The membrane integration of polytopic proteins is coordinated at the endoplasmic reticulum (ER) by the conserved Sec61 translocon, which facilitates the lateral release of transmembrane (TM) segments into the lipid phase during polypeptide translocation. Here we use a site-specific crosslinking strategy to study the membrane integration of a new model protein and show that the TM segments of the P2X2 receptor are retained at the Sec61 complex for the entire duration of the biosynthetic process. This extremely prolonged association implicates the Sec61 complex in the regulation of the membrane integration process, and we use both in vitro and in vivo analyses to study this effect further. TM-segment retention depends on the association of the ribosome with the Sec61 complex, and complete lateral exit of the P2X2 TM segments was only induced by the artificial termination of translation. In the event of the premature release of P2X2 TM1 from the ER translocon, the truncated polypeptide fragment was to found aggregate in the ER membrane, suggesting a distinct physiological requirement for the delayed release of TM segments from the ER translocon site.

In the canonical mammalian secretory protein-targeting pathway, substrates are initially delivered to the endoplasmic reticulum (ER) membrane by a conserved signal recognition particle (SRP)-dependent route (Pool, 2005). SRP delivers precursors to a site termed the ER translocon that forms an aqueous conduit through which proteins traverse the ER membrane (Osborne et al., 2005). In the case of membrane proteins, this site is also laterally gated to allow the co-translocational partitioning of transmembrane (TM) segments into the lipid bilayer (Martoglio et al., 1995). The core mammalian ER translocon comprises Sec61α together with Sec61β and Sec61γ, and a related archaeal structure suggests an architecture by which these components might assemble to provide both a lateral gate and a protein-conducting pore to facilitate the dual function of translocation and membrane integration (Van den Berg et al., 2004). In this structure, the large Sec61α subunit is the major pore-forming constituent and is arranged in two pseudo-symmetrical bundles that might rearrange around a constricted central cavity to open a lateral exit pathway for TM segments (Rapoport, 2007). Although a single heterotrimer is suggested to form the protein-conducting channel through the membrane, it is likely that several Sec61 heterotrimers assemble in the active ER translocon together with multiple accessory components, including TRAM (translocating chain-associated membrane protein), the TRAP complex and PAT-10, during membrane protein integration (Hegde and Kang, 2008). Such an arrangement might contribute to the translocon architecture required to allow the simultaneous accommodation of multiple TM segments during membrane protein biosynthesis (Ismail et al., 2006; Kida et al., 2005; Sadlish et al., 2005).

For single-spanning membrane proteins, lateral dislocation of TM segments from the ER translocon into the lipid phase of the membrane can be correlated with the thermodynamic properties of the side chains present in the segment itself (White and von Heijne, 2008). Hence, more hydrophobic residues appear to promote efficient release from the ER translocon and the incorporation of the TM segment into the membrane (Hessa et al., 2005). In the case of complex polytopic membrane proteins, this model is probably an oversimplification, and multiple additional factors, including TM segment interaction (Buck et al., 2007; Meindl-Beinker et al., 2006; Zhang et al., 2007) and membrane protein assembly (Ismail et al., 2008; Sadlish et al., 2005), have yet to be fully accounted for. A particularly fascinating facet of polytopic membrane protein integration is the phenomenon of selective TM-segment retention at the ER translocon site (Cross and High, 2009). In the case of one well-studied example, opsin, the final TM segment of the protein, appears to remain directly adjacent to the ER translocon for an extended period during biogenesis (Ismail et al., 2008). Only artificial termination of translation was able to trigger the release of the TM segment into the lipid phase and the delay of the C-terminal TM segments of opsin at the Sec61 complex is suggested to facilitate the assembly of two distinct folding domains within the protein (Ismail et al., 2008). More recently, the eighth TM segment in the cystic fibrosis transmembrane conductance regulator (CFTR) protein was found to be specifically delayed at the ER translocon during CFTR biosynthesis, via a process that is dependent upon the presence of a charged residue within this membrane span (Pitonzo et al., 2009). The authors concluded that such retention might facilitate the formation of inter-TM segment contacts in CFTR necessary for the correct folding of the protein, and invoked the active participation of the Sec61 complex in regulating this process (Pitonzo et al., 2009). Thus, although the phenomenon of TM-segment retention at the ER translocon is clearly an important facet of polytopic protein biosynthesis, the generality and importance of this process is presently poorly understood.

The goal of this study was to analyse the behaviour of a novel model precursor to better understand the membrane integration of polytopic proteins. P2X2 (also known as P2RX2) is a member of the ATP-gated family of ion-channels containing two membrane-spanning regions separated by an extended loop region that houses the ATP-binding domain. In addition, the first TM segment contains three potentially unfavourable residues (H33, R34 and Q37), which might cause a thermodynamic barrier to membrane integration (Hessa et al., 2005). Hence, the biosynthesis of the P2X2 receptor presents a distinct, and hitherto unexamined, challenge to the ER translocon, and insights into its membrane integration should expand our limited understanding of polytopic membrane protein biogenesis. We used a cysteine-mediated site-specific crosslinking analysis to monitor the molecular environment of the TM segments of P2X2 during biosynthesis. Remarkably, we found that TM1 and TM2 of P2X2 are simultaneously retained at the ER translocon throughout the synthesis of the entire polypeptide. Premature dislocation of the first TM segment from the translocon and into the ER membrane results in pronounced aggregation of the membrane-integrated polypeptide fragment both in vitro and in vivo. Taken together, these data suggest a distinct physiological basis for selective TM-segment retention at the ER translocon, and we propose that the translocon facilitates TM segment assembly to promote effective membrane protein integration.

In this study, we used a quantitative in vitro crosslinking strategy to closely monitor the molecular environment of the P2X2 TM segments during biosynthesis. Exclusion of a stop codon from mRNA coding transcripts causes the resulting polypeptide to remain anchored to the ribosome peptidyl transferase centre via the peptidyl-tRNA bond (Gilmore et al., 1991). By designing transcripts of different lengths, distinct stages of the biosynthetic pathway can be simulated and probed by the addition of crosslinking reagents. An important consideration for the application of bifunctional crosslinking reagents to membrane integration studies is the specificity of the reactive moiety; adducts will only be formed with partner proteins containing a suitable and accessible amino acid side chain. The ER translocon components Sec61α, Sec61β and TRAM each contain at least one cysteine residue and have all been efficiently crosslinked during previous studies exploiting homobifunctional sulphydryl-reactive crosslinking reagents (Garrison et al., 2005; Heinrich and Rapoport, 2003; Ismail et al., 2008). Thus, in the present study, cysteine residues incorporated into P2X2 in a cysteine-null background allowed for a site-specific crosslinking approach using the membrane-permeable thiol-specific reagent, bismaleimidohexane (BMH).

We initially sought to establish the suitability of P2X2 for this in vitro crosslinking approach and found that P2X2 was first delivered to the Sec61 complex via the canonical SRP-dependent delivery pathway (supplementary material Fig. S1) and subsequently efficiently integrated into the ER membrane (supplementary material Fig. S2). In addition, different stalled nascent P2X2 polypeptides were each found to represent a homogenous population following a protease protection analysis (supplementary material Fig. S2). BMH-dependent adducts with P2X2 were identified by the immunoprecipitation (IP) of candidate proteins in parallel with the HA-tagged nascent chain, and crosslinking to components of the ER translocon was found to be contingent upon a stable peptidyl-tRNA bond and the presence of a cysteine probe site within the nascent P2X2 chain (supplementary material Fig. S3). Quantification of crosslinking efficiency was by phosphorimaging, using a strategy that accounts for any variability in translation efficiency and allows a direct comparison of multiple independent adducts formed with the same nascent chain (supplementary material Fig. S4). Although crosslinking using BMH, or other such reagents, can never achieve complete efficiency, the synchronisation of stalled nascent chains and the analysis of only C-terminally tagged polypeptides ensures that this approach accurately reports the environment of the bulk of the P2X2 chains present in each reaction (see also Daniel et al., 2008; Ismail et al., 2006; Mothes et al., 1997; Pitonzo et al., 2009).

Crosslinking profile of P2X2 TM1 during early-phase biosynthesis

To analyse the molecular environment of P2X2 TM1 during early phase biosynthesis, a scanning BMH-dependent crosslinking analysis was conducted that could identify ER translocon components adjacent to each possible probe position within TM1. In this way, the opportunity for adduct formation with TM1 is maximised to more accurately assess the environment of the TM segment using a bifunctional crosslinking reagent, and the resulting data informed the selection of crosslinking probes for future experiments. Variants of HA-tagged P2X2 truncated to 127 residues and with a cysteine residue incorporated into each position within and several flanking the first TM segment (Fig. 1A) were translated in vitro in the presence of ER-derived microsomes. At this chain length, the first TM segment has engaged the Sec61 complex (supplementary material Figs S2 and S3), and was predicted to have attained a native P2X2 topology (Fig. 1A) (Kowarik et al., 2002). Following crosslinking, adducts formed from each probe position were identified by IP (supplementary material Fig. S3), and quantification revealed a distinct and reproducible crosslinking profile of P2X2 TM1 in the context of a 127-residue integration intermediate. Hence, adducts with Sec61α were formed from each of the probe positions analysed in TM1, although this crosslinking was most evident from the C-terminal region of TM1 and peaked at position 56, where the mean relative crosslinking efficiency was remarkably high (Fig. 1B, Sec61α). In addition to this primary crosslinking, adducts between the nascent P2X2 chain and Sec61α that are secondarily crosslinked to Sec61β (Ismail et al., 2006), were observed for probes at the C-terminal region of TM1 (Fig. 1B, Sec61α × Sec61β; see also supplementary material Fig. S3). The efficiency of formation for these latter adducts is significantly correlated with the formation of the primary Sec61α adducts (P=0.0004, Spearmans) (Fig. 1B, compare Sec61α and Sec61α × Sec61β), and is indicative of an active ER translocon conformation (Kalies et al., 1998). In contrast to the Sec61α adducts, crosslinking to Sec61β and TRAM was primarily from cysteines located at the N-terminal region of P2X2 TM1 (Fig. 1C,D), which is consistent with several previous crosslinking studies (Ismail et al., 2008; Ismail et al., 2006; Laird and High, 1997; Meacock et al., 2002). Sec61β has only one cysteine residue, and therefore a single Sec61β molecule is unable to form BMH-mediated crosslinking adducts with the nascent P2X2 chain and Sec61α simultaneously. Thus, since P2X2 × Sec61α × Sec61β adducts are only observed from C-terminal probe sites, crosslinking to this region of TM1 is most likely from a different cysteine residue of Sec61α than the adducts formed with the P2X2 TM1 N-terminus. In this interpretation, crosslinking to Sec61α from the N-terminal probes in TM1 is competed for by crosslinking to Sec61β, such that the secondary adducts representing P2X2 × Sec61β × Sec61α are limited. Thus, cysteine probes positioned at each end of P2X2 TM1 report distinct ER translocon environments when analysed by crosslinking (Fig. 1B,C,D). This TM1 profile is quantitatively consistent with the crosslinking profile of P2X2-108 chains (data not shown) and qualitatively consistent with the P2X2-88 chains (data not shown), indicating that during the early stages of biosynthesis, TM1 is stabilised with respect to the ER translocon. These data therefore allow for the rational selection of specific probe sites within P2X2 TM1 that reflect the behaviour of the entire span. To this end, it is clear that probes within both the N- and C-terminal regions of TM1 should be analysed in order to gain a comprehensive understanding of TM1 behaviour during later stages of P2X2 biosynthesis. Thus, an obvious candidate for such an extended analysis is the cysteine probe at position 56, because crosslinking from this location is of the highest overall efficiency and therefore likely to be particularly informative in describing the molecular environment of TM1. A second candidate is position 29; a probe at the N-terminus of TM1 that consistently crosslinks to TRAM and Sec61β with good efficiency and also forms adducts with Sec61α, probably via a different cysteine to that targeted by the probe at position 56. We conclude that by combining data derived from these two crosslinking probes, an accurate description of the membrane integration of TM1 can be established during the later stages of P2X2 biosynthesis.

Fig. 1.

Quantitative crosslinking analysis of P2X2-127. (A) The domain arrangement and TM segment sequence of P2X2. Residues shown in bold were mutated to cysteine (Jiang et al., 2001) and analysed by a site-specific crosslinking analysis in this study. In vitro translations of 25 different 127-residue N-terminal fragments of P2X2 were performed in a rabbit reticulocyte lysate system supplemented with ER microsomes and [35S]Met/Cys. Crosslinking was induced by the addition of BMH before immunoprecipitation, to identify adducts with ER translocon components. This analysis was then repeated and adducts quantified by 2D densitometry using AIDA v3.44 and standardised against the nascent P2X2 chain for each transcript (supplementary material Fig. S4). Mean relative crosslinking efficiency to Sec61α (B; grey bars), Sec61α × Sec61β (B; white bars), Sec61β (C) and TRAM (D) are shown in arbitrary units (AU).

Fig. 1.

Quantitative crosslinking analysis of P2X2-127. (A) The domain arrangement and TM segment sequence of P2X2. Residues shown in bold were mutated to cysteine (Jiang et al., 2001) and analysed by a site-specific crosslinking analysis in this study. In vitro translations of 25 different 127-residue N-terminal fragments of P2X2 were performed in a rabbit reticulocyte lysate system supplemented with ER microsomes and [35S]Met/Cys. Crosslinking was induced by the addition of BMH before immunoprecipitation, to identify adducts with ER translocon components. This analysis was then repeated and adducts quantified by 2D densitometry using AIDA v3.44 and standardised against the nascent P2X2 chain for each transcript (supplementary material Fig. S4). Mean relative crosslinking efficiency to Sec61α (B; grey bars), Sec61α × Sec61β (B; white bars), Sec61β (C) and TRAM (D) are shown in arbitrary units (AU).

P2X2 TM1 occupies a different molecular environment during late-phase integration

The integration of P2X2 TM1 was first examined by crosslinking from a probe incorporated at position 56, just flanking the first TM segment. Transcripts encoding P2X2[Q56C] of different chain lengths were generated and translated in an in vitro reaction before crosslinking was induced by the addition of BMH. At a chain length of 127 residues, P2X2[Q56C] was found to be efficiently crosslinked to Sec61α and Sec61α × Sec61β (Fig. 2A, α and αβ; compare lanes 2 and 4 for 127-residue chain), as previously observed (see Fig. 1B). When the chain length was increased to 148 residues, these same adducts were observed, although the efficiency of their formation appeared to be reduced (Fig. 2A, α and αβ; compare lanes 2 and 4 for 148-residue chain). Similarly, the formation of these adducts with Sec61α was further reduced at a chain length of 168 residues, and by 188 residues, only faint primary adducts with Sec61α and an unidentified product of ∼45 kDa were observed (Fig. 2A, α, αβ and asterisk; compare lanes 2 and 4 for 168- and 188-residue chains). At a chain length of 208 residues, the unidentified ∼45 kDa product was observed, albeit faintly, in both samples immunoprecipitated with anti-HA and anti-Sec61α antisera (Fig. 2A, asterisk; compare lanes 2 and 4 for 168-residue chain and lanes 2 and 4 for 208-residue chain). However, this product did not migrate as expected for an ∼22 kDa P2X2 nascent chain crosslinked to Sec61α, and in order to clarify its identity, a sequential IP approach was exploited. Although the adduct was recovered in a control reaction (Fig. 2B, asterisk lane 3), primary IP with anti-HA antiserum followed by secondary IP with anti-Sec61α antiserum failed to recover the product (Fig. 2B, lane 5; supplementary material Fig. S3). Thus, the ∼45 kDa product is not an authentic adduct between the precisely truncated 208-residue P2X2 nascent chain and Sec61α but might instead represent adducts with shorter P2X2 chains resulting from ribosome stacking (Ismail et al., 2006). At chain lengths of 228, 248 and 268, no adducts with Sec61α were observed (Fig. 2A, see 228, 248 and 268). One additional BMH-dependent adduct was consistently present for all chain lengths analysed (Fig. 2A,B, labelled `D'; see also below). Thus, as the P2X2 nascent chain length is increased to simulate later stages of the integration process, crosslinking to Sec61α is diminished, and by 208 residues, it is completely absent. Therefore, either the first TM segment of P2X2 has left the proximity of the ER translocon at this stage of biosynthesis, or alternatively the C-terminal region of TM1 is relocated relative to the ER translocon so as to occupy a molecular environment distinct from that occupied during the early stages of membrane integration.

Fig. 2.

BMH-mediated crosslinking analysis of P2X2[Q56C] integration intermediates. (A) Transcripts encoding HA-tagged P2X2 fragments of varying chain length, each with a single cysteine residue at position 56 were translated in an in vitro system as described in the legend to Fig. 1. Following crosslinking with BMH, samples were subjected to IP with anti-HA (HA), anti-Sec61α (α) or non-related antisera (NR). An aberrant BMH-dependent product is labelled with an asterisk, and a further adduct identified by `D' might result from unstable nascent chains prematurely released from the ribosome (see Fig. 6 and associated text). (B) The translation and crosslinking of P2X2[Q56C]-208HA was repeated and the resulting samples either analysed directed following anti-HA IP (lanes 1 and 3) or subjected to a second round of IP using either anti-HA (lane 3), anti-Myc (lane 4) or anti-Sec61α (lane 5) antisera.

Fig. 2.

BMH-mediated crosslinking analysis of P2X2[Q56C] integration intermediates. (A) Transcripts encoding HA-tagged P2X2 fragments of varying chain length, each with a single cysteine residue at position 56 were translated in an in vitro system as described in the legend to Fig. 1. Following crosslinking with BMH, samples were subjected to IP with anti-HA (HA), anti-Sec61α (α) or non-related antisera (NR). An aberrant BMH-dependent product is labelled with an asterisk, and a further adduct identified by `D' might result from unstable nascent chains prematurely released from the ribosome (see Fig. 6 and associated text). (B) The translation and crosslinking of P2X2[Q56C]-208HA was repeated and the resulting samples either analysed directed following anti-HA IP (lanes 1 and 3) or subjected to a second round of IP using either anti-HA (lane 3), anti-Myc (lane 4) or anti-Sec61α (lane 5) antisera.

TM1 is retained at the ER translocon throughout P2X2 biosynthesis

To ensure a comprehensive understanding of the integration of P2X2 TM1 at the ER translocon, a second probe position in TM1 was selected to represent the N-terminal region of the TM segment. Hence, nascent chains representing P2X2[L29C] of increasing chain lengths were translated in vitro and subjected to a crosslinking analysis. At a chain length of 127 residues, crosslinking adducts between L29C in TM1 and each of Sec61α, Sec61β and TRAM were again detected following BMH treatment by parallel IP reactions using antisera specifically recognising these components (Fig. 3A, α, β and T; compare lanes 2 to 5 for 127-residue chain; compare with Fig. 1). Adducts to ER translocon components were also observed when the chain length was increased to 148, 168, 188 and 208 residues, consistent with the analysis of P2X2[Q56C] (Fig. 3A, α, β and T; compare lanes 2-5 for 168- and 208-residue chains, respectively, and quantification in Fig. 3B). At the longer chain lengths, several unidentified BMH-dependent adducts were also recovered by IP using anti-HA antiserum (Fig. 3A, asterisk; lane 2 for 148- to 335-residue chains). When the nascent chain length was further extended, adducts with each of Sec61α, Sec61β and TRAM persisted, and strikingly, crosslinking to the ER translocon was observed for all nascent chains analysed, including one equivalent to the full-length P2X2 stalled on the ribosome (Fig. 3A, α, β and T; compare lanes 2-5 for 168- to 480-residue chains). Quantification of these adducts indicated that the efficiency of crosslinking of P2X2 TM1 to ER translocon components was relatively consistent for chain lengths between 208 and 480 residues, and is unaffected by the synthesis of TM2 (Fig. 3B). Thus, although the C-terminal region of TM1 appeared to relocate during the synthesis of the luminal loop region in P2X2, the molecular environment of the N-terminal region of P2X2 TM1 was unchanged during late-phase integration. This region of TM1 remained adjacent to the ER translocon throughout the entirety of P2X2 biosynthesis, despite the synthesis of a polypeptide tether between TM1 and the ribosome that was more than 300 residues long.

P2X2 is dislocated from the ER translocon by translational termination

To investigate this remarkable behaviour further, the specificity of adduct formation was examined using a puromycin-release experiment. When P2X2[L29C]-127 was again translated and subjected to crosslinking, adducts with Sec61α, Sec61β and TRAM were observed as previously (Fig. 4A, α, β and T; compare lanes 2 with lanes 3, 4 and 5). When the translation reaction was treated with puromycin and EDTA to release the nascent chain from the ribosome before crosslinking, only an adduct with an ∼15 kDa component was formed, and no adducts with ER translocon components were observed (Fig. 4A, labelled `D'; lanes 6-10) (see also below). Similarly, when a nascent chain representing the entire P2X2 polypeptide with a single cysteine residue at position 29 in TM1 was analysed in the same way, adducts with Sec61α, Sec61β and TRAM were observed as previously, albeit faintly, and were also lost upon puromycin treatment (Fig. 4B, α, β and T; compare lanes 1-5 with lanes 6-10). Thus, when P2X2[L29C] nascent chains were artificially released from the ribosome using puromycin-EDTA to simulate the termination of translation and release of the ribosome from the ER translocon docking site, crosslinking to the ER translocon was lost. This suggests that TM1 has been fully released from the proteinaceous translocon site and into the lipid environment (see supplementary material Fig. S1) and confirms the specificity of the crosslinks to the stalled nascent chains.

Fig. 3.

BMH-mediated crosslinking analysis of P2X2[L29C] integration intermediates. (A) Transcripts encoding HA-tagged P2X2 fragments of varying chain length, each with a single cysteine residue at position 29 were translated in an in vitro system as described in Figs 1 and 2. At a chain length of 127 residues, the migration of adducts between the P2X2 polypeptide and Sec61α that are recovered by IP using anti-Sec61α antiserum (see 127, compare lanes 2 and 3) is distorted by a high concentration of immunoglobulin heavy chain in the reaction (data not shown). (B) Adducts from A were quantified by phosphorimaging as described in the legend to Fig. 1.

Fig. 3.

BMH-mediated crosslinking analysis of P2X2[L29C] integration intermediates. (A) Transcripts encoding HA-tagged P2X2 fragments of varying chain length, each with a single cysteine residue at position 29 were translated in an in vitro system as described in Figs 1 and 2. At a chain length of 127 residues, the migration of adducts between the P2X2 polypeptide and Sec61α that are recovered by IP using anti-Sec61α antiserum (see 127, compare lanes 2 and 3) is distorted by a high concentration of immunoglobulin heavy chain in the reaction (data not shown). (B) Adducts from A were quantified by phosphorimaging as described in the legend to Fig. 1.

TM2 is retained at the ER translocon throughout P2X2 biosynthesis

The first TM segment in P2X2 remains in an unchanged molecular environment throughout the latter stages of biosynthesis, even at nascent chain lengths that correspond to the synthesis, and presumptive presentation to the ER translocon, of the second P2X2 TM segment (see Fig. 3B). To investigate this effect further, the membrane integration of P2X2 TM2 was analysed using a site-specific crosslinking approach. Cysteine residues were introduced into the second TM segment in P2X2 in positions representing the N-terminal, middle and C-terminal regions, to reflect the experiments with P2X2 TM1 (see Fig. 1 and Fig. 5, diagram). Nascent P2X2 chains of 413 residues in length with cysteine residues at position T330, S340 and W350 together with a ΔCys variant, were initially translated in vitro and subjected to BMH-mediated crosslinking. At this chain length, P2X2 TM2 is predicted to be sufficiently far from the ribosome peptidyl transferase centre to engage with the ER translocon.

Crosslinking of P2X2[T330C]-413 resulted in the formation of a rather diffuse adduct that was also recovered by IP using anti-Sec61α antiserum (Fig. 5A, α; compare lanes 2 and 3 for T330C). In addition, crosslinking to an unidentified adduct with an ∼30 kDa component was observed (Fig. 5A, see χ lane 2 for T330C). When the cysteine probe was present in the middle region of TM2, weak crosslinking to both Sec61α and very weak crosslinking to Sec61β was observed (Fig. 5A, α and β; compare lanes 2 and 4 for S340C). BMH crosslinking from the probe positioned at the C-terminus of TM2 resulted in discrete adduct formation with both Sec61α and Sec61β (Fig. 5A, α and β; compare lanes 2 and 4 for W350C), whereas no adducts were observed for the nascent chains that lacked any cysteine residues (Fig. 5A, ΔCys), confirming the specificity of adduct formation. Thus, at this stage of biosynthesis, P2X2 TM2 was found adjacent to both Sec61α and Sec61β and has therefore engaged the ER translocon. At 480 residues long, the second TM segment of P2X2 is separated from the ribosome by a 130 amino acid polypeptide tether, which is predicted to be long enough to allow the free diffusion of the TM span into the lipid phase and away from the ER translocon. Crosslinking of P2X2[T330C]-480 resulted in the formation of a clear adduct, which was also recovered by IP using anti-Sec61α (Fig. 5B, α; compare lanes 2 and 4 for T330C). This adduct was not present when the reaction was treated with puromycin before crosslinking (Fig. 5B, T330C Puromycin; compare lanes 1-4 and 5-7), indicating that the formation of the adduct is not due to stochastic events at the ER membrane. Taken together, these data indicate that TM2 engages the ER translocon and remains at a site adjacent to Sec61α throughout the remainder of P2X2 biosynthesis. Thus, P2X2 TM1 and TM2 occupy the ER translocon simultaneously, are retained at this site throughout biosynthesis, and are only released into the ER membrane by the termination of translation at the ribosome.

P2X2 TM1 fragments aggregate following premature release from the ER translocon

Selective TM-segment retention at the ER translocon is presently a poorly understood process (Ismail et al., 2008), and to better understand the significance of this effect during P2X2 biosynthesis, we sought to analyse the fate of TM1 when TM-segment retention was artificially bypassed. We previously observed the formation of an unidentified adduct with P2X2 TM1 following the puromycin-mediated release of the TM segment from the ER translocon site (Fig. 4A; supplementary material Fig. S3). The migration of these adducts following SDS-PAGE is consistent with the formation of dimers of P2X2 fragments, and to test this hypothesis directly, we used a co-translation of differentially tagged nascent chains and a sequential IP approach (see diagram Fig. 6A). A single prominent adduct was observed following crosslinking of the puromycin-released chains and this product was recovered by sequential IP using antisera specific to the HA- and V5-tagged nascent chains (Fig. 6B, Dimer; compare lanes 2, 6 and 7), indicating that P2X2 does indeed form dimers in the ER membrane following premature release from the ER translocon. Such an assembly is not predicted to occur upon the formation of the functional P2X2 ion channel (Barrera et al., 2005; Jiang et al., 2003; Murrell-Lagnado and Qureshi, 2008), and we therefore investigated the possibility that these dimers might report an aggregation event in the ER membrane. With a single cysteine probe present in the nascent chain, treatment with BMH can only induce the formation of a dimer and the presence of any higher-order oligomers would not be detected (see diagram, Fig. 6A). A second cysteine probe was therefore incorporated into the TM segment and the co-translation and sequential IP analysis repeated (see diagram, Fig. 6B). In this case, dimers of P2X2 TM1 were again observed (Fig. 6B, lane 7), but several larger adducts were also present (Fig. 6B, lane 7), consistent with higher order oligomerisation.

Fig. 4.

P2X2 TM1 is dislocated from the ER translocon on termination of translation. (A) P2X2[L29C]-127 was translated in an in vitro system and treated with either cycloheximide (lanes 1-5) or puromycin (lanes 6-10) before BMH-mediated crosslinking and immunoprecipitation. A puromycin-dependent adduct that is not recognised by anti-translocon antisera is labelled `D' (see Fig. 6 and associated text). (B) P2X2[L29C]-480 was translated, crosslinked and subjected to IP as described in A.

Fig. 4.

P2X2 TM1 is dislocated from the ER translocon on termination of translation. (A) P2X2[L29C]-127 was translated in an in vitro system and treated with either cycloheximide (lanes 1-5) or puromycin (lanes 6-10) before BMH-mediated crosslinking and immunoprecipitation. A puromycin-dependent adduct that is not recognised by anti-translocon antisera is labelled `D' (see Fig. 6 and associated text). (B) P2X2[L29C]-480 was translated, crosslinked and subjected to IP as described in A.

These observations are strongly indicative of the aggregation of P2X2 TM1 fragments, and to examine this striking property further, the behaviour of the fragment following expression in cultured mammalian cells was investigated. Hence, an HA-tagged 208-residue N-terminal fragment of wild-type P2X2 was expressed in HeLa cells and the sedimentation of the polypeptide was analysed by velocity gradient centrifugation of detergent solubilised cellular extracts. A native N-glycosylation site at position 182 in this fragment provides a useful in vivo report of ER membrane integration. Although a small proportion of P2X2-208 was recovered in each fraction of the sucrose gradient, a clear peak was observed in fractions 2-4 with additional material in fraction 8 (Fig. 7A). However, a substantial proportion of both the glycosylated and the non-glycosylated P2X2 fragment was recovered in the pellet fraction (Fig. 7A, labelled P), indicating it was present in very high molecular weight complexes, consistent with aggregation. By contrast, the full-length P2X2 protein was found to co-sediment with markers of 140-232 kDa in fractions 5-7 (Fig. 7B), consistent with the formation of a ∼200 kDa P2X2 trimer that represents the functional ion channel (Murrell-Lagnado and Qureshi, 2008). A fragment comprising only the first TM segment of the opsin protein was also analysed in the same way (Op91), and although a small population of the non-glycosylated Op91 was found in the pellet fraction, most of this polypeptide, and all of the glycosylated population was recovered in the lower molecular weight fractions (Fig. 7C, compare fractions 1-5 with P). Thus, the P2X2 TM1 fragment appears to have a high propensity to aggregate in vivo. Importantly, and in contrast to the opsin fragment, this aggregation very clearly extends to the N-glycosylated polypeptides, and therefore includes the membrane-integrated population of the P2X2 fragment.

Fig. 5.

TM2 is retained at the ER translocon during P2X2 biosynthesis. (A) Transcripts encoding 413-residue truncated fragments of P2X2 incorporating a C-terminal HA tag were generated that included a single cysteine residue at either position 330, 340 or 350 within TM2 or were without cysteine residues (ΔCys). These transcripts were translated and crosslinked in vitro as described in the legend to Figs 1 and 2. An unidentified adduct with P2X2[T330C]-413 is labelled `χ'. (B) Transcripts equivalent to the full-length P2X2 and an HA tag, but lacking a stop codon, were synthesised in vitro and subjected to crosslinking. IP was for either the anti-HA (HA), anti-Sec61α (α) or non-related (NR) antisera.

Fig. 5.

TM2 is retained at the ER translocon during P2X2 biosynthesis. (A) Transcripts encoding 413-residue truncated fragments of P2X2 incorporating a C-terminal HA tag were generated that included a single cysteine residue at either position 330, 340 or 350 within TM2 or were without cysteine residues (ΔCys). These transcripts were translated and crosslinked in vitro as described in the legend to Figs 1 and 2. An unidentified adduct with P2X2[T330C]-413 is labelled `χ'. (B) Transcripts equivalent to the full-length P2X2 and an HA tag, but lacking a stop codon, were synthesised in vitro and subjected to crosslinking. IP was for either the anti-HA (HA), anti-Sec61α (α) or non-related (NR) antisera.

Although the propensity for aggregation of P2X2 TM1 fragments might be potentially damaging to a cell, it is unclear how such aggregation could affect the function of P2X2 ion channels. To investigate this, full-length P2X2 and the TM1-only fragment were co-expressed, and the sedimentation of the two proteins reanalysed when present in the same cells. In the presence of the overexpressed full-length P2X2, the P2X2-208 fragment had a different profile to that observed in its absence. Less of the fragment was apparent in the low molecular weight fractions with detectable levels of polypeptide recovered in most of fractions of the gradient and no obvious peak (Fig. 7D, see fractions 2-11). This fragment was again recovered in the pellet, including the N-glycosylated form of the protein (Fig. 7D,P), indicating that the P2X2-208 fragment was still present in high molecular weight aggregates. In the case of the full-length protein, the sedimentation profile was also now more diffuse, with P2X2 present in each of fractions 3-9, although the peak population was again observed in fraction 6 (Fig. 7E). Most strikingly however, a distinct population of full-length P2X2 was now found in the pellet fraction (Fig. 7E,P), indicating that it was now present in high molecular weight species, consistent with aggregation. This behaviour suggests that should TM1 be prematurely released from the ER translocon during normal P2X2 biosynthesis, it has the potential to assemble incorrectly and/or aggregate. Furthermore, it appears that the presence of TM1-containing fragments can interfere with the assembly of the full-length protein and might promote the aggregation of otherwise functional ion channels. Thus, when taken together with our previous crosslinking analysis, these data suggest that the prolonged retention of TM1 at the ER translocon might be necessary to prevent detrimental effects to the cell resulting from aberrant TM1 assembly during P2X2 receptor biosynthesis.

Fig. 6.

P2X2 TM1 forms aberrant oligomers in the ER membrane. (A) P2X2-208 transcripts incorporating either an HA or V5 tag at the C-terminus and with a single cysteine probe were co-translated in vitro and treated with puromycin to simulate premature release into the ER membrane before crosslinking with BMH. Samples were either analysed directly, by a single round of immunoprecipitation or by sequential immunoprecipitation as indicated. (B) Dual probe P2X2-208 transcripts with cysteine residues at position 52 and 56 were synthesised and analysed as described in A. The P2X2-208 nascent chain is indicated by `208', and putative higher oligomers are indicated.

Fig. 6.

P2X2 TM1 forms aberrant oligomers in the ER membrane. (A) P2X2-208 transcripts incorporating either an HA or V5 tag at the C-terminus and with a single cysteine probe were co-translated in vitro and treated with puromycin to simulate premature release into the ER membrane before crosslinking with BMH. Samples were either analysed directly, by a single round of immunoprecipitation or by sequential immunoprecipitation as indicated. (B) Dual probe P2X2-208 transcripts with cysteine residues at position 52 and 56 were synthesised and analysed as described in A. The P2X2-208 nascent chain is indicated by `208', and putative higher oligomers are indicated.

Fig. 7.

Sucrose gradient centrifugation of in vivo expressed P2X2. (A) Transiently expressed wild-type P2X2-199 incorporating a C-terminal HA tag followed by an amber stop codon was analysed by velocity centrifugation. The TCA-precipitated fractions were probed by immunoblotting using anti-HA antiserum. Glycosylated P2X2-208 is labelled `P2X2-208ψ'. (B) Full-length wild-type P2X2 was analysed as described in A. Detection was by immunoblotting for the C-terminal Glu-Glu epitope. All of the protein was found in the glycosylated form, and is indicated by `P2X2-FL'. (C) A fragment of opsin representing only the first TM segment of the protein was analysed as described in A, and detected by immunoblotting using an anti-opsin antiserum. Doubly glycosylated Op91 is indicated by `OP91ψ'. (D) Full-length P2X2 and the TM1-only P2X2 fragment were co-expressed and analysed as described in A. (E) Samples from D were probed with the anti-Glu-Glu antiserum as described in B. Fractions are as indicated; P indicates the pellet fraction and the sedimentation of native molecular mass (kDa) markers is shown by unfilled triangles.

Fig. 7.

Sucrose gradient centrifugation of in vivo expressed P2X2. (A) Transiently expressed wild-type P2X2-199 incorporating a C-terminal HA tag followed by an amber stop codon was analysed by velocity centrifugation. The TCA-precipitated fractions were probed by immunoblotting using anti-HA antiserum. Glycosylated P2X2-208 is labelled `P2X2-208ψ'. (B) Full-length wild-type P2X2 was analysed as described in A. Detection was by immunoblotting for the C-terminal Glu-Glu epitope. All of the protein was found in the glycosylated form, and is indicated by `P2X2-FL'. (C) A fragment of opsin representing only the first TM segment of the protein was analysed as described in A, and detected by immunoblotting using an anti-opsin antiserum. Doubly glycosylated Op91 is indicated by `OP91ψ'. (D) Full-length P2X2 and the TM1-only P2X2 fragment were co-expressed and analysed as described in A. (E) Samples from D were probed with the anti-Glu-Glu antiserum as described in B. Fractions are as indicated; P indicates the pellet fraction and the sedimentation of native molecular mass (kDa) markers is shown by unfilled triangles.

We selected a novel model polytopic protein and conducted a detailed crosslinking analysis to examine the process of membrane protein integration at the ER translocon. P2X2 has two TM segments separated by an extended lumenal loop region, which is required to form the ATP-binding domain of the ion channel at the cell surface (North, 2002). Given this arrangement, the question of how the individual TM segments are integrated into the lipid phase by the ER translocon is especially salient. In exceptional circumstances, polytopic membrane proteins with extended cytosolic loop regions have been reported to require a second round of targeting by SRP during biosynthesis, suggesting that the ER translocon treats such proteins as two independent polypeptides (Kuroiwa et al., 1996). Similarly, TM-segment integration into the lipid phase for polytopic membrane proteins might occur sequentially when each TM segment is presented to the ER translocon, provided that a suitably extended polypeptide tether to the ribosome is available (Heinrich and Rapoport, 2003; Ismail et al., 2006; Sadlish et al., 2005; Sauri et al., 2005). In this study, the generality of these observations was challenged with a new polypeptide model.

A comprehensive, quantitative scanning analysis of crosslinking from each of the possible probe sites within TM1, and several locations flanking this segment, revealed a distinct profile of BMH-mediated adduct formation. Hence, although the N-terminal region of TM1 was found to crosslink to each of Sec61α, Sec61β and TRAM, the C-terminal probe sites formed adducts primarily with Sec61α. This pattern is consistent with previous crosslinking analyses (Ismail et al., 2008; Ismail et al., 2006; Meacock et al., 2002), and most likely reflects the position and availability of suitable cysteine residues in the partner proteins of the ER translocon. This profile was exploited to inform the rational selection of probe sites within the TM segment that could most accurately represent the entire region. Probes incorporated at position 56 of the nascent P2X2 chain gave rise to particularly high crosslinking efficiency to Sec61α at the early phases of P2X2 biosynthesis. This high efficiency is likely to be, in part, a result of the position of this probe just flanking the C-terminus of the first TM segment, within a potentially more flexible region of the polypeptide. Crosslinking to Sec61α from this location is all the more striking for its complete absence at the later stages of P2X2 biosynthesis. Loss of crosslinking to the ER translocon can be used to report the release of a TM segment from a proteinaceous environment and into the lipid phase of the membrane (Heinrich and Rapoport, 2003; Ismail et al., 2008; Ismail et al., 2006; Martoglio et al., 1995; McCormick et al., 2003; Mothes et al., 1997; Sadlish and Skach, 2004; Sauri et al., 2005); however, in this study we extended our analysis to examine this effect more closely. Strikingly, we found that crosslinking to the ER translocon from a probe incorporated into the N-terminal region of the first TM segment is observed at all stages of P2X2 biosynthesis. Thus, although the first TM segment of P2X2 is anchored to the ER translocon at the N-terminal region, the C-terminal region occupies a distinct environment, resulting either from conformational changes within the P2X2 chain or the ER translocon, or alternatively reflecting the lateral release of this sub-region of TM1 away from the ER translocon. Such rearrangements might be necessary to allow the simultaneous accommodation of the both P2X2 TM segments at the ER translocon, as suggested by this study.

The retention of P2X2 TM1 at the ER translocon is an extreme example of this phenomenon, being apparent during the synthesis of more than 400 amino acids, and supports the idea that a largely unappreciated regulatory function is exhibited by the ER translocon during the membrane integration of polytopic membrane proteins (Ismail et al., 2008; Pitonzo et al., 2009). Hence, although the integration of single-spanning membrane protein TM segments into the ER membrane appears to be strongly influenced by the composition and properties of the TM helix (Heinrich et al., 2000; Hessa et al., 2005), polytopic protein integration presents distinct challenges to the ER translocon and demands additional modulation. The TM segments of polytopic proteins have a high prevalence of functionally essential charged or polar amino acids that could compromise effective membrane partitioning. Thus, the retention of one or more TM segments at the ER translocon to generate TM segment bundles might facilitate partitioning into the lipid phase (Ismail et al., 2008; Meindl-Beinker et al., 2006). In the case of P2X2 TM1, retention at the translocon pending the synthesis of TM2 might allow for the assembly of the two helices into a conformation that masks polar residues within TM1 (H33, R34 and Q37) and TM2 (D349), reducing the thermodynamic cost of membrane partitioning (Zhang et al., 2007). In support of this hypothesis, substitution of this highly conserved Asp for an uncharged residue in TM2 of a closely related P2X family member, caused an assembly defect and loss of function of the ion channel (Duckwitz et al., 2006). Hence, for P2X family members, intramolecular bonding between this residue in TM2 and TM1 (most likely at R34 in P2X2) could be essential for functional assembly of the protein (Murrell-Lagnado and Qureshi, 2008) (Fig. 8). Such an effect strongly underlines the importance of cooperativity between membrane-spanning segments during their integration into the ER membrane (Heinrich and Rapoport, 2003; Lin and Addison, 1995; Skach and Lingappa, 1993; Zhang et al., 2007).

Fig. 8.

Model of P2X2 biosynthesis at the ER translocon. During the early stages of membrane integration, premature release of the TM1 segment from the translocon environment causes potentially toxic aggregation of the polypeptide. The retention of P2X2 TM1 by the ER translocon throughout the entire of biosynthesis prevents this aggregation and allows the formation of intra-molecular bonding between TM1 (at R34) and TM2 (at D349), generating an integration- and assembly-competent P2X2 conformation.

Fig. 8.

Model of P2X2 biosynthesis at the ER translocon. During the early stages of membrane integration, premature release of the TM1 segment from the translocon environment causes potentially toxic aggregation of the polypeptide. The retention of P2X2 TM1 by the ER translocon throughout the entire of biosynthesis prevents this aggregation and allows the formation of intra-molecular bonding between TM1 (at R34) and TM2 (at D349), generating an integration- and assembly-competent P2X2 conformation.

Although the precise molecular details of the mechanism for TM-segment retention at the ER translocon are yet to be clearly defined, it is interesting that we find the polar residues within the first TM segment of P2X2 are retained directly adjacent to TRAM. This component has previously been suggested to perform a TM-chaperoning role during the integration of suboptimal TM helices (Hegde and Kang, 2008; Heinrich et al., 2000). Our data support this model for TRAM function, and further suggest a more specific role for this component in TM-segment retention, consistent with previous studies (Do et al., 1996). Furthermore, the behaviour of P2X2 TM1 is in stark contrast to the first TM segment of two other double-spanning membrane proteins studied previously, which were found to exit the translocon site provided a tether of ∼100 residues to the ribosome was synthesised (Heinrich and Rapoport, 2003; Sauri et al., 2005). In both of these studies, the second TM segment of the protein was either artificial (Heinrich and Rapoport, 2003) or was artificially separated from the first TM segment (Sauri et al., 2005). Hence, the first TM segment of these polypeptides is unlikely to have an implicit requirement or signal for retention, because the need for this process does not arise in the native protein. Thus, when compared with the results presented here, these studies indicate that TM-segment retention at the ER translocon is a selective process (Ismail et al., 2008).

In this study, we wished to explore the potential physiological basis for TM-segment retention at the ER translocon and we therefore examined the post-translocon fate of P2X2. We found that fragments representing only the first TM segment of P2X2 that are prematurely released from the ER translocon site assemble into higher-order oligomers in vitro and display a high propensity to aggregate in vivo (see Fig. 8). This behaviour is entirely consistent with that of P2X5, which was shown to aggregate in vivo when expressed in the absence of the TM2 coding region (Duckwitz et al., 2006). Naturally occurring splice variants of P2X5 and P2X7 have been identified that lack TM2, and these are both non-functional and retained at the ER (Duckwitz et al., 2006; Feng et al., 2006). In the case of the truncated P2X7 molecule, co-expression with full-length P2X7 was shown to antagonise receptor function, most likely by non-productive co-assembly (Feng et al., 2006). We also observed that P2X2 fragment expression promotes the aggregation of the full-length P2X2 molecule. It seems likely therefore, that the presence of P2X2 fragment aggregates in the ER membrane could also cause loss of P2X2 function. In contrast to P2X2, a single-TM opsin fragment examined in the same way was not as prone to aggregation, most likely as a result of its association with a variety of presumptive ER quality control factors following its release from the ER translocon (Crawshaw et al., 2007; Crawshaw et al., 2004).

Thus, the retention of P2X2 TM1 at the ER translocon provides clear physiological advantages during biosynthesis by promoting the correct folding and function of the P2X2 ion channel. In this model, the ER translocon specifically regulates the integration of the P2X2 protein to prevent the potentially damaging effects of TM1 aggregation by facilitating the formation of essential interactions between the two P2X2 TM segments. Our model is consistent with a recent analysis of CFTR biogenesis (Pitonzo et al., 2009) and supports the emerging view that the biogenesis and membrane integration of polytopic proteins can only be partially accounted for by the thermodynamic scale of TM segment partitioning derived for simple membrane proteins (Bernsel et al., 2008; Hessa et al., 2005; Park and Helms, 2008; White and von Heijne, 2008). In particular, this analysis of P2X2 biosynthesis highlights the importance of TM segment cooperativity during membrane integration (Buck et al., 2007; Lin and Addison, 1995; Meindl-Beinker et al., 2006; Skach and Lingappa, 1993; Zhang et al., 2007) and provides direct experimental evidence that the ER translocon facilitates this process.

In vitro transcription and translation

Plasmids encoding rat P2X2 and single-cysteine and Cys-null derivatives lacking native P2X2 N-glycosylation sites in pcDNA3 (Jiang et al., 2001) were used as templates for the generation of mRNA coding transcripts as previously described (Ismail et al., 2008). Translation of these transcripts in vitro was in the presence of ER-derived microsomes, essentially as described previously (Ismail et al., 2008). Artificial release of the nascent chain from the ribosome was induced by the addition of 1 mM puromycin and 6 mM EDTA followed by incubation for 10 minutes at 37°C.

Alkaline extraction and endoglycosidase H treatment

Following isolation of the membrane associated fraction from in vitro translation reactions, peripheral proteins was distinguished from integral membrane proteins by a sodium carbonate solution extraction procedure (Fujiki et al., 1982) and samples were analysed by trichloroacetic acid (TCA) precipitation and SDS-PAGE, as described previously (Callan et al., 2007). Removal of high mannose N-linked glycan chains was performed either on the resuspended Sepharose beads (Genscript Corporation) following immunoprecipitation or directly in a volume-adjusted sample. Initial denaturation was in the presence of 0.5% SDS and 1% β-mercaptoethanol for 30 minutes at 37°C. Endoglycosidase H (500 U) was then added with digestion buffer (0.25 M sodium citrate, pH 5.5, 5 mM PMSF). This reaction was then incubated at 37°C for 30 minutes before preparation for SDS-PAGE.

Crosslinking and immunoprecipitation

To the resuspended membrane fraction, bismaleimidohexane (BMH) prepared in DMSO was added to 1 mM final concentration and incubated at 30°C for 10 minutes. Quenching of the crosslinker was by addition of β-mercaptoethanol (βME) to a final concentration of 5 mM and incubation for a further 5 minutes at 30°C. RNaseA (250 μg/ml) treatment of all crosslinked samples and incubation for 5 minutes at 37°C digested any remaining peptidyl tRNA species in the samples. Samples were then prepared for IP as described previously (Meacock et al., 2002).

Mammalian cell culture and sucrose gradient centrifugation

HeLa cells were grown in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal calf serum (FCS) and L-glutamine (Lonza, Cambridge, UK). Transfection of HeLa cells with mammalian expression vectors encoding full-length P2X2, P2X2-208HA and Op91 in pZeo, each including stop codons, was using Lipofectamine 2000 according to the manufacturer's instructions (Invitrogen). 10 cm dishes of transfected cells were incubated overnight at 37°C and washed twice in phosphate buffered saline (PBS) before solubilisation in 1% C12E8 solublisation buffer [1% C12E8 (w/v), 10 mM Tris-HCl pH 7.6, 140 mM NaCl, 1 mM EDTA, 1 mM PMSF, 1:100, mammalian protease inhibitor cocktail (Sigma)] by gentle agitation at 4°C for 1 hour. The resulting supernatant was then layered over a 12 ml 5-25% (w/v) continuous sucrose gradient [containing 0.1% C12E8 (w/v), 20 mM NaCl, 25 mM Tris-HCl pH 7.4] poured over a 0.5 ml 50% (w/v) sucrose cushion and centrifuged at 192,000 g for 16 hours at 4°C. Fractions (1 ml) were collected and prepared for SDS-PAGE. Protein standards used were: thyroglobulin, 38 μg; ferritin, 25 μg; catalase, 18 μg; lactate dehydrogenase, 24 μg; albumin, 20 μg (Amersham).

SDS-PAGE and sample analysis

SDS-PAGE was performed as described elsewhere (Ismail et al., 2008). Quantitative analysis was carried out using AIDA v3.44 (Raytest Isotopenmessgerate, Straubenhardt, Germany). Crosslinking product intensities were standardised against the intensity of the partner nascent chain to gain measures of relative crosslinking efficiency for each nascent chain. Graphs and statistical analyses were prepared using GraphPad Prism v4.0 for Macintosh (GraphPad Software).

This work was supported by the BBSRC and we acknowledge our colleagues for advice and important reagents. We are indebted to R. Alan North for his support and thank Jim Warwicker, Mark Young and Martin Pool for helpful comments during the preparation of this manuscript.

Barrera, N. P., Ormond, S. J., Henderson, R. M., Murrell-Lagnado, R. D. and Edwardson, J. M. (
2005
). Atomic force microscopy imaging demonstrates that P2X2 receptors are trimers but that P2X6 receptor subunits do not oligomerize.
J. Biol. Chem.
280
,
10759
-10765.
Bernsel, A., Viklund, H., Falk, J., Lindahl, E., von Heijne, G. and Elofsson, A. (
2008
). Prediction of membrane-protein topology from first principles.
Proc. Natl. Acad. Sci. USA
105
,
7177
-7181.
Buck, T. M., Wagner, J., Grund, S. and Skach, W. R. (
2007
). A novel tripartite motif involved in aquaporin topogenesis, monomer folding and tetramerization.
Nat. Struct. Mol. Biol.
14
,
762
-769.
Callan, A. C., Bunning, S., Jones, O. T., High, S. and Swanton, E. (
2007
). Biosynthesis of the dystonia-associated AAA+ ATPase torsinA at the endoplasmic reticulum.
Biochem. J.
401
,
607
-612.
Crawshaw, S. G., Martoglio, B., Meacock, S. L. and High, S. (
2004
). A misassembled transmembrane domain of a polytopic protein associates with signal peptide peptidase.
Biochem. J.
384
,
9
-17.
Crawshaw, S. G., Cross, B. C., Wilson, C. M. and High, S. (
2007
). The oligomeric state of Derlin-1 is modulated by endoplasmic reticulum stress.
Mol. Membr. Biol.
24
,
113
-120.
Cross, B. C. S. and High, S. (
2009
). Membrane protein biosynthesis at the endoplasmic reticulum. In
Protein Transport Into The Endoplasmic Reticulum
(ed. R. Zimmermann). Austin, TX: Landes Bioscience.
Daniel, C. J., Conti, B., Johnson, A. E. and Skach, W. R. (
2008
). Control of translocation through the Sec61 translocon by nascent polypeptide structure within the ribosome.
J. Biol. Chem.
283
,
20864
-20873.
Do, H., Falcone, D., Lin, J., Andrews, D. W. and Johnson, A. E. (
1996
). The cotranslational integration of membrane proteins into the phospholipid bilayer is a multistep process.
Cell
85
,
369
-378.
Duckwitz, W., Hausmann, R., Aschrafi, A. and Schmalzing, G. (
2006
). P2X5 subunit assembly requires scaffolding by the second transmembrane domain and a conserved aspartate.
J. Biol. Chem.
281
,
39561
-39572.
Feng, Y. H., Li, X., Wang, L., Zhou, L. and Gorodeski, G. I. (
2006
). A truncated P2X7 receptor variant (P2X7-j) endogenously expressed in cervical cancer cells antagonizes the full-length P2X7 receptor through hetero-oligomerization.
J. Biol. Chem.
281
,
17228
-17237.
Fujiki, Y., Hubbard, A. L., Fowler, S. and Lazarow, P. B. (
1982
). Isolation of intracellular membranes by means of sodium carbonate treatment: Application to endoplasmic reticulum.
J. Cell Biol.
93
,
97
-102.
Garrison, J. L., Kunkel, E. J., Hegde, R. S. and Taunton, J. (
2005
). A substrate-specific inhibitor of protein translocation into the endoplasmic reticulum.
Nature
436
,
285
-289.
Gilmore, R., Collins, P., Johnson, J., Kellaris, K. and Rapiejko, P. (
1991
). Transcription of full-length and truncated mRNA transcripts to study protein translocation across the endoplasmic reticulum.
Methods Cell Biol.
34
,
223
-239.
Hegde, R. S. and Kang, S. W. (
2008
). The concept of translocational regulation.
J. Cell Biol.
182
,
225
-232.
Heinrich, S. U. and Rapoport, T. A. (
2003
). Cooperation of transmembrane segments during the integration of a double-spanning protein into the ER membrane.
EMBO J.
22
,
3654
-3663.
Heinrich, S. U., Mothes, W., Brunner, J. and Rapoport, T. A. (
2000
). The Sec61p complex mediates the integration of a membrane protein by allowing lipid partitioning of the transmembrane domain.
Cell
102
,
233
-244.
Hessa, T., Kim, H., Bihlmaier, K., Lundin, C., Boekel, J., Andersson, H., Nilsson, I., White, S. H. and von Heijne, G. (
2005
). Recognition of transmembrane helices by the endoplasmic reticulum translocon.
Nature
433
,
377
-381.
Ismail, N., Crawshaw, S. G. and High, S. (
2006
). Active and passive displacement of transmembrane domains both occur during opsin biogenesis at the Sec61 translocon.
J. Cell Sci.
119
,
2826
-2836.
Ismail, N., Crawshaw, S. G., Cross, B. C., Haagsma, A. C. and High, S. (
2008
). Specific transmembrane segments are selectively delayed at the ER translocon during opsin biogenesis.
Biochem. J.
411
,
495
-506.
Jiang, L. H., Rassendren, F., Spelta, V., Surprenant, A. and North, R. A. (
2001
). Amino acid residues involved in gating identified in the first membrane-spanning domain of the rat P2X(2) receptor.
J. Biol. Chem.
276
,
14902
-14908.
Jiang, L. H., Kim, M., Spelta, V., Bo, X., Surprenant, A. and North, R. A. (
2003
). Subunit arrangement in P2X receptors.
J. Neurosci.
23
,
8903
-8910.
Kalies, K. U., Rapoport, T. A. and Hartmann, E. (
1998
). The beta subunit of the Sec61 complex facilitates cotranslational protein transport and interacts with the signal peptidase during translocation.
J. Cell Biol.
141
,
887
-894.
Kida, Y., Mihara, K. and Sakaguchi, M. (
2005
). Translocation of a long amino-terminal domain through ER membrane by following signal-anchor sequence.
EMBO J.
24
,
3202
-3213.
Kowarik, M., Kung, S., Martoglio, B. and Helenius, A. (
2002
). Protein folding during cotranslational translocation in the endoplasmic reticulum.
Mol. Cell
10
,
769
-778.
Kuroiwa, T., Sakaguchi, M., Omura, T. and Mihara, K. (
1996
). Reinitiation of protein translocation across the endoplasmic reticulum membrane for the topogenesis of multispanning membrane proteins.
J. Biol. Chem.
271
,
6423
-6428.
Laird, V. and High, S. (
1997
). Discrete cross-linking products identified during membrane protein biosynthesis.
J. Biol. Chem.
272
,
1983
-1989.
Lin, J. and Addison, R. (
1995
). A novel integration signal that is composed of two transmembrane segments is required to integrate the Neurospora plasma membrane H(+)-ATPase into microsomes.
J. Biol. Chem.
270
,
6935
-6941.
Martoglio, B., Hofmann, M. W., Brunner, J. and Dobberstein, B. (
1995
). The protein-conducting channel in the membrane of the endoplasmic reticulum is open laterally toward the lipid bilayer.
Cell
81
,
207
-214.
McCormick, P. J., Miao, Y., Shao, Y., Lin, J. and Johnson, A. E. (
2003
). Cotranslational protein integration into the ER membrane is mediated by the binding of nascent chains to translocon proteins.
Mol. Cell
12
,
329
-341.
Meacock, S. L., Lecomte, F. J., Crawshaw, S. G. and High, S. (
2002
). Different transmembrane domains associate with distinct endoplasmic reticulum components during membrane integration of a polytopic protein.
Mol. Biol. Cell
13
,
4114
-4129.
Meindl-Beinker, N. M., Lundin, C., Nilsson, I., White, S. H. and von Heijne, G. (
2006
). Asn- and Asp-mediated interactions between transmembrane helices during translocon-mediated membrane protein assembly.
EMBO Rep.
7
,
1111
-1116.
Mothes, W., Heinrich, S. U., Graf, R., Nilsson, I., von Heijne, G., Brunner, J. and Rapoport, T. A. (
1997
). Molecular mechanism of membrane protein integration into the endoplasmic reticulum.
Cell
89
,
523
-533.
Murrell-Lagnado, R. D. and Qureshi, O. S. (
2008
). Assembly and trafficking of P2X purinergic receptors (Review).
Mol. Membr. Biol.
25
,
321
-331.
North, R. A. (
2002
). Molecular physiology of P2X receptors.
Physiol. Rev.
82
,
1013
-1067.
Osborne, A. R., Rapoport, T. A. and van den Berg, B. (
2005
). Protein translocation by the Sec61/Secy channel.
Annu. Rev. Cell Dev. Biol.
21
,
529
-550.
Park, Y. and Helms, V. (
2008
). MINS2: revisiting the molecular code for transmembrane-helix recognition by the Sec61 translocon.
Bioinformatics
24
,
1819
-1820.
Pitonzo, D., Yang, Z., Matsumura, Y., Johnson, A. E. and Skach, W. R. (
2009
). Sequence-specific retention and regulated integration of a nascent membrane protein by the ER Sec61 translocon.
Mol. Biol. Cell
20
,
685
-698.
Pool, M. R. (
2005
). Signal recognition particles in chloroplasts, bacteria, yeast and mammals (review).
Mol. Membr. Biol.
22
,
3
-15.
Rapoport, T. A. (
2007
). Protein translocation across the eukaryotic endoplasmic reticulum and bacterial plasma membranes.
Nature
450
,
663
-669.
Sadlish, H. and Skach, W. R. (
2004
). Biogenesis of CFTR and other polytopic membrane proteins: new roles for the ribosome-translocon complex.
J. Membr. Biol.
202
,
115
-126.
Sadlish, H., Pitonzo, D., Johnson, A. E. and Skach, W. R. (
2005
). Sequential triage of transmembrane segments by Sec61alpha during biogenesis of a native multispanning membrane protein.
Nat. Struct. Mol. Biol.
12
,
870
-878.
Sauri, A., Saksena, S., Salgado, J., Johnson, A. E. and Mingarro, I. (
2005
). Double-spanning plant viral movement protein integration into the endoplasmic reticulum membrane is signal recognition particle-dependent, translocon-mediated, and concerted.
J. Biol. Chem.
280
,
25907
-25912.
Skach, W. R. and Lingappa, V. R. (
1993
). Amino-terminal assembly of human P-glycoprotein at the endoplasmic reticulum is directed by cooperative actions of two internal sequences.
J. Biol. Chem.
268
,
23552
-23561.
Van den Berg, B., Clemons, W. M., Jr, Collinson, I., Modis, Y., Hartmann, E., Harrison, S. C. and Rapoport, T. A. (
2004
). X-ray structure of a protein-conducting channel.
Nature
427
,
36
-44.
White, S. H. and von Heijne, G. (
2008
). How translocons select transmembrane helices.
Annu. Rev. Biophys.
37
,
23
-42.
Zhang, L., Sato, Y., Hessa, T., von Heijne, G., Lee, J. K., Kodama, I., Sakaguchi, M. and Uozumi, N. (
2007
). Contribution of hydrophobic and electrostatic interactions to the membrane integration of the Shaker K+ channel voltage sensor domain.
Proc. Natl. Acad. Sci. USA
104
,
8263
-8268.

Supplementary information