Renovating a double fence with or without notifying the next door and across the street neighbors: why the biogenic cytoplasmic membrane of Gram-negative bacteria display asymmetry?

Abstract

The complex two-membrane organization of the envelope of Gram-negative bacteria imposes an unique biosynthetic and topological constraints that can affect translocation of lipids and proteins synthesized on the cytoplasm facing leaflet of the cytoplasmic (inner) membrane (IM), across the IM and between the IM and outer membrane (OM). Balanced growth of two membranes and continuous loss of phospholipids in the periplasmic leaflet of the IM as metabolic precursors for envelope components and for translocation to the OM requires a constant supply of phospholipids in the IM cytosolic leaflet. At present we have no explanation as to why the biogenic E. coli IM displays asymmetry. Lipid asymmetry is largely related to highly entropically disfavored, unequal headgroup and acyl group asymmetries which are usually actively maintained by active mechanisms. However, these mechanisms are largely unknown for bacteria. Alternatively, lipid asymmetry in biogenic IM could be metabolically controlled in order to maintain uniform bilayer growth and asymmetric transmembrane arrangement by balancing temporally the net rates of synthesis and flip-flop, inter IM and OM bidirectional flows and bilayer chemical and physical properties as spontaneous response. Does such flippase-less or ‘lipid only”, ‘passive’ mechanism of generation and maintenance of lipid asymmetry exists in the IM? The driving force for IM asymmetry can arise from the packing requirements imposed upon the bilayer system during cell division through disproportional distribution of two negatively curved phospholipids, phosphatidylethanolamine and cardiolipin, with consistent reciprocal tendency to increase and decrease lipid order in each membrane leaflet respectively.

Introduction

The envelope of diderm Gram-negative bacteria comprises two distinct membranes: an inner membrane (IM) and an outer membrane (OM) that are separated by a peptidoglycan-containing periplasm. The E. coli envelope consists predominantly of the zwitterionic phosphatidylethanolamine (PE) (~70%), monoanionic phosphatidylglycerol (PG, ~20%) and dianionic tetraacylated cardiolipin (CL, ~5%) with the remaining phospholipids making up the rest (~5%) [1]. This relatively simple phospholipid composition is achieved primarily by de novo synthesis on the IM of the cell envelope via the main intermediate phosphatidic acid (PA) which is then converted either to zwitterionic PE or anionic PG and CL. Plasma membrane in Gram-positive and Gram-negative bacteria is biogenic (i.e. self-synthesizing and self-assembling). However, there is a major difference in the appearance and transmembrane distribution of newly synthesized phospholipids between these organisms. In monoderm Gram-positive bacteria continuously synthesized phospholipids appear first in the cytoplasmic leaflet followed by distribution to the outer leaflet of the cytoplasmic membrane. However, if this process exists compositional phospholipid asymmetry within a single cytoplasmic membrane cannot be explained only by sidedness of lipid synthesis or breakdown which can create only a transient lipid concentration gradient and possible a downhill flow until equilibrium is reached. However in diderm Gram-negative bacteria phospholipids are synthesized on the cytoplasmic leaflet of IM and concomitantly inserted and translocated across the IM and between the IM and OM to reach their final destination [2–4] and to act as biosynthetic precursors for synthesis and modification of major envelope components: lipoproteins (lpp), lipopolysaccharides (LPS) and an enterobacterial common antigen, ECA_PG [5–7]. A recent study provides the first direct evidence that nascent PE is asymmetrically and dynamically distributed across the IM of Escherichia coli and Yersinia pseudotuberculosis [1]. However, how Gram-negative bacteria maintain a lipid asymmetry (biosynthetically, physically, or enzymatically) in the IM is still unknown. Dedicated IM flippases or scramblases have not be genetically and biochemically identified in E. coli.

Maintenance of lipid asymmetry in the IM by biosynthetic trade-off mechanism

Establishment and maintenance of defined membrane lipid topography within the OM and between the IM and OM should involve the coordinated biosynthesis and translocation of individual phospholipids across the IM and from the IM to OM. Since almost all steps of phospholipid biosynthesis occur on the cytoplasmic side of the IM, catalyzed by well-defined biosynthetic machinery, newly synthesized lipid molecules must then be translocated first across the IM and then delivered to the OM. Given that no specific lipid flippases have been identified to flip the phospholipids across the IM in E. coli, other mechanisms may be employed in diderm bacteria. One such mechanism is a dynamic balance between the incorporation of phospholipids, PE and PG into the cytoplasmic leaflet of IM (due to their continuous synthesis) and their ‘removal’ from the periplasmic leaflet (due to the transport of PE to the OM or the utilization in metabolic pathways involving lipid A and periplasmic lipoproteins anchored to periplasmic leaflet of OM). Membrane growth and continuous emergence and consumption of lipids on periplasmic leaflet of the IM would require a constant ‘refill’ of phospholipids to the IM cytoplasmic leaflet.

Synthesis of PE starts with the formation of phosphatidylserine (PS) by a PS synthase on cytoplasmic surface of the IM, followed by its conversion to PE by the IM residing PS decarboxylase (Psd) (Figure 1). The small head group of PE imparts a cone shape to the molecule, and, in membranes, the acyl chains of PE impart lateral pressure that can be released by formation of negatively curved membrane structures. Therefore, from a structural and topological points of view, PE is expected to be localized on cytoplasmic leaflet of IM. However, from a biosynthetic point of view, a large pool of PE should be always present and consumed on periplasmic leaflet of IM (Figure 1) for continuous flow from the outer leaflet of the IM to the periplasmic leaflet of the OM to build up the periplasmic leaflet of OM during cell growth. This may create a steady state concentration gradient of this most abundant lipid resulting in IM lipid asymmetry. The major OM lipoprotein Lpp and several other lipoproteins require PG for their maturation which occurs on the outer leaflet of the IM by the sequential action of three IM residing enzymes. PG serves particularly as a diacylglycerol donor to the N-terminal cysteine thiol group of lipoproteins in reaction catalyzed by the essential enzyme prolipoprotein diacylglyceryl transferase Lgt with its active site facing periplasm. The cells lacking PG are not viable primarily due to lethal accumulation of Lpp at the IM [8]. On periplasmic leaflet of IM, PG also provides its diacylglycerol-phosphate moiety to ECA_PG which finally ends up at cell surface tethered to outer leaflet of OM via phosphoglyceride [7]. On the periplasmic leaflet of the IM, Lnt protein transfers the palmitate moiety normally from sn-1 of PE to the N terminus of the diacylglycerol-linked Lpp precursor resulting in a triacylated mature Lpp however PG can also serve as the donor of fatty acid in the N-acylation of apolipoprotein [9].

Figure 1. Biogenic nature of lipid asymmetry in double membraned Gram-negative bacteria.

A steady state concentration gradient of the lipids across IM resulting in IM lipid asymmetry can be maintained metabolically during their utilization as metabolic precursors at periplasmic leaflet, continuous flow from the outer leaflet of the IM to the periplasmic leaflet of the OM and balanced with retrograde transport of phospholipids from the OM to the IM during cell growth. Figure is adapted from Bogdanov et al. [1] with permission from American Association for the Advancement of Science (AAAS) for manuscripts published under CC BY-NC 4.0 license.

In antibiotic stressed E. coli cells and pathogenic bacteria, EptA/PmrC phosphoethanolamine transferase catalyzes continuous modification of newly synthesized lipid A core by adding a phosphoethanolamine moiety to the 1-phosphate group of the heptose residue in lipid A on the periplasmic surface of the IM [10,11]. In E. coli outgrown in hypoosmotic media, OpgB (MdoB) with its active site on the periplasmic leaflet of the IM catalyzes the transfer of sn-1-phosphoglycerol from PG to anionic osmoregulated periplasmic glucans (formerly known as membrane derived oligosaccharides [12]. Thus the transmembrane asymmetry of PE and PG could also be maintained by a metabolic mechanism, which relies on delicate balance between anabolic and catabolic reactions on both sides of the IM.

The asymmetrical distribution of PE and PG in the IM can be assumed to be an unstable state which can potentially dissipate to a symmetric stable state by constant, near equal ‘refill’ and ‘removal’ of these lipids on the cytoplasmic and periplasmic leaflets of the IM, respectively. To avoid a mass imbalance between the two IM leaflets, rapid flip-flop of lipids across IM is continuously required during growth and division of cells. The asymmetric distribution of PE and PG in IM can result from the balance between biosynthetic buildup on the cytoplasmic leaflet of the IM, from their translocation to the periplasmic leaflet, and from loss on the periplasmic leaflet due to biosynthetic utilization and/or flow to the OM. Unless the rate of translocation across the IM and removal and/or consumption as precursors on the periplasmic leaflet is precisely more than half that of their synthesis rate, an asymmetric IM would always result.

Nascent PE and CL are dynamically redistributed across IM during membrane growth to follow or direct the changes in bacterial shape

Prevailing assumption is that the IM in Gram-negative bacteria contain phospholipids mixed randomly between both leaflets [13]. Although phospholipids are found in both the IM and the OM, the OM is extremely asymmetric, with LPS residing in the outer leaflet and phospholipids (normally PE) in the inner leaflet (Figure 1). The IM of Gram-negative bacteria is composed of glycerol-phosphate based phospholipids and was suspected to be asymmetric too; however, only limited and controversial data were available [14]. The scarcity and uncertainty of information regarding PE distribution in IM of Gram-negative bacteria was due, at least in part, to the complexity of envelope and to the lack of suitable and established methods to validate the sidedness of membrane vesicles and vectorial probes to analyze the distribution and dynamics of this key phospholipid in intact, spheroplasted and fractionated cells. When 2,4,6-trinitrobenzenesulphonic acid (TNBS) was used to label externally localized PE, up to 80–87% of PE in osmotically stable Erwinia carotovora spheroplasts were remained unlabeled, i.e. was therefore facing the cytoplasm. However in spheroplasts-derived IM vesicles with total 65% of PE, only 27% of the PE was found on external surface and 38% on the internal surface PE of these vesicles with unknown sidedness [14].

Sequential utilization of two primary amine-specific vectorial non-penetrating (i.e. outer leaflet specific) TNBS and penetrating (1,5-difluoro-2,4-dinitobenzene (DFDNB)) probes led to the development of a novel approach for determining head-amino and acyl-group asymmetries in the IM of Gram-negative bacteria using either radiometric, spectrophotometric or mass-spectrometric methods. This was the first unambiguous demonstration of PE asymmetry in the IM of Gram-negative bacteria which has long remained unknown and controversial [1]. Using of fully devoid of OM, uniformly oriented inside-out (ISO) membrane vesicles sequentially treated with TNBS and DFDNB allowed us to confidently assess PE asymmetry in the IM of Escherichia coli and Yersinia pseudotuberculosis. Pulse and pulse chase radiolabeling experiments with the strains in which PE can be either eliminated, tightly titrated or controlled temporally allowed us to map the transmembrane distribution of de novo synthesized PE in the IM of initially filamentous E. coli cells while progressively reducing their size upon accumulation of newly synthesized PE or lock PE fluxes in short rod-sized (sausage-shaped) and long filamentous Escherichia coli induced by specific inhibition of cell division by aztreonam, an FtsI (penicillin-binding protein 3)-selective β-lactam antibiotic, which induces cell filamentation and blocks cell division, but still allows cell growth [15]. Unexpectedly but remarkably, the IM of Escherichia coli was found to be highly asymmetric with 75%/25% (cytoplasmic/periplasmic leaflet) distribution of PE in rod-shaped cells and mirrored PE (25%/75% (cytoplasmic/periplasmic leaflet) distribution in filamentous cells suggesting that PE is dynamically redistributed within the IM either reflecting (most likely) or facilitating a change in cell morphology. In accordance with this established steady-state distribution of PE in the IM, an asymmetrical distribution in growing cells would result whenever the net rate of outward movement of new lipid to the outer leaflet of IM is equal to exactly two-thirds of the rate of biosynthesis or vice versa. However surprisingly and counterintuitively, nascent PE first appears on the periplasmic side of the IM in wild-type rod-shaped cells followed by dynamic and disproportional distribution to the cytoplasmic leaflet [1] resulting in final highly asymmetric distribution of PE with approximately one-fourth in the periplasmic leaflet of the IM. This very intriguing finding suggests that presence of initial ample amounts of periplasmic CL may somehow promote retro-translocation and predominant localization of PE in the cytoplasmic leaflet of IM as will be discussed below. It is worthwhile to speculate at this point that critical threshold amounts exist which may be compatible or incompatible for the presence of PE and CL, a two negatively curved phospholipids, in each IM leaflet.

Aztreonam induced filamentous cells contain assembled septal rings which are however unable to constrict due to the inactivation of FtsI which still localizes to the septal ring and subsequently recruits FtsN [15]. PE-lacking cells are filamentous due to loss of proper interaction of FtsZ with membrane nucleation sites resulting in aberrant spiral FtsZ structure [16]. To differentiate between cell morphology and PE levels as a determinant of PE localization and determine the sidedness of PE during its synthesis starting from cells beginning with near-zero amounts to wild-type levels, the strain AT2033 was induced for de novo PE biosynthesis after growth under conditions in which PE was initially absent and cells were filamentous. In this strain, the PE content of the cell can be regulated from near-zero to wild-type level (75%) proportional to the amount of gene inducer anhydrotetracycline (aTC) in the growth medium. Interestingly, these gradual reciprocal changes in distribution of PE/CL amounts between the two IM leaflets during de novo PE biosynthesis coincide with a progressive reduction in cell size (Figure 2B,C). The cytoplasmic leaflet of IM is highly compressed during bacterial cell growth [17]. Theoretically, the PE asymmetry may be not only arisen within the IM but also maintained by its movement from the periplasmic leaflet of the IM to periplasmic leaflet of OM in process which would relieve the lateral pressure within the periplasmic leaflet of IM after predominant buildup of PE on concave cytoplasmic leaflet. Therefore, percentage and localization of PE and CL in each leaflet of IM of cells could be simply adjusted to relax frustrated bilayer and maintain desired spontaneous curvature during uniform bilayer growth.

Figure 2. Transmembrane distribution of PE and CL in rod-shaped and filamentous cells is dynamically and inversely correlated.

(A) Biogenic nature of lipid asymmetry in double membraned Gram-negative bacteria. Although compositional phospholipid asymmetry within the IM can be explained by imbalance of sidedness of PE synthesis, it can rely also on transport of specific lipids between the IM and OM. (B) Transbilayer distribution of PE in the IM as a function of phospholipid composition and cell shape in ISO vesicles prepared from initially filamentous PE-deficient E. coli strain induced for PE biosynthesis. In ISO vesicles, the external and luminal surfaces correspond topologically to the cytoplasmic and periplasmic leaflets of the IM of whole cells, respectively. As the mole fraction of PE increases there is a continuous decrease in disproportionation of PE toward its presence in outer leaflet resulting in an asymmetric redistribution of this lipid from the outer (periplasmic) to inner (cytoplasmic) leaflet of the IM. PE is retrotranslocated, progressively accumulated and exposed to cytoplasm but CL is consistently dissipated from the cytoplasmic leaflet of the IM during de novo PE biosynthesis in cells initially filamentous and lacking PE in concert with progressive reduction in cell size and change in lipid packing order. Cell aliquots were collected at the indicated times after addition of inducer for PE synthesis and converted to ISO vesicles. TLC with derivatized phospholipids from ISO vesicles treated sequentially with non-permeant (2,4,6-trinitrobenzenesulfonic acid (TNBS) followed by fully permeant amino-specific (1,5-difluoro-2,4-dinitobenzene (DFDNB) reagent is shown. Sidedness and dynamic distribution of PE was estimated from the amount of radiolabeled derivatized products of PE: trinitrophenyl-PE (TNP-PE) and dinitrophenyl-PE (DNP-PE) corresponding respectively to amount of PE in outer (cytoplasmic in cells) and luminal (periplasmic in cells) leaflets of uniformly oriented ISO vesicles isolated from ³²P-pulse-labeled cells and sequentially labeled by TNBS (first) and followed by DFDNB (C) Diagrams representing progressive enrichment of outer leaflet of ISO vesicles corresponding topologically to the cytoplasmic side of the IM of cells progressively adopting a rod shape are shown. Microscope images of representative cells outgrown in the presence of inducer anhydrotetracycline (aTC)for 30, 90 and 180 min. Therefore dynamic distribution, percentage and localization of PE and CL in both leaflets of IM should be adjusted to satisfy packing requirements, curvature continuity and maintain a stability of bilayer and desired morphology of E. coli cells. Figure is adapted from Bogdanov et al. [1] with permission from American Association for the Advancement of Science (AAAS) for manuscripts published under CC BY-NC 4.0 license.

Omnis membrana e membrana’: an extention of Virchow’s paradigm to the origin of lipid and protein asymmetry in the biogenic IM

Asymmetric biological membranes can be simply envisioned as a lipid bilayer of asymmetrically distributed different lipids into which membrane proteins are asymmetrically embedded. Why is the biogenic IM asymmetric? Is it because when the membrane proteins synthesized de novo they are inserted already into a preexisting asymmetric membrane and net lipid asymmetry is required to balance the asymmetrically distributed charges of membrane proteins and orient them in accordance with Positive Inside Rule? Or vice versa? In contrast with lipid asymmetries which are assumed to required active maintenance (see previous and next sections), it is widely accepted that the asymmetry of integral membrane proteins can be initiated during their insertion and does not have to be actively maintained due to the enormous energy penalty required to reorient soluble extramembrane domains across the hydrophobic barrier of the membrane [18]. Membrane proteins that are able to adopt co and post-insertionally dual and dynamic topologies are an exception to this and, therefore, challenge this simplistic view [19–21]. Theoretically, specific lipids can be highjacked during insertion and permanently immobilized at high affinity binding sites located predominantly on one side of the inserting protein. If high number of membrane proteins expose their specific lipid-binding sites to either the cis or trans sides of a biogenic membrane, a net asymmetry of lipids between the leaflets of the IM will arise. With the advent of membrane protein crystal structures, specific phospholipids were indeed found at bilayer-embedded halves of the proteins corresponding to just one leaflet of the IM [22,23]. Supporting this idea recent structural bioinformatic analyses reveal a striking pattern of lipid asymmetry between E. coli IM bilayer-embedded halves of the proteins with known highly resolved structures [24]. Not surprisingly, among these lipids is CL, which has unique chemical and structural characteristics including two phosphates and four fatty acid chains and has been implicated in the functions of a number of IM proteins. Remarkably CL binding motifs were found much more readily and consistently in cytoplasmic leaflet region of these structurally unrelated proteins which engage in strong binding with this lipid. Driven in accordance with Positive Inside Rule by an increased number of positively charged residues on the cytoplasmic face of the proteins such CL binding motifs might impact the steady-state ratios of lipids in each leaflet of the IM and contribute to maintenance of their net asymmetry between the leaflets of the IM. Although the inability to completely covalently modify 10% of PE with membrane-impermeant TNBS in solubilized membranes [1] can be explained by clustering of this very abundant lipid around residing proteins, we don’t know how much of ‘free’ and immobilized CL exist in either leaflet of the IM.

Because of the disturbance of the IM bilayer during translocon-assisted or unassisted insertion of membrane proteins one could postulate the ability of intercalating nascent membrane protein domain to disrupt or perturb the lateral and transmembrane continuity of the bilayer structure and therefore reduce the activation energy barrier for transmembrane movement of phospholipids. Such local ‘weakening’ of bilayer rigidity at specific interfacial lipid ‘flip’ or ‘flop’ sites becomes plausible in this scramblase less ‘slip-pop’ mechanism [25]. In accordance with this mechanism phospholipids could occasionally ‘slip through’ the bilayer affecting its preexisting tendency to maintain the asymmetrical state. Such ‘slipping through’ movement is not specific for the phospholipid headgroup, is bidirectional and energy independent, and is solely determined by overall membrane protein structural fold and number of membrane-spanning segments forming the non-specific lipid flip-flop site. While the precise mechanism of this accelerated flip-flop is still unknown, nevertheless α-helical stretches (not β-strands) of subset of proteins may facilitate transiently lipid translocation and scrambling rendering dedicated flippases and scramblases redundant in biogenic membranes.

Scope for glycerol-based phospholipid flippase function in Gram-negative bacteria

Because flip-flop of amphipathic phospholipids in protein-free membranes is thermodynamically unfavorable rapid phospholipd translocation in biological membranes is protein catalyzed. Specific flippases mediate the net transfer of specific phospholipids from one leaflet of a membrane to the other often against a gradient of concentration. Although many flippases that catalyze translocation of different classes of lipids have been identified in eukaryotes [26], so far, no specific glycerophospholipid flippases have been found in the IM of E. coli. MsbA was originally identified as a multicopy suppressor of a thermosensitive htrB(lpxL) mutants of E. coli that accumulated tetraacylated lipid A species and phospholipids in the IM at non-permissive temperatures [27]. Subsequent studies characterized MsbA as lipid A floppase with poor ability to transport of underacylated substrates and highly selective for hexaacylated lipid A [28]. MsbA can poorly flip underacylated lipid A due to mallosteric uncoupling and inefficient stimulation of ATPase activity which is stimulated by binding of mature hexa-acylated lipid A [29]. Therefore, increased levels of MsbA on a multicopy plasmid can directly rescue growth by transporting still reasonable amounts of lipid A molecules for which MsbA has lower affinity and relieving toxic side-effects of underacylated lipid A accumulation in the IM. Alternatively, the ABC multidrug exporter MsbA could restore the balance between the biosynthesis of phospholipids and growth rate, enabling htrB (lpxL) mutants sustain viability and grow at high temperatures, in addition to its role in LPS transport. This conclusion was further validated in vivo by demonstrating that msbA conditional mutants accumulate not only LPS but also phospholipids on the inner leaflet of the IM upon shift to non-permissive temperature, as judged by the accessibility to enzymatic modification and labeling with membrane-impermeable reagents [30]. Originally, it was shown that MsbA is not capable of promoting translocation of phospholipids in either ATP-dependent or -independent manner in vitro [25], suggesting that either MsbA alone is necessary but not sufficient for flip-flop of phospholipids in vitro (e.g. unknown assistant proteins in addition to MsbA are required for phospholipid translocation across the IM in vivo) [31] or MsbA is involved only in LPS lipid A transport and not required at all for phospholipid flip-flop across the IM [32]. Despite this existing controversy the first in vitro evidence for phospholipid floppase activity of MsbA was provided recently [33]. It has been demonstrated that translocation of biotinylated analog of PE into lumen of proteoliposomes requires simultaneously the ATP binding and hydrolysis and presence of the proton gradient. However biotinylated Lipid A transport was maintained in the proteoliposomes with ATP hydrolysis only and not driven by artificially imposed ΔpH. It is still difficult to extrapolate these in vitro results to the in vivo situation since fluorescent phospholipid or biotinylated analogs are structurally different from naturally occurring counterparts and appears to be not a reliable mimic of the native phospholipid [34].

Other Gram-negative bacteria flippases were only postulated based on partial purification and reconstituted activity of the IM total detergent extract measured by the fluorescence stopped flow technique and bovine serum albumin back extraction [35]. Although several flippases that bind and move non-glycerol phosphate bacterial lipids across bacterial IM [36] including the essential protein MurJ acting as flippase for the bactoprenol based lipid-linked peptidoglycan precursor Lipid II were identified, LplT appears to be the first example of a bacterial protein capable of facilitating the rapid retrograde translocation of a monoacylated glycerophospholipids across the cytoplasmic membrane of Gram-negative bacteria [37,38]. Many membrane-stressing events that compromise integrity of OM are accompanied by drastic increase in lysoPE level within Gram-negative envelope. In stressed cells lysoPEs are released as (i) products of PE hydrolysis by detergent-resistant phospholipase A1 encoded by pldA gene (ii) by-products of PagP mediated transfer of palmitoyl groups from PE to glucosamine saccharides of the lipid A and head group of PG as acceptor substrates within the OM. Lnt transfers the fatty acid moiety from the sn-1 position of PE to the N terminus of the Lpp generating mature triacylated lipoprotein and releasing of lysoPE as by-product of reaction. LysoPE can be released into envelope as result of action of exogenous sPLA₂-IIA from host or PLA₂ delivered from invading bacteria via type IV secretion system. LysoPG can be released as result of action of Lgt with active site facing periplasm as described above. Please see [39] for all origins of lysophospholipids in envelope of Gram-negative bacteria. Transiently residing OM lysophospholipids are translocated back to the IM where they can be taken up by LplT, a translocase component of retrograde transport system that recycles lysophospholipids back to the cytoplasmic surface of the IM where they can be reacylated by acyl-glycerophosphoethanolamine (Aas) acyltranferase. LplT/Aas system acts by a uniport mechanism as flippase since its function does not require ATP or the proton motive force [37]. LplT has broad substrate specificity and well suited to rescue PE, PG and CL from their lysoform with comparable kinetics and binding which can be limited to their sn-1 lyso configurations in accordance with [38] or, in contrast, to their sn-2 lyso configurations [37]. The conflicting results could be due to utilization of label-free natural 32P-labeled lysophospholipids [38] versus NBD-labeled fluorescent analog 2-dodeceyl-NBD-GPE in LplT mediated translocation assays [37]. Although second results are in the agreement with the fact that presence and accumulation of lysoPE in stressed envelope are mostly due to activation of detergent-resistant phospholipase A1 encoded by pldA gene on the OM resulting in the formation of free fatty acid and sn-2 acyl lysophospholipid, the fluorescently labeled head group and fatty acid labeled NBD-labeled analogues have the potential to perturb the measured dynamics and energetics of translocation [34]. It is also possible that during preparation of translocation substrates by the hydrolysis of the corresponding diacylated phospholipids with lipase, sn-2 acyl lysophosphospholipids were isomerized to form more stable sn-1 acyl isomer due to intramolecular acyl migration. Once on cytoplasmic side of the IM, the diacyl form of phosphilipid molecule is successfully regenerated by the action of the coupled LplT/acyl-glycerophosphoethanolamine acyltransferase (Aas) recycling system which provides a unique bacterial membrane phospholipid repair mechanism similar to the Lands cycle in eukaryotes.

Although CL trans IM translocator activity has never been reconstituted, the globular periplasmic domain of YejM/PbgA was suggested to be required for regulated increases in the OM CL [40] and contains a conserved and hydrophobic amino acid residues, which were implicated to direct and specific binding and shielding of CL acyls chains during translocation through periplasm [41]. However recently YejM was implicated also in LPS homeostasis independently of CL transport. It was demonstrated simultaneously by several labs that YejM actually functions to adjust OM levels of LPS by preventing degradation of the LPS key biosynthetic enzyme LpxC which catalyzes the first committed step in lipid A biosynthesis [42–44].

These results suggest that the changes in CL levels observed in yejM domain deletions may be secondary to YejM’s regulatory role in LPS biosynthesis.

Two recent cryo-EM structures of the retrograde phospholipid transport system, MlaFEDB at 3.05 Å and 3.3–4.1 Å resolution, revealed electron densities assigned to either one or two diacyl PE molecules trapped in the outward-open cavity formed by MlaE and MlaD [45,46] in the the channel of MlaD, and in the periplasm, suggesting that the Mla system may be capable of transporting one or two diacylated or one four acylated (CL) lipid substrates per each cycle. Although MlaFEDB is involved in phospholipid transport to the IM from the outer leaflet of the OM, purified and reconstituted MlaFEDB is capable of translocation of fluorescent analogs of PE, PG and CL from the outer to the inner leaflet of the IM [47]. This may suggest that the final destination of phospholipids removed from the OM is the inner leaflet of the IM. Alternatively, the OM phospholipids may be inserted into the outer leaflet of the IM and the translocation activity allows balancing of the leaflets by moving half as many phospholipids to the inner leaflet, resulting in a net gain of one phospholipid molecule in each leaflet of the IM. The ability of MlaFEDB to facilitate the retrograde transbilayer movement of non-natural fluorescent analoques of PE, PG and CL across IM is consistent with action of this complex as generic lipid flippase in coordination with its retrograde transport activity.

Why 3? The physiological significance of cls genes multiplicity: making CL at desired bilayer leaflet instead translocating?

E. coli has three genes (clsA (cls), clsB (ybhO) and clsC (ymdC)) that encode CL synthases; however the reason for this multiplicity is still unclear as is the role that each Cls plays in establishing E. coli CL net content and CL and PE distributions across IM. ClsABC catalyzed reactions which are reversible in vivo and in vitro utilizing either two PGs (ClsA), two PGs and/or PE and glycerol (ClsB) [48] and making CL or PG respectively or utilizing PE and PG and making CL (ClsC) [49]. Principal ClsA is present under all growth conditions, while ClsB and ClsC are up-regulated during early stationary phase of growth [49]. Please note that ybhO encoded ClsB catalyzes PGP-independent PG synthesis via conversion of PE and glycerol into PG. Full ClsC activity requires co-transcription from a polycistronic operon containing ymdB (clsC2). How YmdB affects ClsC1 activity is not clear [49]. Based solely on structural bioinformatics ClsB and ClsC contain no predicted transmembrane helices, however both activities were found in IM fraction, suggesting they are membrane-associated proteins [49] (Figure 3).

Figure 3. ‘Slip-pop’ and ‘pop and lock’ mechanism for origin and maintenance of lipid asymmetry in bacterial biogenic IM.

Lipid asymmetry in biogenic IM is maintained by flippase-less mechanism and dynamically regulated by biosynthetic, topological and physical constraints. The driving force for IM lipid asymmetry can arise from the packing requirements imposed upon bilayer system by the opposing forces of two negatively curved phospholipids in both leaflets. The directions of arrows are consistent with tight or loose intermolecular lipid packing within IM leaflet. CL dependent leaflet-specific changes in IM packing order can drive asymmetry of other lipids (PE). The neighboring CL molecules are spaced far enough apart to generate loose packing and discontinuity of the monolayer structure can simply increase the probability of phospholipid molecule other than CL (e.g. PE) to flip spontaneously from one leaflet to the other leaflet by retentive thermodynamic ‘slip-pop’ mechanism and stably lock on other side due to interaction of PG and PE headgroups via intermolecular H₂N–H^···O–PO H-bond formation (by ‘pop and lock’ mechanism). IM asymmetry can be also elegantly achieved and controlled by action of Cls enzymes controlling indirectly PE asymmetry and the distribution and redistribution of CL or its molecular forms between IM leaflets. It is also possible that it is advantageous to the cell to use a CL as a less abundant membrane constituent for adjustment of bilayer physical properties as it can be more easily regulated. Therefore dynamic lipid (re)distribution can be naturally balanced not only metabolically but also by chemical and thermodynamic potentials of individual lipids and therefore ‘passively’ instead of actively (flippases and floppases) maintained.

There is some current controversy surrounding the catalytic site orientation of ClsA, which resides in a large C-terminal globular extramembrane domain. Evidence for a periplasmic orientation of the ClsA active site came from experiments showing that feeding E. coli cells with mannitol resulted in formation of phosphatidylmannitol (PM) and diphosphatidylmannitol (DPM) [50]. Mannitol is taken up by the MtlA transport system and is immediately converted in a coupled reaction to cytoplasmic mannitol-P, which is not a substrate for ClsABC. When cells are grown with 600 mM mannitol, PM and DPM are synthesized in a ClsA-dependent manner suggesting that the active site of ClsA (ClsBC were unknown at the time) faces the periplasm. However recent topological mapping of native cysteines with membrane-impermeant fluorophore Oregon Green Maleimide and subsequent analysis of immunoprecipitated ClsA-His10 suggested that the catalytic domain faces the cytoplasm [51] (Figure 3).

It is generally accepted that a thermodynamic barrier to lipid flip-flop is related to the energetic penalty of moving the hydrophilic headgroup of a phospholipid through the hydrophobic core of the membrane. Theoretically due to thermodynamic unfavourability of large hydrophobic domain and net charge of dianionic headgroup CL may require a flippase to undergo transbilayer movement like that of bulky tetraacyl KDO₂-Lipid A (LPS), which requires the MsbA flippase. If active sites of ClsA are oriented in periplasm while ClsBC are up-regulated to make CL on cytoplasmic leaflet, these enzymes can make CL on either side of IM. Since ClsB and ClsC synthesize CL, presumably located in the cytoplasmic leaflet of the IM, CL may not spontaneously move to the periplasmic leaflet.

At this point, it is worthwhile to suggest that ClsB and ClsC can be turned off to support an established asymmetric distribution of PE (75%/25%, cytoplasmic/periplasmic leaflet) and prevent a symmetric distribution of CL in IM of logarithmically grown rod-sized cells. However cells can continuously utilize PE in reactions with up-regulated ClsB and ClsC in the cytoplasmic leaflet of filamentous cells [1] to support a mirrored PE distribution (25_%/75%, cytoplasmic/periplasmic leaflet) by ‘making more room’ for curved CL by converting cytoplasmic leaflet PE into PG and then to CL during filamentous growth.

Do cells balance distribution of lipids across the IM to satisfy an intermolecular packing of lipids and physical order in each membrane leaflet?

The outer leaflet of ISO vesicles derived from the IM of E. coli corresponds topologically to the cytoplasmic side of the IM. An advanced NR12S fluorescent probe whose fluorescence emission is highly sensitive to the lipid packing order of the outer leaflet of membrane was utilized to know whether transmembrane redistributions of CL and lipid packing order in each leaflet of ISO IM membrane vesicles and mimetic liposomes are interrelated. It was demonstrated that the increase in packing order driven by PE is countered by the increase in disorder driven by CL in ISO vesicles and ISO-mimetic liposomes [1] ruling out the contribution of the bulk protein component to this phenomenon. Independent study of artificial membranes free of proteins demonstrated that the fluidity of the lipid bilayeris increased and its mechanical stability decreased with increase in CL concentration, indicating that CL decreases the packing of the membrane facilitating bilayer structure deformation and remodeling [52]. The reciprocal tendency of PE to increase lipid order and the tendency of CL to decrease packing order may be correlated with tight or looser (less ordered) intermolecular packing of lipids.

Late acyltransferase enzymes LpxL and LpxM sequentially add the fifth (laurate) and sixth fatty (myristate) acids, respectively. LpxM does not function efficiently without the laurate chain in its substrate. It was demonstrated recently that lpxM mutants exhibited decreased transport of LPS and that CL levels mitigate somehow underacylated (penta-acylated) LPS transport across the IM [29]. ClsA appears to be essential in the absence of LpxL [53] and both double clsA lpxL and clsAlpxM mutants [29,53] became conditionally lethal. This conditional lethality can be rescued by suppressor mutations mapped to the msbA gene [29,53] suggesting that flopping of poorly translocated underacylated lipid A molecules stacking in the IM requires CL made particularly by ClsA. Thus ClsA may set a threshold CL level for ‘flippability’ of matured and underacylated lipid A molecules. ClsA and MsbA may act together to balance temporally the net rates of synthesis and translocation of LPS across IM and to OM.

clsAlpxM mutants also became very filamentous [29]. Tetraacylated lipid A molecules are symmetrical and cylindrical while hexacylated molecules are asymmetrical. Due to these differences in 3D geometrical shape hexa- and tetra forms of lipid A tend to adopt non-bilayer inverted type II hexagonal structures and lamellar (bilayer) supramolecular structures respectively [54]. Bilayer prone lipids can stabilized non-bilayer lipids in overall bilayer organization in mixed system like the IM. Additionally, ClsA may play an important role in cell morphogenesis by managing delicate packing balance of two tetraacylated lipids with glucosamine phosphate (lipid A) and glycerophosphate (CL) derived backbones in IM. Conformational duality of underacylated and mature lipid A molecules and their dual balance with non-bilayer prone CL molecules can be considered as one of critical operating packing parameters in biogenesis of cell envelope and cell morphogenesis. To satisfy envelope growth capacity balance of these biogenic events in both leaflets of IM is required in order to keep an IM flat or concave and control rod or filamentous growth especially when Lnt-PE-mediated acylation of Lpp should be continuously maintained due to high demand of peptidoglycan growth as such as continuous growth of OM during filamentation process. Most if not all PE in periplasmic leaflet is required in this case to maintain these two processes during filamentous growth and periplasmic PE cannot contribute to generation of negative curvature in periplasmic leaflet to compensate/counter the generation of negative curvature in cytoplasmic leaflet by action of PE and CL in cytoplasmic leaflet of IM.

Flipping the dogma in bacterial biogenic IM: the maintenance of lipid asymmetry by flippase-less mechanism

Thus, the compositional and physical asymmetries of the bilayer are inherently related and the number of lipids and lipid composition are not conserved properties of each monolayer because two non-bilayer prone lipids PE and CL are distributed dynamically between two leaflets of IM. These results support the existence of tight regulation of PE transmembrane distribution and CL content to maintain a bilayer packing order and propensity toward asymmetric distribution of PE in the IM of E. coli. Although PE is asymmetrically distributed across IM no difference in its acyl group composition in either leaflet of the IM was found [1]. We cannot exclude the possibility that changes in leaflet lipid order caused by leaflet-specific location of CL could be the result of the combined properties of the acyl chain and polar head group composition since the preference for saturated acyl chain for CL synthesized by ClsA which is opposite to the enzymes involved in PE and PG biosynthesis was recently documented [55].

Interestingly, the kinetics and thermodynamics of flip-flop in bilayers are correlated to changes in the lipid packing density as the concentration of one lipid in binary system was changed. An order-of-magnitude difference in the rate was found for net neutral native lipids in Langmuir-Shafer asymmetric bilayer membranes prepared at 28 mN/m and 42 mN/m surface pressure as measured by label-free sum-frequency vibrational spectroscopy employing deuterated lipid species [56]. The measured half-lives(t_1/2) of net neutral lipid flip-flop at 37°C at two extremal surface pressure examined, 28 mN/m and 42 mN/m, were found to be 5.3 min and 56 min, respectively. The order-of-magnitude change in rate results from relatively small (<10%) change in the packing density of the lipids in the bilayer, however, leading to significant increase in the area of activation for native phospholipid flip-flop. Thus loose packing and discontinuity of the monolayer structure can simply increase the probability of phospholipid molecule other than non-flipping CL (e.g. PE) to flip spontaneously from one leaflet to the other leaflet by ‘slip-pop’ mechanism (Figure 3). The measured half-lives (t_1/2) of spontaneous PE lipid flip-flop were found accordingly to be at least 10-times smaller than those of a homologous fluorescent PG derivative [57] in agreement with decrease in the activation energy barrier (corresponding to increase in rate). Therefore, CL-dependent, leaflet-specific changes in IM packing order can drive asymmetry of other lipids (PE). Flipped or flopped PE may reside stably on one side of IM due to interaction of PG and PE headgroups via intermolecular H₂N–H^···O–PO H-bonding hydrogen-bond formation with PE being a main H-bond donor [58]. The privileged association of PE with PG presumably may act as a thermodynamic trap that prevents return/translocation of PE back to opposite leaflet of bilayer.

Asymmetry can be triggered and harnessed in binary liposomal system strongly permitting H-bonding interactions between phosphate and amine groups of adjacent phospholipids [59]. Intriguingly CL dependent leaflet-specific changes in IM packing order can drive asymmetry of other lipids (PE) by actions of different Cls on either side of membrane (Figure 3). Asymmetric distribution of lipids in biogenic IM can be generated and maintained in the absence of flippase and floppases due to property of IM to synthesize on desired leaflet a lipid lacking ability for spontaneous transbilayer movement (newly synthesized or preexisting CL) and lipid which is able to flip spontaneously (e.g. PE and PG) due to retentive thermodynamic mechanisms.

Lipid asymmetry can be released ether globally and uniformly or locally and irregularly.

The E. coli IM, where the concentration of CL made by ClsA in logarithmically grown cells is relatively low, may exhibit local regions of compositional and physical transbilayer asymmetries. Considering an existence of dual balance between two non-bilayer forming phospholipids CL and PE two important notes should be taken into account. The IM PE content is roughly 50% of the total (75%) given the high PE content of the cytoplasmic leaflet of the OM bilayer (Figure 1). Since only lipids with unsaturated fatty acid chains and therefore with relatively small size of headgroup tend to form a non-bilayer, only symmetric unsaturated PE molecules (representing one third of total E. coli PE pool) tend to adopt non-lamellar structure in the IM. Distribution of CL synthesized by ClsB and ClsC in stationery grown cells is unknown. Predominant localization of CL made by ClsA at E. coli cell poles or division sites is documented [51] and can be due to spontaneous tendency of highly intrinsically curved CL to form homogeneous clusters in the opposing leaflets of IM: at the bacterial poles (inner leaflet) or at division sites (outer leaflet) [60].

Conclusions and perspectives

Different phospholipids synthesized on cytoplasmic leaflet of IM are consumed on the outer (periplasmic) surface of the IM as precursors in LPS or lpp biogenesis and translocated or retrotranslocated across the IM [1] and between the IM and OM to maintain an asymmetries of IM [1] and OM [2–4], respectively. Obviously, an establishment and maintenance of defined membrane lipid asymmetry within the IM and OM should involve the balanced and coordinated biosynthesis and translocation of individual phospholipids across the IM and between the IM and OM. This balance is likely to be different in short rod-sized and long filamentous cells because of different relative areas of the IM and OM and different dynamics of membrane growth. This may explain the opposite asymmetric distribution of PE in their IMs. ‘Passive’ asymmetric distribution of lipids in the biogenic IM may be maintained in the absence of flippase and floppases by (i) different rate of biosynthesis and turnover of individual lipids; (ii) consumption of lipids as precursor on periplasmic leaflet of the IM; (iii) by lipid inter-membrane flows within the envelope; (iv) the orientation of active sites of multiple lipid synthesizing, degrading and remodeling enzymes; (v) retentive thermodynamic mechanisms that trap one non-bilayer prone lipid (CL) in one leaflet of the bilayer and allow another non-bilayer lipid (PE) to ‘flip’ across the bilayer; (vi) sequestration and immobilization of anionic or zwitterionic lipids at high affinity binding sites on asymmetric proteins; and/or (vi) dynamic relationships between IM compositional and physical asymmetries. This asymmetrical status quo can be affected by insertion of newly synthesized protein promoting lipid translocation and other factors intimately regulated at different cellular circumstances (cell division, starvation-induced dormancy, adaptation to antibiotics, different abiotic and biotic stresses, etc.).

Still much remains to be discovered and several controversies remain to be resolved. If the asymmetric IM is dispensable for cell viability it may be difficult or even impossible to identify genetically flippase candidates by searching for synthetic lethal interactions and loss of function mutations unless new cellular circumstances in stressed cells will be uncovered. However, the biochemical reconstitution and search for in vivo conditions which can modulate an IM asymmetry represent a very reasonable starting point. Although PE is asymmetrically distributed across the IM, currently it is not known whether PG or CL are evenly or asymmetrically distributed across and along IM. Whether individual CL acyl group asymmetry exists in IM is still unknown as well. Thus new approaches for mapping anionic lipid bilayer asymmetry are required in order to develop a complete picture of the molecular requirements for lipid translocation across the IM. Any difference in CL transmembrane distribution maybe related to differential roles of the ClsABC in generation and maintenance of lipid asymmetry of IM. The results presented in this review raise additional questions to be addressed in the future. How are the compositional (head and acyl groups), molecular (combination of acyls with different length and unsaturation within the same lipid class of molecules) and physical (lipid order and packing) lipid asymmetries regulated enzymatically in biogenic cytoplasmic membranes? The search for putative general lipid flippase activities should be continued. An identification of the mechanism (flippase-free or flippase-guided) controlling the asymmetric distribution of lipids across the bacterial IM is a guiding factor for the future research focused on diderm bacteria.

Summary.

• Lipid asymmetry in IM can be metabolically controlled to balance temporally the net rates of asymmetrical synthesis and subsequent flip-flop to adjust IM bilayer chemical and physical properties as spontaneous response.

• Compositional and physical asymmetries in biogenic IM are intrinsically coupled. An asymmetric effect of one non-bilayer prone phospholipid (PE) cannot be disconnected from the asymmetric effect of another (CL). Lipid asymmetry in biogenic IM is a consequence of biophysical and molecular properties of PE and CL and corresponding retentive thermodynamic mechanisms that dictate their ability to ‘flip or not to flip’ across the bilayer spontaneously.

• De novo synthesis of lipids and retrograde translocation of PE across IM is the dominant mechanism by which cells control IM lipid packing order in vivo.

• Nascent non-bilayer prone PE and CL are dynamically (re)distributed in IM to follow or direct the changes in bacterial morphology.

• De novo synthesized and inserted membrane proteins have an intrinsic capacity to immobilize and cluster specific lipids on one side of IM or facilitate their translocation and therefore scramble preexisting phospholipid asymmetry.

Abbreviations

DFDNB

1,5-difluoro-2,4-dinitobenzene

DPM

diphosphatidylmannitol

IM

inner membrane

ISO

inside-out

LPS

lipopolysaccharides

OM

outer membrane

PE

phosphatidylethanolamine

PG

phosphatidylglycerol

PM

phosphatidylmannitol

PS

phosphatidylserine

TNBS

2,4,6-trinitrobenzenesulphonic acid