. suggested that these natural and synthetic compounds all target a similar binding site on the Sec61α translocon to regulate translocation (Fig. 3). Due to significant differences in their chemical structure and potency though, each of these natural compounds produces a distinct inhibitory profile .
Since the translocon machinery evolved to accept a broad range of targeting signals, it is surprising to see a selective inhibition of only a small subset of translocation substrates for some of the inhibitors. Nevertheless, such a selective suppression of cell surface receptor expression can have many novel therapeutic and antiviral applications. Hegde and Kang proposed that, in absence of accessory translocon components, the interaction between Sec61 and most signal peptides is intrinsically unstable . Only a limited set of ‘strong’ sequences are able to engage the channel on their own, for a sufficiently long duration to allow for complete insertion of the elongating chain, and subsequent translocation of the protein. All other targeting signals (‘weak’ sequences) are assumed to have a very low basal translocation efficiency. In this model, the occurrence of many Sec61-associated protein complexes (e.g., TRAP, TRAM, Sec62/63, BiP and OST) that dynamically assist the translocon, is the key concept that regulates substrate-specific translocation efficiency. Selective inhibition of translocation is possible too, either through (allosteric) destabilization of channel/signal peptide interactions or through stabilization of the translocon channel in a closed state. Additional regulatory factors likely exist, as the current knowledge is mostly obtained from a very limited set of model translocation systems (e.g., dog pancreas cells, model yeast and bacterial systems). Properties of Sec-dependent signal peptides are thus linked to the observed variation in selectivity (the inhibitory profiles) of these gating inhibitors.
Selective inhibition of signal peptide topology inversion
The small molecule cyclotriazadisulfonamide (CADA) is a synthetic macrocycle (Fig. 2) that showed antiviral activity against a broad range of human immunodeficiency virus (HIV) strains and human herpes virus-7 (HHV-7). Analysis of receptor expression showed that CADA treatment induced a downregulation of the cell surface- and intracellular CD4 levels. Since both HIV and HHV-7 use CD4 as the primary receptor for cell entry, this down-modulation of CD4 is responsible for CADA’s inhibition of viral entry .
CADA selectively inhibits the cell surface expression of human CD4 (hCD4) on a post-transcriptional level. The compound interacts with the signal peptide of hCD4 and it prevents co-translational translocation of the pre-protein chain across the ER membrane . Instead, the affected precursor protein chains end up in the cytosol where they are degraded by the proteasome. Targeting of RNCs to the ER translocon is not affected by the compound. The hCD4 signal peptide initially inserts head-on (Nexo/Ccyt) into the translocon, and inverts to a looped topology (Ncyt/Cexo) upon chain elongation (transitioning, Fig. 1). Early models assumed that signal peptides generally insert into the translocon with a looped topology, but head-on insertion and a dynamic topology inversion has been described for signal anchor proteins [56, 57]. Vermeire et al. suggest that CADA interferes with the mandatory inversion (transitioning) of hCD4’s signal peptide, and thereby prevents translocation of the chain.
Furthermore, the down-modulating effect of CADA seems to be selective for hCD4 and the membrane glycoprotein sortilin. As previously shown for hCD4, CADA also inhibits the co-translational translocation of sortilin in a signal peptide-dependent way . The effects of CADA on sortilin were less pronounced as compared to hCD4 though, and sortilin appears to be a secondary substrate of CADA. Importantly, expression of the homologous mouse CD4 protein is not affected by CADA. Mutagenesis of the CD4 signal peptide identified the central hydrophobic signal peptide region as crucial for CADA sensitivity, with a lesser contribution from the C-terminal SP region .
The binding site of CADA is not known, but quantitative structure–activity relationship studies of CADA analogues suggest a two-site binding model for these compounds [59, 60], and surface plasmon resonance experiments showed a selective, non-covalent interaction between CADA and the human CD4 signal peptide .
Another HUN-7293 derivative, CAM741 (Fig. 2), also interferes with the co-translational translocation of only a limited set of substrates in a signal peptide-selective way. VCAM-1 [61, 62] and VEGF  are reported CAM741 targets. Blocked polypeptide chains are directed towards the cytosol and degraded by the proteasome [36, 61]. Their cytosolic accumulation also induces the unfolded protein response . The compound CAM741 prevents correct insertion of the VCAM-1 SP into the translocon . Using a diagnostic amino-terminal glycosylation tag, the topology of the VCAM-1 SP was determined during the early post-targeting phase: it initially inserts head-on (Nexo/Ccyt) into the translocon channel and reorients upon polypeptide chain elongation. However, the amino-terminus does not enter the ER lumen in the presence of CAM741. Systematic analysis of VCAM-1 signal peptide mutants identified residues in the SP C-region, h-region and the first residue of the VCAM-1 mature domain as crucial elements for full CAM741 sensitivity . VEGF-1 mutagenesis revealed a different pattern, as mutagenesis of leucines in the N-terminal SP region and hydrophobic residues in the h-region resulted in a loss of compound sensitivity .
Signal peptides with increased hydrophobicity are able to escape CAM741 activity, while reducing the hydrophobicity of VCAM-1 and VEGF-1 SP’s h-region increases the inhibitory effect of CAM741 and vice versa [62, 63]. Hydrophobicity is a major determinant of TM segment recognition at the translocon , and signal peptide recognition at the translocon is also dependent on hydrophobic interactions. Targeting signals were shown to interact with a specific hydrophobic patch on the cytosolic side of Sec61, after which they intercalate between the channel helices . Mutations in the signal peptide that increase its hydrophobicity are thus better at opening the channel (they are “stronger” signal peptides according to Hegde’s theorem of translational regulation ) and this should allow them to overcome the translocational block imposed by CAM741 more easily.
After more than 30 years of study, it is clear that SRP-dependent co-translational translocation is highly efficient. However, it is now also evident that a significant fraction of all ER-targeted proteins do not utilize the SRP pathway for targeting, and/or do not even rely on the Sec61 translocon for translocation [65, 66, 67]. Membrane proteins with large translocated domains or multiple TM domains are usually constrained to the SRP-dependent co-translational pathway, but other physical limitations can restrict SRP recognition. In bacteria and yeast, SRP recognition requires targeting signals with sufficiently high hydrophobicity. SPs with low or moderate hydrophobicity are targeted towards the Sec61-dependent post-translational pathway . Mammalian SRP does not differentiate between SPs of different hydrophobicity, while microorganisms likely favor post-translational translocation to support higher rates of protein secretion, as this pathway does not use up the limited pool of ribosomes .
Another one of these cytosolic factors is the calcium-binding protein calmodulin, that was shown to maintain the small protein preprocecropin A (64 amino acids) in a translocation-competent state inside the cytosol, by selective binding to the signal peptide . CaM also binds to the cytosolic N terminus of Sec61α, which is proposed as the targeting mechanism for these peptides.
In a screen for heat shock cognate 70 (Hsc70, a member of the Hsp70 chaperone family)-interacting compounds, NSC 630668-R/1 (referred to as R/1) was identified as an inhibitor of Hsc70 ATPase activity (Fig. 4). R/1 almost completely inhibits in vitro post-translational translocation of pre-pro-α-factor (ppαF) in yeast microsomes at a concentration of 300 µM, with an IC50 of 6 µM .
Ophiobolin A and E6 Berbamine are specific antagonists of CaM, and thus disrupt the translocation competence of small-secreted proteins (SSPs) that depend on CaM during the targeting phase towards the ER .
Driving forces in post-translational translocation
The polymerization of peptides in the ribosome provides a directional driving force during co-translational translation. GTP is hydrolyzed during the ribosomal chain elongation step, which pushes the preprotein through the ribosomal exit tunnel and translocon pore  (Fig. 1). However, roughly 70 residues still remain inside the tunnel and channel after completion of translation  and this chain can move freely up and down through the channel. Cells employ a secondary mechanism to complete the translocation of this free-moving substrate: after sufficient downwards diffusion (due to Brownian motion) of the chain through Sec61 and towards the lumen, the chaperone BiP can bind to the exposed chain segment. Once bound to BiP, the chain segment cannot diffuse back to the cytosolic side. This starts a cycle, where stepwise binding of additional BiP molecules acts as a ‘molecular ratchet’ that pulls nascent chains towards the lumen [75, 76] (Fig. 1).
Sec61-dependent post-translational translocation (Fig. 4) requires the Sec62/63 complex and the lumenal BiP chaperone for directional protein movement through the translocon channel towards the lumen , similar to the function of BiP in co-translational translocation. Sec62 is an integral membrane protein that forms a stable complex with Sec63 and associates with ER-bound ribosomes, where it binds near the ribosomal exit tunnel  (Fig. 1). These proteins are highly abundant in the mammalian ER . A study by Reithinger et al. used the yeast model system to show that Sec62 is required in addition to SRP for the targeting and translocation of uncleaved signal anchor sequences with moderate hydrophobicity . Sec62 is also required for post-translational translocation of SSPs, an important class of SRP-independent proteins . Due to their small size (< 160 aa), the signal peptide of these SSPs remains (partially) buried inside the ribosomal exit tunnel after completion of translation, which prevents recognition by SRP . Additionally, Sec62 is suggested to function as a receptor that detects Ca2+ leakage through Sec61 and facilitates CaM recruitment , which in turn mediates Ca2+-dependent closure of the channel . Interestingly, the different functional studies and the location of Sec62 near the ribosome suggests that both (co- and post-) translocational systems overlap at this site.
The BiP chaperone has multiple functions during both co- and post-translational translocation
ER-resident Hsp40/DnaJ-like proteins ERdj1 and Sec63 (ERdj2) are integral membrane proteins that recruit the lumenal Hsp70 member BiP (termed Kar2p in yeast) to the translocon (Figs. 1, 4a) [77, 84, 85, 86]. BiP consists of a substrate binding domain and a nucleotide-binding domain and can perform several functions during protein translocation: (1) BiP is proposed to drive translocon gating from the closed to the open state during early stages of translocation, as it was found to assist with insertion of precursor peptides into the channel . The chaperone can bind to loop 7 of Sec61α, which forms the hinge region between both halves of the channel, and this binding energy facilitates insertion of ‘weak’ nascent peptides which are otherwise unable to open the channel by themselves [7, 88]. (2) It closes off the lumenal side of Sec61 and returns the channel to a closed state to prevent Ca2+ efflux after translocation has ended [88, 89]. (3) BiP acts as a molecular ratchet to complete translocation of pre-proteins in both co- and post-translational Sec61-dependent translocation (Figs. 1, 4a) [76, 90]. The substrate-binding domain of BiP has affinity for unfolded hydrophobic oligopeptides  and this affinity is regulated by its nucleotide-binding domain. The ATP-bound state of BiP has low substrate affinity while the ADP-bound state has high affinity . Recruitment of BiP to the translocon complex allows Sec63 and ERdj1 to activate BiP’s ATPase activity with their lumenal J domain. This converts the bound ATP to ADP and thus increases the substrate binding affinity, which allows BiP to bind peptides as they emerge from the translocon. After completion of translocation, release of bound proteins from BiP requires exchange of ADP for ATP, which is performed by two lumenal nucleotide exchange factors: Grp170 and Sil1 .
SecA-dependent post-translational translocation
Bacteria (and chloroplasts in plants) uniquely contain SecA as part of their translocation systems. SecA recognizes targeting signals in cytoplasmic pre-proteins and functions as an essential motor protein that mechanically drives post-translational translocation, as it uses sequential cycles of substrate clamping and ATP-dependent domain movement to push the preprotein through the SecYEG channel [14, 93]. Interestingly, mature protein regions are also bound to a (currently unknown) site on SecA and can target the preprotein independent of their signal peptide, but this only occurs after allosteric activation of SecA by a bound signal peptide [94, 95].
Inhibition of the translocational driving force
In addition to the previously described actions, mycolactone also depletes BiP and this may affect the translocation of other proteins indirectly . Co-translational translocation of BiP itself could be inhibited by mycolactone too, resulting in lower luminal BiP levels, and further amplifying the inhibition of translocation for BiP-dependent substrates.
The macrocycle valinomycin, isolated from an Actinomycete culture, is a known potassium anionophore . Interestingly, it was identified as a down-regulator of BiP expression and induces cell death in cancer cells with glucose starvation . The chaperone function of BiP protects cells during ER stress and supports the unfolded protein response. Reduced BiP levels in valinomycin-treated cancer cells can thus result in cell death under ER stress conditions, which are typical for solid tumor environments where nutrient supply is limited due to the poor vascularization. Valinomycin is also a signal peptide-specific inhibitor of hamster prion protein (PrP) translocation . Inefficient translocation of PrP leads to cytosolic accumulation and misfolding of the protein, resulting in cytotoxic protein aggregates. Interestingly, human PrP was not affected by valinomycin and this selectivity was due to small differences in the signal peptides of these two homologous proteins. The downregulation of BiP is suggested as a mechanism for the selectivity of the compound, as BiP dependency is linked to targeting properties of the signal peptide: ‘weak’ signal peptides require assistance from accessory factors for translocon gating, which is one of the proposed functions of BiP.
Two derivatives of R/1, MAL3-39 and MAL3-101, were shown to inhibit the in vitro post-translational translocation of ppαF substrates, but these derivatives are less active and only inhibit, respectively, 45 and 30% of translocation at 300 µM . While R/1 affects both the innate and Hsp40-stimulated ATPase activity of Hsp70 chaperones, MAL3-39 and MAL3-101 only inhibit the J domain-mediated stimulation of Hsp70 ATPase activity. It is suggested that these two derivatives enter the ER lumen, where they modulate the functions of BiP/Kar2p and Sec63, and therefore, affect post-translational translocation.
Bactericidal ATPase inhibitors
Overview of different inhibitors and modulators of translocation
Affected translocation pathways
Virulence factors and retrograde transport inhibitors
Prevents NC transfer from SRP to Sec61 
Sec61-dependent co-translocational import 
Induces UPR 
Induces an irreversible conformational change in Sec61α 
Broad effect on Sec61-dependent co-translational translocation, selective inhibition of Sec61-dependent post-translational translocation. SSPs are less affected. 
ER retrotranslocation of immunogenic peptides 
Hsp70 ATPase inhibitor (BiP/Kar2p) 
Post-translational translocation in yeast ER 
J domain-mediated Hsp70 ATPase activity 
Post-translational translocation in yeast ER 
J domain-mediated Hsp70 ATPase activity 
Post-translational translocation in yeast ER 
VSG_117 transport into Trypanosoma brucei ER 
Calmodulin antagonist 
Calmodulin-dependent post-translational translocation of small proteins 
Calmodulin antagonist 
Calmodulin-dependent post-translational translocation of small proteins 
Equisetin and CJ-21058
SecA ATPase inhibitor 
SecA-dependent post-translational translocation 
VSG_117 transport into Trypanosoma brucei ER 
Rose bengal and erythrosin B
SecA ATPase inhibitor 
SecA-dependent post-translational translocation 
P97-A4, P87-A4, 17D9, P91-E9, 16F6
Inhibits signal peptide binding to SecA 
SecA-dependent post-translational translocation 
SecA ATPase inhibitor 
SecA-dependent post-translational translocation 
Cyclic depsipeptides and triaza compounds
Traps NC TMDs at the cytosolic side of the Sec61α lateral gate 
ICAM-1 and VCAM-1 translocation 
Traps NC TMDs at the cytosolic side of the Sec61α lateral gate 
Affects most secreted proteins, but only a minority of integral membrane proteins 
Prevents correct insertion of VCAM-1 NCs into the translocon 
Stabilizes the Sec61α in a closed conformation 
Selective inhibition of co-translational translocation 
Broad-spectrum inhibition of translocation 
Targets Sec61/SecY, similar but not identical to cotransin 
Sec-dependent co- and post-translational translocation 
K+ anionophore 
Down-regulates BiP 
Signal-peptide-specific inhibitor of hamster PrP translocation 
Interferes with SP topology inversion inside the Sec61 translocon 
CJ-21058, a derivative of the fungal antibiotic equisetin was discovered as the first inhibitor of E. coli SecA ATPase activity . It also showed antibacterial activity against multidrug-resistant Staphylococcus aureus and Enterococcus faecalis. The fluorescein analogs rose Bengal and erythrosin B are able to inhibit the in vitro translocation of proOmpA through inhibition of the SecA ATPase activity. These compounds are predicted to occupy the ATP binding site in SecA . Inhibition of ATPase activity was indeed competitive at low ATP concentrations, however, more recent data shows that rose bengal non-competitively inhibits the translocation activity of SecA at high ATP concentrations . They have both bacteriostatic and bactericidal effects. Five compounds (termed P97-A9 family in Fig. 4) were identified in a small molecule screening as inhibitors of SecA translocase activity . These compounds target the signal peptide binding site in SecA, a conserved and essential feature in SecA from both Gram positive and Gram negative bacterial species, and also showed weak antimicrobial activity. Recently discovered analogues of bisthiouracil [105, 106] and the bistriazole compound SCA-21  are more potent inhibitors of the ATPase activity and SecA-dependent protein translocation. Moreover, they are effective against methicillin-resistant S. aureus strains. For most of these SecA inhibitors, permeability of the outer membrane in Gram negative bacteria was required for the antimicrobial activity . These novel antibacterial mechanisms are promising solutions for the urgent problem of multidrug-resistant pathogens.
Alternative targeting pathways
A route that functions in parallel with the SRP and TA pathways, termed SND (SRP-independent targeting), was recently discovered in Saccharomyces cerevisiae [110, 111]. Three proteins were identified (Fig. 5): Snd1 is located in the cytosol and predicted to interact with the ribosome, where it may act as the receptor for hydrophobic targeting signals. Snd2 and Snd3 form a complex in the ER membrane, together with the translocon, and could act as targeting receptors. Though only described in yeast, a human ortholog of Snd2 exists (hSnd2) and its function as a membrane receptor was recently confirmed . The SND pathway was originally shown to serve as a backup targeting system for both SRP-dependent and TRC40-dependent pathways in yeast . Recent evidence also showed that the TRC40 targeting pathway is not essential for integration of TA proteins in human cells, as membrane integration of TA proteins in TRC40 knockouts can be complemented by both the SND and SRP pathways . This supports the idea that SRP is likely able to recognize some targeting signals inside the ribosomal exit tunnel, near the end of TA protein translation, instead of outside the ribosome [114, 115, 116].
Inhibitors of the TRC40 pathway
Calmodulin inhibits ER membrane insertion of mammalian TA proteins in multiple translocation pathways, likely due to binding of CaM to the targeting signals as in Sec-dependent post-translational translocation, masking them for recognition. This function of CaM has been suggested as a regulatory mechanism for the TRC40 pathway .
TRC40 is also involved in ER targeting of SSPs  and glycosylphosphatidylinositol (GPI)-anchored proteins . Targeting of these proteins is thus apparently facilitated by multiple SRP-independent pathways. Mycolactone is an interesting inhibitor in this regard because it only partially affects translocation of SSPs. Some SSP pre-proteins that normally use the Sec61 pathway are able to escape the translocational block; in the presence of mycolactone, they are redirected to alternative pathways such as the TRC40 pathway. Importantly, hydrophobicity of the signal peptide and properties of the mature domain were shown to affect mycolactone sensitivity of SSPs .
Species-specific signal peptides
The GPI-anchored surface protein VSG_117 protein from Trypanosoma brucei, a protozoan parasite responsible for human African trypanosomiasis (sleeping sickness), is used for immune evasion in the bloodstream. Interestingly, post-translational translocation of VSG_117 is inhibited by MAL3-101, equisetin and CJ-21058 , three ATPase-related inhibitors (Table 1).
Trypanosomatida diverged early from other eukaryotes and they developed a surprisingly different SRP complex . Importantly, T. brucei signal peptides are incompatible with the eukaryotic post-translational translocation pathway . Properties of the hydrophobic region of the targeting signals appear to determine compatibility with the mammalian translocation system, likely due to their different interactions with the unique trypanosomal SRP. Most trypanosomal signal peptides are also able to use multiple translocation pathways, but they depend heavily on the SRP-independent post-translational translocation pathway for the production of their GPI-anchored proteins. Sec71, a non-essential post-translational translocon component in yeast which is not present in mammalian cells, is also present in T. brucei and essential for their survival .
Compounds like MAL3-101 might, therefore, offer a novel method for the treatment of trypanosomal infection. Selective inhibition of only trypanosomal post-translational translocation is possible due to the differences in the composition of host and parasite signal peptides and the corresponding translocon complexes. This key therapeutic concept of signal peptide-selective translocation inhibition was previously demonstrated in mammalian cells for the HUN-7293/cotransin family and CADA compounds (Table 1).
Since the discovery of the ubiquitous Sec-dependent protein translocation pathway, various translocation inhibitors have been discovered (Table 1). Furthermore, recent reports have uncovered a significant redundancy between the different protein translocation pathways, in which properties of the targeting signals determine the preference for the selected translocation system. In this review, we highlighted how multiple stages in the different translocation pathways can be modulated or even inhibited, leading to either selective or broad inhibition of protein translocation. In general, most of the translocation inhibitors described here affect the recognition, chaperoning or function of the signal peptides.
Based on the characteristic inhibitory profile of each translocation inhibitor, there is some optimism that modulation of protein translocation can be exploited for the development of new therapeutic and antimicrobial applications. Recent CRISPR/Cas9-based screenings of host factors involved in viral replication revealed an important role of translocon-associated components [123, 124]. For cotransin, it has already been shown that it limits proteostasis of enveloped viruses such as influenza virus, human immunodeficiency virus and dengue virus . In addition for mycolactone, one might expect a similar broad antiviral effect as it influences translocation of a broad range of proteins. Screening for translocation inhibitors has resulted in the discovery of eeyarestatin I and apratoxin A with anticancer properties, and a new class of broad-spectrum antibacterial compounds is being developed based on inhibition of SecA. Notwithstanding the long road to go for translocation inhibitors to become therapeutics, in the mean time they are valuable as research tools to decipher the remaining mysteries of protein translocation across membranes.
This work was supported by the KU Leuven (GOA no. 10/014 and PF/10/018) and the FWO (No. G.485.08).
Compliance with ethical standards
Conflict of interest
The authors declare that they have no conflict of interest.
- 1.Uhlen M, Fagerberg L, Hallstrom BM, Lindskog C, Oksvold P, Mardinoglu A, Sivertsson A, Kampf C, Sjostedt E, Asplund A, Olsson I, Edlund K, Lundberg E, Navani S, Szigyarto CA, Odeberg J, Djureinovic D, Takanen JO, Hober S, Alm T, Edqvist PH, Berling H, Tegel H, Mulder J, Rockberg J, Nilsson P, Schwenk JM, Hamsten M, von Feilitzen K, Forsberg M, Persson L, Johansson F, Zwahlen M, von Heijne G, Nielsen J, Ponten F (2015) Proteomics. Tissue-based map of the human proteome. Science 347(6220):1260419. https://doi.org/10.1126/science.1260419 PubMedGoogle Scholar
- 4.von Heijne G (1985) Signal sequences. The limits of variation. J Mol Biol 184(1):99–105Google Scholar
- 5.Osborne AR, Rapoport TA, van den Berg B (2005) Protein translocation by the Sec61/SecY channel. Annu Rev Cell Dev Biol 21:529–550. https://doi.org/10.1146/annurev.cellbio.21.012704.133214 PubMedGoogle Scholar
- 12.Green ER, Mecsas J (2016) Bacterial secretion systems: an overview. Microbiol Spectr. https://doi.org/10.1128/microbiolspec.VMBF-0012-2015 PubMedCentralGoogle Scholar
- 14.Park E, Rapoport TA (2012) Mechanisms of Sec61/SecY-mediated protein translocation across membranes. Annu Rev Biophys 41:21–40. https://doi.org/10.1146/annurev-biophys-050511-102312 PubMedGoogle Scholar
- 21.Wang Q, Mora-Jensen H, Weniger MA, Perez-Galan P, Wolford C, Hai T, Ron D, Chen W, Trenkle W, Wiestner A, Ye Y (2009) ERAD inhibitors integrate ER stress with an epigenetic mechanism to activate BH3-only protein NOXA in cancer cells. Proc Natl Acad Sci USA 106(7):2200–2205. https://doi.org/10.1073/pnas.0807611106 PubMedPubMedCentralGoogle Scholar
- 24.Cross BC, McKibbin C, Callan AC, Roboti P, Piacenti M, Rabu C, Wilson CM, Whitehead R, Flitsch SL, Pool MR, High S, Swanton E (2009) Eeyarestatin I inhibits Sec61-mediated protein translocation at the endoplasmic reticulum. J Cell Sci 122(Pt 23):4393–4400. https://doi.org/10.1242/jcs.054494 PubMedPubMedCentralGoogle Scholar
- 25.McKibbin C, Mares A, Piacenti M, Williams H, Roboti P, Puumalainen M, Callan AC, Lesiak-Mieczkowska K, Linder S, Harant H, High S, Flitsch SL, Whitehead RC, Swanton E (2012) Inhibition of protein translocation at the endoplasmic reticulum promotes activation of the unfolded protein response. Biochem J 442(3):639–648. https://doi.org/10.1042/BJ20111220 PubMedPubMedCentralGoogle Scholar
- 27.Aletrari MO, McKibbin C, Williams H, Pawar V, Pietroni P, Lord JM, Flitsch SL, Whitehead R, Swanton E, High S, Spooner RA (2011) Eeyarestatin 1 interferes with both retrograde and anterograde intracellular trafficking pathways. PLoS ONE 6(7):e22713. https://doi.org/10.1371/journal.pone.0022713 PubMedPubMedCentralGoogle Scholar
- 29.Hall BS, Hill K, McKenna M, Ogbechi J, High S, Willis AE, Simmonds RE (2014) The pathogenic mechanism of the Mycobacterium ulcerans virulence factor, mycolactone, depends on blockade of protein translocation into the ER. PLoS Pathog 10(4):e1004061. https://doi.org/10.1371/journal.ppat.1004061 PubMedPubMedCentralGoogle Scholar
- 30.Grotzke JE, Kozik P, Morel JD, Impens F, Pietrosemoli N, Cresswell P, Amigorena S, Demangel C (2017) Sec61 blockade by mycolactone inhibits antigen cross-presentation independently of endosome-to-cytosol export. Proc Natl Acad Sci USA 114(29):E5910–E5919. https://doi.org/10.1073/pnas.1705242114 PubMedPubMedCentralGoogle Scholar
- 35.Foster CA, Dreyfuss M, Mandak B, Meingassner JG, Naegeli HU, Nussbaumer A, Oberer L, Scheel G, Swoboda EM (1994) Pharmacological modulation of endothelial cell-associated adhesion molecule expression: implications for future treatment of dermatological diseases. J Dermatol 21(11):847–854PubMedGoogle Scholar
- 37.Westendorf C, Schmidt A, Coin I, Furkert J, Ridelis I, Zampatis D, Rutz C, Wiesner B, Rosenthal W, Beyermann M, Schulein R (2011) Inhibition of biosynthesis of human endothelin B receptor by the cyclodepsipeptide cotransin. J Biol Chem 286(41):35588–35600. https://doi.org/10.1074/jbc.M111.239244 PubMedPubMedCentralGoogle Scholar
- 40.Klein W, Westendorf C, Schmidt A, Conill-Cortes M, Rutz C, Blohs M, Beyermann M, Protze J, Krause G, Krause E, Schulein R (2015) Defining a conformational consensus motif in cotransin-sensitive signal sequences: a proteomic and site-directed mutagenesis study. PLoS ONE 10(3):e0120886. https://doi.org/10.1371/journal.pone.0120886 PubMedPubMedCentralGoogle Scholar
- 41.Junne T, Wong J, Studer C, Aust T, Bauer BW, Beibel M, Bhullar B, Bruccoleri R, Eichenberger J, Estoppey D, Hartmann N, Knapp B, Krastel P, Melin N, Oakeley EJ, Oberer L, Riedl R, Roma G, Schuierer S, Petersen F, Tallarico JA, Rapoport TA, Spiess M, Hoepfner D (2015) Decatransin, a new natural product inhibiting protein translocation at the Sec61/SecYEG translocon. J Cell Sci 128(6):1217–1229. https://doi.org/10.1242/jcs.165746 PubMedPubMedCentralGoogle Scholar
- 49.Baron L, Paatero AO, Morel JD, Impens F, Guenin-Mace L, Saint-Auret S, Blanchard N, Dillmann R, Niang F, Pellegrini S, Taunton J, Paavilainen VO, Demangel C (2016) Mycolactone subverts immunity by selectively blocking the Sec61 translocon. J Exp Med 213(13):2885–2896. https://doi.org/10.1084/jem.20160662 PubMedPubMedCentralGoogle Scholar
- 54.Vermeire K, Zhang Y, Princen K, Hatse S, Samala MF, Dey K, Choi HJ, Ahn Y, Sodoma A, Snoeck R, Andrei G, De Clercq E, Bell TW, Schols D (2002) CADA inhibits human immunodeficiency virus and human herpesvirus 7 replication by down-modulation of the cellular CD4 receptor. Virology 302(2):342–353PubMedGoogle Scholar
- 55.Vermeire K, Bell TW, Van Puyenbroeck V, Giraut A, Noppen S, Liekens S, Schols D, Hartmann E, Kalies KU, Marsh M (2014) Signal peptide-binding drug as a selective inhibitor of co-translational protein translocation. PLoS Biol 12(12):e1002011. https://doi.org/10.1371/journal.pbio.1002011 PubMedPubMedCentralGoogle Scholar
- 60.Chawla R, Van Puyenbroeck V, Pflug NC, Sama A, Ali R, Schols D, Vermeire K, Bell TW (2016) Tuning side arm electronics in unsymmetrical cyclotriazadisulfonamide (CADA) endoplasmic reticulum (ER) translocation inhibitors to improve their human cluster of differentiation 4 (CD4) receptor down-modulating potencies. J Med Chem 59(6):2633–2647. https://doi.org/10.1021/acs.jmedchem.5b01832 PubMedGoogle Scholar
- 61.Besemer J, Harant H, Wang S, Oberhauser B, Marquardt K, Foster CA, Schreiner EP, de Vries JE, Dascher-Nadel C, Lindley IJ (2005) Selective inhibition of cotranslational translocation of vascular cell adhesion molecule 1. Nature 436(7048):290–293. https://doi.org/10.1038/nature03670 PubMedGoogle Scholar
- 63.Harant H, Wolff B, Schreiner EP, Oberhauser B, Hofer L, Lettner N, Maier S, de Vries JE, Lindley IJ (2007) Inhibition of vascular endothelial growth factor cotranslational translocation by the cyclopeptolide CAM741. Mol Pharmacol 71(6):1657–1665. https://doi.org/10.1124/mol.107.034249 PubMedGoogle Scholar
- 77.Lang S, Benedix J, Fedeles SV, Schorr S, Schirra C, Schauble N, Jalal C, Greiner M, Hassdenteufel S, Tatzelt J, Kreutzer B, Edelmann L, Krause E, Rettig J, Somlo S, Zimmermann R, Dudek J (2012) Different effects of Sec61alpha, Sec62 and Sec63 depletion on transport of polypeptides into the endoplasmic reticulum of mammalian cells. J Cell Sci 125(Pt 8):1958–1969. https://doi.org/10.1242/jcs.096727 PubMedPubMedCentralGoogle Scholar
- 78.Muller L, de Escauriaza MD, Lajoie P, Theis M, Jung M, Muller A, Burgard C, Greiner M, Snapp EL, Dudek J, Zimmermann R (2010) Evolutionary gain of function for the ER membrane protein Sec62 from yeast to humans. Mol Biol Cell 21(5):691–703. https://doi.org/10.1091/mbc.E09-08-0730 PubMedPubMedCentralGoogle Scholar
- 82.Linxweiler M, Schorr S, Schauble N, Jung M, Linxweiler J, Langer F, Schafers HJ, Cavalie A, Zimmermann R, Greiner M (2013) Targeting cell migration and the endoplasmic reticulum stress response with calmodulin antagonists: a clinically tested small molecule phenocopy of SEC62 gene silencing in human tumor cells. BMC Cancer 13:574. https://doi.org/10.1186/1471-2407-13-574 PubMedPubMedCentralGoogle Scholar
- 83.Erdmann F, Schauble N, Lang S, Jung M, Honigmann A, Ahmad M, Dudek J, Benedix J, Harsman A, Kopp A, Helms V, Cavalie A, Wagner R, Zimmermann R (2011) Interaction of calmodulin with Sec61alpha limits Ca2+ leakage from the endoplasmic reticulum. EMBO J 30(1):17–31. https://doi.org/10.1038/emboj.2010.284 PubMedGoogle Scholar
- 88.Schauble N, Lang S, Jung M, Cappel S, Schorr S, Ulucan O, Linxweiler J, Dudek J, Blum R, Helms V, Paton AW, Paton JC, Cavalie A, Zimmermann R (2012) BiP-mediated closing of the Sec61 channel limits Ca2+ leakage from the ER. EMBO J 31(15):3282–3296. https://doi.org/10.1038/emboj.2012.189 PubMedPubMedCentralGoogle Scholar
- 101.Sugie Y, Inagaki S, Kato Y, Nishida H, Pang CH, Saito T, Sakemi S, Dib-Hajj F, Mueller JP, Sutcliffe J, Kojima Y (2002) CJ-21,058, a new SecA inhibitor isolated from a fungus. J Antibiot (Tokyo) 55(1):25–29Google Scholar
- 108.Jin J, Hsieh YH, Cui J, Damera K, Dai C, Chaudhary AS, Zhang H, Yang H, Cao N, Jiang C, Vaara M, Wang B, Tai PC (2016) Using chemical probes to assess the feasibility of targeting SecA for developing antimicrobial agents against Gram-negative bacteria. ChemMedChem 11(22):2511–2521. https://doi.org/10.1002/cmdc.201600421 PubMedPubMedCentralGoogle Scholar
- 110.Aviram N, Ast T, Costa EA, Arakel EC, Chuartzman SG, Jan CH, Hassdenteufel S, Dudek J, Jung M, Schorr S, Zimmermann R, Schwappach B, Weissman JS, Schuldiner M (2016) The SND proteins constitute an alternative targeting route to the endoplasmic reticulum. Nature 540(7631):134–138. https://doi.org/10.1038/nature20169 PubMedPubMedCentralGoogle Scholar
- 117.Hassdenteufel S, Schauble N, Cassella P, Leznicki P, Muller A, High S, Jung M, Zimmermann R (2011) Ca2+ -calmodulin inhibits tail-anchored protein insertion into the mammalian endoplasmic reticulum membrane. FEBS Lett 585(21):3485–3490. https://doi.org/10.1016/j.febslet.2011.10.008 PubMedPubMedCentralGoogle Scholar
- 119.Patham B, Duffy J, Lane A, Davis RC, Wipf P, Fewell SW, Brodsky JL, Mensa-Wilmot K (2009) Post-translational import of protein into the endoplasmic reticulum of a trypanosome: an in vitro system for discovery of anti-trypanosomal chemical entities. Biochem J 419(2):507–517. https://doi.org/10.1042/BJ20081787 PubMedPubMedCentralGoogle Scholar
- 120.Liu L, Ben-Shlomo H, Xu YX, Stern MZ, Goncharov I, Zhang Y, Michaeli S (2003) The trypanosomatid signal recognition particle consists of two RNA molecules, a 7SL RNA homologue and a novel tRNA-like molecule. J Biol Chem 278(20):18271–18280. https://doi.org/10.1074/jbc.M209215200 PubMedGoogle Scholar
- 123.Marceau CD, Puschnik AS, Majzoub K, Ooi YS, Brewer SM, Fuchs G, Swaminathan K, Mata MA, Elias JE, Sarnow P, Carette JE (2016) Genetic dissection of Flaviviridae host factors through genome-scale CRISPR screens. Nature 535(7610):159–163. https://doi.org/10.1038/nature18631 PubMedPubMedCentralGoogle Scholar
- 124.Zhang R, Miner JJ, Gorman MJ, Rausch K, Ramage H, White JP, Zuiani A, Zhang P, Fernandez E, Zhang Q, Dowd KA, Pierson TC, Cherry S, Diamond MS (2016) A CRISPR screen defines a signal peptide processing pathway required by flaviviruses. Nature 535(7610):164–168. https://doi.org/10.1038/nature18625 PubMedPubMedCentralGoogle Scholar
- 125.Heaton NS, Moshkina N, Fenouil R, Gardner TJ, Aguirre S, Shah PS, Zhao N, Manganaro L, Hultquist JF, Noel J, Sachs D, Hamilton J, Leon PE, Chawdury A, Tripathi S, Melegari C, Campisi L, Hai R, Metreveli G, Gamarnik AV, Garcia-Sastre A, Greenbaum B, Simon V, Fernandez-Sesma A, Krogan NJ, Mulder LCF, van Bakel H, Tortorella D, Taunton J, Palese P, Marazzi I (2016) Targeting viral proteostasis limits influenza virus, HIV, and dengue virus infection. Immunity 44(1):46–58. https://doi.org/10.1016/j.immuni.2015.12.017 PubMedPubMedCentralGoogle Scholar
- 126.Wang Q, Shinkre BA, Lee JG, Weniger MA, Liu Y, Chen W, Wiestner A, Trenkle WC, Ye Y (2010) The ERAD inhibitor Eeyarestatin I is a bifunctional compound with a membrane-binding domain and a p97/VCP inhibitory group. PLoS ONE 5(11):e15479. https://doi.org/10.1371/journal.pone.0015479 PubMedPubMedCentralGoogle Scholar
Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.