Go your own way: membrane-targeting sequences

Abstract

Membrane-targeting sequences, connected targeting mechanisms, and co-factors orchestrate primary targeting of proteins to membranes.

Introduction

Within eukaryotic cells, distinct functions are distributed across intracellular membrane-bounded organelles. Each of these organelles has a discrete pool of proteins, around 30% of which are membrane-bounded (Krogh et al., 2001). Membrane proteins occupy a wide variety of functions, including channels, pores, receptors, or may be involved in various metabolic processes and signal transduction pathways (Hedin et al., 2011). Most of these proteins within the organelles, even endosymbiotic-derived organelles that contain their own genome, originate from nuclear genes, and are initially translated in the cytosol. A total of 99% of mitochondrial proteins (Rehling et al., 2004) and more than 95% of chloroplast proteins (Soll, 2002) are encoded by nuclear DNA and imported into the organelles. From this cytosolic pool, it is imperative that a protein reaches the correct compartment, and a multitude of mechanisms exist to ensure this specific targeting takes place. These mechanisms largely require the recognition of sequences or motifs within (1) the peptide as it is being synthesized (co-translational) or (2) the mature protein (post-translational). These are the two main targeting distinctions that determine the mechanism and sequences involved; however, targeting between different organelles, as well as potentially multiple membranes within an organelle increases the requirement for distinct signals and mechanisms for each membrane. Additionally, although membrane targeting is an evolutionarily conserved problem, for some organelles little work has been done on plant models. Plant cells also have the extra requirements of chloroplast and tonoplast membranes. This review aims to give an overview on specific signal sequences and the mechanisms involved in membrane targeting with a focus on plant cells specifically where data are available and comparing with knowledge from other eukaryotic organisms (Figure 1 andTable 1).

Figure 1.

Schematic summary of general routes of import for membrane proteins. Numbers correspond with signal sequence details in Table 1. (1A) Co-translational targeting of signal-anchored proteins destined for the secretory pathway, mediated via the SRP. (1B) Post-translational targeting of tail-anchored proteins, similarly for the secretory pathway, mediated by the GET pathway and other mechanisms. (1C) COPII-mediated anterograde transport to the Golgi bodies. (1D) COPI-mediated retrograde transport, utilizing the dilysine ER retention motif. (1E) Vesicular transport from Golgi to the plasma membrane (PM). (2A) Targeting of outer chloroplast membrane proteins via multiple mechanisms. TOC membrane mediates import of both soluble stromal proteins and is also involved in the insertion of outer chloroplast membrane proteins. (2B) TIC membrane is again required for soluble stromal protein trafficking, but also that of inner chloroplast membrane proteins. (2C) Thylakoid membrane protein insertion, post- or co-translational both require csSRP. (3A and 3B) analogous to the TOC–TIC mechanism for chloroplast, the TOM and TIM mitochondrial membranes mediate insertion of both soluble matrix-targeted proteins, and also are involved in the insertion of mitochondrial membrane proteins. (4) Targeting of PMPs, either from ER-derived pER or direct from the cytosol. (5) Inner nuclear envelope targeting requires translocation from ER/outer nuclear envelope to the inner via multiple mechanisms, some requiring NLS, others not. (6) Targeting of tonoplast membrane proteins, similar to peroxisomes, can occur via an ER-derived pathway or directly from the cytosol.

Table 1.

Select examples of typical signal sequences utilized in targeting to specific organelle membranes. Numbers correspond with pathway illustrated in Figure 1

Number in Figure1	Target membrane	Protein type	Source	Typical signal sequence/targeting motif	Cleavable	Examples
1A	ER	SA – Type I	Cytosol	Signal peptide—N(+)-hydrophobic-C(polar)	Yes
		SA – Others		SA—approximately 20 hydrophobic residues	No
1B		Tail-anchor		Tail-anchor—approximately 20 hydrophobic residues at C-terminus	No
1C		Any	Golgi	C-terminal dilysine (–KKXX–COOH)	No
1D	Golgi		ER	Default (moderate length TMD)
1E	PM		Golgi	Default (long TMD)
2A	Chloroplast Outer	SA	Cytosol	<0.4 WW hydrophobicity TMD and C-terminal basic residues	No	Toc64
		Tail-anchor		TA of variable length and hydrophobicity and C-terminal basic residues	No	Toc33, Toc34
		β-barrel		Bipartite signal—transit peptide, polyglutamine stretch	Yes	Toc75
2B	Chloroplast inner		Cytosol	Transit peptide (Ser/Pro rich—TMD)		Tic40
2C	Thylakoid		Cytosol	L18 motif and DPLG–		LHCP
			Stroma	NA		Cytochrome b5
3A	Mitochondria outer	SA	Cytosol	Moderate hydrophobicity and C-terminal basic residues
		Tail-anchor		<17 aa TMD and C-terminal basic residues
		β-barrel		NA
3B	Mitochondria inner		Cytosol	NA
4	Peroxisome		Cytosol	Dilysine, –YLSQLQQHPLRT–, 2 basic clusters		PMP22, APX
			ER	>18 aa TMD and C-terminal –RKRMK–		Pex16
5	Inner nuclear membrane		ER	Nuclear localization signal or inner nuclear membrane sorting motif
6	Tonoplast		Cytosol	NA		CBL2
			ER	Dileucine or tyrosine motif (–YTRL–)		Ptr2, Fruct4

Advances

• Example sequences are available that allow a protein of interest to target a membrane of choice for basic research and biotechnological approaches.

• Not all GET homologs have been identified in plants, suggesting pathway divergence for the insertion of tail-anchored proteins.

• Chaperone protein complexes comprising specific Hsp compositions could be important for membrane protein targeting; receptors capable of binding the highly conserved C-terminus of Hsp70/Hsp90 are found on each organelle.

• The main role of targeting factors may be to increase targeting efficiency by preventing folding and aggregation.

• Recently discovered putative pathways targeting proteins to the inner nuclear membrane, peroxisome membrane, and tonoplast offer insight into the variety and complexity of targeting, and identify signals relevant for certain pathways.

Co-translational targeting in the secretory pathway

The most common mechanism for the targeting of proteins to the membranes of the secretory pathway is the use of a signal recognition particle (SRP) and subsequent SRP receptor localized in the endoplasmic reticulum (ER) membrane (Alberts et al., 2019) [Figure 1(1A)]. The SRP pathway is conserved across all domains of life, with prokaryotes utilizing a similar mechanism for plasma membrane targeting. In short: as the polypeptide is being translated, the SRP recognizes signal sequences that occur either as an N-terminal cleavable single peptide (SP), or an internal non-cleavable signal anchor (SA). It has been shown that binding to the nascent peptide requires a combination of RNA and various small polypeptides. Guanosine triphosphate (GTP)ase functionality ensures unidirectionality of targeting (Akopian et al., 2013). Initial binding of SRP in eukaryotes has been suggested to occur via three stages. An initial low-affinity preferential binding to translating ribosomes that takes place regardless of the peptide being synthesized (Flanagan et al., 2003), followed by an increased affinity between SRP and the ribosome nascent chain complex (RNC) when a SA sequence is present in the ribosomal exit tunnel. This primes the RNC for the final step whereby the SRP interacts directly with the SA and forms the strongest affinity with the complex, to allow efficient translocation to the ER (Berndt et al., 2009). Once bound to the RNC, the SRP halts translation and mediates the translocation of the complex to the ER. Here, it transitions to preferential binding of the SRP receptor in a GTP-dependent manner to ensure unidirectionality of movement. Interaction between the SRP and its receptor relinquishes SRP’s hold on the RNC, which now resumes translation when attached to the translocon SEC61 (Keenan et al., 2001). From here, the protein will be synthesized as an integral membrane protein.

Co-translational signal sequences are not only responsible for the targeting of the protein to the ER but are also integral for the protein assuming correct topology, and thus the type of membrane protein will have differing signal sequences. The structure of the SP has been well characterized and comprised three domains with an N-terminal positively charged region (1–5 amino acids), followed by a hydrophobic core (7–15 amino acids), and a C-terminal polar region (3–7 amino acids; von Heijne, 1990). SAs comprising the transmembrane domains (TMDs) again are very generic across proteins, consisting simply of around 20 hydrophobic residues. Typically, whichever flanking region contains positively charged residues will result in that terminus facing the cytosol (Hartmann et al., 1989). For multispanning proteins, this rule applies to the first SA in the protein, and the presence of latter SA and stop-transfer (ST) sequences (another hydrophobic sequence), will generate the correct topology. Other factors such as the folding state (Denzer et al., 1995) and the N-terminal charge: hydrophobic domain length ratio (Sakaguchi et al., 1992) are also implicated in the generation of correct topology. SA membrane proteins are divided into different types (Figure 2); Type I are single spanning membrane proteins that contain a cleavable SP, which orients the N-terminus to be lumenal at the translocon complex and attaches to the bilayer prior to being cleaved by relevant peptidases (von Heijne, 1990). Following this, an internal ST sequence prevents further translocation of the protein, and the C-terminus of the polypeptide is subsequently synthesized in the cytosol. Types II and III membrane proteins only contain a SA sequences, and are therefore single membrane spanning proteins with no cleavable peptide. Type II is orientated via the presence of basic residues N-terminal of the TMD, and the N-terminus of the protein residing in the cytosol. Type III is the opposite, with a luminal facing N-terminus. This is an example of the positive inside rule with inside referring to the cytosol (von Heijne, 1990). Types IVA and IVB are multi-spanning membrane proteins, with the first TMD following the same rules as other transmembrane proteins. Type IVA proteins contain an N-terminal cytosolic facing portion and Type IVB a C-terminal one. This rule of orientation is generally well accepted; however, recent studies have highlighted more specific sequence features that determine topology, such as for example the cysteine and tyrosine inside preference (Baker et al., 2017).

Figure 2.

Illustration of types I, II, III, and IV membrane proteins that are co-translationally inserted into the ER via the SRP. Type I shows the presence of both an N-terminal cleavable SP and an internal SA domain. The other proteins only contain an SA domain and are oriented via the presence of basic residues, either C- or N-terminal of the TMD. ST signals are where the protein is reinserted into the membrane to introduce a subsequent TMD following translation of the first SA. Type IVA proteins are defined by an N-terminal cytosolic facing portion.

Without further mechanisms, proteins bound to ER membranes would be transported to the plasma membrane [Figure 1 (1E)] or tonoplast via anterograde vesicular transport. Therefore, as well as having an initial signal sequence to ensure targeting to the ER, it is also imperative the polypeptide contains a retention sequence [Figure 1 (1D)]. Plant ER membrane proteins utilize the C-terminally positioned dilysine (KKXX) motif (Jackson et al., 1990). These signals are present across Eukarya, and found in ER-resident plant proteins such as glycerol-3-phosphate acyltransferases 8 and 9 (Gidda et al., 2009) and tomato Cf-9 (Benghezal et al., 2000). In mammalian cells, ER membrane proteins such as the p24 family of proteins (Contreras et al., 2004) and others (Zerangue et al., 2001) interact with components of the Coat Protein I (COPI) vesicles such as coatomer subunits (Cosson and Letourner, 1994), as well as the ADP-ribosylation factor (ARF) GTPase directly via the dilysine motif; this could explain how this motif is responsible for formation of COPI vesicles and subsequent retrograde transport. Other retention signals include the Arginine RXR motif found in type II ER membrane proteins, the hydrophobic pentapeptide motif, and the atypical dilysine KK-COOH motif (reviewed in Gao et al., 2014). Similarly, the C-terminal Golgi body retention signal KXD/E was first identified in the Arabidopsis (Arabidopsis thaliana) Endomembrane proteins (EMP) family of proteins and has since been found in Golgi-resident proteins across Eukarya (Gao et al., 2012). It is speculated that the composition of COPI coatomer isoforms and distinct subpopulations of COPI vesicles are responsible for the capacity for KKXX and KXD/E motifs to sequester membrane proteins in distinct compartments (Donohoe et al., 2007). It is also suggested that longer TMD’s favor localization to latter parts of the secretory pathway and truncating a TMD can result in re-targeting to earlier compartments (Yang et al., 1997).

Post-translational targeting

All protein targeting to organelles without the use of the SRP-mediated pathway occurs in a post-translational manner after translation is complete. This is the case for all targeting to organelles other than the ER including chloroplasts, mitochondria and peroxisomes and some ER proteins. This targeting is less well understood than the co-translational pathways and for example, the N-terminal signal sequences that direct proteins into the chloroplast and mitochondria are interestingly very similar, and some signal sequences are capable of dual organellar targeting (Chew et al., 2003).

Tail-anchored proteins as a model system to study post-translational protein targeting

In the absence of specific signal peptides, tail-anchored (TA) proteins rely on a single C-terminal TMD to act both as a targeting sequence as well as a membrane anchor. TA proteins are produced in full cytosolically, before translocation to the required membrane where they are anchored via their C-terminus, with the active N-terminal domain facing the cytosol (High and Abell, 2004; Borgese et al., 2007) [Figure 1 (1B)]. In Arabidopsis, there are 454 TA protein genes with predicted 520 splice variants. The majority of these TA proteins are of unknown localization (75%) and function (66%). Around 13% localize to the ER, 5% each to plastids and mitochondria (Kriechbaumer et al., 2009). Likewise, in rice and potato, the majority of predicted TA proteins (508 and 912) are of unknown localization (Manu et al., 2018). Around 10–20% localize to mitochondria and plastids each, and around 3% localize to the secretory pathway (Manu et al., 2018). Analysis of the targeting of specific TA proteins has identified features rather than specific signal sequences that are responsible for targeting to specific organelles. This includes the length of the TMD and charged residues in the flanking regions (Hwang et al., 2004; Maggio et al., 2007; Wattenberg et al., 2007; Kriechbaumer et al., 2009). In addition, PEX19 is capable of recognizing specific sequence motifs in the tail-anchor of PEX26 leading to its peroxisomal targeting. As there are not many peroxisomal proteins known, it is unclear whether this is a general mechanism (Halbach et al., 2006). The most extensively studied TA proteins cytochrome b5 family proteins (CYB5). Different CYB5 isoforms localize to the mitochondrion or chloroplast outer membranes or the ER. One distinguishing characteristic that determines their localization appears to be their hydrophobicity profile which could potentially influence the interaction between the TMD with the differing contents of the lipid bilayers (Pedrazzini, 2009). Depending on organism, protein and target organelle, hydrophobicity scores, length of TMD, and frequencies of specific residues such as Leucine within the targeting sequence are all impactful on the localization of the protein (Manu et al., 2018).

TA membrane insertion pathways

In yeast and mammalian cells, membrane insertion of TA proteins is suggested to be mediated via the guide entry of tail-anchor (GET) pathway (Stefanovic and Hedge, 2007; Favaloro et al., 2008). In short, the fully synthesized protein is shielded from cytosolic degradation via the pre-targeting complex consisting of Sgt2, Get4, and Get5. These transfer the TA protein to Get3, which then shuttles the complex to Get1/2. Get3 consists of a homodimer with a nucleotide binding domain, Zn²⁺, and TA protein-binding domains. Nucleotide exchange and hydrolysis of ATP occurs during transfer to the Get1/2 complex. The TA is inserted into the membrane and ADP is released (Denic et al., 2013). Despite strong understanding in yeast, the same pathway cannot be extrapolated completely to other branches of eukaryotic life. The components of the plant pre-targeting complex are currently not completely understood but are presumed to consist of currently unidentified Get2/5 and SGt2 homologs. However, multiple putative Get3 paralogs in Arabidopsis, which appear to be responsible for the differential targeting to the ER (Get3a), plastids (Get3b), and mitochondria (Get3c; Xing et al., 2017). As the expression of the GET pathway proteins causes pleiotropic growth defects, alternative pathways for TA insertion and additional functions of GET in plants are suggested (Xing et al., 2017). It is suggested that Get3b and Get3c are unlikely to be involved in TA insertion, whereas GET3a is capable of interacting with Get1 and Get4 at the ER to mediate insertion into the bilayer (Zhuang et al., 2017).

Insertion of TA proteins in eukaryotes has also been reported to occur spontaneous (Brambillasca et al., 2006) or via chaperone protein-mediated mechanisms (Rabu et al., 2008). In plants, a class of chaperone receptors bearing tetratricopeptide repeat domains and capable to specifically bind the highly conserved C-terminus of heat shock protein (Hsp70) and/or Hsp90 can be found on each organelle (Schlegel et al., 2007). This might implicate these chaperone receptors in a universal role in protein targeting (reviewed in Kriechbaumer et al., 2012) but the role of these chaperone receptors is not fully understood; chaperone receptors have been shown to increase the efficiency of protein targeting in plants and yeast (Qbadou et al., 2006; Young et al., 2003; Kriechbaumer and Abell, 2012). It is not clear if they also contribute to targeting specificity. It is suggested that chaperones might provide targeting specificity by recognizing features of the bound protein they deliver and form unique signatures through distinct combinations of chaperones, which are then recognized by membrane receptors at the destination organelle. As an example, post-translational targeting of the M13 procoat into ER microsomal membranes requires specifically the cytosolic Hsp70, Hsc70 and is not interchangeable with other Hsp70s different chaperone classes (Zimmermann et al., 1988). This chaperone specificity could be due to sequences in M13 but also due to specific ER receptors.

Chloroplast membrane targeting

Perhaps, the most complicated topic pertains to that of chloroplast targeting, with three membranes consisting of the outer chloroplast envelope, inner chloroplast envelope, and the thylakoid membrane. Each of these have a distinct protein and lipid composition, with the majority of proteins coming from nuclear-encoded genes and thus require specific mechanisms and pathways of insertion depending on the nature of the protein itself. Soluble stromal targeted proteins rely on N-terminal transit peptides (TPs) of variable lengths and sequences, with little specification other than that of a low number of acidic residues (Bruce, 2000); however, more specific subgroups and motifs related to specific aspects of targeting within the TP have been identified (Lee et al., 2008). Import into the stroma is mediated by GTP-dependent activity of two translocon components—translocon of the outer envelope (TOC) and translocon of the inner envelope (TIC). The mechanism in which TPs are targeted and imported to the stroma is reviewed in Li and Chiu (2010).

Outer chloroplast membrane proteins

The mechanism for outer chloroplast membrane protein integration will vary depending on the type of protein and its location (Kim et al., 2019). SA proteins do not contain a TP, and instead are post-translationally targeted via the cytosolic Ankyrin repeat-containing protein 2 (AKR2; Bae et al., 2008), which recognizes the SA at the ribosomal exit tunnel and acts as a chaperone during transport [Figure 1 (2A)]. ARK2 subsequently targets the protein on the outer envelope membrane (OEM) by interaction with specific lipid components of the OEM such as monogalactosyldiacylglycerols (MGDG) and phosphatidylglycerols (PG). The insertion into the membrane is then aided by Toc75 (Bae et al., 2008; Kim et al., 2014). Interaction with ARK2 occurs via the signal domain, consisting of a TMD of moderate hydrophobicity (<0.4 on W&W scale), as well as a C-terminal positively charged region (CPR) consisting of a minimum of three arginine or lysine residues (Lee et al., 2011). One example that includes these sequence examples is that of Toc64, whose TMD and C-terminal dilysine residues are essential and sufficient for chloroplast targeting. It is speculated that the positive residues prevent SRP binding and trafficking to the ER during translation, and the mature TMD is specific for interaction with AKR2 or other chaperones for correct targeting (Lee et al., 2004).

Chloroplast targeting of TA membrane proteins appears to occur in a similar manner to that of SA proteins, with a requirement of ARK2 and Hsp70 acting as chaperones and Toc75 for insertion (Kim et al., 2019). The signal sequence again consists of a TMD followed by a C-terminal sequence. However, some proteins such as Toc33 and Toc34 require an intact N-terminal GTPase domain, while others such as OEP9 only require an intact TMD and CPR. Hydrophobicity levels appear to have little effect on this targeting so the specifics of the targeting sequence are relatively plastic between individual proteins (Dhanoa et al., 2010). The exact mechanism behind these distinct targeting pathways is unclear. Due to the localization of GET3 isoforms at endosymbiotic organelles, as well as their incapability of rescuing AtGet3a mutant defects in root hair growth, it is suggested that Get3b may be required for TA protein targeting to plastids, and Get3C to mitochondria (Zhuang et al., 2017).

The final category of OEM chloroplast proteins is that of the β-barrels. Seemingly, none of them contain TPs and their method of targeting is not well categorized. One exception is that of Toc75, which is unique in that it contains a bipartite signal, consisting of a generic transit peptide followed by polyglutamine stretch (Tranel and Keegstra, 1996). This allows Toc75 to be translocated from the cytosol with the N-terminus reaching the stroma, where the TP can then cleave, retaining Toc75 in the intermembrane space. Following this, the polyglutamine then stretch is cleaved, an essential step for Toc75 entering the OEM (Inoue et al., 2005). How Toc75 is specifically integrated into the membrane following this processing is still unclear, nor is much understood about how the majority of β-barrel proteins are targeted to the OEM.

Inner chloroplast membrane proteins

Similar to that of the outer membrane, multiple targeting pathways have been suggested for inner chloroplast membrane proteins (Lee et al., 2017) [Figure 1(2B)]. The majority of inner envelope membrane (IEM) proteins contain TPs, and thus are likely translocated via TOC–TIC machineries. During the translocation at the TIC complex, either the TMDs of the mature protein being translocated can signal the release and insertion from the TIC complex via a ST mechanism, or the protein is fully translocated to the stroma (Viana et al., 2010). Membrane proteins such as Tic40 and Tic110 that reach the stroma must undergo post-import insertion, and are suggested to use the SECretory2 (SEC2) complex (Li et al., 2017). Tic40 contains a bipartite sorting signal comprising both a TMD and a serine/proline rich domain N-terminal to the TMD, although this is not sufficient to target other proteins to the IEM upon fusion of this signal sequence (Tripp et al., 2007).

Thylakoid membrane proteins

Essential components of the photosynthetic machinery are thylakoid membrane proteins. These can either be encoded by the nuclear genome and imported posttranslationally to the stroma prior to insertion into the thylakoid membrane, or are encoded via the chloroplast genome and inserted co-translationally via a chloroplast SRP (csSRP), which is distinct from that of its cytosolic counterpart (Henry, 2010) [Figure 1(2C)]. Thylakoid membrane proteins that undergo post-translational insertion proteins are imported to the membrane either via the cpSPR, or spontaneous insertion (Aldridge et al., 2009). Interestingly, csSRP is not only required for co-translational insertion, but is also utilized for post-translational insertion. One example is that of the light-harvesting chlorophyll-binding protein (LHCP), which when in the stroma forms a complex with cpSRP43 and cpSRP54 subunits (Falk and Sinning, 2010). Interaction occurs between cpSRP43 and an 18 amino acid motif in LHCP called L18, which contains four residues (–DPLG–) essential for this interaction and subsequent targeting (Tu et al., 2000; Stengel et al., 2008). Following this, the complex binds to the thylakoid membrane-bounded cpFtsY acting as the target (Kogata et al., 1999). LHCP is integrated into the membrane via the integrase activity of albino3 (ALB3), a homologue of the yeast mitochondrial Oxa1p (Moore et al., 2000). An SRP/ALB3-independent pathway of spontaneous insertion has been shown for other non-LHCP, nuclear encoded thylakoid membrane proteins such as components of the Tat translocase, photosystems I and II, although specific details regarding the mechanism or signal sequences are not well understood (Woolhead et al., 2001; Shünemann, 2007).

Similarly, the csSRP molecule is also essential for the insertion of co-translationally transported proteins derived from the chloroplast genome (Hristou et al., 2019). Although interaction with csSRP and thylakoid membrane proteins has been demonstrated for proteins such as cytochrome b6 (Króliczewski et al., 2016) and the D1 protein of photosystem II (Kim et al., 1991), no specific signals have been identified asides from general hydrophobicity-mediated interaction with csSRP54 and the nascent chain leaving the ribosome (van Wijk and Nilsson, 2002). The general import process operates largely in the same manner as described for LHCP, requiring the integrase activity of ALB3 and other protein components such as cpFtsY. Mutants affecting this pathway usually result in a reduction in chloroplast content and varying severity (Henry et al., 2010).

Targeting of mitochondrial membrane proteins

With both being derived from endosymbiotic organisms, and consisting of at least outer and inner membranes, much of the targeting to mitochondrial membranes occurs via analogous mechanisms to that of the TOC–TIC translocons. In the mitochondria, these are translocase of the outer membrane (TOM) and translocase of the inner membrane (TIM; Neupert, 1997; reviewed in Wiedemann and Pfanner, 2017) [Figure 1 (3)]. Typically, matrix destined proteins will contain pre-sequences, the mitochondrial equivalent of transit peptides, which are identified and then cleaved by the TOM–TIM translocon following successful translocation. Mitochondrial membrane proteins are mostly derived from nuclear encoded genes (Schleiff and Becker, 2011), and come in various topologies and structures thus relying on multiple signals and pathways for correct targeting. Being analogous to the chloroplast, it is often difficult to discriminate or predict a signal that may result in preferential localization to either chloroplast or mitochondria, and little work has been done specifically on plant mitochondrial targeting, instead heavily relying on information derived from mammalian and yeast studies. Distinguishing the factors that determine SA protein localization between chloroplast and mitochondrion, outer membranes have proven difficult due to the similarities in their target sequences. Likewise, for chloroplast SA proteins, a moderate hydrophobicity TMD and a minimum of three basic residues C-terminal of the TMD are enough to prevent targeting to the ER (Lee et al., 2011). This is consistent in mammalian and yeast mitochondrial outer membrane proteins such as Tom70, although yeast mitochondrial proteins can tolerate basic residue substitution to serine, but not acidic residues (Waizenegger et al., 2003). Beyond this, specific SA signal sequences that determine between chloroplast and mitochondria have not been identified.

A similar issue is present for that of TA outer mitochondrial membrane proteins, with the majority of the work being done in mammalian and yeast systems. These proteins are similarly targeted as with any other organelle, requiring a C-terminal TMD, which acts both as the signal and the anchor for the fully mature protein to be inserted into the membrane. Factors of the signal sequence that determine between ER and mitochondria targeting have been well established. Mitochondrial TA proteins typically contain a shorter hydrophobic segment and require basic residues flanking this TMD, such as the 17 amino acid long TMD with C-terminal twin arginine residues observed in mammalian Vesicle-Associated Membrane Protein (VAMP) 1-B that is essential for its unique mitochondrial targeting (Isenmann et al., 1998). Other examples include that of Tom20, which again requires both a TMD of moderate hydrophobicity and flanking basic residues to result in the outer membrane targeting (Kanaji et al., 2000). This flanking of basic residues is effective enough for BCL-xl with 2-RK-C-terminal of the TMD to localize endogenously to the outer mitochondrial membrane, whereas BCL-2 with –HK– only one basic amino acid is targeted across other membranes (Kaufmann et al., 2003). This di-basic motif is also consistent with plant mitochondrial proteins, with CYB5–D exclusively localizing to the outer mitochondrial membrane that is not present in ER localized CYB5 isoforms (Hwang et al., 2004).

Little is understood about β-barrel mitochondrial proteins, besides the fact that like chloroplast β-barrel proteins, most of them if not all do not contain cleavable targeting sequences, nor specific signals. Tom40 is imported to the TOM complex via the interaction with Tom22 and Tom70 and insertion is based on the formation of specific intermediates based on several factors surrounding the complex (Rapaport and Neupert, 1999). Interestingly, bacterial β-barrel proteins are capable of localizing to outer mitochondrial membranes when expressed in eukaryotic cells (Walther et al., 2009). Together, this could suggest that evolutionary conserved aspects of the whole protein structure are more important for targeting rather than specific sequences or cleavable peptides. Initial binding to Tom22 and Tom70 is a consistent first step for any insertion into mitochondrial membranes or further translocation. Little has been described for the import of inner mitochondrial membrane proteins, besides that of ADP/ATP carriers. Following binding with Tom70 and Tom22, Tim9 and Tim10 bind to the hydrophobic segments of the carrier proteins and shuttle the carrier protein to the TIM22 complex where insertion to the inner membrane occurs (Sirrenberg et al., 1996). Aside from general hydrophobic interactions, no specific sequences have been highlighted as essential for this specific targeting, and once again more complex factors involving the full length protein and components of multiple TMDs, as well as even membrane charge are vital for the insertion of inner membrane proteins (Truscott et al., 2003).

Peroxisomal targeting

The targeting of peroxisomal membrane proteins (PMPs) is mediated via two pathways, either imported directly from the cytosol or via an ER-derived compartment (Mayerhofer, 2016) [Figure 1 (4)]. Exploring the example of Arabidopsis PMP22, four membrane peroxisome targeting signal motifs have been identified for targeting to the peroxisomal membrane (Murphy et al., 2003). Two motifs acting in parallel are a dilysine motif at residues 7 and 8, as well as a –YLSQLQQHPLRTK– motif between residues 14 and 26. This is consistent with the presence of a Y-x₃-l-x₃-P-x₃-K found in mammalian PMP22, which also acts as a targeting signal (Pause et al., 2000). Two basic clusters, –KIQIRR– (amino acid 49–54) and –KGKK– (amino acid 82–85) were also found to be important for membrane targeting, and similar motifs can be found in peroxisomal ER (pER)-derived membrane proteins such as the –RKRMK– motif found in ascorbate peroxidases (APX), which functions as a signal sequence. The sorting signals for peroxisomal membrane-bounded APX are within its C-terminal tail (Mullen and Trelease, 2000). Individually, each motif is not sufficient to allow targeting to the peroxisome, and there is a more complex mechanism involving the correct folding, association with chaperones and receptors, as well as the correct topology of the TMDs and motifs to ensure correct targeting (Murphy et al., 2003). The pER subdomain was first identified by the presence of labeled PMP showing signal in domains of the ER, as well as a lack of labeling occurring in peroxisomes when secretion was blocked via brefeldin A treatment (Mullen et al., 1999). It was later understood that this targeting to the pER and, subsequently, the peroxisomes requires the presence of a positively charged motif (–RKRMK–) at the very C-terminus of the protein, as well as a highly hydrophobic TMD at least 18 amino acids long (Mullen and Trelease, 2000). Similarly, other PMPs have been shown to localize to the pER as well as peroxisomes, such as Arabidopsis PEX16; here both TMDs and clusters of basic residues are required for localization (Karnik and Trelease, 2007).

Nuclear membrane targeting

Targeting of soluble nucleoplasm destined proteins has been well characterized. Nuclear localization signals (NLSa) are recognized by importin proteins, and subsequent RanGTP cycles ensure unidirectional transport through the nuclear pore complex (NPC), which connects the inner nuclear membrane (INM) and outer nuclear membrane (ONM; Pouton et al., 2007). Being contiguous with the ER, targeting of membrane proteins to the outer nuclear envelope is thought to only require specific targeting mechanisms that are relevant for ER targeting . Conversely, multiple mechanisms have been suggested for the targeting of INM proteins (Figure 1.5). Nonsignal-mediated methods include the diffusion retention model, which suggests that membrane proteins can simply diffuse from one membrane to the other via the NPC and are maintained in the INM via interaction partners found within the nucleoplasm (Ungricht et al., 2015). Another putative mechanism is the formation of vesicles that traffic between the ONM and INM, but this is not well characterized and not widely observed (Johnson and Baines, 2011). Other mechanisms include the utilization of NLSs and/or other INM sorting motifs; specific metazoan and yeast examples of proteins and signals, as well as more details on these mechanisms are reviewed in (Katta et al., 2014). Recently, it has been shown that this NLS-mediated pathway is utilized in plants, with the fusion of yeast SV40 monopartate NLS (–PKKKRKV–; Kalderon et al., 1984) as well as bipartite and plant specific NLSs, fused to a tail-anchored ER protein result in localization to the INM (Groves et al., 2019).

Tonoplast targeting

Without alternative signals that result in the retention of membrane proteins targeted to the secretory pathway, the default end point is the plasma membrane. However, the most common route for tonoplast membrane proteins is via the secretory pathway via multiple distinct routes including Golgi-dependent and -independent routes (reviewed in Rojas-Pierce, 2013) [Figure 1(6)]. Specific motifs and sequences that have been identified to discriminate and target effectively to either protein storage or lytic vacuole and avoid plasma membrane targeting. Early studies of bean α-Tonoplast intrinsic protein identified that the sixth TMD and 18 C-terminal residues could target a reporter protein to the tonoplast in tobacco (Nicotiana tabacum) cells (Höfte and Chrispeels, 1992). Other larger cytosolic domains of proteins such as VAMP7 (Uemura et al., 2005) and AtPK1 (Maîtrejean et al., 2011) are also capable of signaling tonoplast targeting, and fusions of these cytosolic domains can redirect PM localized proteins to the tonoplast (Pedrazzini et al., 2013). Beyond larger domains found in tonoplast residing proteins, specific domains are featured such as dileucine ([D/E]X3-5L[L/I]) motifs in AtPTR2, 4, and 6 (Komarova et al., 2012), as well as ESL1- (Yamada et al., 2010) and tyrosine-based (YTRL) domains found in AtFruct4 (Jung et al., 2011). These signals are conserved across Eukarya for vacuole targeting in yeast or endosome targeting in mammals and are typically found in close proximity to either TMDs or termini. They are recognized by the adapter group of proteins, which results in vesicle formation in post-Golgi trafficking to the tonoplast (Bonifacino and Traub, 2003). As well as these secretory pathway-derived routes, targeting directly from the cytosol has been documented for calcineurin B-like (CBL) 2 proteins, on which targeting is mediated by a “tonoplast targeting sequence” consisting of the 19 most N-terminal residues. This sequence was sufficient to localize a reporter and other CBL proteins to the tonoplast as part of a fusion protein (Tang et al., 2012).

Prospects and outlook

Membrane proteins can occupy a variety of functions that are essential in producing a functional organelle with compartmentalized functions. Therefore, targeting of proteins to specific membranes is highly important and regulated via a multitude of mechanism. Targeting sequences mostly comprise hydrophobic segments that are either cleaved as part of targeting or both initiate targeting and subsequently exist as TMDs of the mature, embedded protein. Minor variations in the properties of these hydrophobic segments, properties of the flanking residues, as well as the presence of other motifs that may interact with specific chaperones or translocon components most likely enable targeting to specific membranes.

Exploiting membrane protein targeting signals and mechanism can be of great value to biotechnological approaches. Often the cytoplasm is an inhospitable environment for protein accumulation and due to the presence of vacuoles, cytoplasmic space in plant cells especially is limited. Hence, targeting of proteins for biotechnological aspects such as enzyme complexes for production or recombinant protein products to organellar membranes, as for example the ER, chloroplasts, tonoplast, or oil bodies, can be of advantage. Membrane anchoring of enzyme complexes can also be beneficial for complex assembly and stability. TA proteins have the potential to colocalize functional enzyme complexes such as the mitochondrial TOM complex or to specific organelle surfaces. Furthermore, half-life and yields of proteins can be increased with TA addition as shown for tobacco expression of the HIV gene product and antiviral factor Nef (Barbante et al., 2008). This is suggested to be due to Nef being less susceptible to cytosolic degradation processes. Targeting of enzyme cascades to the ER may result in proteins directly accumulating in storage bodies formed within the ER lumen (seeds in cereal crops) or be transported to specialized protein storage vacuoles (legume seeds). Such research could have great potential for manipulating or increasing protein and lipid productivity in plants. Plant cells have been successfully glycol-engineered to mimic human glycosylation pathways for production of recombinant proteins. Examples include the generation catalytic domains of mammalian N-acetylglucosaminyltransferase (GnT) paired with plant fucosyltransferase localization signals, which target the protein to the Golgi apparatus allowing human-like N-glycosylation of recombinant protein (Nagels et al., 2011). Furthermore, mammalian like N- and O-glycosylation of Human erythropoietin was achieved via the transient expression in Nicotiana benthamiana of multiple Golgi localized mammalian and Drosophila enzymes such as α2,6-sialytransferase, proving the system is viable for potential pharmaceutical production of recombinant proteins (Castilho et al., 2012).

Outstanding questions

• Besides targeting sequences, what are the consequences of fully folded proteins and the presence of other domains on targeting?

• How essential are chaperone complexes for targeting and what do they comprise?

• How are membrane proteins integrated into membranes, via soluble intermediates or a translocon complex?

• For many yeast or mammalian targeting pathways, equivalent mechanisms and homologous proteins have been identified in plants. Exemptions are the GET pathway and mitochondrial membrane targeting. Further work is needed to elucidate the mechanisms behind plant membrane protein targeting, rather than filling gaps with knowledge from yeast/mammalian studies.

• What is the mechanism of β-barrel protein targeting for outer chloroplast and mitochondrial membranes?

• Are whole organism or in vitro approaches with reduced components preferential?

Despite lot of achievements and advances in research and biotechnology, important questions on membrane protein targeting general as well as specifically in plant systems still remain to be addressed (see Outstanding Questions Box).

 

S.W. and V.K. wrote the manuscript. S.W designed the figures.

The author responsible for distribution of materials integral to the findings presented in this article in accordance with the policy described in the Instructions for Authors (https://academic.oup.com/plphys) is: Verena Kriechbaumer (vkriechbaumer@brookes.ac.uk).