Adaptation of the periplasm to maintain spatial constraints essential for cell envelope processes and cell viability

  1. Eric Mandela
  2. Christopher J Stubenrauch
  3. David Ryoo
  4. Hyea Hwang
  5. Eli J Cohen
  6. Von L Torres
  7. Pankaj Deo
  8. Chaille T Webb
  9. Cheng Huang
  10. Ralf B Schittenhelm
  11. Morgan Beeby
  12. JC Gumbart  Is a corresponding author
  13. Trevor Lithgow  Is a corresponding author
  14. Iain D Hay  Is a corresponding author
  1. Infection & Immunity Program, Biomedicine Discovery Institute and Department of Microbiology, Monash University, Australia
  2. Interdisciplinary Bioengineering Graduate Program, Georgia Institute of Technology, United States
  3. School of Materials Science and Engineering, Georgia Institute of Technology, United States
  4. Department of Life Sciences, Imperial College London, United Kingdom
  5. Monash Proteomics & Metabolomics Facility, Department of Biochemistry and Molecular Biology, Biomedicine Discovery Institute, Monash University, Australia
  6. School of Physics, Georgia Institute of Technology, United States
  7. School of Biological Sciences, The University of Auckland, New Zealand

Abstract

The cell envelope of Gram-negative bacteria consists of two membranes surrounding a periplasm and peptidoglycan layer. Molecular machines spanning the cell envelope depend on spatial constraints and load-bearing forces across the cell envelope and surface. The mechanisms dictating spatial constraints across the cell envelope remain incompletely defined. In Escherichia coli, the coiled-coil lipoprotein Lpp contributes the only covalent linkage between the outer membrane and the underlying peptidoglycan layer. Using proteomics, molecular dynamics, and a synthetic lethal screen, we show that lengthening Lpp to the upper limit does not change the spatial constraint but is accommodated by other factors which thereby become essential for viability. Our findings demonstrate E. coli expressing elongated Lpp does not simply enlarge the periplasm in response, but the bacteria accommodate by a combination of tilting Lpp and reducing the amount of the covalent bridge. By genetic screening, we identified all of the genes in E. coli that become essential in order to enact this adaptation, and by quantitative proteomics discovered that very few proteins need to be up- or down-regulated in steady-state levels in order to accommodate the longer Lpp. We observed increased levels of factors determining cell stiffness, a decrease in membrane integrity, an increased membrane vesiculation and a dependance on otherwise non-essential tethers to maintain lipid transport and peptidoglycan biosynthesis. Further this has implications for understanding how spatial constraint across the envelope controls processes such as flagellum-driven motility, cellular signaling, and protein translocation

Editor's evaluation

In this study, Mandela et al. investigate the response of cells to a lengthened version of the periplasmic protein Lpp (Lpp+21). An abundant protein present in high abundance in cells, Lpp tethers the outer membrane to the peptidoglycan layer and is implicated in maintaining the distance between the two. Combining genetics, proteomics and simulations, the authors determine that lengthening Lpp does not change the spatial organization of the periplasm due to apparently compensatory effects. Together, these data highlight the importance of periplasmic organization to cell viability and the resilience of the systems that maintain it.

https://doi.org/10.7554/eLife.73516.sa0

Introduction

Gram-negative bacteria have a cell envelope composed of two membranes sandwiching between them an aqueous space called the periplasm, in which an essential structural layer of peptidoglycan (PG) resides. The outer membrane is critical to cell growth and these bacteria face challenges to their cell biology in terms of membrane protein assembly and lipid-transport pathways that must traverse the distance from the inner membrane (IM) to the outer membrane (OM) (Silhavy et al., 2010). Recent work investigating the spatial demands for assembly of proteins into the outer membrane has shown precincts of active protein integration into the membrane can deliver new material to the growing outer membrane (Gunasinghe et al., 2018) and that random planar movement from these precincts drives the observed non-uniform distributions of the major proteins of the outer membrane (Rassam et al., 2015; Ursell et al., 2012). By contrast to these protein components that diffuse to the outer membrane, elements of the LPS-transport machinery (e.g. the Lpt complex) (Freinkman et al., 2012; Sperandeo et al., 2017; Sperandeo et al., 2011; Ekiert et al., 2017; Shrivastava and Chng, 2019) span the OM and IM in order to fulfil their function in delivering lipid components to the outer membrane.

The PG layer is a fundamental aspect of the cell envelope, and it must be dynamically remodeled to allow growth as well as the assembly and transit of trans-envelope structures. PG synthesis and remodeling is a complex process with high levels of redundancy at various steps, involving at least 50 enzymes in E. coli (Pazos et al., 2017). The Penicillin-binding proteins (PBPs) are the core components responsible for the periplasmic biosynthesis of peptidoglycan. There are multiple PBP complexes including the two core, semi-redundant PBP complexes PBP1a and PBP1b embedded in the IM which are activated by interactions with lipoproteins LpoA and LpoB embedded in the OM. Thus, the activation of PG synthesis by these enzymes is spatially regulated, serving as a self-limiting molecular ruler to modulate PG thickness (Typas et al., 2011). Cells must possess either a functional PBP1a or PBP1b system for growth (Paradis-Bleau et al., 2010; Typas et al., 2010).

Trans-envelope complexes such as lipopolysaccharide (LPS) transit pathways (Freinkman et al., 2012; Sperandeo et al., 2017; Sperandeo et al., 2011; Ekiert et al., 2017; Shrivastava and Chng, 2019) and the protein translocation and assembly module (the TAM) (Selkrig et al., 2015; Selkrig et al., 2012; Shen et al., 2014) are also spatially constrained by the need to reach across from the IM to OM in order to function. The PG layer is covalently attached to the outer membrane by Braun’s lipoprotein (Lpp), with two recent papers addressing whether extending the length of Lpp would impact on (i) flagellar function given that the flagellum spans both membranes (Cohen et al., 2017) and (ii) signal-transduction systems that span the OM to IM (Asmar et al., 2017). In both cases, extending the coiled-coil structure of Lpp by 21 residues (Lpp+21) was found to be the longest form that supported close to normal growth (Cohen et al., 2017; Asmar et al., 2017). Imaging of these Lpp+21 strains in both studies showed that the total periplasmic width had been stretched ~3–4 nm (Cohen et al., 2017; Asmar et al., 2017). This being the case, the Lpp+21 model would provide a powerful experimental system to study how processes like OM biogenesis and PG biosynthesis can be maintained under a spatial stress on the cell wall.

To understand how trans-envelope processes in E. coli adapt to the presence of an enlarged periplasm, a combination of phenotypic analysis, proteomics, molecular dynamics and a synthetic lethal screen was employed to identify and characterize factors needed to maintain viability in the Lpp+21 strain of E. coli. The genetic screen demanded synthetic growth phenotypes from an array of mutants each lacking a gene that, while non-essential in wild-type E. coli (Baba et al., 2006), is essential in the Lpp+21 strain. These genes fall into three functional categories: PG biosynthesis and remodeling, LPS biosynthesis and PG-outer membrane linkage. We show that previously non-essential proteins involved in bridging the gap between the OM and PG become essential in the context of the Lpp+21 strain background. These include previously known PG binding OM proteins (OmpA and Pal) as well as proteins previously not known to play an active role in linking the OM and PG (TolC and YiaD). We observed a thicker more diffuse or heterogeneous PG layer in the Lpp+21 strain and whole cell proteomics revealed that in response to an increased length of Lpp, E. coli increases the levels of a range of cell envelope proteins involved in PG turnover. We discuss the outcomes in terms of how the PG-outer membrane linkage functionalizes the periplasm, the evolutionary constrains in place to maintain this functionality, and the specific activity of Lpp in contributing to the load-bearing function of the OM.

Results

Resilience and growth of the Lpp+21 strain

A phylogenetic assessment of Lpp lengths across diverse bacterial lineages showed a very narrow window of protein size (Figure 1A), with Lpp being 78 residues in most species of bacteria including E. coli. Lpp lengths of 99 residues or more are at the upper end of the natural range for this protein and, in nature, these longer Lpp proteins are found in the genus Geobacter. The previously described Lpp+21 isoform expressed in E. coli therefore sits near the upper physiological range observed among bacteria. We introduced a gene encoding the Lpp+21 isoform into an E. coli background suitable for genetic screens (Figure 1—figure supplement 1) and confirmed the size of the protein by SDS-PAGE of bacterial cell extracts (Figure 1B). Enteric bacteria like E. coli can thrive in hyperosmotic environments compared to most laboratory growth media, such as the human gut by maintaining the periplasm and the cytoplasm in an iso-osmotic state (Ingraham and Marr, 1996; Stock et al., 1977). This is achieved by adjusting the solute concentration in the cell compartments by influx or efflux of water. By doing so, osmolality contributes immensely to the architectural aspects of cellular compartments (Stock et al., 1977; Cayley et al., 2000). In this study sorbitol was used to mimic these physiological osmotic conditions. Previous studies have demonstrated that the periplasmic volume increased rapidly in response to increased osmolyte levels in the external medium. This phenomenon can reduce the cytoplasmic volume by about 30%, thereby constricting the IM inwards, and substantially increasing the periplasmic volume by around 300% (Stock et al., 1977; Cayley et al., 2000; Pilizota and Shaevitz, 2013). The Lpp+21 isoform had little impact on growth of E. coli. On minimal growth medium, growth rates of the Lpp+21 strain of E. coli were equivalent to the isogenic wild-type E. coli (Figure 1C). This was likewise true on growth media osmotically balanced with concentrations of sorbitol up to 1.0 M (Figure 1D), and on rich (LB) medium with or without sorbitol (Figure 1D).

Figure 1 with 1 supplement see all
Phenotypes of E. coli cells encoding the Lpp+21 isoform.

(A) Non-redundant Lpp sequences were identified (Materials and methods) and the protein length charted on the x-axis. The number of non-redundant sequences showing that length is shown on the y-axis. The location of Lpp and the lengthened Lpp+21 are indicated. (B) Whole cell lysates were prepared from the indicated strains and subject to SDS-PAGE and immunoblot analysis with anti-Lpp antibodies and anti-OmpA antibodies. OmpA serves as a loading control. (C) The JW5028 – Keio BW25113 strain with kan gene replacing a pseudogene background and isogenic Lpp+21 strain (Figure 1—figure supplement 1) were grown over 24 hr. The growth medium is M9, containing the indicated concentration of sorbitol as an osmolyte. (D) Growth rates for the same strains were measured in rich (LB) growth media with and without sorbitol over 20 hr. (E) The periplasmic width distribution of the indicated strains in hyperosmotic conditions. The histogram depicts the frequency with which a given distance is observed between the OM and PG. (F) Subtomogram averages of cell envelopes in hyperosmotic conditions. Measurements from EM views evaluate the distance between OM and PG in the Lpp+21 strain. While the PG in the wild-type strain is a uniform thin electron dense layer, the PG layer in the Lpp+21 strain is more diffuse and thicker. Each panel represents averages of the subtomogram cell envelope section binned into the sizes shown.

To establish the extent to which the periplasm had been remodeled in the Lpp+21 strain of E. coli, we compared the periplasms of WT and Lpp+21 strains using electron cryotomography in which cells are preserved in a frozen-hydrated, near-native state. To discern peptidoglycan, which was indistinct in previous studies, we increased the signal to noise ratio in images by calculating subtomogram averages instead of inspecting individual tomograms. As expected, the average distance from the middle of the IM density to the middle of the OM density was increased in both strains under hyperosmotic conditions when compared to previously reported data where cells were grown in standard laboratory media (Asmar et al., 2017). As shown in Figure 1E, the averaged distance from the IM to the OM were somewhat greater in the Lpp+21 strain (32–36 nm) compared to the isogenic wild-type strain (30–32 nm) in line with previous Lpp+21 periplasmic width measurements. Previous studies could not discern peptidoglycan; whereas here, we were able discern peptidoglycan in subtomogram averages. The distance to the center of the PG density from the OM was slightly increased in the Lpp+21 strain (Figure 1F), although by less than the anticipated ~3 nm. The PG morphology was also qualitatively different: in the wild-type strain a uniform dark PG layer could be observed in the images, whereas in the Lpp+21 strain a broader and more diffuse PG layer was evident, suggesting heterogeneity in both the density and thickness of the PG.

Cell envelope response to elongated Lpp

To determine the adaptive response to changing the distance constraint between OM and PG, quantitative whole cell proteomics was applied to evaluate the Lpp+21 strain. Triplicate samples of the wild-type and Lpp+21 strains were processed for analysis by mass spectrometry and we sought to identify those proteins where the steady-state level increases or decreases three-fold or more (Log2 fold change of ±1.6) in the Lpp+21 strain (Figure 2A; Supplementary file 1). The level of the oligopeptide transporter subunits (OppB, OppC, OppD and OppF) are substantially increased in the Lpp+21 strain compared to the wild-type (Supplementary file 2). This suggests an increase in PG turnover and an overall increased capacity to recycle PG components and is consistent with the concomitant increase in AmiC, one of the two major amidases involved in PG remodeling. In addition, proteins implicated in diverse stress-responses (cold shock proteins CspG, CspA, CspI, and YdfK, as well as envelope stress protein ZraP and redox stress protein YfcG) were observed at increased steady-state levels in Lpp+21 strain (Figure 2A, Supplementary file 2). The greatest decreases were seen in the steady state levels of the GatZABCD proteins involved in galactitol phosphotransferase system and DHAP synthesis (Figure 2A). The gatABCD genes have been shown to be responsive to factors that change E. coli cell surface tension (Domka et al., 2007) and Lpp+21 has been reported to significantly decrease cell stiffness (Mathelié-Guinlet et al., 2020).

Figure 2 with 3 supplements see all
The Lpp+21 cells have altered outer membranes and increased blebbing.

(A) Heat map of the significant proteomic differences observed between the wild-type and Lpp+21 mutant. Blue boxes indicate a relative reduction and red indicates a relative increase in protein level, centred on the average of the replicate samples. The grouping of the proteins is based on the similarity of the change in expression observed. (B) The Lpp+21 mutant has an overall reduction in the level of periplasmic proteins. (C) The Lpp+21 mutant has an increase in protein secreted via OMV blebbing. (D) SDS sensitivity profiles of the Lpp+21 mutant compared to the wild-type in increasing concentration of SDS in LB (solid) media. Representative data are shown from experiments performed in triplicate.

A decrease was seen in the steady-state level of the Lpp+21 isoform in the mutant, to approximately one-eighth the level of Lpp in the wild-type strain (Figure 2—figure supplement 2). This is consistent with the relative abundance of Lpp and Lpp+21 observed in SDS-PAGE analysis of cell extracts from the two strains (Figure 1F). However, despite the relative decrease, Lpp+21 remains as a highly abundant component of the OM-PG linkage factors given that Lpp is present at up to 106 protein molecules per wild-type cell (Silhavy et al., 2010; Li et al., 2014).

The mass spectrometry data was processed to allow for an analysis of sub-cellular proteomes (Loos et al., 2019; Figure 2—figure supplement 2). An initially puzzling observation was that the Lpp+21 strain has a 12% overall reduction of total periplasmic protein compared to wild type (Figure 2B). This was calculated as the proportion of the summed intensity from identified proteins predicted to reside in the periplasm in the STEPdb: G, E, F2, F3, I annotations (Loos et al., 2019; Figure 2—figure supplement 2). Both Lpp+21 strains and null Lpp strains of E. coli are softer as previously adjudged by atomic force microscopy (Mathelié-Guinlet et al., 2020), and several factors that increase the softness of E. coli are also implicated in an increased outer membrane vesicle (OMV) production (Cohen et al., 2017; Sonntag et al., 1978; Schwechheimer et al., 2014; Rojas et al., 2017). To address whether the measured depletion of periplasmic content reflects an increased production of OMVs, extracts measuring the amount of total protein in the OMV fraction were normalized to OD 600 (Figure 2C, Figure 2—figure supplement 2). This confirmed that the presence of Lpp+21 promotes approximately 10-fold more total protein associated with the OMV fraction, reflecting increased OMV production. The overall level of OM proteins associated with the cells was maintained constant (Figure S3) but a small OM integrity defect was evident from an increased sensitivity to SDS (Figure 2D).

Lpp+21 can be accommodated in the periplasm, but other factors become essential

To directly address the altered phenotype induced by Lpp+21, we established a synthetic genetic array for factors in E. coli that become essential in order to maintain viability of the Lpp+21 strain (Figure 3—figure supplement 1). The screen demands synthetic growth phenotypes from an array of mutants each lacking a single gene that, while non-essential in wild-type E. coli, become essential in the Lpp+21 strain of E. coli. The endogenous lpp gene was replaced by a gene encoding Lpp+21 in a isogenic library of 3818 E. coli mutants, each of which lacks a non-essential gene. Growth on rich medium allowed the rescue of the new library for array into a format suitable for high-throughput screening with a Singer RoToR robotics platform (Figure 3—figure supplement 1). Phenotypic analysis was thereafter scored for growth by comparing the growth of the isogenic mutants in the Lpp background (Figure 3A) with the equivalent mutants in the Lpp+21 background (Figure 3B). Each of the genes that displayed a noticeable phenotype in these analyses are presented in Table 1.

Figure 3 with 3 supplements see all
Factors that become essential to mediate OM-PG linkage in Lpp+21 E. coli.

(A) The growth phenotype in M9 minimal media (0.5 M sorbitol) of single gene knock outs that exhibit essentiality on Lpp+21 background (Table 1). (B) The growth phenotype in M9 minimal media (0.5 M sorbitol) of double gene mutants. The mutants are results of Hfr Cavalli lpp:lpp+21 cat crossed with 22 KanR recipients shown in panel A (Materials and methods). The double mutants are indicated and are arranged in four biological replicates (each having four technical replicates). (C) Sequence similarity network of domain (Pfam PF00691) containing proteins from across the Proteobacteria. Each circle represents a protein from a representative proteome (rp35) containing the PF00691 domain, connected by lines with a length imparted by their similarity score as defined by EFI - Enzyme Similarity Tool (Gerlt et al., 2015), with a cutoff of 30. Proteins are colored by their taxonomic class and the approximate location of the E. coli K12 six PF00691 proteins is indicated. (D) Synthetic lethal phenotype of the drug efflux mutants in the absence of selective antibiotics in M9 minimal media condition. Representative data are shown from experiments performed in biological triplicate.

Table 1
Essential genes in the Lpp+21 strain.
Cellular processLpp+21-essential genesFunction
LPS biosynthesisgalUUDP-glucose metabolic process
gmhB/yaeDADP-L-glycero-β-D-manno-heptose biosynthetic process
rfaD/hldD/waaDADP-L-glycero-β-D-manno-heptose biosynthetic process
Peptidoglycan biosynthesis, turnover, and remodelinglpoA, lpoB,Regulators of PG synthases
mrcA, mrcBPG synthases
ldcAL,D-carboxypeptidase involved in PG recycling/turnover
nlpDRegulator of AmiC PG hydrolase
sltLytic PG transglycosylase, degradation of uncrosslinked glycan strands
PG-OM linkageacrB, mdtL*, mdtK*, ydhl*IM components of drug efflux pumps.
tolCOM component of drug efflux pump
ompAβ-barrel protein with PG binding domain
palLipoprotein with PG binding domain
yiaDLipoprotein with PG binding domain
  1. *

    Putative IM drug efflux machinery, TolC independent with as yet unknown OM component (Bay et al., 2017).

The only cytoplasmic factor identified in our screen, YraN is predicted to be a Holliday-junction resolvase related protein, and we therefore speculate that this mutant failed to resolve the merodiploid condition transient in the introduction of the lpp+21 condition to the background strain, making the yraN mutant a technique-relevant artefact of the screen. This being the case, only functions performed in the periplasm were recovered as essential to viability for the Lpp+21 strain.

Most of the components of the LPS biosynthetic machinery are essential genes in E. coli and are thus not represented in the library of non-essential genes. Several non-essential genes in the LPS biosynthetic pathway that are in the library, become essential to the Lpp+21 strain (Table 1): the core LPS biosynthesis factors GalU, GmhB and RfaD were shown to be essential in the Lpp+21 strain.

An essential role for keeping the OM-PG distance

Several genes encoding proteins that could play roles in anchoring the PG within the cell envelope were identified as essential in the Lpp+21 background. Independently, none of the major proteins bridging the OM and PG are essential for growth in E. coli (Baba et al., 2006) and all are therefore represented in the library. In a Lpp+21 background the genes encoding the β-barrel protein OmpA and the lipoprotein Pal become essential (Table 1). PG-binding domain PF00691 is common to these proteins: appended to a beta-barrel in OmpA, but to a lipoyl anchor in Pal, and is also conserved in other proteins across diverse Gram-negative bacteria. In E. coli, there are four additional proteins containing this PG-binding domain and these were mapped in a sequence similarity network analysis (Figure 3C). A protein of unknown function, YiaD, is present and it too is essential in a Lpp+21 background (Table 1). We suggest, therefore, that this protein plays a substantive role in OM-PG linkage. The remaining three proteins: MotB, LafU and YfiB, are more divergent to the OmpA/Pal/YiaD cluster. Neither motB, lafU nor yfiB displayed a synthetic phenotype with Lpp+21, and it has been suggested previously that motB, lafU, and yfiB are not expressed at detectable levels under laboratory conditions (Li, 2013).

Detecting genes encoding drug-efflux pumps as important for growth of the Lpp+21 strain was initially surprising. Either the absence of the inner membrane proteins AcrB or the OM component TolC caused a reduction of growth in the Lpp+21 genetic background (Table 1). When antibiotic selection was removed by plating the mutants on medium without chloramphenicol, the synthetic growth defects were observed in the absence of drug selection (Figure 3D), indicating that this synthetic phenotype is not the result of a decreased drug efflux activity. The trans-envelope AcrAB-TolC multidrug efflux pump has been shown to traverse through the PG and interact directly with PG at several defined sites (Xu et al., 2012; Kim et al., 2008; Jo et al., 2019; Shi et al., 2019; Gumbart et al., 2021), and as loss of the core AcrB and TolC components became essential, we suggest that this system could be acting as an additional OM-PG linkage that becomes essential in a Lpp+21 background. Together with the observation that OmpA, Pal, YiaD, and TolC are also essential in the Lpp+21 genetic background, these data suggest that functions that maintain local areas of closer contact between the OM and PG are essential for viability.

To compare the behaviors of Lpp and Lpp+21, both with and without an OmpA monomer present, molecular dynamics (MD) simulations were run for 200 ns each (see Materials and methods for details). The Lpp lipoprotein from E. coli is a triple coiled-coil that is anchored to the inside face of the outer membrane by its N-terminal acyl group with a length equating to approximately 7.5 nm (Shu et al., 2000). The experiments were established to test the scenario for Lpp trimers or Lpp+21 trimers in the absence or presence of an OmpA tether between the patch of OM and patch of PG (Figure 4). These MD simulations only represent a portion of the membrane, and as such we could not account for potential heterogeneity in the protein and LPS concentrations between a WT and Lpp+21 cell. For three of the systems: Lpp only, Lpp with OmpA, and Lpp+21 only, tilting was marginal, with tilt angles of 76.9 ± 4.7°, 75.5 ± 4.7°, and 82.8 ± 2.9°, respectively (all numbers from the last 100 ns of the 200-ns trajectory). These angles are in agreement with previous simulations of Lpp alone (~80°) and Lpp with an OmpA monomer (~75°) PG (Gumbart et al., 2021; Boags et al., 2019; Samsudin et al., 2017). In initial simulations of Lpp+21 with OmpA, the non-covalent connection between OmpA and PG was quickly disrupted as Lpp+21 extended from its kinked state. Therefore, the simulation was repeated with an enforced OmpA-PG connection. Lpp+21 was observed to both straighten and tilt within the first 100 ns; the tilt angle measured for the last 100 ns was 49.4 ± 2.3°.

Figure 4 with 1 supplement see all
Final states of the OM-PG linkage from MD simulations.

(A–B). A patch of OM with the LPS molecules depicted in orange (lipid A moiety) and yellow (core oligosaccharides), and the phospholipids in the inner leaflet of the OM depicted in gray. The PG layer (blue for glycans and green for peptide crosslinks) is attached to the OM via two trimers of Lpp (A), or a trimer of Lpp and the PG-binding domain of OmpA (red). The β-barrel anchor of OmpA is shown embedded in the OM. (C–D) Equivalent scenarios formed with Lpp+21 trimers. The distances shown are calculated from the inner face of the OM to the centre of the PG layer and represent the average over the last 100 ns of a 200-ns simulation.

The distance between the OM inner leaflet phosphorus atoms and the PG sugars was measured in each scenario. In the presence of Lpp, the distances with and without OmpA were similar at 8.3 ± 2.1 nm and 8.1 ± 1.2 nm, respectively. This was not true for the other scenarios where the distance for Lpp+21 alone was 11.6 ± 1.3 nm, but for Lpp+21 with OmpA, the distance was reduced significantly to 8.7 ± 2.7 nm. Thus, we observe that PG-binding proteins like OmpA can counteract the increased distance imposed by Lpp+21, inducing it to tilt significantly in accommodation. We also compared our simulations to the distances that were measured by EM (centre of the OM to centre of the PG). In wild-type E. coli (i.e. Lpp+ OmpA), the centre-centre distance in the simulations is 10.7±2.2 nm (Figure 4B), similar to the 9.7–10.8 nm measured in intact cells (Figure 1F). The centre-centre distance measured in the simulation of Lpp+21 tilted by the presence of OmpA (11.0±2.6 nm), fits the observed distances of 10.8–12.8 nm much better than the distance that would be created by a perpendicular Lpp+21 (14.2±1.3 nm).

Discussion

We observed that over a range of osmotic conditions, and in nutrient-rich or nutrient-poor media, growth rates of the Lpp+21 strain of E. coli were equivalent to the isogenic wild-type E. coli, suggesting bacteria can adapt to the presence of the extended Lpp+21. We observed a moderate increase to the OM-IM space similar to that previously reported (Cohen et al., 2017; Asmar et al., 2017; Mathelié-Guinlet et al., 2020) and reviewed (Mathelié-Guinlet et al., 2020; Asmar and Collet, 2018). Three adaptive features were expressed as phenotypes in the Lpp+21 strain: (i) the steady-state level of the Lpp+21 tether was reduced eightfold compared to the level of Lpp in the isogenic wild-type strain, and other tethers that enforce a wild-type distance: OmpA, Pal, YiaD, and TolC, became essential factors in the Lpp+21 condition, (ii) structures that depend on a wild-type OM-IM distance, such as the LPS transport system, continued to function but key components of the system became essential for cell viability, and (iii) the PG network took on characteristics of dysregulated synthesis and all components of an otherwise redundant PG biosynthesis pathway become essential to viability. In response to Lpp elongation, we demonstrated a reduction in Lpp levels, as shown by the quantitative proteomic presented here, and the apparent tilting of the elongated Lpp as suggested by our molecular dynamics simulations. The reduction in the copy number of the elongated Lpp isoform observed in the quantitative proteomics was not due to a transcriptional downregulation of the lpp+21 gene, as RNAseq data demonstrated a moderate increase in the level of lpp+21 transcripts in the mutant (Supplementary file 6). Previous studies have not noted a marked reduction in Lpp+21, but these studies also did not directly measure the levels of the protein (Asmar et al., 2017). One interpretation of the reduction of the level of Lpp+21 observed here is that the Lpp+21 isoform has a higher turnover in the periplasm, either via an inherently lower stability or driven by a targeted degradation due to its influences on the cell envelope structure. Taken together, this suggests a reinterpretation of previous conclusions drawn from experiments with elongated Lpp isoforms that assumed that Lpp+21 was present at the same amount as Lpp, and that Lpp+21 sits perpendicular to the membranes such that the 21 extra amino acids always stretch the periplasmic width by a constant amount (Cohen et al., 2017; Asmar et al., 2017; Mathelié-Guinlet et al., 2020).

Tethers that enforce the distance constraint between the OM and PG layer became essential factors in the Lpp+21 condition

The lipoyl-N-terminus of Lpp is integrated into the OM, and the C-terminus of a protomer of the Lpp trimer is covalently linked to the PG layer. Stereochemically, only one unit of a Lpp trimer can be covalently attached to the PG layer, and this stoichiometry has been observed experimentally (Shu et al., 2000; Boags et al., 2019; Inouye et al., 1972). This makes each Lpp trimer an important bridge between the OM and PG layers of the cell wall, but it has not been clear whether the core role of Lpp is acting in compression or tension. In other words (Miller and Salama, 2018), is the role of Lpp to be a supportive brace to keep the OM away from the PG, or a binding anchor to bring the OM as closely as possible to the PG? In a wild-type E. coli scenario, molecular dynamics simulations show that the flexible linkers in PG-binding proteins like OmpA allow for adjustments in positioning the PG relative to the OM. In the presence of Lpp, the flexible linker of the PG binding domain of OmpA needs to extend further from the OM, as the distance between the OM and PG increases when Lpp sits perpendicular to the OM (Boags et al., 2019). In the absence of Lpp, OmpA determines the distance between the OM and PG (Boags et al., 2019; Samsudin et al., 2017) but is not essential for cell viability (Babu et al., 2014; Babu et al., 2011). Taken together with the data presented here from characterization of the Lpp+21 phenotypes, we suggest that Lpp functions as a brace to keep the OM away from the PG layer.

The genetic screen showed that each one of the OM-located PG-binding factors Pal, YiaD, and OmpA are essential (as is TolC) for viability of the Lpp+21 strain of E. coli, while previous high-throughput synthetic genetic screens demonstrate none of Pal, YiaD, OmpA nor TolC are essential or negatively affect the growth in an Lpp null background (Babu et al., 2014; Babu et al., 2011). The overlap in the distance constraints measured for cells expressing Lpp and Lpp+21 suggests that much of the Lpp+21 is in the highly tilted (49.4 ± 2.3°) form observed in the MD simulations. The enforced tilt and reduction in steady state levels of Lpp+21 would severely impact its bracing force, with the measurable consequence that membrane integrity (SDS resistance) is diminished, OmpA and other tethers become essential to viability, and the OM is highly permissive to OMV formation. Furthermore, the discovery of YiaD in the genetic screen for essential factors is significant. Our proteomic assessment and previously published data (Li et al., 2014) demonstrated that the copy number of YiaD is much lower than that of either Pal or OmpA, making it not immediately clear how a minor component of the membrane could be exerting such an essential role. Previously, YiaD was suggested to be a factor regulating OMP biosynthesis by the BAM complex (Tachikawa and Kato, 2011). Structural analysis shows YiaD to be highly similar to the PG-binding domains of Pal and OmpA (Ishida et al., 2014). We suggest that the primary function of YiaD is to mediate OM-PG linkage, and that this indirectly impacts OMP biogenesis.

Continued function of LPS transport in localized regions of the periplasm

The bridges needed to mediate LPS transport appear to be susceptible to disruption caused by Lpp+21 in E. coli. LptA is subject to degradation if the Lpt complex is compromised (Sperandeo et al., 2011), with LptA-LptC and LptA-LptD interactions proposed as key quality control steps in the assembly of the Lpt complex (Sperandeo et al., 2017). That the Lpp+21 cells have sufficient LPS in the OM to maintain membrane integrity is supported by our observations of the only minor increase in SDS sensitivity of the strain and observations by others of only minor changes in vancomycin sensitivity (Mathelié-Guinlet et al., 2020). These findings are consistent with observations through electron microscopy that regions of the periplasm in Lpp+21 cells are maintained with OM-PG distances reflective of the wild-type condition, which would permit LPS transport to the OM.

Components of the PG biosynthesis pathways become essential to cell viability

In E. coli, the weave of the PG-layer is maintained in a uniform, open state through quality control mechanisms that depend on the regulators in the OM (LpoA and LpoB) being able to permeate it to contact the synthetases (PBP1a and PBP1b) in the IM. It has been hypothesized in this way that the OM lipoproteins may serve as a molecular ruler to modulate PG thickness, maintaining a single layer of PG equidistant from the OM layer under normal conditions (Typas et al., 2011). The PG synthases encoded by mrcA and mrcB are redundant in the sense that mrcA mutant strains and mrcB mutant strains are each viable (Cho et al., 2016; Mueller et al., 2019; Suzuki et al., 1978). However, in the Lpp+21 strain both mrcA and mrcB (as well as their OM lipoprotein partners) are essential for viability, indicative of a compromised capability to build the PG-layer. Mutations designed to impact on these interactions lead to transient deposition of ‘high-density PG’ and ‘multi-layered PG’ through dysregulation of the synthetases (Typas et al., 2011). The broader and more diffuse morphology of the PG-layer observed by electron microscopy in the Lpp+21 strain could suggest a similar phenotype associated with transient or local impacts on the OM-PG distance (Typas et al., 2011). It is worth noting Δlpp ΔmrcB mutants have previously shown a moderate growth defect (Chen et al., 2014) and that this defect could be alleviated with the introduction of the lpp+21 gene (Asmar et al., 2017), whereas in the present study we observed a synthetic lethal phenotype in the ΔmrcB lpp+21 mutants. This discrepancy is potentially due differences in growth conditions used (LB vs hyperosmotic minimal media) imparting additional periplasmic stress or differences in the strain background (DH300 vs BW25113). Indeed, the ΔmrcB lpp+21, ΔmrcA lpp+21, ΔlpoA lpp+21, and ΔlopB lpp+21 mutants in the current study could grow on LB media but not on minimal media (Figure 4—figure supplement 1).

An essential requirement was also placed on PG-layer remodeling, whereby the PG-binding factor NlpD was found to be essential in Lpp+21 cells: its function is in modulating the activity of the amidase AmiC (Stohl et al., 2016; Yang et al., 2018) to remodel PG strands, and AmiC was observed at increased steady-state levels in the Lpp+21 strain. Taken together with the increase in oligopeptide transporter subunits (OppB, OppC, OppD, and OppF) in the Lpp+21 strain to recycle PG precursors across the IM, our results suggest a clearance of the malformed PG caused by dysregulation of the PG synthases is a crucial adaptation in the Lpp+21 strain.

Stiffness and connection of OM-PG

The concept of bacterial cell stiffness has emerged as a means to understand the physical parameters that define how readily bacteria can respond to major environmental changes (Rojas et al., 2017; Rojas and Huang, 2018; Hwang et al., 2018). Measurements by AFM have revealed a characteristic stiffness in Gram-negative bacterial cells that is contributed by load-bearing outer membrane and its attachment to the underlying PG layer (Rojas et al., 2018). In E. coli, mutants lacking Lpp, Pal, or OmpA are softer than wild-type cells (Rojas et al., 2018), and cells expressing the Lpp+21 isoform are also softer than wild-type cells (Mathelié-Guinlet et al., 2020). Taken together with our observation that the Lpp+21 isoform display a broader and more diffuse PG layer morphology in the subtomogram averages, this further supports the proposition that the OM is a major contributor to cell stiffness (Rojas and Huang, 2018; Rojas et al., 2018).

Although Lpp length alone does not appear to dictate periplasm width, indeed the Pseudomonas aeruginosa Lpp protein is only four amino acids longer than that of E. coli but its periplasm has been measured as ~3 nm larger under identical conditions (Matias et al., 2003), the length of Lpp seen in E. coli is conserved in a range of bacterial lineages. The species from the genus Geobacter naturally express longer Lpp proteins of 99 or more residues, equivalent to Lpp+21 in length. Geobacter species have a complicated periplasm housing ‘electron conduits’ formed of transmembrane and periplasmic redox proteins, in order to transfer electrons to the external surface of the bacterial cell (Jiménez Otero et al., 2018; Santos et al., 2015). There have been no direct measurements of the width the periplasm of Geobacter, but it would be interesting to see if these naturally longer Lpp forms correlate with an enlarged periplasm. That species of Geobacter serve as an exception to what is otherwise a strict rule about the length of Lpp, and thus the structurally enforced distance constraint between the OM and PG layer, raises interesting questions about whether OM softness, OMV production or increased OM permeability might assist the unusual biological functions of the OM and periplasm in bacteria other than E. coli.

Materials and methods

Bacterial strains and growth conditions

Request a detailed protocol

E. coli BW25113 was the parental strain of all the recipient strains used in this study. JW5028, a derivative of BW25113, containing a kanamycin resistance marker in place of a pseudogene (Gagarinova et al., 2012), was used as the wild-type for this study. Strains were grown in LB broth or M9 minimal media broth supplemented with 1 mM MgSO4, 0.1 mM CaCl2, 1.12 mM thiamine and 0.2% (w/v) glucose for a defined minimal media. To osmotically stabilize the growth medium, sorbitol was supplemented to a final concentration of 0.5 M. For overnight cultures, strains were grown in LB broth overnight at 37°C under continuous agitation. Subculturing was done by diluting the saturated culture 1:100 using new media. The cells were then grown to mid-exponential growth phase (OD600 = 0.5–0.6) at 37°C under shaking. Culture media was supplemented with antibiotics for plasmid selection and maintenance or selection of mutants at the following concentrations: 100 μg/ml ampicillin, 30 μg/ml kanamycin, 34 μg/ml chloramphenicol. Fifteen g/l agar was added to media before autoclaving when solid media was required (Figure 3—figure supplement 1).

Construction of Lpp+21 mutant

Request a detailed protocol

The endogenous lpp gene was replaced with the extended lpp+21 gene previously described (Cohen et al., 2017; Asmar et al., 2017) with minor modifications (Figure 1—figure supplement 1). First the replacement was done in the donor E. coli Hfr Cavalli cells using the λ –red recombination system (Datsenko and Wanner, 2000). A gene block was sourced (Integrated DNA Technologies) containing extra 21 amino acid residues (three heptad repeats) inserted between codon 42 and 43 of E. coli Lpp. The gene block also contained 50 bp DNA flanking 5’ and 3’ ends of the lpp+21 gene. On the 5’ end, the extension was homologous to DNA sequence upstream of lpp, while on the 3’ end, the extension was homologous to the cat gene. The gene block was combined with the cat gene by Gibson assembly, and the resulting PCR fragment was used to replace lpp. The new Lpp+21 strain was selected by plating on medium containing chloramphenicol. Chromosomal lpp+21 was moved into E. coli BW25113 background by mating with kanamycin-resistant Keio collection strain JW5028, described above, to generate a double mutant (Baba et al., 2006; Gagarinova et al., 2012; Figure 3—figure supplement 1). The mutation was verified by PCR described (Figure 1—figure supplement 1) and sequencing.

SDS-PAGE and immunoblotting

Request a detailed protocol

Cells grown in M9 minimal media (0.5 M sorbitol) and normalized by OD600nm, lysed in Laemmli SDS-loading dye and separated in 15% acrylamide gels and transferred to 0.45 μm hydrophobic Immobilon-P PVDF membrane (Merck Millipore). Immunoblotting was as described previously (Baba et al., 2006). Rabbit primary antibodies; α-Lpp antibody (kindly provided by T. Silhavy) and α-OmpA were diluted 1:400,000 and 1: 30,000, respectively in 5% skim milk, TBST. The membranes were incubated with goat, α-rabbit IgG, HRP-conjugated secondary antibody (Sigma; 1: 20,000 in 5% skim milk, TBST), and washed with TBST. Detection was by enhanced chemiluminescence with ECL prime western blotting detection reagent (GE Healthcare Life Sciences), visualized using Super RX-N film (Fujifilm).

Outer membrane vesicle purification and quantification

Request a detailed protocol

Overnight cultured cells, grown in LB without antibiotics, were washed twice in 1 x M9 salts then subcultured in 500 ml M9 minimal media supplemented with 0.5 M sorbitol (1:1000 dilution). The strains were grown to late logarithmic phase without antibiotics, OD600 ≈ 0.9 and spun down to collect culture supernatant. Collected culture supernatant were then processed for OMVs isolation and purification using differential ultracentrifugation technique as discussed previously (Deo et al., 2018). OMVs were washed twice in PBS to remove sorbitol then quantified using a bicinchoninic acid assay kit (Thermo Scientific CST#23225).

SDS sensitivity

Request a detailed protocol

Streaks were made on LB (solid) media with 0% SDS – 5% SDS. After 16 hr of incubation at 37°C, the minimum SDS concentration inhibiting growth was obtained by analyzing growth on each concentration. The streaks were done in duplicate and repeated three times.

Proteomics

Request a detailed protocol

Saturated overnight cultures were washed twice in 1 x M9 salts and diluted 1:100 in 10 ml M9 minimal media supplemented with 0.5 M sorbitol. Cultures were further grown to logarithmic phase, collected by centrifugation and washed using PBS buffer. The cell pellet was homogenised in 4% SDS, 100 mM Tris, pH 8.1 and boiled at 95 °C for 10 min. The lysate was then sonicated with a Bioruptor Pico (Diagenode) and protein concentration was determined with Bicinchoninic Acid assay (BCA, Thermo Fisher). SDS was removed with chloroform/methanol, the protein was digested by trypsin overnight and the digested peptides were purified with ZipTips (Agilent). Using a Dionex UltiMate 3000 RSLCnano system equipped with a Dionex UltiMate 3000 RS autosampler, an Acclaim PepMap RSLC analytical column (75 µm x 50 cm, nanoViper, C18, 2 µm, 100 Å; Thermo Scientific) and an Acclaim PepMap 100 trap column (100 µm x 2 cm, nanoViper, C18, 5 µm, 100 Å; Thermo Scientific), the tryptic peptides were separated by increasing concentrations of 80% ACN / 0.1% formic acid at a flow of 250 nl/min for 120 min and analyzed with a QExactive Plus mass spectrometer (Thermo Scientific). The instrument was operated in the data dependent acquisition mode to automatically switch between full scan MS and MS/MS acquisition. Each survey full scan (m/z 375–1575) was acquired in the Orbitrap with 60,000 resolution (at m/z 200) after accumulation of ions to a 3 × 106 target value with maximum injection time of 54ms. Dynamic exclusion was set to 30 s. The 20 most intense multiply charged ions (z ≥ 2) were sequentially isolated and fragmented in the collision cell by higher-energy collisional dissociation (HCD) with a fixed injection time of 54ms, 15,000 resolution and automatic gain control (AGC) target of 2 × 105.

The raw data files were analyzed using MaxQuant software suite v1.6.5.0 (Cox and Mann, 2008) against Andromeda search engine (Cox et al., 2011) for protein identification and to obtain their respective label-free quantification (LFQ) values using in-house standard parameters. The proteomics data was analyzed using LFQ-Analyst (Shah et al., 2020) and the analysis of the data quality analysis is presented in Figure 2—figure supplement 3. Due to the 21 amino acid insertion in the Lpp+21 isoform, the relative levels of Lpp in the mutant had to be assessed manually. Only the unique peptide (IDQLSSDVQTLNAK) shared between the two isoforms was used to quantify the levels of Lpp and Lpp+21 in the wild-type and mutant strain, respectively (Figure 2—figure supplement 2).

To estimate the total relative amount of proteins from the various subcellular compartments, the raw intensities from peptides identified from proteins from different subcellular locations were summed and divided by the total summed intensity from all peptides. Subcellular locations annotations were applied from the STEPdb 2.0 (Loos et al., 2019) where proteins designated F1, A, R, and N were classified as cytoplasmic; B was designated inner membrane; H, X, and F4 were designated outer membrane / extracellular; and I, G F2, F3, and E were designated as periplasmic.

Sequence similarity network analysis

Request a detailed protocol

Proteobacterial proteins containing the Pfam domain (PF00691) were extracted from the Representative Proteome 35% co-membership rpg-35 group (Chen et al., 2011) and a sequence similarity network was generated with the EFI Enzyme Similarity Tool (Gerlt et al., 2015). This network was visualized with Cytoscape (Shannon et al., 2003) with a similarity score cutoff of 30. Each protein is represented by a colored circle node and each similarity match above the similarity score cutoff is represented by an edge between nodes with the length determined by the similarity score.

Lpp length distribution across bacterial species

Request a detailed protocol

To determine the amino acid length distribution of Lpp in Gammaproteobacteria (Supplementary file 4), amino acid sequences were sourced from the InterPRO database (version 81.0) (Mitchell et al., 2019) using the Interpro Family tag - Murein-lipoprotein (IPR016367). Filtered Lpp sequences were then concatenated into representative nodes (at least >90% sequence similarity) using the online available amino acid Initiative-Enzyme Similarity Tool (EFI-EST) (Gerlt et al., 2015).

Synthetic genetic interaction array

Request a detailed protocol

The Lpp+21 isoform was transferred to each of the Keio collection clones by conjugation as described (Figure 3—figure supplement 1). First, the Hfr chloramphenicol resistant Lpp+21 strain was arrayed in 384-colony density on LB agar containing chloramphenicol using the Singer rotor HAD (Singer Instruments, United Kingdom). Similarly, the Keio collection arrayed in 384-colony density was pinned on LB agar plates containing kanamycin and incubated overnight at 37°C. Using the Singer rotor HDA, the Hfr Lpp+21 strain and the Keio collection clones from the 384-colony density were then co-pinned onto LB agar plates and incubated at 37°C for 16 hr. Following conjugation, the colonies were transferred to LB agar with kanamycin (selection 1) at the same colony density and incubated at 37°C for 16 hr. To select for double mutants (selection 2), colonies from the intermediate selection were pinned on LB agar with both kanamycin and chloramphenicol and incubated at 37°C for 14 hr. For assessment of synthetic genetic interaction in nutrient-limited media, the double mutants generated were replica pinned in M9 minimal media at the same density and incubated at 37°C for 25–30 hr. Images were acquired using Phenobooth (Singer Instruments, United Kingdom) for analysis. Images were manually screened to cross-reference recipient plate images to the final double antibiotic selection plates images. Candidate synthetic lethal or growth-compromised mutants were then subjected to another round of screening in the same conditions as previously identified (mini-screen) for validation. Four biological replicates were included that were further arrayed in four technical replicates. Mutants were confirmed by PCR. Where further phenotypic screening of mutants was conducted, independent isogenic knock-out mutants were generated using the PCR and λ-red based homologous recombination method (Datsenko and Wanner, 2000).

Since the Keio collection yiaD mutant has been identified as containing a potential duplication event (Yamamoto et al., 2009), the candidate yiaD synthetic lethal interaction was confirmed through independently constructing a yiaD mutant in the BW25113 strain background (Figure 3—figure supplement 2). The lpp+21 variant was subsequently generated in this mutant as described above. As with the Keio yiaD mutant, this strain demonstrated synthetic lethality on M9 media.

Two colony PCR reactions (Datsenko and Wanner, 2000; Yamamoto et al., 2009; Figure 3—figure supplement 1) confirmed the identity of all candidate double mutants, using a set of primers flanking the lpp gene, and a set of primers flanking the kanamycin gene (Supplementary file 5).

Preparation of electron cryo-microscopy samples, data collection, and analysis

Request a detailed protocol

Strains were grown aerobically in M9 minimal media (0.5 M sorbitol) until an OD600 of 0.6 was reached. Cells were collected by spinning at 6000xg for 5 min and resuspended to an OD600 of ≈ 12. Cryo-EM, data collection, and analysis were performed similarly to previous studies (Cohen et al., 2017; Asmar et al., 2017), except using 3-D subtomogram averages derived from whole-cell cryotomograms instead of projection images so as to discern peptidoglycan. Tilt series of WT and Lpp+21 strains was acquired on an FEI Krios operating at 300 keV with a Gatan K2 direct detector and energy filter with a 20 eV slit with a tilt range of±60° using 3° increments and reconstructed using IMOD. Subtomograms were picked manually using 3DMOD along the length of all non-polar periplasm and averaged using PEET.

Generation of simulation systems

Request a detailed protocol

Initially, two systems were generated: the OM and PG, as previously detailed (Hwang et al., 2018), with two copies of wild-type Lpp and with two copies of Lpp+21. For wild-type Lpp, we used the homo trimer from PDB 1EQ7 (Shu et al., 2000). For Lpp+21, a monomer was first built using I-TASSER (Yang and Zhang, 2015). Next, the trimer of Lpp+21 was built using the wild-type Lpp trimer as a template, further optimized using Targeted Molecular Dynamics (TMD) for one ns. For both Lpp and Lpp+21, the proteins were anchored in the OM via N-terminal acylation while the C-terminus of one copy from each trimer was covalently linked to the PG. The systems generated were prepared for equilibration using the following steps for one ns each: (1) minimization for 10,000 steps, (2) melting of lipid tails, (3) restraining only the PG and the protein, and (4) restraining the PG and the protein backbone. Both systems were equilibrated for 200 ns.

For each of the two systems (Lpp and Lpp+21), a new system was constructed with one Lpp trimer removed and OmpA inserted into the OM. The full-length OmpA structure was taken from Ortiz-Suarez et al., 2016. The periplasmic domain region of the OmpA (clamp, hereafter) was lowered to the PG by shortening the distance between the PG and the clamp for 20 ns. After the clamp was lowered, it was clenched to the nearest DAP residue by using two different distance collective variables (Fiorin et al., 2013) between the centre of mass of residue 242 or 256 from OmpA and that of a nearby PG DAP residue to maintain the connection for 110 ns. In the case of Lpp+21, PG was first pulled toward the OM to match the distance between them in wild-type Lpp, after which the clamp of OmpA was lowered and clenched for 110 ns. The wild-type Lpp/OmpA system was equilibrated for 200 ns; OmpA stayed bound to the PG without colvars. The Lpp+21/OmpA system was also equilibrated for 200 ns, but colvars were needed to maintain the OmpA-PG interaction.

Molecular dynamics (MD)

Request a detailed protocol

All-atom molecular dynamics simulations were performed using NAMD 2.11 (Phillips et al., 2005) and the CHARMM36m (Huang et al., 2017) and CHARMM36 (Klauda et al., 2010) force-field parameters for proteins and lipids, respectively, with the TIP3P-CHARMM water model (Jorgensen et al., 1983). Unless otherwise stated, all MD simulations were performed under a periodic boundary condition with a cut-off at 12 Å for short-range electrostatic and Lennard-Jones interactions with a force-based switching function starting at 10 Å. For long-range electrostatic calculations, the particle-mesh Ewald method (Darden et al., 1993) with a grid spacing of at most 1 Å was used for long-range electrostatic calculations. Bonds between a heavy atom and a hydrogen atom were maintained to be rigid, while all other bonds remain flexible. Unless otherwise stated, each system was equilibrated under an isothermal-isobaric ensemble (NPT) at 310 K and 1 bar, with a timestep of 4 fs after hydrogen mass repartitioning (Balusek et al., 2019). A Langevin thermostat with a damping coefficient of 1 ps–1 was used for temperature control and a Langevin piston was used for pressure control. VMD was used for all visualization and analysis (Humphrey et al., 1996).

Data availability

All data generated from this study is supplied in the relevant supplemental files.

References

  1. Book
    1. Ingraham JL
    2. Marr AG
    (1996)
    Effects of temperature, pressure, pH, and osmotic stress on growth
    In: Neidhardt FC, editors. Escherichia Coli and Salmonella: Cellular and Molecular Biology. ASM Press. pp. 1–12.
    1. Kim S
    2. Yum S
    3. Jo WS
    4. Lee BL
    5. Jeong MH
    6. Ha NC
    (2008)
    Expression and biochemical characterization of the periplasmic domain of bacterial outer membrane porin TdeA
    Journal of Microbiology and Biotechnology 18:845–851.

Decision letter

  1. Petra Anne Levin
    Reviewing Editor; Washington University in St. Louis, United States
  2. Bavesh D Kana
    Senior Editor; University of the Witwatersrand, South Africa
  3. Daniel O Daley
    Reviewer; Stockholm University, Sweden

In the interests of transparency, eLife publishes the most substantive revision requests and the accompanying author responses.

Decision letter after peer review:

[Editors’ note: the authors submitted for reconsideration following the decision after peer review. What follows is the decision letter after the first round of review.]

Thank you for submitting your work entitled "Periplasm homeostatic regulation maintains spatial constraints essential for cell envelope processes and cell viability" for consideration by eLife. Your article has been reviewed by 3 peer reviewers, and the evaluation has been overseen by a Reviewing Editor and a Senior Editor. The reviewers have opted to remain anonymous.

We are sorry to say that, after consultation with the reviewers, we have decided that your work will not be considered further for publication by eLife.

All three reviewers appreciated the genetic approach to understand how changes in the extended Lpp+21 variant modulates periplasmic space. At the same time, however, there was consensus that many of the results were preliminary and mechanistic insight was generally missing. We hope you will find the detailed comments provided by all three reviewers helpful.

Reviewer #1:

The outer-membrane anchored lipoprotein Lpp is the most abundant protein in E. coli and about one third of the Lpp molecules are covalently attached to peptidoglycan, which is important to maintain the outer membrane impermeable to SDS (anionic detergent). Lpp forms a homotrimer made of α-helical monomers, hence, the abundant connections between the peptidoglycan and the outer membrane via Lpp are thought to maintain a constant distance between both envelope layers, which is likely important for trans-envelope processes. In two previous papers, an extended version of Lpp, Lpp+21, was used to study flagellar assembly and signal transduction through an artificially widened periplasm but it remained unclear what other processes are impaired in the Lpp+21 expressing cells. Mandela and co-workers provide now a comprehensive characterization of the strain. Different from previous papers, they conclude that lengthening of Lpp only slightly affects the widths of the periplasm under hyperosmotic condition, but there is a more pronounced effect on the mechanical properties of the outer membrane. The expression of Lpp+21 induces several stress responses and renders specific components of the peptidoglycan remodelling pathway essential.

I find this an interesting and important study and would like to make the following specific comments:

Main points:

1. I am concerned that some of the phenotypes of the Lpp+21 expressing strain are caused by the ~8-fold reduced copy number, compared to the copy number of native Lpp in wild-type cells. With 8-fold reduced amount of Lpp, other outer membrane-peptidoglycan connecting proteins are likely more abundant than Lpp+21 and capable of maintaining a normal distance between both layers, explaining why the authors did not observe a more increased width in the Lpp+21 strain. It would be helpful to try to compare strains with similar amounts of Lpp+21 or Lpp and find out which phenotypes, proteome changes and changes in essentiality of genes are caused by the reduced amount of Lpp+21, and which are caused by the extension of the protein.

2. Line 113 and Figure 1F, numbers of periplasmic widths. The numbers measured by the authors are not numbers of periplasmic widths, as they include also half of the cytoplasmic membrane and half of the outer membrane (measurements were take from the middle of each membranes). Hence, these numbers are different to published data of periplasmic widths (for example, from the Beveridge lab), which could confuse the readers. Hence, I suggest to explain the difference to published numbers in the text and refrain from using the term 'periplasmic widths' for their measurements.

3. Line 115, PG morphology. It is hard to see whether the PG morphology is changed (Figure 1F) because the images of the wt are generally darker than those of the mutant. Can the authors apply some sort of image analysis to support their claims that there is a difference? However, it is also not clear what insights about the PG layer one can possibly obtain from transmission EM images and the discussion about possible effects is very speculative (lines 300-302).

4. Figure 1F. It might be interesting to compare the Lpp+21 strain to a mutant lacking Lpp?

5. Figure 4C/D, model. Are they sure that the attachment of Lpp+21 to peptidoglycan is as efficient as the attachment of wt Lpp? If many Lpp+21 trimers were not attached to peptidoglycan, this would explain the almost normal distance between outer membrane and peptidoglycan, the strong vesicle production (Figure 2C) and the higher SDS sensitivity (Figure 2D). Have the authors considered simulations with Lpp+21 that is not attached to peptidoglycan, would it penetrate the layer?

Reviewer #2:

The outer membrane lipoprotein Lpp provides the only covalent link known to exist between the outer membrane and the peptidoglycan cell wall in E. coli. It has been previously shown that elongating the coiled-coil structure of Lpp increases the distance between the outer membrane and the cell wall, increasing the overall width of the periplasm. Mandela et al. investigate the effects of increasing the length of Lpp by 21 (Lpp+21) residues using proteomics, a genetic screen for factors that become essential when Lpp is elongated, and molecular dynamics. Based on their results, the authors propose that factors that maintain the distance constraint between the outer membrane and the cell wall become essential. However, they do not provide clear insight into why other factors were identified in their experiments.

The authors set out to investigate an interesting question and present a significant amount of data. However, there are important concerns about the study.

1) Important phenotypes reported in the study are significantly different from data previously reported by others. Although the authors acknowledge some (not all) of the discrepancies in the Discussion, they do not address the differences. These differences relate to the distribution, uniformity, and overall values of the distance between the outer membrane and the peptidoglycan layer, the impact that adding 21 residues to Lpp has on Lpp protein levels, and the reported synthetic lethality with mrcB. It would be necessary to address why these differences exist to validate this study.

2) The authors provide a list of genes and proteins from the experiments that they conducted using proteomics and genetics. However, they do not further explore the relationship or mechanism linking those factors to Lpp+21. As a result, the work is limited to reporting correlations and speculative links between factors and lengthening Lpp. None of the proposed connections between Lpp+21 and the identified factors are investigated. As a result, it is unclear whether the effect results from a direct or indirect link to the presence of Lpp+21, or even whether they are the result of increasing the distance from the OM to the cell wall, or simply because the levels of Lpp+21 are significantly decreased in comparison to those of wild-type Lpp in the parent strain.

3) The authors conducted a synthetic lethal screen with non-essential genes and lpp+21. The screen relied on conjugation to construct double mutants. However, the authors have not demonstrated that the recipient single mutant strains are not defective in conjugation, which would lead to the wrong conclusion.

4) The conclusions that OmpA becomes essential in the presence of lpp+21 because it anchors the outer membrane to the cell wall is plausible, but requires further testing, since that is not the only role that OmpA plays.

5) The authors incorrectly conclude that "the" (i.e. "all") non-essential genes involved in LPS synthesis become essential in the lpp+21 strain in the Results and Discussion sections. There are several non-essential genes involved in LPS synthesis that were not identified in this work. Moreover, the authors do not explain why changing LPS structure lead to synthetic lethality. It is also unclear why galU, which is required for making the outer core, would be essential but genes encoding for enzymes required for the synthesis of the middle core are not when they function upstream in the LPS biosynthesis pathway.

6) The authors propose that the deletion of some factors leads to synthetic lethality because they are involved in making complexes that link the inner and outer membranes. The authors suggest that the reason why pumps were identified is because they form a complex linking these membranes. However, AcrA forms a pump with AcrB and TolC but acrB was not identified in the screen. It would be good to know what is the evidence that MdtK, MdtL and YdHI form pumps that connect the two membranes.

Recommendations for the authors:

1) Some of the crucial data presented here are different from data previously reported by others. The authors acknowledge this in the Discussion but provide no explanation, which is very problematic. The most salient differences relate to the phenotypes caused by the presence of the lpp+21 allele. The distribution of the width of the periplasm in both the wild-type and the lpp+21 strains is larger than previously reported (e. g. https://pubmed.ncbi.nlm.nih.gov/29257832/. ). In addition, Asmar et al. (PMID: 29257832) reported that levels of the Lpp+21 protein are similar to those of wild-type Lpp. Here, the authors report that they are significantly lower. Given that when lpp is deleted, blebs are formed and there is less uniformity, it is possible that some (if not most) of the phenotypes presented here are the result of cells having less Lpp than wild type cell, not a longer Lpp protein. An additional discrepancy is that Asmar et al. PMID: 29257832 reported that an lpp+21 mrcB strain grows normally, while here the authors report they are synthetic lethal. It is necessary to address why these differences exist to validate this study.

2) The authors report synthetic lethality between a set of alleles and lpp+21. The screen relied on conjugation to make double mutants. It is unclear whether any of these interactions are specific to lpp+21 or not. Are they also observed with δ-lpp? More importantly, the authors should perform a parallel screen in which the Hfr donor carries a wild-type lpp allele linked to the chloramphenicol-resistance marker to ensure that the recipients are not defective in conjugation or resolving the merodiploid state, as suggested for the yraN mutant. This is especially relevant for mutants altering LPS synthesis since alterations to LPS structure can affect conjugation (and transduction).

3) The authors incorrectly state that the non-essential genes involved in LPS synthesis become essential in the lpp+21 strain in the Results and Discussion sections. There are several genes encoding enzymes for the synthesis of LPS core and two acyltranferases that synthesize lipid A that are not essential in wild type that are not in the list of synthetic lethal genes with lpp+21. Moreover, why would changing LPS structure lead to this synthetic lethality? Furthermore, it is peculiar that galU, which is required for making the outer core of LPS, would be essential but genes encoding for enzymes required for the synthesis of the middle core are not (they would also lack the outer core).

4) The authors propose that the deletion of some factors leads to synthetic lethality because they are involved in making complexes that link the IM and OM. The authors suggest that the reason why pumps were identified is because they form a complex linking the IM and OM. That is true for TolC-dependent pumps. AcrA forms a pump with AcrB and TolC, so why was acrB not identified in the screen? Or tested separately to check their claim? The authors should test the δ-acrB allele to test their model. In addition, what is the evidence that MdtK, MdtL and YdHI form pumps that connect the IM and OM? References should be provided.

5) The authors suggest that the synthetic lethality observed with ompA results from the loss of an OM factor that interacts with the cell wall. This is a reasonable proposal, but OmpA is also the most abundant OM β-barrel protein. To test if, as suggested by the authors, the lethality caused by the ompA null results from the loss of an OM-PG connection, the authors should test an allele of ompA that still produces the OM β-barrel domain but lacks the C-terminal domain that contains the PG-binding domain. Related to this, both Pal and OmpA are very abundant proteins at the OM that interact with the PG layer. Is YiaD also very abundant? The prediction is that it would be for its loss to lead to synthetic lethality when both Pal and OmpA are present. I think the authors might be able to provide information on YiaD's abundance from their proteomics analysis.

6) Figure 3A shows that ycaL does not grow in the wild-type background either, so it should be removed from Table 1 and text.

7) The Discussion is highly speculative, and the arguments are not well formulated to strongly support conclusions. Here are some examples:

– Lines 265-266: I do not understand the argument supporting the authors' suggestion that "Lpp functions as a brace to keep the OM away from the PG layer" as opposed to functioning to ensure that the OM is not too far from the PG. The loss of Lpp leads to formation of OM blebs and wide distribution of cell width. This is problematic to cells. more than increasing width according to previously published studies.

– Lines 282-283: "The bridges needed to mediate LPS transport appear to be susceptible to disruption caused by Lpp+21 in E. coli." What is the evidence for this? To my knowledge, there is no evidence supporting this statement shown in this manuscript or elsewhere.

– Line 307: What is the evidence for "malformed PG"?

Reviewer #3:

The manuscript by Mandela et al. deals with the size of the periplasmic space using an Lpp mutant that has been extended by 21 amino acids. The manuscript mainly focuses on the results of a screen in which they introduce the Lpp+21 protein into the Keio collection to determine viability. There are some interesting findings, including the importance of cell-wall synthesis enzymes and of LPS synthesis and OMV production, but these findings are not followed up on to glean some insight into the processes of cell envelope construction. Moreover, some of the messages in the manuscript are not sufficiently supported (such as "homeostatic regulation" in the title, which I do not think the authors have shown ).

– Line 73: the wording had been stretched from ~9.5 nm to ~11.5 nm, and it is very confusing. In Asmar et al., the periplasmic width in Lpp+21 increases from ~24-25 nm to ~28 nm, which is neither in the range quoted above, nor is the difference.

– The first main result seems to be that Lpp+21 does not change the width of the periplasm, at least by as much as was reported in Asmar et al. I have several serious concerns about this section:

– (1) on line 112-113, the statement "periplasmic width were only slightly greater" is a misstatement of the data; the distributions shown in the figure are almost surely statistically different. Also, a shift of 2-4 nm can be incredibly biologically important, so saying "only slightly greater" is misleading. To make any quantitative statement, the authors would need to perform independent biological replicates of this cryoEM experiment, and then ask whether there is a consistent shift between WT and Lpp+21 as there is obviously a lot of cell-to-cell variation.

– (2) The authors do not provide any explanation for the quantitative difference with the Asmar paper, even though the same EM group (Beeby) was doing the measurements in each case.

– Line 114-115: There is a large amount of variation in the periplasmic thickness across cells, so one cannot draw any conclusions from the isolated images in Figure 1F. If the authors want to make any statements about the distance from PG to the OM, they need to quantify this. The same holds true for statements about PG morphology in the following lines – as it stands these statements seemed cherry-picked. The conclusion on line 316 about how the cell is softer despite having a thicker, non-homogeneous PG is not justified in the paper.

– The abstract states that "the periplasm cannot be widened by engineering approaches." Putting aside the points above about the Lpp+21 construct, this statement is not justified – the authors have only tried one way to change periplasmic width, and there are several groups that have shown that periplasmic size can be altered (see e.g. work from Matthias Heinemann's lab).

– The authors have a whole section called "homeostasis in the periplasm", but it’s difficult to understand what homeostatic process they have uncovered. To do so would involve perturbing some quantity and then seeing the process by which it recovers. I think the closest the paper comes is in lines 307-308, but these are just hypotheses.

Several findings seem too preliminary without follow-up experiments:

– (1) It is interesting that their genetic screen indicates a compromised ability to build the PG, but this conclusion would be greatly bolstered by follow-up experiments such as single-cell growth during CRISPRi depletion and/or HPLC measurements of cell-wall composition. This last point on line 248 about PG synthesis is confusing – why would LpoA/B become individually essential? Without follow-up studies, e.g. using CRISPRi to deplete proteins and studying the phenotypes during depletion, the screen seems preliminary.

– (2) It is interesting that LPS synthesis becomes more essential, but this finding does not appear to be further studied. It is unclear whether LPS levels decreased in the Lpp+21 strain.

– (3) The data suggests that the phenotypes of AcrAB/TolC are due to reduced PG binding, but it is not proven that this is the case.

– The molecular dynamics simulations are also intriguing, but do not provide strong evidence to back up their claims. (1) Were the simulations taking into account differences in protein and LPS concentration in the outer membrane? At the very least, these points should be mentioned as potentially important factors. (2) Evidence needs to be presented that any of these parameters on line 217-219 have stabilized. (3) Moreover, steered molecular dynamics simulations need to be carried to ensure that the system is not stuck in a local minimum based on the initial conditions. (4) It is unclear from the writing what the authors are trying to accomplish with the MD simulations. In large part, this is because the authors have not made clear what their model is for the detrimental effects of Lpp+21.

– The initial motivation regarding the conservation of Lpp length seems insufficient – first, there is no reason to believe that different bacterial species would have periplasms of the same size. Second, for the outlier (a Geobacter species), the authors do not show any evidence that this outlier does have a periplasm of a different size. Therefore, this entire motivation appears to distract from the other data in the paper.

Additional major concerns:

– Line 265: if Lpp acts as a brace, what happens in a ∆lpp strain to OM/PG distances?

– Line 270: I don't see how this is consistent, since there is no hypothesis for what the consequence would be for bracing vs. bridging.

– Line 273: This point about the tilting of Lpp seems to be a simple steric argument, that does not require MD? Or am I missing something? Also, this is an overinterpretation of their data – they have not linked the tilt at all to membrane integrity. This could easily be due to the difference in Lpp levels.

– In fact, they should measure the LPS levels in Lpp+21 cells, which would likely be a very important determinant of permeability and stiffness.

– Line 292: This statement about the PG being uniform has not been shown, and is unnecessary for this paper in any case.

[Editors’ note: further revisions were suggested prior to acceptance, as described below.]

Thank you for resubmitting your work entitled "Periplasm homeostatic regulation maintains spatial constraints essential for cell envelope processes and cell viability" for further consideration by eLife. Your revised article has been evaluated by Bavesh Kana (Senior Editor) and a Reviewing Editor.

Notably, all three reviewers felt that the manuscript has been improved but there are some important remaining issues that need to be addressed prior to accepting the manuscript for publication. In particular, all three reviewers felt that substantial revisions are required to both the text and the title to better reflect the nature of the results and address the limitations of the experiments. Additionally, the reviewers were not convinced that the results provide strong evidence for a dedicated homeostatic regulatory mechanism governing periplasmic dimensions, and requested that this point be significantly toned down in the revised manuscript. Any revisions should also address other significant reviewer concerns as outlined below including but not limited to the possibility that Lpp+21 is simply unstable and thus degraded as well as discrepancies between the findings fo this study and those of Asmar et al.

Reviewer #1:

I have read through the revised version of the manuscript and the authors responses to the first round of reviews.

The study by Mandela et al. investigates how E. coli maintains periplasmic homeostasis. They do this by studying the adaptive response of cells when a lengthened version of the Lpp protein (Lpp21+) is used to tether the outer membrane with the peptidoglycan layer. Lpp21+ has been used previously as a model system, and in these studies, it was largely assumed to increase the distance between the outer membrane and peptidoglycan. In the presented study, the authors use proteomics, a synthetic lethal screen, and molecular dynamic simulations to probe the adaptive response. They note that lengthening Lpp does not change the spatial constrain in the periplasm, but that a number of adaptive responses can be observed, which help to understand how periplasmic homeostasis is maintained.

On the whole, the authors have addressed the vast majority of comments and concerns raised by the 3 referees in the first round of review. This has improved the manuscript. However, two points remain unresolved, which the editor needs to consider.

1. The initial review noted that the study "is limited to reporting correlative associations with the observed phenotypes, and definitive mechanistic evidence would be needed to support the conclusions." This point was strongly refuted by the authors in the revision.

On this point I agree with the initial review. The study has used omics approaches and MD simulations to probe the adaptive response of the cells to the presence of Lpp21+. As such, the conclusions are largely correlative and definitive mechanistic evidence is not provided. However, I do not see this as a 'deal-breaker' for publication. Rather I feel that the study provides plausible explanations and hypotheses as to how E. coli maintains periplasmic homeostasis, which is valuable information for the community.

2. The initial review raised concern that some of the phenotypes of the Lpp21+ expressing strain are caused by the ~8-fold reduced copy number, rather than the extension in protein linker length. On this point I agree with the initial review. Since both occur, the authors need to be more conservative in their conclusions. Or they need to experimentally tease apart the two possibilities. This has not been sufficiently dealt with in the revision.

Related to the point above, the initial review also noted that discrepancies with previous studies had not been resolved in the revised text. For example, the fact that Asmar et al. (PMID: 29257832) reported that levels of the Lpp+21 protein are similar to those of wild-type Lpp. This has not been sufficiently dealt with in the revision.

Finally, a concern that I had, related to the point above is that the study assumes that the 8-fold reduction in levels of Lpp21+ are an adaptive response that helps maintain periplasmic homeostasis (see abstract). They do not exclude the possibility that the engineered protein is subject to proteolysis because it is not native.

To summarise, I think that the study is an interesting and important contribution to the field, but that the authors need to be more conservative in the conclusions that they draw. For example, they need to provide some discussion as to the limitations of their study, and the differences with other published work (point 2).

Reviewer #2:

The authors addressed my concerns about the initial manuscript and submitted a substantially improved revised version. In particular, they now clarify that the 8-fold reduced cell copy number of Lpp+21 could be responsible of some of the phenotypes observed. My remaining doubt is that the observed changes in cell envelope properties (softer outer membrane, increased blebbing) and the altered essentiality of genes could be similar in cells that lack Lpp or that have 8-fold less wild-type Lpp. Overall, I am now more positive about the revised manuscript as it clarifies the question about the periplasmic widths of cells with an elongated Lpp, which has not been fully addressed in previous papers. The work by Mandela et al. provides an interesting and important analysis of how a Gram-negative bacterium maintains a largely functional cell envelope (biogenesis) upon severe disturbance in the linkage between the two key components, the peptidoglycan layer and outer membrane.

Reviewer #3:

The revision by Mandela et al. addresses some but not all of the comments and suggestions that I and the other reviewers introduced in the our initial review. In particular, they have clarified the differences (or lack thereof) with Asmar et al. in terms of the periplasmic width changes due to Lpp+21. Moreover, they have helped to clarify some of the differences between their experiments and others such as the growth defect in ∆mrcB with lpp mutants, though I was surprised that they did not simply do the experiment to test whether their explanation (difference in growth media) was the case.

One of my major criticisms in the initial review persists – that there are interesting conclusions that emerge from the screen of the Keio collection, but none of these results are followed up on, and hence I am left with the question of to what extent any of these phenotypes are direct. I had suggested investigating one in detail using e.g. CRISPRi knockdown in order to examine single-cell phenotypes during depletion of a now-essential protein in the lpp+21 background, but this was not done. Thus, while I find the topic of their paper generally interesting and well motivated, I am left unsatisfied with some of the major conclusions of their paper.

Moreover, many of the conclusions of their paper remain to be fully justified. For instance: in their abstract:

"…impacts the load-bearing capacity of the outer membrane": this is not something that they show to be true in their paper, it is an inference from another paper, and there is nothing in this work that directly addresses load bearing. I don't think they have shown adequately that the combination of tilting and reduced Lpp abundance is the cause of reduced load bearing.

– "E. coli homeostatically counteracts periplasmic enlargement by tilting Lpp and reducing Lpp abundance": I still don't see how there is any homeostasis here, since they are now emphasizing that Lpp+21 does in fact change periplasmic width. Also they have not shown that the reduced Lpp abundance is a cause – what happens if you increase Lpp+21 abundance via inducible expression.

– Line 158: note that I do not believe it has been established that stiffness of E. coli is always associated with OMV production – I would not make claims based on this hypothesized connection.

– Figure 2 title: there is no data here that shows that Lpp+21 cells have a softened OM.

https://doi.org/10.7554/eLife.73516.sa1

Author response

[Editors’ note: the authors resubmitted a revised version of the paper for consideration. What follows is the authors’ response to the first round of review.]

Reviewer #1:

The outer-membrane anchored lipoprotein Lpp is the most abundant protein in E. coli and about one third of the Lpp molecules are covalently attached to peptidoglycan, which is important to maintain the outer membrane impermeable to SDS (anionic detergent). Lpp forms a homotrimer made of α-helical monomers, hence, the abundant connections between the peptidoglycan and the outer membrane via Lpp are thought to maintain a constant distance between both envelope layers, which is likely important for trans-envelope processes. In two previous papers, an extended version of Lpp, Lpp+21, was used to study flagellar assembly and signal transduction through an artificially widened periplasm but it remained unclear what other processes are impaired in the Lpp+21 expressing cells. Mandela and co-workers provide now a comprehensive characterization of the strain. Different from previous papers, they conclude that lengthening of Lpp only slightly affects the widths of the periplasm under hyperosmotic condition, but there is a more pronounced effect on the mechanical properties of the outer membrane. The expression of Lpp+21 induces several stress responses and renders specific components of the peptidoglycan remodelling pathway essential.

I find this an interesting and important study and would like to make the following specific comments:

Main points:

1. I am concerned that some of the phenotypes of the Lpp+21 expressing strain are caused by the ~8-fold reduced copy number, compared to the copy number of native Lpp in wild-type cells. With 8-fold reduced amount of Lpp, other outer membrane-peptidoglycan connecting proteins are likely more abundant than Lpp+21 and capable of maintaining a normal distance between both layers, explaining why the authors did not observe a more increased width in the Lpp+21 strain. It would be helpful to try to compare strains with similar amounts of Lpp+21 or Lpp and find out which phenotypes, proteome changes and changes in essentiality of genes are caused by the reduced amount of Lpp+21, and which are caused by the extension of the protein.

We also conclude that the phenotypes of the Lpp+21 expressing strain are caused by the ~8-fold reduced copy number and the ability to tilt Lpp+21 to minimize its “stretch” on the OM-PG distance.

To clarify, we now state explicitly in the Abstract and Introduction, that we perturbed (enlarged) the main bridge between the OM and PG and show by electron microscopy that E. coli does NOT simply enlarge the periplasm in response, but maintains homeostasis by a combination of (a) tilting the (Lpp) bridge to an angle, and (b) reducing the amount of the (Lpp) bridge, keeps the periplasm from expanding. By genetic screen we identified all of the genes in E. coli that become essential in order to enact this homeostasis, and by quantitative proteomics discovered that very few proteins need to be up- or down-regulated in steady-state levels in order to enact this homeostasis.

2. Line 113 and Figure 1F, numbers of periplasmic widths. The numbers measured by the authors are not numbers of periplasmic widths, as they include also half of the cytoplasmic membrane and half of the outer membrane (measurements were taken from the middle of each membranes). Hence, these numbers are different to published data of periplasmic widths (for example, from the Beveridge lab), which could confuse the readers. Hence, I suggest to explain the difference to published numbers in the text and refrain from using the term 'periplasmic widths' for their measurements.

We appreciate this point, though our measurements are methodologically consistent with those used in the two key papers – Cohen et al. (2017) and Asmar et al. (2017) – we are typically making comparisons with. We now make it clear the reported measurements represent the distance from the middle of the IM density to the middle of the OM density in the subtomograms.

3. Line 115, PG morphology. It is hard to see whether the PG morphology is changed (Figure 1F) because the images of the wt are generally darker than those of the mutant. Can the authors apply some sort of image analysis to support their claims that there is a difference? However, it is also not clear what insights about the PG layer one can possibly obtain from transmission EM images and the discussion about possible effects is very speculative (lines 300-302).

We have modified the sentence on the PG morphology in Line 115 and Lines 300-302. We agree that since this observation is not important to our conclusions, it should be kept as a nominal observation.

4. Figure 1F. It might be interesting to compare the Lpp+21 strain to a mutant lacking Lpp?

The envelope phenotype of a lpp null mutant has been address by EM in multiple papers (eg. Cohen, E. J. et al. 2017 , Sonntag, I. et al. 1978 , and Schwechheimer et al. 2014 ) – demonstrating an increase in OM blebbing. We now make mention of the phenotype (and cite these papers) and thank the reviewer for the suggestion.

5. Figure 4C/D, model. Are they sure that the attachment of Lpp+21 to peptidoglycan is as efficient as the attachment of wt Lpp? If many Lpp+21 trimers were not attached to peptidoglycan, this would explain the almost normal distance between outer membrane and peptidoglycan, the strong vesicle production (Figure 2C) and the higher SDS sensitivity (Figure 2D). Have the authors considered simulations with Lpp+21 that is not attached to peptidoglycan, would it penetrate the layer?

It has been demonstrated the Lpp+21 isoform does not affect the cross linking to PG (Cohen, E. J. et al. (2017) and Asmar et al. (2017)), thus it seems to be not biologically relevant to do this simulation.

Reviewer #2:

The outer membrane lipoprotein Lpp provides the only covalent link known to exist between the outer membrane and the peptidoglycan cell wall in E. coli. It has been previously shown that elongating the coiled-coil structure of Lpp increases the distance between the outer membrane and the cell wall, increasing the overall width of the periplasm. Mandela et al. investigate the effects of increasing the length of Lpp by 21 (Lpp+21) residues using proteomics, a genetic screen for factors that become essential when Lpp is elongated, and molecular dynamics. Based on their results, the authors propose that factors that maintain the distance constraint between the outer membrane and the cell wall become essential. However, they do not provide clear insight into why other factors were identified in their experiments.

The authors set out to investigate an interesting question and present a significant amount of data. However, there are important concerns about the study.

1) Important phenotypes reported in the study are significantly different from data previously reported by others. Although the authors acknowledge some (not all) of the discrepancies in the Discussion, they do not address the differences. These differences relate to the distribution, uniformity, and overall values of the distance between the outer membrane and the peptidoglycan layer, the impact that adding 21 residues to Lpp has on Lpp protein levels, and the reported synthetic lethality with mrcB. It would be necessary to address why these differences exist to validate this study.

We accept the criticism and can make a more direct and clear comparison in rewriting the text: no previous study has evaluated the reduced level of Lpp+21 protein by any line of investigation, nor has any previous study taken into account whether the Lpp+21 protein might be tilted in the periplasm. We have now revised the text in the Results section around Figure 1 more direct language to explain why we needed to evaluate these features. We also are more explicit in the text concerning the need to re-examine by EM enough sections to be representative of the whole cell scenario for periplasmic width. We have done all of the appropriate controls, using immunoblotting and quantitative proteomics and structural biology to accurately represent the experimental model. We now explicitly state that this is the first time that such controls have been done, and make clear that we anticipate our explanations will change the conclusions of five previous papers that assumed that Lpp+21 was present at the same amount as Lpp, and that Lpp+21 always sits perpendicular to the membranes and thereby the 21 extra amino acids always stretch the periplasmic width by a constant amount. This assumption is incorrect, as we evidence here for the first time.

2) The authors provide a list of genes and proteins from the experiments that they conducted using proteomics and genetics. However, they do not further explore the relationship or mechanism linking those factors to Lpp+21. As a result, the work is limited to reporting correlations and speculative links between factors and lengthening Lpp. None of the proposed connections between Lpp+21 and the identified factors are investigated. As a result, it is unclear whether the effect results from a direct or indirect link to the presence of Lpp+21, or even whether they are the result of increasing the distance from the OM to the cell wall, or simply because the levels of Lpp+21 are significantly decreased in comparison to those of wild-type Lpp in the parent strain.

We disagree that our study is limited to reporting correlations and speculative links between factors and Lpp+21. There are no correlations in this paper. We perturbed (enlarged) the main bridge between the OM and PG and show by electron microscopy that E. coli does NOT simply enlarge the periplasm in response, but can maintain homeostasis by a combination of (a) tilting the bridge to an angle, and (b) reducing the amount of the bridge, to keep the periplasm from expanding. These are documented facts, not correlations. To test the hypothesis that a collection of genes of known function would become essential in order to enact this homeostasis, we established a genetic screen and identified a set of genes defining five known functional categories all relevant to periplasmic width. By applying quantitative proteomics we discovered that very few proteins need to be up- or down-regulated in steady-state levels in order to enact this homeostasis. No correlations are measured or drawn in the paper.

As to the last point, the three options mentioned are the same thing: the presence of Lpp+21 was previously assumed to increase the distance from the OM to the cell wall, and we show that the levels of Lpp+21 are significantly decreased in comparison to those of wild-type Lpp in the parent strain. We document, quantitatively, what happens in this situation, correcting the record.

3) The authors conducted a synthetic lethal screen with non-essential genes and lpp+21. The screen relied on conjugation to construct double mutants. However, the authors have not demonstrated that the recipient single mutant strains are not defective in conjugation, which would lead to the wrong conclusion.

We appreciate this point and have clarified the sentence in the Methods and in the Results that deals with this issue. The donor lpp+21 single mutant is clearly not defective in conjugation as multiple mutants of this strain could be generated and confirmed (if the parent was defective in conjugation we would expect no antibiotic resistant colonies to be generated). It is unreasonable to suggest we individually assess every single mutant in the Keio collection (~4000 strains) for conjugation defects (this has never been required of any of the multiple widely cited published eSGA screens, all using this established method relying on conjugation for gene knock out – e.g. PMID: 18677321; 24586182; 22125496). Furthermore – If any of the single mutants were defective in conjugation they would appear as “false positives” in the synthetic screen (i.e they would not grow on the second antibiotic). Once we had identified synthetic phenotypes in the primary conjugation-based screen, we then made “clean” mutants that did not rely on conjugation (ie via the λ red method which utilises transformation) to be certain of the outcomes documented here. We thank the reviewer for seeking clarity on this point.

4) The conclusions that OmpA becomes essential in the presence of lpp+21 because it anchors the outer membrane to the cell wall is plausible, but requires further testing, since that is not the only role that OmpA plays.

We must respectfully disagree. While ompA deletion mutants have been implicated genetically in various screens, the only biochemically valid function for OmpA is anchoring the outer membrane to the cell wall. Even if OmpA were to play other roles, the OM-PG anchor role and any other OmpA roles are not mutually exclusive.

The important point in this paper is that OmpA is not essential for viability in an Lpp strain, but becomes essential in the Lpp21 background.

5) The authors incorrectly conclude that "the" (i.e. "all") non-essential genes involved in LPS synthesis become essential in the lpp+21 strain in the Results and Discussion sections. There are several non-essential genes involved in LPS synthesis that were not identified in this work. Moreover, the authors do not explain why changing LPS structure lead to synthetic lethality. It is also unclear why galU, which is required for making the outer core, would be essential but genes encoding for enzymes required for the synthesis of the middle core are not when they function upstream in the LPS biosynthesis pathway.

We appreciate this point and have clarified the sentence in the Results and Discussion sections dealing with the LPS biosynthesis pathway, to remove this ambiguity. While there are non-essential genes involved in LPS synthesis become essential in the lpp+21 strain, we did not intend to imply that all non-essential genes involved in LPS synthesis must become essential in the lpp+21 strain.

6) The authors propose that the deletion of some factors leads to synthetic lethality because they are involved in making complexes that link the inner and outer membranes. The authors suggest that the reason why pumps were identified is because they form a complex linking these membranes. However, AcrA forms a pump with AcrB and TolC but acrB was not identified in the screen. It would be good to know what is the evidence that MdtK, MdtL and YdHI form pumps that connect the two membranes.

We now note that acrB was not identified in this screen, and cite the papers that implicate MdtK, MdtL and YdHI with pumps that connect the two membranes. YdhI – while it is unknown function it is in the same operon as tripartite pump components YdhJK, and has homology to AaeX [also in same operon as tripartite pump component AaeAB] and homology to YtcA [which is also in same operon as tripartite pump components MdtNO, possibly MdtNOP]

Recommendations for the authors:

1) Some of the crucial data presented here are different from data previously reported by others. The authors acknowledge this in the Discussion but provide no explanation, which is very problematic. The most salient differences relate to the phenotypes caused by the presence of the lpp+21 allele. The distribution of the width of the periplasm in both the wild-type and the lpp+21 strains is larger than previously reported (e. g. https://pubmed.ncbi.nlm.nih.gov/29257832/. ). In addition, Asmar et al. (PMID: 29257832) reported that levels of the Lpp+21 protein are similar to those of wild-type Lpp. Here, the authors report that they are significantly lower. Given that when lpp is deleted, blebs are formed and there is less uniformity, it is possible that some (if not most) of the phenotypes presented here are the result of cells having less Lpp than wild type cell, not a longer Lpp protein. An additional discrepancy is that Asmar et al. PMID: 29257832 reported that an lpp+21 mrcB strain grows normally, while here the authors report they are synthetic lethal. It is necessary to address why these differences exist to validate this study.

In their paper, Asmar et al. (PMID: 29257832, https://pubmed.ncbi.nlm.nih.gov/29257832/) show a width distribution (Figure 2B) that is similar to ours with most widths of the Lpp+21 cells overlap with the wild-type cells in both papers, but have selected the microraphs (Figure 2A) only for the increased width distributions. These select micrographs are the basis on which their conclusions are drawn: this is appropriate for their conclusion of “Can Lpp+21 exert an effect on periplasmic width”, but would not be appropriate for our study which asks “Is the effect of Lpp+21 uniform across the bacterial cell, or is it buffered by homeostasis”. There is a moderate increase in in the distribution of widths among both the WT and lpp+21 strain compared to the Asmar paper (~18-35 vs 20-46 and 2139 vs 24-50) but given the differences in data collections methodology (we use whole-cell cryotomograms instead of projection images) and the osmotic conditions of the growth media this is not unexpected or concerning. The trend in both studies is similar with our data showing a longer distribution on the upper end.

In their paper, Asmar et al. (PMID: 29257832, https://pubmed.ncbi.nlm.nih.gov/29257832/) conduct the viability experiments (i.e. address the phenotype of a mrcB strain of E. coli) in rich LB medium. The synthetic lethal screen in our work is conducted on minimal medium. Furthermore, there are differences in the strain backgrounds used. We have made this point more clear in the text of the revised paper.

In their paper, Asmar et al. (PMID: 29257832) do not show data on the levels of the Lpp+21 protein, and do not measure this relative to the wild-type Lpp. The only data dealing with this point is presented as Figure S5 which does demonstrate that all of the Lpp isoforms have similar PG connectivity, but does not address the steady-state level of Lpp versus Lpp+21. The loading of the gel have been made to set the same level of Lpp protein isoform in each lane, with no controls used to interpret the relative levels of Lpp and Lpp+21.

2) The authors report synthetic lethality between a set of alleles and lpp+21. The screen relied on conjugation to make double mutants. It is unclear whether any of these interactions are specific to lpp+21 or not. Are they also observed with δ-lpp? More importantly, the authors should perform a parallel screen in which the Hfr donor carries a wild-type lpp allele linked to the chloramphenicol-resistance marker to ensure that the recipients are not defective in conjugation or resolving the merodiploid state, as suggested for the yraN mutant. This is especially relevant for mutants altering LPS synthesis since alterations to LPS structure can affect conjugation (and transduction).

We appreciate this point and have clarified the sentence in the Methods and in the Results that deals with this issue. The eSGA library creation and screening methods used here are established techniques (e.g. PMID: 18677321; 24586182; 22125496). If any of the single mutants were defective in conjugation they would appear as “false positives” in the synthetic screen. That is why, once we had identified synthetic phenotypes in the primary conjugation-based screen, we then made “clean” mutants that did not rely on conjugation (ie by transformation) to be certain of the outcomes documented here.

3) The authors incorrectly state that the non-essential genes involved in LPS synthesis become essential in the lpp+21 strain in the Results and Discussion sections. There are several genes encoding enzymes for the synthesis of LPS core and two acyltranferases that synthesize lipid A that are not essential in wild type that are not in the list of synthetic lethal genes with lpp+21. Moreover, why would changing LPS structure lead to this synthetic lethality? Furthermore, it is peculiar that galU, which is required for making the outer core of LPS, would be essential but genes encoding for enzymes required for the synthesis of the middle core are not (they would also lack the outer core).

We have clarified the sentence in the Results and Discussion sections dealing with the LPS biosynthesis pathway.

4) The authors propose that the deletion of some factors leads to synthetic lethality because they are involved in making complexes that link the IM and OM. The authors suggest that the reason why pumps were identified is because they form a complex linking the IM and OM. That is true for TolC-dependent pumps. AcrA forms a pump with AcrB and TolC, so why was acrB not identified in the screen? Or tested separately to check their claim? The authors should test the δ-acrB allele to test their model. In addition, what is the evidence that MdtK, MdtL and YdHI form pumps that connect the IM and OM? References should be provided.

We now note that acrB was not identified in this screen, and cite the papers that implicate MdtK, MdtL and YdHI with pumps that connect the two membranes.

5) The authors suggest that the synthetic lethality observed with ompA results from the loss of an OM factor that interacts with the cell wall. This is a reasonable proposal, but OmpA is also the most abundant OM β-barrel protein. To test if, as suggested by the authors, the lethality caused by the ompA null results from the loss of an OM-PG connection, the authors should test an allele of ompA that still produces the OM β-barrel domain but lacks the C-terminal domain that contains the PG-binding domain. Related to this, both Pal and OmpA are very abundant proteins at the OM that interact with the PG layer. Is YiaD also very abundant? The prediction is that it would be for its loss to lead to synthetic lethality when both Pal and OmpA are present. I think the authors might be able to provide information on YiaD's abundance from their proteomics analysis.

On the point of OmpA function, we must respectfully disagree. While ompA deletion mutants have been implicated genetically in various screens, the only biochemically valid function for OmpA is anchoring the outer membrane to the cell wall.

On the point of YiaD we agree and have modified the sentence. The quantitative proteomics suggests that YiaD is present at 5% the level of Pal, similar to the 7% amounts found by Li et al. (https://doi.org/10.1016/j.cell.2014.02.033).

6) Figure 3A shows that ycaL does not grow in the wild-type background either, so it should be removed from Table 1 and text.

We have made this correction and thank the reviewer.

7) The Discussion is highly speculative, and the arguments are not well formulated to strongly support conclusions. Here are some examples:

– Lines 265-266: I do not understand the argument supporting the authors' suggestion that "Lpp functions as a brace to keep the OM away from the PG layer" as opposed to functioning to ensure that the OM is not too far from the PG. The loss of Lpp leads to formation of OM blebs and wide distribution of cell width. This is problematic to cells. more than increasing width according to previously published studies.

– Lines 282-283: "The bridges needed to mediate LPS transport appear to be susceptible to disruption caused by Lpp+21 in E. coli." What is the evidence for this? To my knowledge, there is no evidence supporting this statement shown in this manuscript or elsewhere.

– Line 307: What is the evidence for "malformed PG"?

We have revised the text in the Discussion as suggested, and to address (and clarify) the points of speculation. We appreciate the feedback.

Reviewer #3:

The manuscript by Mandela et al. deals with the size of the periplasmic space using an Lpp mutant that has been extended by 21 amino acids. The manuscript mainly focuses on the results of a screen in which they introduce the Lpp+21 protein into the Keio collection to determine viability. There are some interesting findings, including the importance of cell-wall synthesis enzymes and of LPS synthesis and OMV production, but these findings are not followed up on to glean some insight into the processes of cell envelope construction. Moreover, some of the messages in the manuscript are not sufficiently supported (such as "homeostatic regulation" in the title, which I do not think the authors have shown ).

We note that the manuscript is not aimed at the processes of cell envelope construction, which are already understood in great detail. The paper aims to address whether and how the periplasmic volume is maintained within a very narrow range by self-regulating to maintain stability while adjusting to perturbed conditions i.e. homeostasis.

– Line 73: the wording had been stretched from ~9.5 nm to ~11.5 nm, and it is very confusing. In Asmar et al., the periplasmic width in Lpp+21 increases from ~24-25 nm to ~28 nm, which is neither in the range quoted above, nor is the difference.

We apologise for this typo, and have corrected the text to reflect the data presented in the cited papers.

– The first main result seems to be that Lpp+21 does not change the width of the periplasm, at least by as much as was reported in Asmar et al. I have several serious concerns about this section: (1) on line 112-113, the statement "periplasmic width were only slightly greater" is a misstatement of the data; the distributions shown in the figure are almost surely statistically different. Also, a shift of 2-4 nm can be incredibly biologically important, so saying "only slightly greater" is misleading. To make any quantitative statement, the authors would need to perform independent biological replicates of this cryoEM experiment, and then ask whether there is a consistent shift between WT and Lpp+21 as there is obviously a lot of cell-to-cell variation.

– (2) The authors do not provide any explanation for the quantitative difference with the Asmar paper, even though the same EM group (Beeby) was doing the measurements in each case.

1. We appreciate the point and have corrected the text. We agree with the reviewer, and a major motivation of our work – and the conclusions we can draw – comes from what is biologically important. The paper has been revised to fix this.

2. We initiated the collaboration with Prof. Morgan Beeby in order to be sure that the evaluation we did here would have consistency to the previous work by Asmar et al., and to enable us to have proper control and quantitation in our study. We have revised the text to explicitly state that the primary data is in agreement, but that our conclusions cover all of the observations, and are not biased by a simplistic use of an “average value” approach. See also explicit comments to Reviewer 2 on this point, including area under the graph calculations that have been added into the revised manuscript.

– Line 114-115: There is a large amount of variation in the periplasmic thickness across cells, so one cannot draw any conclusions from the isolated images in Figure 1F. If the authors want to make any statements about the distance from PG to the OM, they need to quantify this. The same holds true for statements about PG morphology in the following lines – as it stands these statements seemed cherry-picked. The conclusion on line 316 about how the cell is softer despite having a thicker, non-homogeneous PG is not justified in the paper.

We appreciate this point and have revised the statements accordingly. As we noted to Reviewer 1, the sentences on the PG morphology in Lines 114-115 (and Lines 300-302). We agree that since this observation is not important to our conclusions, it should be kept as a nominal observation. We do over quantified measurements of the mid PG to mid OM distance in Figure 1F.

We have taken the reviewer’s advice on the conclusion on line 316 about how the cell is softer despite having a thicker, non-homogeneous PG and revised the conclusion accordingly.

– The abstract states that "the periplasm cannot be widened by engineering approaches." Putting aside the points above about the Lpp+21 construct, this statement is not justified – the authors have only tried one way to change periplasmic width, and there are several groups that have shown that periplasmic size can be altered (see e.g. work from Matthias Heinemann's lab).

We accept the reviewer’s point and have modified the Abstract to remove the sentence.

– The authors have a whole section called "homeostasis in the periplasm", but it’s difficult to understand what homeostatic process they have uncovered. To do so would involve perturbing some quantity and then seeing the process by which it recovers. I think the closest the paper comes is in lines 307-308, but these are just hypotheses.

To address this point we have changed the sub-heading, added an explanatory sentence to the Introduction and revised the Discussion to make this more clear.

The logic provided is that we perturbed (enlarged) the main bridge between the OM and PG and show by electron microscopy that E. coli does NOT simply enlarge the periplasm in response: in some places the periplasm is enlarged but in the majority of the periplasm width is maintained (homeostasis) by a combination of (a) tilting the bridge to an angle, and (b) reducing the amount of the enlarged bridge, to keep the periplasm from expanding. We note in discussing the genetic screen that it identified all of the genes in E. coli that become essential in order to enact this homeostasis, and state that by quantitative proteomics very few proteins need to be up- or downregulated in steady-state levels in order to enact this homeostasis.

Several findings seem too preliminary without follow-up experiments:

– (1) It is interesting that their genetic screen indicates a compromised ability to build the PG, but this conclusion would be greatly bolstered by followup experiments such as single-cell growth during CRISPRi depletion and/or HPLC measurements of cell-wall composition. This last point on line 248 about PG synthesis is confusing – why would LpoA/B become individually essential? Without followup studies, e.g. using CRISPRi to deplete proteins and studying the phenotypes during depletion, the screen seems preliminary.

While we appreciate that these further investigations would be telling for PG biogenesis, the purpose of the genetic screen in this paper is simply to identify those genes in E. coli that become essential in order to enact the maintenance of periplasmic width. The paper as presented represents 5-person years of work by a new investigator and his first PhD student, both of whom have now moved on to new positions. We agree that interesting questions remain, but they sit beyond the scope of this study, and beyond the technology available to me in the new laboratory.

– (2) It is interesting that LPS synthesis becomes more essential, but this finding does not appear to be further studied. It is unclear whether LPS levels decreased in the Lpp+21 strain.

With respect, we would note again that while interesting questions remain they sit beyond the scope of this study.

– (3) The data suggests that the phenotypes of AcrAB/TolC are due to reduced PG binding, but it is not proven that this is the case.

We agree with the reviewer (and Reviewer 1) on this point, we have revised the text to both note that the quantitative proteomics suggests that TolC (and AcrA and AcrB) are present at 5% the level of OmpA, and to make clear that this hypothesis has not been tested.

– The molecular dynamics simulations are also intriguing, but do not provide strong evidence to back up their claims. (1) Were the simulations taking into account differences in protein and LPS concentration in the outer membrane? At the very least, these points should be mentioned as potentially important factors. (2) Evidence needs to be presented that any of these parameters on line 217-219 have stabilized. (3) Moreover, steered molecular dynamics simulations need to be carried to ensure that the system is not stuck in a local minimum based on the initial conditions. (4) It is unclear from the writing what the authors are trying to accomplish with the MD simulations. In large part, this is because the authors have not made clear what their model is for the detrimental effects of Lpp+21.

We address these points by number.

1) Because we are simulating only a very tiny portion of the membrane, we did not account for heterogeneity in the protein and LPS concentrations. This is now noted in the revised manuscript.

2) Plots of the angles over time, from which the averages and standard deviations were derived, are now included as a supplemental figure. We also modified how we measured the angle to correct for an error due to wrapped coordinates; we note this primarily affected the Lpp+21 w/OmpA simulation. The new angles (average ± standard deviation), taken from the last 100 ns of the 200-ns simulations, are as

follows:

Lpp 76.9 ± 4.7

Lpp w/OmpA 75.5 ± 4.7

Lpp+21 82.8 ± 2.9

Lpp+21 w/OmpA 49.4 ± 2.3

To check for convergence, we ran the Lpp simulation for an additional 200 ns (data not shown). The angle from the last 200 ns is 78.3 +/- 4.8°, nearly identical (within 1.4°) to that from the 100-200 ns window. Thus, judging by the plots and the this extended run, we believe the angles are converged within 100 ns.

3) As part of another study, we recently carried out Steered MD and free-energy calculations for Lpp, the latter of which we note were computationally expensive at over 1 μs in total. Specifically, we determined the free energy as a function of the distance between the cell wall and outer membrane with two Lpp trimers bridging them. As can be seen in Gumbart et al.,. Lpp positions peptidoglycan at the AcrA-TolC interface in the AcrAB-TolC multidrug efflux pump. Biophys. J. In press., the free energy minimum agrees with the position from our equilibrium simulations. We also note that there are no barriers nor other features in the PMF that would indicate it is trapped in a local minimum in our equilibrium simulations. Therefore, we feel justified in concluding that our other systems are also unlikely to be trapped in local minima in the equilibrium simulations.

Gumbart, Ferreira, Hwang, Hazel, Cooper, Parks, Smith, Zgurskaya, Beeby. Lpp positions peptidoglycan at the AcrA-TolC interface in the AcrAB-TolC multidrug efflux pump. Biophys. J. In press.

We note that three of the authors, Gumbart, Hwang, and Beeby, are common between both papers.

1) The purpose of the simulations is to show plausible ways that connections of Lpp, Lpp+21, and OmpA with the cell wall are accommodated. We show that in WT E. coli, Lpp and OmpA are complementary to one another. However, the preferred lengths of Lpp+21 and OmpA are in conflict with one another and that accommodation of Lpp+21 requires tilting.

– The initial motivation regarding the conservation of Lpp length seems insufficient – first, there is no reason to believe that different bacterial species would have periplasms of the same size. Second, for the outlier (a Geobacter species), the authors do not show any evidence that this outlier does have a periplasm of a different size. Therefore, this entire motivation appears to distract from the other data in the paper.

We appreciate the point as raised, and we have now modified the text around this accordingly, but we would keep this analysis in the paper: it accurately portrays our motivation to initiate the study, but we have been more clear that the question we wondered at was “IF these elongated Lpp isoforms DO act as read-outs on periplasmic width then WHY is it that most Lpp lengths are equivalent to that of E. coli. Thus, in the revised Results text we make clear that we did not assume that different bacterial species would have periplasms of the same size, but started with the question: Over what size range does Lpp range across species? Secondly, in the revised text in the Discussion we make clear that we do not assume that Geobacter has an extraordinary periplasm, but simply that we will be open to the possibility up until direct analysis is done on Geobacter.

Additional major concerns:

– Line 265: if Lpp acts as a brace, what happens in a ∆lpp strain to OM/PG distances?

– Line 270: I don't see how this is consistent, since there is no hypothesis for what the consequence would be for bracing vs. bridging.

We have removed the discussion on this point, so as to remove the speculation.

– Line 273: This point about the tilting of Lpp seems to be a simple steric argument, that does not require MD? Or am I missing something? Also, this is an overinterpretation of their data – they have not linked the tilt at all to membrane integrity. This could easily be due to the difference in Lpp levels.

The discussion on this point has been clarified at Line 273 and we have revised the discussion to make clear BOTH the tilting of Lpp+21 and the difference in steady state levels between Lpp+21 and Lpp are consistent means by which periplasmic width can be maintained in this system.

– In fact, they should measure the LPS levels in Lpp+21 cells, which would likely be a very important determinant of permeability and stiffness.

– Line 292: This statement about the PG being uniform has not been shown, and is unnecessary for this paper in any case.

We appreciate the points raised and have modified the sentence on the PG morphology in Line 115 (and from lines 300-302). We agree that since this observation is not important to our conclusions, it should be kept as a nominal observation and not be over-stated.

[Editors’ note: what follows is the authors’ response to the second round of review.]

Reviewer #1:

I have read through the revised version of the manuscript and the authors responses to the first round of reviews.

The study by Mandela et al. investigates how E. coli maintains periplasmic homeostasis. They do this by studying the adaptive response of cells when a lengthened version of the Lpp protein (Lpp21+) is used to tether the outer membrane with the peptidoglycan layer. Lpp21+ has been used previously as a model system, and in these studies, it was largely assumed to increase the distance between the outer membrane and peptidoglycan. In the presented study, the authors use proteomics, a synthetic lethal screen, and molecular dynamic simulations to probe the adaptive response. They note that lengthening Lpp does not change the spatial constrain in the periplasm, but that a number of adaptive responses can be observed, which help to understand how periplasmic homeostasis is maintained.

On the whole, the authors have addressed the vast majority of comments and concerns raised by the 3 referees in the first round of review. This has improved the manuscript. However, two points remain unresolved, which the editor needs to consider.

1. The initial review noted that the study "is limited to reporting correlative associations with the observed phenotypes, and definitive mechanistic evidence would be needed to support the conclusions." This point was strongly refuted by the authors in the revision.

On this point I agree with the initial review. The study has used omics approaches and MD simulations to probe the adaptive response of the cells to the presence of Lpp21+. As such, the conclusions are largely correlative and definitive mechanistic evidence is not provided. However, I do not see this as a 'deal-breaker' for publication. Rather I feel that the study provides plausible explanations and hypotheses as to how E. coli maintains periplasmic homeostasis, which is valuable information for the community.

We thank the reviewer for their thoughtful assessment of our revised manuscript. While this study does not provide definitive answers to some biological observations, we agree that the present study expands our understanding of lpp length and we attempt to offer data-driven hypotheses to the observations. In our current revision, we aim to better document the potential limitations of our findings.

2. The initial review raised concern that some of the phenotypes of the Lpp21+ expressing strain are caused by the ~8-fold reduced copy number, rather than the extension in protein linker length. On this point I agree with the initial review. Since both occur, the authors need to be more conservative in their conclusions. Or they need to experimentally tease apart the two possibilities. This has not been sufficiently dealt with in the revision.

Related to the point above, the initial review also noted that discrepancies with previous studies had not been resolved in the revised text. For example, the fact that Asmar et al. (PMID: 29257832) reported that levels of the Lpp+21 protein are similar to those of wild-type Lpp. This has not been sufficiently dealt with in the revision.

In their paper, Asmar et al. (PMID: 29257832) do not directly quantitate data on the levels of the Lpp+21 protein and do not measure this relative to the wild-type Lpp. The only data dealing with this point is presented as Figure S5B, which does demonstrate that all of the Lpp isoforms have similar PG connectivity but does not address the steady-state levels of Lpp versus Lpp+21. The method of loading of the gel is unclear, with no loading controls used to interpret the relative levels of Lpp and Lpp+21 – no quantification of the levels of Lpp isoforms was presented. Furthermore, in the figure there is arguably a minor reduction in the levels of the two longer versions.

Due to this apparent discrepancy, we ensured that we used appropriate loading controls in our gels and WBs (Figure 1B), and furthermore, we confirmed our observed reduction with proteomics.

While, we cannot rule out that minor differences in methodology used in the construction of the lpp+21 "knock-in" mutant strains in our study and the Asmar study may result in expression differences or transcript stability, our transcriptomics data do not support this. The transcriptomics shown in table S6 do not show a reduction in lpp transcripts compared to WT. The resulting sequence and structure of the Lpp+21 protein expressed from the modified genes should be identical in both studies.

We have better clarified the differences in the revised manuscript.

Finally, a concern that I had, related to the point above is that the study assumes that the 8-fold reduction in levels of Lpp21+ are an adaptive response that helps maintain periplasmic homeostasis (see abstract). They do not exclude the possibility that the engineered protein is subject to proteolysis because it is not native.

We agree with the reviewer on this point – there is a possibility that the longer "synthetic" version of Lpp is more unstable and/or degraded more readily. It’s worth noting that the transcription level of lpp was not reduced (table S6), suggesting that either: (i) the protein is less "stable" or; (ii) the protein is "targeted" for degradation in response to the stretching or; (iii) that the modified transcript is translated less efficiently.

We have clarified this in the discussion of the revised manuscript.

To summarise, I think that the study is an interesting and important contribution to the field, but that the authors need to be more conservative in the conclusions that they draw. For example, they need to provide some discussion as to the limitations of their study, and the differences with other published work (point 2).

We have now clarified and amended the discussions to better describe the limitation of the present study and attempt to outline the key differences with previous studies.

Reviewer #2:

The authors addressed my concerns about the initial manuscript and submitted a substantially improved revised version. In particular, they now clarify that the 8-fold reduced cell copy number of Lpp+21 could be responsible of some of the phenotypes observed. My remaining doubt is that the observed changes in cell envelope properties (softer outer membrane, increased blebbing) and the altered essentiality of genes could be similar in cells that lack Lpp or that have 8-fold less wild-type Lpp. Overall, I am now more positive about the revised manuscript as it clarifies the question about the periplasmic widths of cells with an elongated Lpp, which has not been fully addressed in previous papers. The work by Mandela et al. provides an interesting and important analysis of how a Gram-negative bacterium maintains a largely functional cell envelope (biogenesis) upon severe disturbance in the linkage between the two key components, the peptidoglycan layer and outer membrane.

We thank the reviewer for their thoughtful assessment of our revised manuscript. We agree that we can not rule out the effect that reduction in Lpp copy number may have on the cells. In revision, we have clarified that this is a limitation of the data, but it is an important finding due to the potential implication on previous studies using the Lpp+21 isoform.

Reviewer #3:

The revision by Mandela et al. addresses some but not all of the comments and suggestions that I and the other reviewers introduced in the our initial review. In particular, they have clarified the differences (or lack thereof) with Asmar et al. in terms of the periplasmic width changes due to Lpp+21. Moreover, they have helped to clarify some of the differences between their experiments and others such as the growth defect in ∆mrcB with lpp mutants, though I was surprised that they did not simply do the experiment to test whether their explanation (difference in growth media) was the case.

We have now conducted this experiment and included it in as a supplemental figure. In our synthetic screen the initial selections were done on LB media with antibiotics before testing in minimal media. Both the ΔmrcB lpp+21 and ΔmrcA lpp+21 were able to be recovered from the initial double antibiotic selection on LB and could grow on LB agar plates, albeit with an apparent minor growth defect (Figure 3—figure supplement 3).

One of my major criticisms in the initial review persists – that there are interesting conclusions that emerge from the screen of the Keio collection, but none of these results are followed up on, and hence I am left with the question of to what extent any of these phenotypes are direct. I had suggested investigating one in detail using e.g. CRISPRi knockdown in order to examine single-cell phenotypes during depletion of a now-essential protein in the lpp+21 background, but this was not done. Thus, while I find the topic of their paper generally interesting and well motivated, I am left unsatisfied with some of the major conclusions of their paper.

While we agree that it would be interesting, the proposed experiment is beyond the scope of the present study. However, it could form the basis of a subsequent manuscript. Many studies using the KEIO mutant library do not (and realistically could not) follow up on all "hits" with a conditional single cell knockdown experiments. This represents a significant amount of work, and we cannot recall any study where synthetic lethal screens were routinely investigated via such silencing methods.

We confirmed all “hits” by generating independent mutants of all relevant double knockouts.

Moreover, many of the conclusions of their paper remain to be fully justified. For instance: in their abstract:

"…impacts the load-bearing capacity of the outer membrane": this is not something that they show to be true in their paper, it is an inference from another paper, and there is nothing in this work that directly addresses load bearing. I don't think they have shown adequately that the combination of tilting and reduced Lpp abundance is the cause of reduced load bearing.

We have revised the relevant text to avoid making this claim throughout.

– "E. coli homeostatically counteracts periplasmic enlargement by tilting Lpp and reducing Lpp abundance": I still don't see how there is any homeostasis here, since they are now emphasizing that Lpp+21 does in fact change periplasmic width.

We have revised the title, abstract and discussions to removed and/or toned down any reference to a "homeostatic" regulatory mechanism throughout the manuscript and have sought to place more emphasis on the cellular "adaptation" or "response" to the Lpp+21 protein.

Also, they have not shown that the reduced Lpp abundance is a cause – what happens if you increase Lpp+21 abundance via inducible expression.

In our study there was a reduction in the steady state protein level of Lpp+21 compared to Lpp. We have also demonstrated that this was not due to reduced transcription of lpp+21, indeed transcriptomic data (table S6) showed an increased level of lpp+21 transcripts.

– Line 158: note that I do not believe it has been established that stiffness of E. coli is always associated with OMV production – I would not make claims based on this hypothesized connection.

We are not suggesting that all things that effect cell stiffness affect OMV production, just that there is a documented correlation between reduced cell stiffness and OMV production. Thus we sought to investigate OMV production in our lpp mutant strain. We have clarified the relevant sentence to better represent our intent.

– Figure 2 title: there is no data here that shows that Lpp+21 cells have a softened OM.

The title to better represent the data.

https://doi.org/10.7554/eLife.73516.sa2

Article and author information

Author details

  1. Eric Mandela

    Infection & Immunity Program, Biomedicine Discovery Institute and Department of Microbiology, Monash University, Clayton, Australia
    Contribution
    Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Visualization, Writing - original draft, Writing – review and editing
    Contributed equally with
    Christopher J Stubenrauch
    Competing interests
    No competing interests declared
  2. Christopher J Stubenrauch

    Infection & Immunity Program, Biomedicine Discovery Institute and Department of Microbiology, Monash University, Clayton, Australia
    Contribution
    Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Funding acquisition, Software, Supervision, Visualization, Writing - original draft, Writing – review and editing
    Contributed equally with
    Eric Mandela
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-4388-3184
  3. David Ryoo

    Interdisciplinary Bioengineering Graduate Program, Georgia Institute of Technology, Atlanta, United States
    Contribution
    Formal analysis, Investigation, Methodology, Software, Visualization
    Competing interests
    No competing interests declared
  4. Hyea Hwang

    School of Materials Science and Engineering, Georgia Institute of Technology, Atlanta, United States
    Present address
    NIH Center for Macromolecular Modeling and Bioinformatics, Beckman Institute for Advanced Science and Technology, Department of Biochemistry, Center for Biophysics and Quantitative Biology, University of Illinois at Urbana-Champaign, Urbana, United States
    Contribution
    Formal analysis, Investigation, Methodology, Visualization
    Competing interests
    No competing interests declared
  5. Eli J Cohen

    Department of Life Sciences, Imperial College London, London, United Kingdom
    Contribution
    Formal analysis, Investigation, Methodology, Software, Visualization
    Competing interests
    No competing interests declared
  6. Von L Torres

    Infection & Immunity Program, Biomedicine Discovery Institute and Department of Microbiology, Monash University, Clayton, Australia
    Contribution
    Formal analysis, Investigation, Methodology, Visualization
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-3387-1112
  7. Pankaj Deo

    Infection & Immunity Program, Biomedicine Discovery Institute and Department of Microbiology, Monash University, Clayton, Australia
    Contribution
    Formal analysis, Investigation, Methodology
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-6947-5317
  8. Chaille T Webb

    Infection & Immunity Program, Biomedicine Discovery Institute and Department of Microbiology, Monash University, Clayton, Australia
    Contribution
    Data curation, Formal analysis, Software, Supervision
    Competing interests
    No competing interests declared
  9. Cheng Huang

    Monash Proteomics & Metabolomics Facility, Department of Biochemistry and Molecular Biology, Biomedicine Discovery Institute, Monash University, Clayton, Australia
    Contribution
    Formal analysis, Investigation, Methodology, Supervision, Visualization
    Competing interests
    No competing interests declared
  10. Ralf B Schittenhelm

    Monash Proteomics & Metabolomics Facility, Department of Biochemistry and Molecular Biology, Biomedicine Discovery Institute, Monash University, Clayton, Australia
    Contribution
    Data curation, Formal analysis, Investigation, Methodology, Software, Supervision, Visualization
    Competing interests
    No competing interests declared
  11. Morgan Beeby

    Department of Life Sciences, Imperial College London, London, United Kingdom
    Contribution
    Formal analysis, Investigation, Methodology, Software, Visualization, Writing – review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-6413-9835
  12. JC Gumbart

    School of Physics, Georgia Institute of Technology, Atlanta, United States
    Contribution
    Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Software, Visualization, Writing – review and editing
    For correspondence
    gumbart@physics.gatech.edu
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-1510-7842
  13. Trevor Lithgow

    Infection & Immunity Program, Biomedicine Discovery Institute and Department of Microbiology, Monash University, Clayton, Australia
    Contribution
    Conceptualization, Data curation, Visualization, Investigation, Funding acquisition, Project administration, Software, Supervision, Writing - original draft, Writing – review and editing
    For correspondence
    trevor.lithgow@monash.edu
    Competing interests
    No competing interests declared
  14. Iain D Hay

    School of Biological Sciences, The University of Auckland, Auckland, New Zealand
    Contribution
    Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Funding acquisition, Project administration, Software, Software, Supervision, Visualization, Writing - original draft, Writing – review and editing
    For correspondence
    iain.hay@auckland.ac.nz
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-8797-6038

Funding

Australian Research Council (FL130100038)

  • Trevor Lithgow
  • Iain D Hay

National Institutes of Health (R01-GM123169)

  • JC Gumbart

National Institutes of Health (R01-AI052293)

  • JC Gumbart

National Science Foundation

  • JC Gumbart

Marsden Fund (UOA1907)

  • Iain D Hay

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

We are grateful to Rebecca Bamert and Jonathan Wilksch for critical comments on the manuscript. The authors thank Andrea Nans and Peter Rosenthal at the Francis Crick Institute for electron microscopy services, and support staff at the Monash Proteomic & Metabolomic Facility. We acknowledge research support from the Australian Research Council (FL130100038 to TL and IDH). EM was recipient of a Monash Research Scholarship. JCG acknowledges support from the US National Institutes of Health (R01-GM123169 and R01-AI052293). Computational resources were provided through XSEDE (TG-MCB130173), which is supported by the US National Science Foundation (NSF; ACI-1548562). This work also used the Hive cluster, which is supported by the NSF (1828187) and is managed by PACE at the Georgia Institute of Technology.

Senior Editor

  1. Bavesh D Kana, University of the Witwatersrand, South Africa

Reviewing Editor

  1. Petra Anne Levin, Washington University in St. Louis, United States

Reviewer

  1. Daniel O Daley, Stockholm University, Sweden

Publication history

  1. Preprint posted: January 13, 2021 (view preprint)
  2. Received: September 1, 2021
  3. Accepted: January 21, 2022
  4. Accepted Manuscript published: January 27, 2022 (version 1)
  5. Version of Record published: February 8, 2022 (version 2)

Copyright

© 2022, Mandela et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 1,323
    Page views
  • 233
    Downloads
  • 6
    Citations

Article citation count generated by polling the highest count across the following sources: PubMed Central, Crossref, Scopus.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Open citations (links to open the citations from this article in various online reference manager services)

Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)

  1. Eric Mandela
  2. Christopher J Stubenrauch
  3. David Ryoo
  4. Hyea Hwang
  5. Eli J Cohen
  6. Von L Torres
  7. Pankaj Deo
  8. Chaille T Webb
  9. Cheng Huang
  10. Ralf B Schittenhelm
  11. Morgan Beeby
  12. JC Gumbart
  13. Trevor Lithgow
  14. Iain D Hay
(2022)
Adaptation of the periplasm to maintain spatial constraints essential for cell envelope processes and cell viability
eLife 11:e73516.
https://doi.org/10.7554/eLife.73516

Further reading

    1. Cell Biology
    2. Stem Cells and Regenerative Medicine
    Terren K Niethamer, Lillian I Levin ... Edward E Morrisey
    Research Article

    Following acute injury, the capillary vascular bed in the lung must be repaired to reestablish gas exchange with the external environment. Little is known about the transcriptional and signaling factors that drive pulmonary endothelial cell (EC) proliferation and subsequent regeneration of pulmonary capillaries, as well as their response to stress. Here, we show that the transcription factor Atf3 is essential for the regenerative response of the mouse pulmonary endothelium after influenza infection. Atf3 expression defines a subpopulation of capillary ECs enriched in genes involved in endothelial development, differentiation, and migration. During lung alveolar regeneration, this EC population expands and increases the expression of genes involved in angiogenesis, blood vessel development, and cellular response to stress. Importantly, endothelial cell-specific loss of Atf3 results in defective alveolar regeneration, in part through increased apoptosis and decreased proliferation in the endothelium. This leads to the general loss of alveolar endothelium and persistent morphological changes to the alveolar niche, including an emphysema-like phenotype with enlarged alveolar airspaces lined with regions that lack vascular investment. Taken together, these data implicate Atf3 as an essential component of the vascular response to acute lung injury that is required for successful lung alveolar regeneration.

    1. Cell Biology
    2. Microbiology and Infectious Disease
    Emma M Briggs, Catarina A Marques ... Keith R Matthews
    Research Article Updated

    African trypanosomes proliferate as bloodstream forms (BSFs) and procyclic forms in the mammal and tsetse fly midgut, respectively. This allows them to colonise the host environment upon infection and ensure life cycle progression. Yet, understanding of the mechanisms that regulate and drive the cell replication cycle of these forms is limited. Using single-cell transcriptomics on unsynchronised cell populations, we have obtained high resolution cell cycle regulated (CCR) transcriptomes of both procyclic and slender BSF Trypanosoma brucei without prior cell sorting or synchronisation. Additionally, we describe an efficient freeze–thawing protocol that allows single-cell transcriptomic analysis of cryopreserved T. brucei. Computational reconstruction of the cell cycle using periodic pseudotime inference allowed the dynamic expression patterns of cycling genes to be profiled for both life cycle forms. Comparative analyses identify a core cycling transcriptome highly conserved between forms, as well as several genes where transcript levels dynamics are form specific. Comparing transcript expression patterns with protein abundance revealed that the majority of genes with periodic cycling transcript and protein levels exhibit a relative delay between peak transcript and protein expression. This work reveals novel detail of the CCR transcriptomes of both forms, which are available for further interrogation via an interactive webtool.