Bacterial rhomboid proteases mediate quality control of orphan membrane proteins

Abstract

Although multiprotein membrane complexes play crucial roles in bacterial physiology and virulence, the mechanisms governing their quality control remain incompletely understood. In particular, it is not known how unincorporated, orphan components of protein complexes are recognised and eliminated from membranes. Rhomboids, the most widespread and largest superfamily of intramembrane proteases, are known to play key roles in eukaryotes. In contrast, the function of prokaryotic rhomboids has remained enigmatic. Here, we show that the Shigella sonnei rhomboid proteases GlpG and the newly identified Rhom7 are involved in membrane protein quality control by specifically targeting components of respiratory complexes, with the metastable transmembrane domains (TMDs) of rhomboid substrates protected when they are incorporated into a functional complex. Initial cleavage by GlpG or Rhom7 allows subsequent degradation of the orphan substrate. Given the occurrence of this strategy in an evolutionary ancient organism and the presence of rhomboids in all domains of life, it is likely that this form of quality control also mediates critical events in eukaryotes and protects cells from the damaging effects of orphan proteins.

Keywords: Shigella; intramembrane proteolysis; membrane protein complexes; quality control; rhomboid.

Figure EV1. Amino acid sequence alignments of S. sonnei GlpG (GlpG_Ss) with E. coli GlpG (GlpG_Ec), and S. sonnei Rhom7 (Rhom7_Ss) with P. stuartii AarA (AarA_Ps)

The single amino acid difference between the two versions of GlpG is highlighted by a purple arrow. The potential active site residues of Rhom7, Ser¹³³ and His¹⁸⁷, are highlighted by red arrows. Alignment performed by Clustal Omega.

Figure 1. GlpG and Rhom7 are active rhomboids

1. Topology of GlpG, Rhom7 and the artificial substrate with a maltose‐binding protein (MBP) domain, triple‐FLAG tag (3xFLAG), the TMD of P. stuartii TatA and a thioredoxin domain (Trx).

2. Western blot analysis (probing with an anti‐FLAG mAb) to detect cleavage of the artificial substrate by wild‐type (WT)/inactive (S²⁰¹A) GlpG encoded on pBAD33 with/without arabinose (Ara.).

3. Western blot analysis to detect cleavage of the artificial substrate by wild‐type (WT)/modified (S¹³³A or H¹⁸⁷A) Rhom7 encoded on pBAD33 with/without arabinose (Ara.).

4. Activity of Rhom7 with/without its 7^th TMD and/or C‐terminal domain.

Data information: In (B‐D), rhomboid substrates that are uncleaved, cleaved by GlpG or cleaved by Rhom7 are marked by black, red and blue arrows, respectively. Controls, empty pBAD33 (empty) and wild‐type Shigella sonnei (Ss). RecA, loading control. Source data are available online for this figure.

Figure EV2. GlpG and Rhom7 do not influence the growth of S. sonnei under various conditions

1. A, B

   Wild‐type S. sonnei or S. sonnei ΔglpGΔrhom7 were grown in liquid LB in the presence of oxygen at 37°C with shaking at 180 rpm. Growth was assessed by measuring the optical density of cultures at 600 nm (OD₆₀₀) (A) and colony‐forming units of bacteria (CFU) recovered from the same samples (B).

2. C, D

   Bacteria were grown anaerobically in LB supplemented with 10 mM KNO₃ with shaking at 180 rpm. Growth was assessed by measuring the OD₆₀₀ as above (C) and CFU of bacteria recovered from the same cultures (D).

3. E, F

   Comparisons of cellular respiration (NADH production) of S. sonnei ΔglpGΔrhom7 (green) to wild‐type S. sonnei (red) (E), or of S. sonnei ΔglpGΔrhom7 + glpG (green) to wild‐type S. sonnei (red) (F) in conditions provided by PM1‐10 of the Biolog MicroArrays. Areas of overlap are coloured yellow, while differences are shown as patches of red or green. Wild‐type S. sonnei seemed to respire better than S. sonnei ΔglpGΔrhom7 + glpG in pH 9.5 + anthranilic acid (black box in E). However, this result was not reproducible.

Data information: In (A‐D), data are presented as mean ± SD from three independent experiments.

Figure EV3. GlpG and Rhom7 do not affect S. sonnei survival following oxidative stress

1. Resistance of wild‐type S. sonnei or S.sonnei ΔglpGΔrhom7 to H₂O₂. S. sonnei was grown in LB to an OD₆₀₀ of ≈ 0.5 and then pre‐incubated in 0 or 200 μM H₂O₂ for 20 min, before incubation in 0 or 2 mM H₂O₂ for 15 min. Bacterial survival was measured by plating samples to solid media. The percentage survival was calculated as the ratio of CFU recovered from H₂O₂‐exposed bacteria compared to bacteria not exposed to H₂O₂.

2. Resistance of wild‐type S. sonnei or S. sonnei ΔglpGΔrhom7 to paraquat (PQ). S. sonnei was grown as above and then pre‐incubated in 0 or 200 μM PQ for 15 min before incubation in 0 or 10 mM PQ for 45 min; samples were then plated to solid media to measure bacteria survival. The percentage survival was calculated as the ratio of CFU recovered from PQ‐treated bacteria to those that were untreated.

Data information: In (A) and (B), data are presented as mean ± SD from three independent experiments.

Figure EV4. T3SS‐mediated secretion of Ipas by S. sonnei is rhomboid‐independent

T3SS‐mediated secretion of effectors by wild‐type S. sonnei and S. sonnei ΔglpGΔrhom7. Bacteria were grown to exponential phase (OD₆₀₀ ≈ 0.5), and Congo red (final concentration, 200 μg/ml) was added to bacteria to induce secretion. Supernatants were analysed by SDS–PAGE and silver staining. Sizes of a molecular weight marker are shown in kDa.

Figure 2. Identification of potential GlpG and Rhom7 substrates

1. Workflow of the bioinformatic identification of candidate rhomboid substrates.

2. A library of artificial substrates harbouring TMDs from 44 candidate rhomboid substrates.

3. Western blot analysis of candidate substrates reproducibly cleaved by GlpG from the screen.

4. Western blot analysis of candidate substrates reproducibly cleaved by Rhom7 from the screen.

5. Venn diagram summarising putative rhomboid substrates identified by the screen.

Data information: In (C and D), rhomboid substrates that are uncleaved, cleaved by GlpG or cleaved by Rhom7 are marked by black, red and blue arrows, respectively. Controls, inactive enzymes and wild‐type Shigella sonnei (Ss). RecA, loading control. Source data are available online for this figure.

Figure 3. HybA is a physiologic substrate of GlpG

1. Schematic of HybA and HybO fusions.

2. Western blot analysis (probing with an anti‐FLAG mAb) to detect cleavage of HybO with (+)/without (−) chromosomal (native) or pBAD33‐encoded rhomboids (plasmid).

3. Western blot analysis to detect cleavage of HybA with (+)/without (−) chromosomal (native) or pBAD33‐encoded rhomboids (plasmid).

4. Western blot analysis to detect cleavage of HybA^G296F by endogenous or pBAD33‐encoded rhomboid.

5. Western blot analysis to detect cleavage of HybA^P300A by endogenous or pBAD33‐encoded rhomboid.

Data information: In (B‐E), rhomboid substrates that are uncleaved, cleaved by GlpG or cleaved by Rhom7 are marked by black, red and blue arrows, respectively. Wild‐type Shigella sonnei, Ss. Wild‐type (WT)/inactive (SAHA: alanine substitution of the catalytic serine and histidine residues) enzymes were pBAD33‐encoded (plasmid) in S. sonnei ΔglpGΔrhom7. RecA, loading control.Source data are available online for this figure.

Figure EV5. Identification of HybA cleavage site by GlpG

N‐terminal sequencing of cleaved chromosomally encoded sfCherry‐3xFLAG‐tagged HybA by chromosomally encoded GlpG in S. sonnei in the absence of HybB. Bacteria were grown anaerobically in LB supplemented with 0.5% fumarate, 0.5% glycerol to OD₆₀₀ = 0.5 before lysis. HybA was purified by anti‐FLAG affinity chromatography prior to analysis by SDS–PAGE. Two bands only present in the GlpG (+) sample with molecular weight consistent with cleaved HybA were subject to N‐terminal sequencing and gave the same result.

Figure 4. GlpG specifically targets orphan HybA and does not influence hydrogenase‐2 activity

1. Bacterial two‐hybrid analysis with HybA and/or HybB fused chromosomally with the T25 and T18 domains of B. pertussis CyaA, respectively, in the absence of endogenous CyaA. Bacteria were grown on LB agar containing 20 μg/ml X‐gal at 37°C for 14 h in the presence/absence of O₂. Scale bar, 1 cm.

2. Western blot analysis to detect HybA cleavage in S. sonnei Δrhom7 by wild‐type (+) or inactive (S²⁰¹A) GlpG expressed chromosomally (native) or from pUC19 (plasmid) with (+)/without (−) HybB. HybA was expressed from its native locus or a plasmid (pHybA).

3. Quantification of the ratio of cleaved/uncleaved HybA in strains with (+) or without (−) HybB. Mean ± S.D. of three experiments. *P < 0.05; ****P < 0.0001 (one‐way ANOVA).

4. Western blot analysis to detect GlpG‐mediated cleavage of native C‐terminally sfCherry‐3xFLAG‐tagged HybA at indicated times after blocking protein translation by the addition of chloramphenicol at T₀ in the absence of HybB (−).

5. Western blot analysis to detect GlpG‐mediated cleavage of tagged HybA at times after blocking protein translation by the addition of chloramphenicol at T₀ in the presence of HybB (+).

6. Growth of bacteria lacking hydrogenases (ΔhyaA‐F ΔhycE +/− ΔhybO‐G) or GlpG (ΔglpG), or expressing uncleavable HybA (hybA ^G296F) in 5% H₂

7. H₂ consumption by S. sonnei. Bacteria were grown aerobically in LB overnight and then diluted into M9 minimal media supplemented with 0.5% fumarate, 12.5 μg/ml nicotinic acid and 0.2% casamino acids in a sealed glass chromatography vial. The headspace was purged with 10% H₂/90% argon, and cultures were incubated at 37 °C with shaking at 180 rpm for 9 h. H₂ in the headspace was sampled and measured by gas chromatography.

8. Shigella sonnei growth in 10% H₂ with plasmid‐expressed wild‐type or non‐functional (pglpG ^SAHA) GlpG.

Data information: P‐values were calculated by two‐way ANOVA (F, H). Bars represent the mean ± SD n = 3 (F‐H); *P < 0.05; ****P < 0.001. In (B, D, and E), HybA that is uncleaved or cleaved by GlpG is marked by black and red arrows, respectively. RecA, loading control. Source data are available online for this figure.

Figure 5. GlpG and Rhom7 also target orphan components of formate dehydrogenases

1. Schematic of FdoH and FdnH fusions.

2. Western blot analysis (probing with an anti‐FLAG mAb) to detect cleavage of FdoH cleavage in S. sonnei Δrhom7 with (+)/without (−) FdoI.

3. Western blot analysis to detect cleavage of FdnH in S. sonnei ΔglpG with (+) or without (−) chromosomal Rhom7 (native) or wild‐type (+)/inactive (SAHA) Rhom7 expressed from pBAD33 in S. sonnei ΔglpGΔrhom7 with (+) or without (−) FdnI.

4. Alignments of HybA, FdoH and FdnH in S. sonnei with Providencia stuartii TatA and with their homologues from phylogenetically diverse organisms (S. sonnei, Salmonella enterica, Yersinia enterocolitica, Proteus mirabilis, Pseudomonas aeruginosa and Burkholderia pseudomallei) highlighting conserved glycine (green) and proline (orange) residues.

5. Western blot analysis to detect cleavage of wild‐type (WT) or uncleavable (P³⁰⁰A) HybA by active (+) or inactive (S²⁰¹A) chromosomal GlpG in the presence (+) or absence (−) of HybB.

6. Western blot analysis to detect cleavage of wild‐type (WT) or modified (P²⁵⁹A) FdoH by active (+) or inactive (S²⁰¹A) chromosomal GlpG in the presence (+) or absence (−) of FdoI.

7. Western blot analysis to detect cleavage of wild‐type (WT) or modified (P²⁵⁹A) FdnH with active (+) or inactive (SAHA) Rhom7 expressed from pBAD33 with (+) or without (−) FdnI.

Data information: In (B, C, E‐G), rhomboid substrates that are uncleaved, cleaved by GlpG or cleaved by Rhom7 are marked by black, red and blue arrows, respectively. RecA, loading control.Source data are available online for this figure.

Figure 6. Rhomboid cleavage licenses further degradation of substrates

1. A

   Western blot analysis (probing with an anti‐V5 mAb) to detect degradation of N‐terminally V5‐tagged wild‐type (WT) or modified (P³⁰⁰A) HybA in S. sonnei Δrhom7 chromosomally expressing wild‐type (+) or inactive (S²⁰¹A) GlpG with (+) or without (−) HybB.

2. B

   Degradation of V5‐tagged HybA at times after blocking protein translation at T₀ in the presence (+) or absence (−) of HybB.

3. C

   Western blot analysis of N‐terminally V5‐tagged wild‐type (WT) or modified (P²⁵⁹A) FdoH in S. sonnei Δrhom7 chromosomally expressing wild‐type (+) or inactive (S²⁰¹A) GlpG with (+)/without (−) FdoI.

4. D

   Western blot analysis of N‐terminally V5‐tagged wild‐type (WT) or modified (P²⁵⁹A) FdnH in S. sonnei Δrhom7 with wild‐type (+) or inactive (SAHA) Rhom7 expressed from pBAD33 with (+)/without (−) FdnI.

5. E, F

   Degradation of N‐terminally V5‐tagged wild‐type (WT) or modified (P²⁵⁹A) FdoH (E) or FdnH (F) in S. sonnei Δrhom7 with wild‐type (+) or inactive (S²⁰¹A, SAHA) GlpG expressed chromosomally (native) or from pUC19 (plasmid) without FdoI or FdnI (−), respectively, +/− exposure to 400 μM CuCl₂ for 30 min.

Data information: Rhomboid substrates that are uncleaved, cleaved by GlpG or cleaved by Rhom7 are marked by black, red and blue arrows, respectively. Degradation products post‐rhomboid cleavage are marked by green arrows. RecA, loading control.Source data are available online for this figure.

Figure 7. Rhomboids prevent aggregation of orphan substrates in the inner membrane

1. Western blot analysis probing the localisation and status of plasmid‐encoded N‐terminally V5‐tagged wild‐type (WT) or modified (P³⁰⁰A) HybA in S. sonnei Δrhom7ΔhybB with wild‐type (+) or inactive (SAHA) GlpG expressed from pUC19. Whole‐cell lysate (Whole cell.), Soluble (Sol.), Membrane (detergent‐solubilised, Mem.) and the Aggregate (Agg.) fractions are shown. HybA that is uncleaved, cleaved only by GlpG and further degraded post‐GlpG cleavage is marked by black, red and green arrows, respectively.

2. Model of rhomboid‐mediated quality control by selectively targeting orphan components of multiprotein respiratory complexes.

Source data are available online for this figure.