Revealing the mechanisms of membrane protein export by virulence-associated bacterial secretion systems

Abstract

Many bacteria export effector proteins fulfilling their function in membranes of a eukaryotic host. These effector membrane proteins appear to contain signals for two incompatible bacterial secretion pathways in the same protein: a specific export signal, as well as transmembrane segments that one would expect to mediate targeting to the bacterial inner membrane. Here, we show that the transmembrane segments of effector proteins of type III and type IV secretion systems indeed integrate in the membrane as required in the eukaryotic host, but that their hydrophobicity in most instances is just below the threshold required for mediating targeting to the bacterial inner membrane. Furthermore, we show that binding of type III secretion chaperones to both the effector's chaperone-binding domain and adjacent hydrophobic transmembrane segments also prevents erroneous targeting. These results highlight the evolution of a fine discrimination between targeting pathways that is critical for the virulence of many bacterial pathogens.

Fig. 1

Type III-secreted transmembrane proteins a Type III-secreted transmembrane proteins harbor an N-terminal type III secretion signal and a chaperone binding domain (CBD). They are targeted by cognate chaperones to the sorting platform (cytoplasmic components) of the T3SS injectisome. They are secreted into the host cell and incorporated into the host membrane. SipB (green) is a hydrophobic translocator of the SPI-1 T3SS of Salmonella, forming pores for T3SS effector translocation. SipB is targeted by the chaperone SicA. SseF (orange) and SseG (green) are effector proteins of the SPI-2 T3SS and targeted by the chaperone SscB. Both proteins interact with endosomal compartments e.g., the Golgi network. The Tir protein (blue) of EPEC is secreted through the T3SS and inserted into the host membrane in order to act as receptor for the adhesin intimin. Tir is targeted by the chaperone CesT. b T3SS substrates were analyzed using the full protein scan of the ΔG predictor (window: 18–35 aa, length correction: OFF). The histogram shows the distribution of the number of TMS of type III-secreted transmembrane proteins. CBD chaperone binding domain, IM inner membrane, OM outer membrane, SPI-1 Salmonella pathogenicity island 1, SPI-2 Salmonella pathogenicity island 2, TMD transmembrane domain, TMS transmembrane segment

Fig. 2

Prediction and experimental validation of membrane integration of TMS of T3SS substrates a Distribution of the calculated membrane integration propensity (ΔG_app) of TMS of T3SS substrates (red) compared to previously published values for ΔG_app of regular transmembrane- and soluble proteins, respectively. For each protein, only its lowest ΔG_app of any given sequence window is shown (ΔG predictor settings: window size: 18–35 aa, length correction: ON). b Principle of the TMS insertion assay, see results and methods sections for description. c Relative membrane insertion of the indicated TMS was analyzed in E. coli BW25113 with the TMS insertion assay. ± ara indicates induction of expression of the respective test construct. The fate of the Lep-LacY chimera was assessed by SDS PAGE, Western blotting and immunodetection of the Lep P2 domain. A representative result of three independent experiments is shown. The fraction of inserted Lep-LacY (f_I) was calculated from the ratio of uncleaved to cleaved (indicated by asterisk) Lep-LacY, corrected for different degradation rates of the individual fragments according to Öjemalm et al.. Abbreviations: ara: arabinose; Lep: leader peptidase; Lep-LacY (T+I): translocated and integrated membrane helix; Lep-LacY (T_small): small translocated form of Lep-LacY. IM inner membrane, TMS transmembrane segment

Fig. 3

Membrane targeting potential of TMS of type III-secreted transmembrane proteins (a) Calculation of ∆G_app for the SRP-targeting window of 12–17 aa (∆G_app^SRP) for transmembrane proteins of type III-secreted transmembrane proteins (red) compared to E. coli transmembrane proteins. The classification of E. coli membrane proteins is according to Schibich et al.: SRP substrates (dark gray), non SRP substrates (middle gray) and substrates, in which the first TMS was skipped by SRP (light gray). b, c Relevant TMS of T3SS substrates were assessed for their inner membrane targeting potential and membrane integration in a S-35 Met-based pulse-chase targeting assay using inverted leader-peptidase (Lep-inv, b) and ProW Nt/TM1/P2 (c), respectively. The principle of the assays is shown on the left. Membrane-inserted Lep-inv/ProW Nt/TM1/P2 were cleaved into a smaller fragment by exogenously added proteinase K. Lep-inv/ProW Nt/TM1/P2 that fails to insert into the membrane is not affected by proteinase K. Cleavage can be detected by immunoprecipitation. For assessment of targeting, the H1 segment of Lep-inv or ProW Nt/TM1/P2, respectively, was exchanged against the indicated segment of interest. Proteins were expressed in E. coli MC4100 from rhamnose-inducible plasmids. After spheroplasting and addition of proteinase K, proteins of interest were immunoprecipitated and analyzed by SDS PAGE and autoradiography of 35-S. The outer membrane protein OmpA served as control for successful proteinase K digestion. Band X (indicated by asterisk) was used as a control for intact spheroplasts. A representative result of three independent experiments is shown. IM inner membrane, TMS transmembrane segment

Fig. 4

Analysis of membrane integration of SseF in dependence of its chaperone SscB (a) Salmonella ΔsscB, ΔsseF, ΔsseG triple mutants were grown under SPI-2-inducing conditions and complemented with SseF^FLAG with or without the chaperone SscB from a rhamnose-inducible low-copy number plasmid. Crude membranes were prepared and treated with 8 M urea for 1 h at room temperature as indicated. SseF content was then analyzed by SDS PAGE, Western blotting and immunodetection with anti-FLAG and anti-YidC (inner membrane control) antibodies. b As in a but showing expression of indicated SseF^FLAG mutants with increased hydrophobicity of the first TMS. A representative result of three independent experiments is shown. Abbreviation: CM: crude membranes

Fig. 5

Protein-protein interaction analysis of SseF and SscB by in vivo photocrosslinking a Protter visualization of SseF presenting predicted TM topology, chaperone binding domain (blue), and positions analyzed by in vivo photocrosslinking (orange). b Immunodetection of SseF^FLAG in whole cell lysates with and without UV irradiation (upper panel). pBpa mutations are denotated as “X”. T3SS-dependent secretion of mutated SseF was analyzed by SDS PAGE, Western blotting and immunodetection of TCA-precipitated SseF^FLAG in culture supernatants (lower panel). c As in b but showing SseF^FLAG in a ΔsscB deletion background. d As in b but showing crosslinking position V73 of SseF in mutants harboring various leucine residues on the indicated positions of the first TMS of SseF. A representative result of three independent experiments is shown. CBD chaperone binding domain, WC whole cell lysates, sup culture supernatant, TMS transmembrane segment

Fig. 6

Model of targeting discrimination of secreted transmembrane proteins The semi-hydrophobic nature of most TMS of type III-secreted and type IVB-secreted transmembrane proteins prevents recruitment of SRP and erroneous targeting to the Sec-translocon. In addition, specific T3SS chaperones bind TMS of type III-secreted transmembrane proteins and thus shield these segments from wrong targeting factors and aggregation. T4BSS can also accept strongly hydrophobic substrates from within the bacterial inner membrane. HM host membrane, IM inner membrane, OM outer membrane