Key contributions of a glycolipid to membrane protein integration

Abstract

Regulation of membrane protein integration involves molecular devices such as Sec-translocons or the insertase YidC. We have identified an integration-promoting factor in the inner membrane of Escherichia coli called membrane protein integrase (MPIase). Structural analysis revealed that, despite its enzyme-like name, MPIase is a glycolipid with a long glycan comprising N-acetyl amino sugars, a pyrophosphate linker, and a diacylglycerol (DAG) anchor. Additionally, we found that DAG, a minor membrane component, blocks spontaneous integration. In this review, we demonstrate how they contribute to Sec-independent membrane protein integration in bacteria using a comprehensive approach including synthetic chemistry and biophysical analyses. DAG blocks unfavorable spontaneous integrations by suppressing mobility in the membrane core, whereas MPIase compensates for this. Moreover, MPIase plays critical roles in capturing a substrate protein to prevent its aggregation, attracting it to the membrane surface, facilitating its insertion into the membrane, and delivering it to other factors. The combination of DAG and MPIase efficiently regulates the integration of membrane proteins.

1. Introduction

The integration of proteins into membranes is an important cellular event. Membrane proteins such as receptors, transporters, and enzymes are functional only when their structures are properly positioned in membranes. To facilitate protein transport across membranes, proteins called translocons act as channels. The Sec heterotrimer is the central component of the translocon. In eukaryotes, Secα, Secβ, and Secγ form Sec61, and in prokaryotes SecY, SecE, and SecG form SecYEG.^1)-3) This heterotrimeric unit is highly conserved across species, indicating a common molecular basis for the translocation of secretory proteins and integration of membrane proteins. However, translocons are functionally promiscuous, dynamically associating with various partner proteins and oligomerizing into different complexes.

We have discovered a unique factor named membrane protein integrase (MPIase), which mediates integration in Escherichia coli (E. coli).⁴⁾ Despite its nomenclature suggesting enzymatic activity, MPIase is a glycolipid (Fig. 1a).⁵⁾ Our membrane protein integration assay utilizing an in vitro translational system revealed that MPIase possesses integration-promoting ability. Molecular biological approaches using E. coli mutants demonstrated its significance in vivo.^6),7) MPIase is characterized as a glyco-glyceropyrophospholipid; glycolipids with pyrophosphorylated diacylglycerol (DAG) are extremely rare.⁸⁾ Notably, MPIase has a long glycan comprising N-acetyl amino sugars. We have investigated the contribution of these structural features to its integration-promoting ability using a combination of synthetic chemistry and biophysical analyses.^9)-14)

Fig. 1

Structures of (a) MPIase and (b) DAG.

Understanding membrane properties is important to elucidate the integration mechanism, especially the differences between eukaryotic and prokaryotic biomembranes.^15),16) Phosphatidylcholine (PC) and phosphatidylethanolamine (PE) are predominant in eukaryotic and prokaryotic membranes, respectively. Although various phospholipids such as phosphatidylglycerol (PG), phosphatidylserine (PS), phosphatidylinositol, and PE are found in eukaryotic membranes, bacterial membranes are less diverse, with PG and cardiolipin. The presence of sphingomyelin and cholesterol, which form microdomains called rafts, distinguishes eukaryotic membranes from bacterial membranes. Consequently, there are significant differences in the physicochemical properties between these membrane types. In addition to these bulk membrane lipids, biomembranes contain trace components. DAG (Fig. 1b), which acts as a second messenger to stimulate protein kinase C in eukaryotic cells, is a trace component in bacterial cells. The function of DAG in bacteria was unclear, because a similar signaling system is absent in bacterial membranes. We found that DAG suppresses membrane protein integration by regulating the properties of bacterial membranes.¹⁷⁾

In this review, we elucidate the opposing impacts of MPIase and DAG on membrane properties and discuss how these lipids regulate membrane protein integration in bacterial membranes.

2. Factors involved in membrane protein integration systems

2.1. Signal hypothesis in eukaryotes

The mechanism of membrane protein integration is similar to Blobel’s signal hypothesis,¹⁸⁾which explains how proteins are targeted to organelles, such as the endoplasmic reticulum (ER).^1),19) In eukaryotes, a cleavable signal peptide tag (comprising 16–30 amino acid residues) is located at the N-terminus of the protein, directing it toward the ER membrane (Fig. 2a). As a ribosome synthesizes the signal peptide, the signal recognition particle (SRP) binds to the ribosome-nascent chain complex (RNC). The SRP, along with the RNC, is directed to the SRP receptor (SR) located on the membrane surface. Upon transfer of the RNC–SRP complex from the SR to the translocon, the signal sequence is inserted into the translocon. Simultaneously, the polypeptide chain undergoes co-translational elongation through the translocon. In cases of secretion, a signal peptidase cleaves the signal peptide either during or after protein translocation to produce the mature protein.

In cases of membrane protein integration, the initial transmembrane (TM) region plays a key role instead of the signal peptide in the secretion model. Unlike the signal peptide, the TM region is not cleaved by a peptidase. As the ribosome synthesizes the polypeptide chain, the hydrophobic portion is laterally extruded into the lipid bilayer, while the hydrophilic part is released outside the membrane. This process is mediated by conformational changes in the translocon. For proteins with multiple TM regions, these steps are performed iteratively, resulting in the formation of intricate structures within the lipid bilayer.

Fig. 2

(a) Schematic models of representative translocation pathways in the endoplasmic reticulum (ER). SRP: signal recognition particle; SR: SRP receptors; SPase: signal peptidase. (b) Several translocation or integration pathways in the E. coli inner membrane (IM). (i) Sec-dependent secretory protein translocation pathway. (ii) Sec-dependent membrane protein integration pathway. (iii) Sec-independent membrane protein integration pathway. (i–iii) Although MPIase contributes to these pathways in E. coli, the localization of MPIase in the IM is unknown. (c) Examples of Sec-independent proteins. Pf3 coat: major coat protein of bacteriophage Pf3; M13 procoat: preprotein of the major coat protein of bacteriophage M13; F₀-c: F₀F₁ ATPase subunit c in E. coli; and DgkA: diacylglycerol kinase in E. coli. The cleavage of the M13 procoat at 23 amino acid residues from the N-terminus generates the mature M13 coat protein. Each protein possesses basic amino acid residues in the cytoplasm according to the positive-inside rule. (ChemBioChem 25, e202300808 modified.)

2.2. Sec-dependent protein transport in E. coli

Sec-dependent pathways are fundamentally dedicated in bacteria^2),20),21); however, for the translocation of less hydrophobic proteins destined for the outer membrane, the signal sequence is post-translationally recognized by the molecular chaperone SecB, instead of the eukaryotic SRP (Fig. 2b-i). The preprotein is then led to the motor protein SecA, which pushes the protein into the SecYEG pore in an ATP-dependent manner. The preprotein penetrates the inner membrane (IM) through a complex consisting of SecYEG, SecDF(yajC), and YidC, then, a signal peptidase cleaves the signal peptide, releasing the extramembrane region.

In the case of membrane protein integration into the E. coli IM, the Sec-dependent pathway utilizes SecYEG along with the bacterial equivalent of SRP and SR (Ffh/4.5S RNA and FtsY). YidC, an additional proteinaceous factor, facilitates the lateral translocation of TM segments into the lipid bilayer (Fig. 2b-ii).^22),23)

2.3. Sec-independent transport in E. coli

A Sec-independent pathway operates for a subset of small membrane proteins and tail-anchored (TA) membrane proteins whose TM region is located at the C-terminus (Fig. 2b-iii). Because the TM region of these proteins emerges from the ribosomal tunnel after translation is complete, they cannot undergo co-translational insertion. In eukaryotes, ribosome-associating proteins serve as chaperones for TA membrane proteins^24),25); however, such chaperones have not been identified in prokaryotes. Thus, the Sec-independent pathway in E. coli was thought to be spontaneous, facilitated by hydrophobic interactions between the TM region of the protein and the acyl chains of membrane lipids.²⁶⁾ However, it was later revealed that this process is not entirely spontaneous; instead, YidC plays a critical role in mediating protein integration into the membrane.²⁷⁾

2.4. Functions of insertase YidC

YidC was identified as a factor that interacts with the Sec-translocon and shares homology with Oxa1p in mitochondria and Alb3p in chloroplasts, both of which are crucial for membrane protein integration.^28)-31) In E. coli, YidC is essential for viability and facilitates the integration of both Sec-dependent and Sec-independent proteins. Depletion of YidC primarily affects the insertion of Sec-dependent membrane proteins, whereas it has little effect on the export of secretory proteins.

Sec-independent substrate proteins become more dependent on YidC as the hydrophilicity of the protein increases. YidC has a hydrophilic groove in the membrane, and its short TM region locally thins the lipid bilayer.^23),32) Consequently, a hydrophilic region of a substrate protein is believed to be captured in the groove, facilitating its passage through the lipid bilayer. Hence, YidC is recognized as an insertase.^29),33) Although several studies have confirmed the role of YidC in Sec-independent integration, they may have overlooked the effects of improper spontaneous insertion into artificial (proteo)liposomes.

2.5. Positive-inside rule

The positive-inside rule is a feature of the amino acid sequence of membrane proteins.^34)-36) According to this rule, membrane proteins tend to have positively charged amino acid residues on the cytoplasmic side near the TM region. The electrical potential inside the cell is kept negative, and therefore, the abundance of positively charged amino acids on the inside of the cell membrane is thought to increase electrical stability. Small proteins that are substrates for Sec-independent integration in E. coli follow the positive-inside rule (Fig. 2c).

3. Factors regulating membrane protein integration in E. coli

3.1. Blocker of spontaneous insertion

Historically, the impact of translocons and YidC has been assessed using the protease protection assay,³⁷⁾ which employs an in vitro translational system and inverted inner membrane vesicles (INVs) derived from E. coli mutants with genetically modified factors. Because INVs contain various endogenous components, isolating the role of translocons alone would be challenging. To simplify the system and exclude factors other than translocons, Nishiyama et al. utilized artificial (proteo)liposomes prepared from E. coli phospholipids (EPLs) instead of INVs. However, the in vitro translational system employing EPL liposomes did not faithfully replicate the integration process.³⁸⁾ For instance, although mannitol permease (MtlA) is a Sec-dependent membrane protein with six TM regions, it spontaneously integrated into EPL liposomes without a translocon or YidC. This was in sharp contrast to INVs, where such spontaneous integration was rigorously inhibited, suggesting that INVs contain unidentified components that regulate integration. Consequently, DAG was found to inhibit spontaneous integration.¹⁷⁾ The addition of DAG markedly suppressed the spontaneous integration of MtlA into EPL liposomes, suggesting that bacterial membranes use DAG to prevent disordered spontaneous integration. In eukaryotic cell membranes, cholesterol inhibits the membrane binding and insertion of antimicrobial peptides, fitting between lipids under the large head groups of phospholipids.³⁹⁾ Spontaneous insertion of membrane proteins is also suppressed by cholesterol.⁴⁰⁾ Bacterial membranes lack cholesterol; therefore, DAG may function analogously to cholesterol. In bacterial membranes, the DAG content (typically 2–3 w/w%) is significantly lower than the cholesterol content (20–30 mol%) in eukaryotic membranes. In our assay system, 5 w/w% of DAG sufficiently blocked spontaneous integration.

3.2. Membrane protein integration-promoting factor

When DAG was present in the membranes, both Sec-dependent and -independent integrations halted, even in the presence of YidC and the Sec-translocon.³⁸⁾ This indicates the presence of another factor that facilitates integration. Nishiyama et al. found that the addition of E. coli IM extract to DAG-containing liposomes reinstated membrane protein integration activity, strongly suggesting the presence of an integration-promoting factor referred to as MPIase.⁴⁾

When E. coli INVs were treated with urea, followed by extraction with sodium cholate, precipitation with acetone, and subsequent solubilization in a detergent, the resulting extract exhibited activity in Sec-independent integration assays. Notably, this activity was observed in the absence of detectable levels of Sec factors, YidC, or SRP. Additionally, the extract stimulated both Sec-dependent integration of MtlA and preprotein translocation of OmpA. Hence, MPIase plays a critical role in a wide range of protein transport activities across membranes in E. coli.³⁸⁾

3.3. Membrane protein integration assay

The extract was further fractionated through various column chromatography systems, and the activity of each fraction was assessed using the protease protection assay (Fig. 3a).^4),41) The protease protection assay was conducted using an in vitro translational system containing essential factors for transcription, translation, aminoacylation, and energy regeneration.⁴²⁾ In this system, a mutant of the bacteriophage Pf3 coat protein with three additional leucine residues in the TM region (termed 3L-Pf3 coat) was synthesized with ³⁵S-methionine (Fig. 3b). This protein has been used as a model substrate for Sec-independent integration because membrane potential is not required for its integration due to its increased hydrophobicity.²⁶⁾

Fig. 3

(a) Procedures for the membrane protein integration activity assay. (ChemBioChem 25, e202300808 modified.) (b) Sequence of the substrate protein used in the integration activity assay. Truncated derivatives (Pf3_24 and Pf3_27) were used for physicochemical analysis. The isotope labelled amino acid residues are shown in green and basic amino acids are in blue; TM regions are underlined. (c) Net integration percentage of MPIase. Approximately 40% of 3L-Pf3 coat integrated in the E. coli phospholipid (EPL) liposomes without DAG. In the presence of 5 w/w% of DAG in EPL liposomes, integration was significantly inhibited. The addition of MPIase restored integration in a dose-dependent manner.

When 3L-Pf3 coat was synthesized in the presence of EPL liposomes containing 5 w/w% of DAG and the test compound, the synthesized protein was integrated into the liposomes if the compound was active (Fig. 3a-i). After translation, the reaction mixture was divided into two fractions: one fraction was digested with protease K and the other was not (Fig. 3a-ii). The fraction without protease provided the total amount of synthesized proteins (A), whereas the one with protease yielded the amount of the integrated protein (B), because proteins outside the membrane were digested (Fig. 3a-iii). The percentage of net integration was determined using the ratio of (B) to (A) assessed by autoradiography on SDS-PAGE (Fig. 3a-iv). Normalized integration values were calculated by subtracting control (DAG/EPL) percentages from the net integration percentages in each experiment.

3.4. Purification of MPIase

The integration activity was significantly enhanced during purification. Notably, liquid–liquid partition chromatography removed amphiphilic contaminants, such as lipopolysaccharide (LPS). An active fraction was obtained as a 7 kDa band on a urea/SDS-gel or as a single spot in thin-layer chromatography (TLC) analysis. Approximately 20% of 3L-Pf3 coat integrated into the liposomes when 5 w/w% of purified MPIase was added to DAG-containing EPL liposomes (Fig. 3c).⁴⁾ Furthermore, the integration of 3L-Pf3 coat increased in a dose-dependent manner (0.05–5 w/w% of EPL). Moreover, stoichiometric analysis revealed that the number of integrated protein molecules was higher than the number of MPIase molecules, indicating that MPIase catalytically promoted the integration.

4. Structural determination of MPIase

4.1. Composition analysis

Because of the presence of LPS in the extract and the attenuation of activity upon proteinase digestion, it was hypothesized that MPIase contains both an LPS moiety and a proteinaceous moiety.³⁸⁾ However, MPIase was successfully separated from LPS by liquid–liquid partition chromatography. Furthermore, no peaks corresponding to standard amino acids were observed in the amino acid analysis of the highly purified MPIase, indicating that MPIase is not a proteinaceous molecule (Fig. 4a). Only two peaks were detected: one for glucosamine and the other for ammonia. We considered that some amino sugars were decomposed by hydrolysis to produce ammonia, as only glucosamine was detected.

Fig. 4

(a) Amino acid analysis of standard amino acids and hydrolyzed MPIase. (b) ¹H-NMR of MPIase. (c) ³¹P-NMR of MPIase. (d) Q-TOF-MS of the hydrophobic fraction obtained from the hydrolysis of MPIase. (Nat. Commun. 3, 1260 modified.)

¹H and ¹³C nuclear magnetic resonance (NMR) spectroscopy demonstrated the presence of sugars and lipids in MPIase. Notably, multiple signals detected at δ ～2 ppm in ¹H-NMR were identified as N- and O-acetyl (Ac) groups (Fig. 4b).⁵⁾ Additionally, two peaks at δ -11.3 and -13.0 ppm in ³¹P-NMR suggested the presence of a pyrophosphate diester, which was assigned as a linker between a glycan chain and a lipid moiety (Fig. 4c). Hydrolysis of MPIase with aqueous HF yielded a lipid anchor moiety and a glycan. The lipid anchor was identified as DAG by quadrupole time-of-flight mass spectrometry (Q-TOF-MS), and the fatty acids of DAG were found to be similar to those of membrane lipids in E. coli (16:0, 16:1, 18:0, and 18:1 etc.), although the composition varied between culture lots (Fig. 4d). Thus, we concluded that MPIase is a glyco-glyceropyrophospholipid.⁸⁾

4.2. Constituent sugars of the glycan

The structure of MPIase glycan was determined through a combination of instrumental analyses and synthetic studies. A series of broad peaks in matrix-assisted laser desorption/ionization (MALDI)-TOF-MS of MPIase suggested the presence of repeating units with various modifications (Fig. 5a).⁵⁾ The subdivided peaks, differing by 42 mass units, were attributed to different degrees of acetylation in the molecule, as indicated by NMR. The sizes of the typical units were estimated to be 608 or 650 mass units based on the intervals between the peak apexes. MS/MS analysis suggested that the unit with m/z 608 contains three components with 187, 203, and 217 mass units, whereas the unit with m/z 650 contains 187, 245, and 217 mass units (Fig. 5b). This may indicate partial acetylation of the 203 mass unit component.

Fig. 5

(a) MALDI-TOF-MS of MPIase in the negative-linear mode. (b) MS/MS spectrum from m/z 608 in the positive-reflecting mode. (c) MALDI-TOF MS of the glycan part of hydrolyzed MPIase in the negative-linear mode. 2,5-Dihydroxybenzoic acid was used as the matrix. (Nat. Commun. 3, 1260 modified.)

The 203 mass unit component was identified as N-acetyl-glucosamine (GlcNAc) in its dehydrated form, since glucosamine was detected in the amino acid analysis. The 187 mass unit component was presumed to be a deoxy form of N-acetyl-aminohexose. The methyl signal at δ 1.10 ppm in the ¹H-NMR indicated the presence of a 6-deoxy sugar, and correlation with an amide proton in nuclear Overhauser effect spectroscopy (NOESY) suggested the presence of 4-acetamido group. The 217 mass unit component would be an oxidized form of N-acetyl-aminohexose.

Constituent sugars analysis by gas chromatography (GC)-MS after methanolysis gave three peaks, supporting that MPIase mainly contains trisaccharides. Comparison of NMR and GC-MS with synthetic authentic samples showed that the 187 mass unit component was N-acetyl-4-aminofucose (Fuc4NAc) and the 203 mass unit component was N-acetyl-mannosaminuronate (ManNAcA). GC-MS analyses of the glycosides obtained through solvolysis in (S)-(+)-2-butanol revealed that all sugars had d-configuration.

4.3. Sequence and repeating number of trisaccharide units

The sequence of the trisaccharide units was estimated using two-dimensional (2D)-NMR. We synthesized some substructures and compared their ¹³C-NMR spectra to confirm the linkage mode. We concluded that the repeating trisaccharide is →3)-α-d-Fuc4NAc-(1→4)-β-d-ManNAcA-(1→4)-α-d-GlcNAc-(1→. The reducing end of the glycan was determined as GlcNAc, because reduction of the glycan produced 2-amino-2-deoxyglucitol. In addition, based on ¹H-NMR, approximately 30% of the GlcNAc was 6-O-acetylated.

The number of repeating units was analyzed by MALDI-TOF-MS of the hydrolyzed glycan moiety. Mild hydrolysis with cold NaOH produced a simple glycan without O-Ac and lipid modifications. Because its MALDI-TOF-MS spectra showed regular intervals with 608 mass units, we presumed that MPIase is a mixture of homologous molecules with the number of repeating units ranging from 7 to 14, but mainly 9–11 (Fig. 5c).

Thus, we determined the structure of MPIase as shown in Fig. 1a.⁵⁾ Notably, MPIase contains more than 30 N- and O-Ac groups. Digestion of the acetoamide linkages by a non-selective proteinase may attenuate its activity. Another remarkable feature is that the trisaccharide unit of MPIase is the same as that of enterobacterial common antigen (ECA), which is found in the outer membranes of enterobacteria (Fig. 6a).^43)-45) However, the number of repeating units of ECA (18–55) is longer and more diverse than that of MPIase. ECA has a monophosphate linker instead of a pyrophosphate linker in MPIase. Notably, despite its similar structure, ECA has no membrane protein integration activity.

Fig. 6

(a) Structure of ECA. (b) Biosynthetic pathway of ECA. (c) Putative biosynthetic pathway of MPIase. (d) Putative biosynthetic intermediates of MPIase. (Front. Chem. 12, 1353688 modified.)

5. Biosynthesis of MPIase

5.1. Biosynthesis of ECA

ECA biosynthesis has been studied and was reviewed elsewhere.^8),45) The ECA trisaccharide repeating unit is constructed on a lipid carrier called undecaprenyl phosphate (Und-P) (Fig. 6b).⁴⁶⁾ Und-P serves as a universal lipid carrier necessary for synthesizing glycans in E. coli. The biosynthetic genes for ECA are located in the wec operon.⁴⁷⁾ Initially, WecA catalyzes the synthesis of the lipid I_ECA by utilizing uridine diphosphate (UDP)-GlcNAc and Und-P, where GlcNAc is linked to Und via a pyrophosphate.⁴⁸⁾ Subsequently, stepwise addition of UDP-ManNAcA by WecG and dTDP-Fuc4NAc by WecF leads to the formation of lipid II_ECA and lipid III_ECA, respectively.^49),50)

Although lipid III_ECA is synthesized in the IM inner leaflet facing the cytoplasm, it is transferred to the IM outer leaflet by the flippase WzxE.⁵¹⁾ Here, the trisaccharide units are polymerized on the Und-PP carrier.^52),53) The resulting long glycan is transferred to a new lipid carrier, DAG, with a phosphate linker. Recent studies have identified PG as the lipid donor; however, the enzyme that catalyzes this reaction is unknown.⁵⁴⁾ O-acetylation occurs non-stoichiometrically at the hydroxy group at the 6-position of GlcNAc.⁵⁵⁾ Although mature ECA is present in the outer membrane, the trafficking mechanisms are not fully understood.

ECA is not essential for enterobacteria, thus its precise function remains unclear. However, its biosynthesis affects the biosynthesis of other glycans, such as O-antigen, peptidoglycan, and capsule. Therefore, inhibition of ECA biosynthesis midway is toxic to cells due to the loss of the common Und-P pool.^51),53),56)

5.2. Biosynthesis of MPIase

The trisaccharide moieties of MPIase and ECA are identical. However, MPIase is produced even in ECA-deficient E. coli, whereas MPIase depletion did not inhibit ECA biosynthesis.⁵⁷⁾ This indicates that the biosynthesis of MPIase is different from that of ECA. During studies on ECA biosynthesis, depletion of Fuc4NAc resulted in the accumulation of the lipid II_ECA, in which the disaccharide is attached to Und-PP.⁴⁶⁾ Additionally, DAG pyrophosphate (DGP)-disaccharide, in which the disaccharide is attached to DAG through a pyrophosphate, was identified (Fig. 6c,d). This suggests that the biosynthetic intermediate for MPIase directly uses DAG instead of Und.⁸⁾ There are no other examples in which DAG was initially used as a carrier.

Therefore, we hypothesized that GlcNAc-PP-DAG (compound I) is the first product for MPIase biosynthesis (Fig. 6d). Identifying the enzymes responsible for the production of compound I revealed the involvement of CdsA and its paralog, YnbB.^6),58),59) Generally, CdsA catalyzes the biosynthesis of a common intermediate of membrane phospholipids, called cytidine diphosphate DAG (CDP-DAG). Depletion of CdsA not only disrupted phospholipid biosynthesis but also depleted MPIase. Although the yeast mitochondrial CDP-DAG synthase, Tam41p,⁶⁰⁾ compensated for phospholipid biosynthesis, it did not suppress MPIase depletion.⁶⁾ MPIase depletion induced abnormal protein membrane transport and inhibition of bacterial growth. Thus, the significance of MPIase in protein integration was shown in vivo. Furthermore, expression of both CdsA and YnbB is controlled by temperature, and upregulation of these genes at low temperatures (< 25℃) increased MPIase production, probably because cells stimulate protein translocation to counteract reduced membrane flexibility in cold conditions.^61)-63)

When INVs prepared from CdsA-overexpressing mutants were reacted with phosphatidic acid, compound I was detected by liquid chromatography-MS in a GlcNAc phosphate (GlcNAc-P)- and cytidine triphosphate (CTP)-dependent manner.⁶⁾ Moreover, in reactions containing the pH-sensitive mutant CdsA8, CDP-DAG was biosynthesized and then compound I, suggesting that CDP-DAG is converted into compound I by incorporating GlcNAc-P. Introduction of the eukaryotic homologue of CdsA, Cds1p, with Tam41p into CdsA-deficient E. coli compensated for MPIase biosynthesis and promoted cell growth.

It is considered that E. coli has a complete set of biosynthetic genes for MPIase, as the knockout mutant for each gene involved in ECA biosynthesis expressed MPIase at the wild-type (WT) level. However, the enzymes responsible for condensing the second sugar and subsequent reactions in MPIase biosynthesis remain unknown.

6. Synthesis of MPIase analogs

MPIase is the first glycolipid known to be involved in protein translocation. Purified MPIase induced membrane integration of proteins; however, we cannot rule out the possibility that trace amounts of impurities contributed to this activity. Moreover, the mechanism by which MPIase inserts the substrate protein into the membrane is unclear. Because the molecular composition of natural MPIase is heterogeneous, including variations in the number of O-Ac groups, length of glycan, or types of fatty acids, precise analysis of the mode of action using natural MPIase may be challenging. Therefore, we synthesized structurally defined analogs of MPIase.⁹⁾

6.1. Retrosynthetic strategy for mini-MPIase-3

During structural determination, we synthesized several substructures of the trisaccharide. Chemical syntheses of the trisaccharide and hexasaccharide moieties of ECA, which are identical to those of MPIase, have been reported.^64),65) However, neither included the phospholipid moiety. Therefore, we synthesized a minimal structural unit termed mini-MPIase-3 (1) as our first target, which consisted of a trisaccharide and a pyrophospholipid. This required the synthesis of unusual sugars (Fuc4NAc and ManNAcA), construction of a β-mannosaminide linkage, selective O-acetylation at the 6-OH of GlcNAc, and introduction of a labile pyrophospholipid. The design for the synthesis of 1 is outlined in Scheme 1. The labile pyrophospholipid was introduced at a later stage. To construct the pyrophospho-linkage, a phosphoramidite-possessing DAG (2) and trisaccharyl phosphate (3) were designed. Myristic acid (C14:0) was chosen for the fatty acid component of DAG (4), because it can be conveniently synthesized and its phase transition temperature is comparable to that of natural membrane lipids.

Scheme 1.

Retrosynthetic strategy for mini-MPIase-3 (1).

The key trisaccharide intermediate (5) was protected by benzyl (Bn)-type groups, which can be removed by hydrogenation in neutral conditions to avoid basic hydrolysis of the 6-OAc on GlcNAc, DAG, and pyrophosphate. Properly protected monosaccharide building blocks were designed as follows. The GlcNAc acceptor (7) had a tert-butyldiphenylsilyl (TBDPS) group at O6 and an allyl group at O1, both selectively cleavable upon introduction of an Ac group or phosphorylation. The ManNAcA moiety was synthesized from a Glc moiety containing 2-O-benzoyl (Bz) (8). The Fuc4N donor (9), equipped with a selectively removable p-methoxybenzyl (PMB) group for subsequent glycan elongation, was synthesized from a Glc unit (10).

6.2. Synthesis of mini-MPIase-3

The synthesis was carried out in accordance with Scheme 2.⁹⁾ Condensation of the GlcNAc acceptor (7) and the Glc donor (8) selectively yielded the desired β-disaccharide (11) due to the neighboring group participation effect. Subsequently, 2-OBz was converted to 2-N₃ (13) via an intermediate (12), accompanied by configurational inversion. Removal of the benzylidene acetal moiety and regioselective acetylation at 6-OH of 13 provided the disaccharide acceptor (6). The glycosylation of the Fuc4N donor (9) with 6 afforded the trisaccharide (14) with good α-anomer selectivity (α/β = 7/1). Deprotection of the 6-OAc group of the ManN moiety followed by oxidation of the 6-OH gave the uronate (15). The TBDPS group of the GlcNAc residue was successfully converted to an Ac group, yielding the 6-OAc GlcNAc derivative (16). Reduction of the two azide groups of 16 was achieved using a combination of SnCl₂, PhSH, and Et₃N,^64),66) followed by N-acetylation to produce the acetoamide (17). Selective deprotection of allyl glycoside at the reducing end provided the trisaccharide hemiacetal (5), which was subsequently combined with commercially available diallylphosphoramidite to yield the phosphoric ester (18). Elimination of the allyl esters afforded trisaccharyl phosphate (3).

Scheme 2.

(a) Synthesis of mini-MPIase-3 (1). (i) TMSOTf, MS4A, CH₂Cl₂, 0℃ to RT, 74% (β). (ii) (1) NaOMe, MeOH, RT. (2) Tf₂O, pyr./CH₂Cl₂, 0℃. (iii) NaN₃, DMF, 120℃, 46% (3 steps). (iv) (1) p-TsOH, EtOH, 60℃. (2) AcCl, Et₃N, CH₂Cl₂, 0℃, 57% (2 steps). (v) 9 (5 equiv.), NIS, TfOH, MS4A, CH₂Cl₂, -20 to 0℃, 86% (α/β = 7/1). (vi) (1) NaOMe, MeOH, RT. (2) TEMPO, DIB, CH₂Cl₂/H₂O, RT. (3) BnBr, Cs₂CO₃, DMF, RT, 75% (3 steps). (vii) (1) TBAF, THF, RT. (2) Ac₂O, pyr., RT, 77% (2 steps). (viii) (1) SnCl₂, PhSH, Et₃N, MeCN/MeOH/CH₂Cl₂. (2) Ac₂O, MeOH, 89% (2 steps). (ix) (1) Ir(COD)(Ph₂MeP)₂PF₆, THF. (2) sat. NaHCO₃ aq., I₂, THF, 86% (2 steps). (x) (1) diallyl diisopropylphosphoramidite (4 equiv.), Tetrazole, CH₂Cl₂, -20℃ to RT. (2) TBHP, CH₂Cl₂, -40 to 0℃, 55% (α). (xi) Pd(PPh₃)₄, Et₂NH, THF, RT, 95%. (xii) (1) 2 (1.5 equiv.), 4,5-dicyanoimidazole, MeCN, RT. (2) TBHP, MeCN, RT, 52%. (xiii) Pd-black, H₂, MeOH/THF, RT, 75%. (b) Synthesis of phosphoramidite bearing DAG (2). (xiv) (1) BnOH, Pyr., THF, -78℃ to RT. (2) N, N-diisopropylamine, THF, -10℃ to RT. (xv) 4 (1 equiv.), 4,5-dicyanoimidazole, MeCN, 0℃ to RT, 60% (3 steps).

The fully protected trisaccharyl pyrophospholipid (19) was obtained by the coupling of 3 with 2, followed by oxidation. Subsequently, the Bn-type protective groups were globally deprotected via hydrogenolysis, resulting in the trisaccharyl pyrophospholipid with the 6-OAc group on GlcNAc, namely mini-MPIase-3 (1).

6.3. Synthesis of mini-MPIase analogs

Mini-MPIase-3 (1) exhibited integration activity (see 7.1); therefore, we synthesized various analogs for structure–activity relationship studies (Fig. 7).¹⁰⁾ We synthesized trisaccharide variants with modified functional groups: mini-MPIase-3 (6-OH) (22), mini-MPIase-3 (6-F) (23), mini-MPIase-3 (6-OMe) (24), mini-MPIase-3 (6-OBz) (25), mini-MPIase-3 (COOMe) (26), glycan length variants: mini-MPIase-6 (27), mini-MPIase-9 (28), variants for the pyrophosphate moiety: mini-ECA-3 (29), mini-ECA-6 (30), Trisac-DAG (31), anchorless variants: Trisac-P (32), Hexsac-P (33), Nonasac-P (34), and Trisac-P (6-OH) (35). An anchorless pyrophosphate analog Trisac-PP (36) was designed for docking simulation (DS) but not synthesized. Anchorless derivatives possessing a long glycan, including polysac-P (37), polysac-P (6-OH) (38), and polysac (39), were prepared from natural MPIase.

Fig. 7

Structures of MPIase and its synthetic analogs.

MPIase consists of repeating units, so glycan elongation was performed based on the trisaccharide, with some modifications to the protective groups compared with the strategy used for 1 (Scheme 3). To construct the challenging α-glucosaminide linkage, the GlcNAc moiety had an N₃ group at C2 with no neighboring group participation effect, along with bulky substituents at O3 and O6.⁶⁷⁾ Both the trisaccharide donor (40) and acceptor (41) were derived from the common intermediate (42), where selective deprotection of an allyl group at the non-reducing end or a methoxyphenyl (MP) group at the reducing end facilitated orthogonal conversion.

Scheme 3.

Retrosynthetic strategy for various MPIase analogs.

7. Structure–activity relationship study

7.1. Membrane protein integration by mini-MPIase-3

The membrane protein integration assay demonstrated that natural MPIase can integrate approximately 20% of the 3L-Pf3 coat into EPL liposomes containing DAG (Fig. 8a). Mini-MPIase-3 (1) exhibited weak but significant activity in a dose-dependent manner, conclusively proving that a glycolipid can serve as an integration-promoting factor.⁹⁾ Furthermore, we confirmed that 1 mediated the integration of other Sec-independent substrate proteins, such as WT Pf3 coat and M13H5 phage procoat protein, into DAG/EPL liposomes. Notably, 1 contains the minimum active structure although the glycan length of 1 is approximately one tenth that of natural MPIase.

Fig. 8

(a)–(e) Membrane protein integration activity of MPIase and its analogs. The designated amount of test compound was incorporated into DAG (5 w/w%)/EPL liposomes. All values were obtained from at least three independent experiments. (a) Net integration activity of MPIase and 1. (b) Net integration activity of mini-MPIase-3 analogs (22–26) with the equivalent dose (0.76 nmol/tube) to 5 w/w% of 1 in EPL. (c) Normalized integration activity of elongated analogs (27, 28). (d) Comparison of the net integration activity of phosphate variants (0.76 nmol/tube). (e) Normalized integration activity with the addition of anchorless analogs into liposomes including 1. (f) Structural requirements of MPIase; (i) a compact and hydrophobic group for R¹ at GlcNAc C6, (ii) the carboxyl group of ManNAcA, (iii) the pyrophosphate group, and (iv) the lipid anchor are required. (v) Both integration activity of MPIase analogs and integration-enhancing effects of anchorless analogs improve according to the number of trisaccharide units. (Chemistry 29, e202300437 modified.)

7.2. Activities of mini-MPIase analogs

We examined the net integration percentages of the mini-MPIase analogs, each with the same number of molecules (0.76 nmol/tube), corresponding to 5 w/w% of 1 against EPLs (Fig. 8b).¹⁰⁾ First, we compared the effect of substituents at the 6-position on GlcNAc, because this position of natural MPIase is partially acetylated. The 6-OH analog (22) exhibited reduced activity compared with 1, whereas the values of the 6-F analog (23) and the 6-OMe analog (24) were almost comparable. The hydrophobic interaction at this position may contribute to the integration activity. However, the activity of the 6-OBz analog (25) was significantly reduced, suggesting that a bulky hydrophobic substituent is unfavorable. The value of the methyl ester analog (26) was decreased, indicating that the carboxy group is also involved in the integration activity. Improved activity through glycan elongation was demonstrated by 27 and 28 in a dose-dependent manner (Fig. 8c). The effect of glycan length was more pronounced at lower concentrations.

7.3. Activities of mini-ECA analogs

We compared the pyrophosphate moiety variants (Fig. 8d). When comparing the trisaccharide analogs, the activity decreased as the number of phosphate groups in the linker decreased. The significance of the pyrophosphate was more evident for the hexasaccharide analogs; the integration value of 27 was higher than that of 1, while the value of 30 was almost the same as that of 29. In contrast to pyrophosphate-type analogs, long sugar chains appear to be undesirable in monophosphate-type analogs, since ECA has no membrane protein integrating activity.

7.4. Activities of anchorless analogs

Next, the role of the lipid anchor moiety in MPIase was investigated. When an MPIase derivative (37) lacking the lipid moiety was subjected to the integration assay containing control liposomes (DAG/EPL), no integration activity was observed even at high concentrations of 1.0 mg/mL (data not shown). This suggests that MPIase must be anchored in the membrane.⁹⁾

In contrast, when the liposomes contained 5 w/w% of 1, the integration of 1 was significantly enhanced by 37 (Fig. 8e). Such a synergistic effect was not observed with the addition of other derivatives lacking the 6-OAc group (38) or the phosphate group (39). These results suggest that the glycan moiety interacts with the protein using these functional groups. We hypothesize that 37 maintains the integration-competent structure of proteins by preventing their aggregation.

We then investigated the enhancing effect of synthetic anchorless analogs on the activity of 1.^9),10) We found that the addition of 34 notably increased the activity. In contrast, the normalized integration values of 32 and 33 were not significantly different from those without additives. These analogs would be too short to cover the hydrophobic region of the protein.

7.5. Structure–activity relationship

The results of the structure–activity relationship are summarized in Fig. 8f.^8),10) Mini-MPIase-3 (1) encompasses an essential structure for activity, where both a compact hydrophobic group on GlcNAc C6 and a uronate on ManNAcA are important. Integration activity correlates with the number of trisaccharide units, with longer glycans being more efficient. However, even a single trisaccharide unit displayed weak integration ability when located at the membrane. The pyrophosphate linker was found to be significant, with activity decreasing as the number of phosphate groups decreased. The significance of the pyrophosphate group was more pronounced with longer glycans.

Anchoring in the membrane by the phospholipid is critical for integration. Although anchorless analogs alone did not integrate proteins into the membrane, long glycans enhanced integration by preventing protein aggregation. This chaperone-like activity requires a phosphate at the reducing end and an acetyl group on GlcNAc C6. Contrary to the case placed in the membrane, a single trisaccharide unit is insufficient and at least three are required for the chaperone-like activity.

8. Intermolecular interactions between MPIase and a substrate protein

We hypothesized that MPIase operates through two steps of intermolecular interactions, a chaperone-like activity that folds a protein and an integration activity that embeds a protein in the membrane. Hydrophobic proteins generated from ribosomes immediately aggregate without assistance. MPIase may serve like a chaperone to prevent protein aggregation by capturing the nascent protein. In this section, we have verified the direct interactions between MPIase and the substrate protein using several physicochemical techniques. The methods used here have been described elsewhere.⁶⁸⁾

8.1. Anti-aggregation effect

8.1.1. Gel filtration experiments

Our hypothesis that the glycan moiety of MPIase interacts with the substrate protein was verified by gel filtration experiments.⁵⁾ In the absence of 37, the aggregates of the 3L-Pf3 coat eluted in the void fraction. In contrast, in the presence of 37, the protein was eluted in lower molecular weight fractions, indicating that the 3L-Pf3 coat and 37 formed a soluble complex. Such an anti-aggregation effect was absent for 38 or 39.

8.1.2. Secondary structure alteration

Chaperones help ensure correct folding and prevent misfolding or aggregation of substrate proteins. We investigated whether MPIase has such chaperone-like effects.¹³⁾ We used Pf3 coat without the 3L-substituent as a substrate, because the severe hydrophobicity of the 3L-Pf3 coat hampered sample preparation. Even a dilute solution of Pf3 coat in a buffer without MPIase resulted in fluffy insoluble matter; however, dispersion of the Pf3 coat in an MPIase-containing buffer did not generate visible aggregates. Since MPIase prevented aggregation, we measured circular dichroism (CD) spectra to detect alterations in secondary structure. The CD spectra were significantly different in the presence and absence of MPIase (Fig. 9a). In MPIase-free buffer, the Pf3 coat exhibited no definite secondary structure, whereas the characteristic α-helical structure was observed in MPIase-containing buffer. These results demonstrate that MPIase prevented the aggregation by altering the secondary structure of a substrate protein. Given the results of gel filtration experiments, it is likely that at least phosphate and O-acetyl groups contribute to the structure alteration. It is unclear whether the lipid moiety of MPIase is required.

Fig. 9

(a) CD spectra of the Pf3 coat with and without MPIase. (b) SPR sensorgrams of natural MPIase, mini-MPIase-3 (1), and Trisac-DAG (31) to the immobilized Pf3 coat. (c) The on-off rate map of MPIase, 1, and liposomes including 1. (d) Relative intensities of STD signals of Trisac-P (32) and Trisac-P (6-OH) (35) with Pf3 coat. OAc: 6-O-acetyl group on GlcNAc; NAc(F,M): N-acetyl groups on Fuc4NAc and ManNAcA; NAc(G): N-acetyl group on GlcNAc; and Me(F): methyl group on Fc4NAc. (e) Heatmap of frequencies in docking simulation shown over the atoms of Trisac-PP (36) and amino acid sequence of Pf3 coat. (f) Arginine side chain signal regions of the 2D ¹H–¹⁵N ssNMR spectra of Pf3_27 in EPL liposomes including 1 or mini-ECA-3 (29). (ACS Chem. Biol. 17, 609 modified.)

8.2. Kinetic analysis of the interactions

Surface plasmon resonance (SPR) is used to monitor molecular binding in real time. Thus, SPR was conducted to obtain affinity profiles of the interactions between MPIase and the Pf3 coat (Fig. 9b).¹³⁾ The Pf3 coat was immobilized on the SPR flow cells of the sensor chip, and glycolipid analytes were flowed over the cells. Direct interaction was verified by the concentration-dependent binding of natural MPIase to Pf3 coat. The dissociation constant (K_D) was determined with a steady-state affinity model (2.6 ± 0.6 μM). Synthetic analogs were then compared. Mini-MPIase-3 (1) interacted with Pf3 coat in a concentration-dependent manner; however, 31 had a weaker response than 1. The apparent K_D value of 31 (136.7 ± 32.7 μM) was larger than that of 1 (2.1 ± 1.6 μM), demonstrating the contribution of a pyrophosphate group.

Although the K_D values of MPIase and 1 were comparable, the on-off rate map revealed differences in binding kinetics between the two glycolipids (Fig. 9c).^13),69) MPIase exhibited faster association and dissociation with Pf3 coat compared with 1. A longer glycan may contribute to faster contact with the protein. When the liposome incorporating 1 was flowed as an analyte, its apparent association and dissociation rates were accelerated compared with those in solution. A high local concentration of 1 within membranes may induce fast interactions, potentially facilitated by the assembly of trisaccharides resembling the elongated glycan chain of MPIase.

8.3. Epitope mapping of the interaction

We then investigated the interfaces of the interactions between the glycan of MPIase and a protein at the atomic level.

8.3.1. Acetyl group on GlcNAc

Saturation transfer difference (STD)-NMR is used to detect transient binding of small molecule ligands to macromolecular receptors. We used STD-NMR to identify interaction sites.¹³⁾ Given the significance of the 6-OAc group in GlcNAc indicated by the integration assay, we compared trisaccharide analogs devoid of a lipid moiety, 32 and 35, using STD-NMR. Figure 9d summarizes the normalized STD values of representative signals upon saturation with Pf3 coat. We observed that 6-OAc and N-Ac on GlcNAc exhibited higher values than others, indicating that the substrate protein is attracted to 6-OAc on GlcNAc, thus influencing the chemical environment around GlcNAc. However, STD signals were also detected from most of the protons of the trisaccharide, suggesting that Pf3 coat and the trisaccharide can bind flexibly in various modes and adopt possible conformations. Therefore, the interaction is not as strongly restricted as the lectin–glycan interaction.

8.3.2. Docking simulation

Computational simulation (in silico) offers a distinct analytical approach, whereas in vitro experimental analysis of protein–ligand complexes may encounter challenges in sample preparation and selection of optimal conditions. The STD-NMR results suggested that the ligand is highly flexible and that there is no obvious pocket on the target protein. This suggests the presence of multiple docking poses rather than a single stable docking pose. The docking function in the MOE software can search for favorable binding conformations of medium-sized ligands, such as glycans, using physics-based scoring functions.⁷⁰⁾ Thus, virtual truncated ligands were subjected to docking simulations (DS) with the Pf3 coat, and the ensemble of docking poses was statistically analyzed. The heatmap illustrates the frequency of interaction between each atom of Trisac-PP (36) and each amino acid residue of Pf3 coat (Fig. 9e).¹³⁾ Frequent contacts indicate hydrophobic interactions between acetyl groups and hydrophobic residues, as well as electrostatic interactions between pyrophosphate and basic residues. Analogs lacking 6-OAc on GlcNAc or pyrophosphate exhibited decreased contact in all combinations, emphasizing the significance of these functional groups in the MPIase–protein interaction.

8.3.3. Interaction sites in the protein

We used SPR to confirm the contacts indicated by DS between the glycan and hydrophobic residues or basic amino acids of the Pf3 coat protein. The binding of MPIase to Pf3 coat mutants lacking each region decreased compared with that of intact Pf3 coat. In addition, the interaction between Pf3 coat and MPIase was smaller in a high salt buffer (650 mM NaCl) than that in conventional conditions (150 mM NaCl), because the electrostatic interactions may be disturbed in high salt conditions. Thus, MPIase captures the hydrophobic region and the basic amino acids in substrate proteins.

8.3.4. Pyrophosphate linker and basic amino acid residues

As basic amino acids are commonly present in membrane proteins (positive-inside rule), we focused on basic amino acid residues in the Pf3 coat. A Pf3 coat analog, termed Pf3_27 (Fig. 3b), was selectively ¹⁵N-labeled at the Arg20 and Lys23 side chains and incorporated into EPL liposomes. The solid-state NMR (ssNMR) signal of Arg20 protons exhibited a downfield shift upon addition of 1 to the liposomes, which disappeared in high salt conditions (Fig. 9f).¹²⁾ In contrast, the monophosphate group in 29 interacted with the protein as weakly as bulk lipids.¹⁰⁾ Therefore, the shift indicated that the Arg20 residue interacts with the pyrophosphate in MPIase through intermolecular hydrogen bonding. The downfield shift value was smaller than that of a typical ligand–receptor complex, indicating that binding between the basic amino acid and pyrophosphate in MPIase is weak and transient in the membrane. Thus, MPIase can easily detach from the protein.

9. Intermolecular interactions between MPIase and membranes

We investigated the subsequent step of protein insertion into the membrane. Historically, spontaneous insertion of antimicrobial peptides into the membrane has been extensively studied, with findings indicating that membrane morphology and physicochemical properties influence insertion.⁷¹⁾ In this section, we elucidate the physicochemical properties of membranes containing DAG and/or MPIase, followed by an assessment of the contribution of MPIase to protein insertion.

9.1. Alteration of the physicochemical properties of membranes

Disruption of the lipid bilayer or pore formation in the membrane may interfere with protein integration. However, neither DAG nor MPIase induced any change in membrane morphology or membrane fusion at the concentrations employed in the assay.¹¹⁾ This was confirmed by ³¹P ssNMR, cryo-transmission electron microscopy, and cobalt–calcein assays. The physicochemical properties of the membrane were then evaluated. A summary of the measurements and techniques employed is outlined in Table 1 and Fig. 10a.^{11),12),68),72)} DAG notably decreased the mobility of the acyl chains, particularly within the membrane core. In addition, the flip-flop of DAG hindered lateral diffusion and filled membrane defects, thereby impeding protein access to the membrane interior. Conversely, MPIase counteracted the effects of DAG. The flexible glycan of MPIase made the membrane surface flexible, increased mobility in the membrane core, and mitigated the flip-flop motion of DAG.

Table 1

Alteration of physicochemical properties of EPL membranes in the presence of DAG ± MPIase

Property	Method of measurement	Result
		+DAG	+MPIase/DAG
Acyl chain order	2H quadrupole splittinga	Order	Disorder
Acyl chain mobility	DHP fluorescence anisotropy	Slow	High
Head-group mobility	31P T1relaxation timea	Little effect	High
Head-group packing	Laurdan fluorescence	Tight	Loose
DAG flip-flop	NBD-DAG quenchingb	Fast	Decrease
Lateral diffusion	FRAPc	Decrease	Little effect
Sugar chain mobility	13C CP and DPa,d	–	Fast

^assNMR.

^bNBD-DAG, nitrobenzoxadiazole-labeled diacylglycerol.

^cFRAP, fluorescence recovery after photobleaching.

^dCP, cross polarization; DP, direct polarization.

Fig. 10

(a) Schematic model of the physicochemical properties of the membrane influenced by DAG and/or MPIase on EPL membranes. Blue double-headed arrows show the rate of flip-flop motion of DAG. Red double-headed arrows show the degree of the lateral diffusion rate of membrane lipids. The flexibility of lipids in the membrane core region is shown as a red gradient. (b) The procedure for inserting Pf3_24 into membranes. (i) Pf3_24 (dark gray) solubilized in 1,2-diheptanoyl-sn-glycero-3-phosphocholine (DHPC) was mixed with liposomes composed of various lipid types. (ii) The mixture was incubated for 30 min. (iii) Then, it was diluted with a buffer solution until the DHPC concentration reached below its critical micelle concentration. (iv) The supernatant containing DHPC was removed after centrifugation. The precipitates were subjected to ¹⁵N-ssNMR. (c) ¹⁵N-ssNMR spectra of Pf3_24. (top) Pf3_24 fully reconstituted into PC liposomes. (middle) Pf3_24 in aggregates. The sample was prepared according to the procedure shown in Fig. 10b without using liposomes. (bottom) The sample was prepared according to the procedure shown in Fig. 10b with PC liposomes. The pink line at 125.6 ppm corresponds with a β-strand conformation, and the blue line at 119.6 ppm corresponds with random-coil and/or α-helical conformations. (d) Correlation between membrane insertion efficiency values and fluorescence anisotropy values of 1,6-diphenyl-1,3,5-hexatriene (DPH) at 37℃ in various types of membranes. The black line is the regression line obtained from global linear fits performed on bulk phospholipids. The values for ELP liposomes containing 5 mol% DAG (blue), 5 mol% Lyso-PC with DAG (green), and 1 mol% MPIase with DAG (red) were not included in the fitting. (Sci. Rep. 12, 12231 modified.)

9.2. Protein insertion efficiency affected by membrane lipids

The spontaneous insertion of antimicrobial peptides into membranes is influenced by the charge and spontaneous curvature of membrane lipids.⁷¹⁾ To assess the impact of various membrane lipids, we measured the insertion efficiency of a model protein into liposomes composed of various lipids.

Typically, the TM regions of membrane proteins tend to adopt α-helical or β-strand conformations. CD spectra and ssNMR spectra revealed that a TM region of the Pf3 coat, termed Pf3_24 (Fig. 3b), adopts an α-helical conformation upon insertion into the lipid bilayer.¹²⁾ The correlation between ¹⁵N-NMR chemical shift and protein secondary structure is well-established.⁷³⁾ For instance, the averaged ¹⁵N chemical shift values for valine residues are approximately 123.27 ppm for β-strands, 119.66 ppm for random coils, and 119.53 ppm for α-helices.⁷⁴⁾ Thus, we measured the ¹⁵N ssNMR of Val and Gly selectively ¹⁵N-labeled Pf3_24 upon mixing with liposomes (Fig. 10b). Without DAG, part of the proteins was inserted into the liposomes, while the rest formed aggregates.¹²⁾ The ¹⁵N chemical shift value of Pf3_24 embedded in liposomes was approximately 120 ppm, confirming its α-helical structure (Fig. 10c). In contrast, in a buffer without liposomes, aggregates of Pf3_24 yielded a prominent signal at 125 ppm and weaker signals at 120 ppm, suggesting that the aggregates adopted β-strand-rich structures, with minor conformers of random coils and/or α-helices. When Pf3_24 was mixed with the liposomes, the spectrum represented a weighted sum of the former and latter spectra, with the coefficient of the former spectrum corresponding to the insertion efficiency.

Figure 10d (black circles) illustrates a strong correlation between the resulting insertion efficiency of Pf3_24 into various membranes and the fluorescence anisotropy values. Fluorescence anisotropy measured with a fluorescent probe, 1,6-diphenyl-1,3,5-hexatriene (DPH), reflects the flexibility of the acyl chains of lipids.⁷⁵⁾ Since PE, a cone-shaped lipid with a relatively small head group, induces lateral pressure inside the membrane, EPL membranes mainly comprising PE exhibited higher rigidity compared with PC membranes, resulting in a significantly lower insertion efficiency. In the presence of DAG (5 mol% to EPL), as the mobility of the core region was markedly reduced, the insertion was completely blocked (blue circle). Conversely, upon adding MPIase (1 mol%) to DAG-containing EPL liposomes (red circle), the insertion efficiency exceeded the expected value from the correlation plot. The addition of lyso-PC (5 mol%), an inverted-cone shaped lipid similar to MPIase, to DAG-containing EPL liposomes restored both membrane mobility and insertion efficiency (green circle); however, the correlation plotted on the regression line suggests that the effect of lyso-PC on insertion can be explained solely by alterations in the physicochemical properties of membranes. We hypothesize that the difference between MPIase and lyso-PC is attributed to the chaperone-like activity of the MPIase glycan. MPIase may capture a protein by its glycan and insert it into the loosened membrane.

10. Cooperative activity of MPIase with other integration factors

10.1. Cooperative activity with YidC

Although YidC is recognized as an insertase,^28),29) in our in vitro translational assay using DAG-containing proteoliposomes, YidC did not exhibit integration activity in the absence of MPIase. Conversely, MPIase promoted membrane protein integration independently of YidC.^17),38) However, when combined with MPIase, YidC accelerated membrane protein integration, particularly when the amount of substrate protein was increased (Fig. 11a).^76)-78) Thus, we hypothesize that YidC alone cannot overcome the blockade imposed by DAG, and that MPIase functions at the initial step of protein integration into the membrane, while YidC facilitates complete penetration at a subsequent stage. This proposed mechanism implies a functional interplay between MPIase and YidC. Consistent with this, MPIase expression was elevated following YidC depletion, strongly suggesting that MPIase partially compensates for the integration deficiency. Furthermore, a direct interaction between MPIase and YidC was confirmed through co-precipitation of MPIase with YidC in pull-down analyses.⁷⁹⁾

Fig. 11

(a) Normalized integration values of MPIase analogs reconstructed with or without YidC in substrate excess conditions. (Chemistry 29, e202300437 modified.) (b) Schematic model showing assistance of YidC and proton motive force (PMF). (c) SecYEG side-by-side dimers. SecY: purple; SecE: yellow; SecG: blue. (Proc. Natl. Acad. Sci. U.S.A. 110, 9734.)

Crystal structure analysis revealed that the TM helices of YidC form a hydrophilic groove that is accessible to both the lipid bilayer and the cytoplasm.^23),80),81) This structure suggests that YidC may receive substrate proteins, potentially facilitated by electrostatic interactions between positively charged residues within the YidC cavity and the negatively charged residues in substrate proteins. Such interactions may accelerate the dissociation of the protein from MPIase to complete integration.

The synergistic effect was validated in synthetic MPIase analogs longer than hexasaccharides (27 and 28) in conditions of substrate protein overproduction (Fig. 11a).^10),78) The potential for a direct interaction between MPIase and YidC may increase with glycan elongation. Carboxylic acids within MPIase might interact with numerous basic amino acids present in the highly flexible C1 region of YidC. Crystal structure analysis has revealed that the C1 region is situated approximately 2 nm from the IM on the cytoplasmic side^80),81); thus, a longer glycan may be required for this interaction. Alternatively, elongated glycan may promote the formation of functional domains on the membrane, potentially enhancing interaction and integration efficiency. Although direct evidence is lacking, it is plausible that multiple MPIase molecules accumulate on the membrane surface and form a domain where YidC may further augment cooperative efficiency.

10.2. Effect of proton motive force

The proton motive force (PMF) is defined as the sum of the proton gradient difference between the inside and outside of the membrane (ΔpH) and the membrane potential component (ΔΨ). PMF is known to affect membrane transport. The PMF serves as an energy source for biological processes such as intracellular substance transport. However, basic interactions between substrate proteins, MPIase, and YidC occurred even in the absence of a PMF, because the PMF was not imposed in our integration assay. Nevertheless, the function of YidC was powered by the PMF (Fig. 11b).⁷⁹⁾ Because the negative charge in the N-terminal region of the membrane protein is translocated to the periplasmic space through the positively charged cavity of YidC, PMF may accelerate such electrostatic interactions. Furthermore, the PMF stimulated MPIase-dependent integration in conditions of elevated substrate synthesis or increased negative charges in the N-terminal region of N-out topology substrate proteins. These findings suggest the sequential and cooperative roles of MPIase, YidC, and the PMF in the catalytic cycle of membrane protein integration.

10.3. Interaction with SecYEG

MPIase stimulates SecYEG-dependent integration and SecA-dependent preprotein translocation when co-reconstituted with SecYEG.^38),82) Similar to Sec-independent integration, MPIase may facilitate the membrane insertion of preprotein signal peptides. It is plausible that MPIase interacts with SecYEG, because MPIase coprecipitated with SecYEG.⁸²⁾ Additionally, we found that MPIase modifies the structure of the SecYEG dimer. Crystal structure analysis suggested that SecYEG forms a dimer in which SecE lies at the dimer interface (back-to-back structure).^83),84) Our cross-linking experiments with SecYEG overexpression conditions supported the formation of a SecE dimer. However, in the WT, the SecG dimer (side-by-side structure) was predominant (Fig. 11c).⁸²⁾ We hypothesize that SecYEG sufficiently interacts with MPIase in WT, whereas most SecYEG is unable to interact with MPIase when SecYEG is overproduced.

SecG stimulated preprotein translocation by undergoing a cycle of topology inversion, thereby promoting smoother structural changes in SecA.⁸⁵⁾ This inversion of SecG occurs exclusively when SecYEG adopts the side-by-side structure in the presence of MPIase.⁸²⁾ Given the promiscuous nature of the translocon in oligomerizing with dynamic associations in different complexes with various partner proteins, it is likely that MPIase plays a role in these interactions.

Fig. 12

Schematic showing the mechanism of membrane protein integration supported by MPIase. Because DAG inhibits spontaneous integration, a substrate protein immediately aggregates without MPIase. MPIase prepares the membrane environment, suppresses protein aggregation, targets proteins to the membrane surface, inserts the protein into the membrane, and delivers it to YidC.

11. Integration mechanism mediated by MPIase

We have evaluated the functions of MPIase in Sec-independent membrane protein integration in E. coli at the atomic level by various approaches. Figure 12 shows a schematic diagram of the integration by MPIase.^10),68),72) (1) The physicochemical property of the E. coli IM is arranged by its lipid components. Although DAG is a minor component, it significantly reduces membrane mobility, especially in the core region, and exerts a rapid flip-flop that hinders contact with substrate proteins inside the membrane. MPIase restores the mobility of the membrane core and promotes the formation of unfilled hydrophobic spaces, facilitating integration. (2) Without MPIase, hydrophobic substrate proteins cannot insert into the membrane and rapidly aggregate. In contrast, MPIase captures the nascent protein with its long glycans through multiple interactions using N- and O-Ac groups and carboxylic acids. (3) The protein alters its secondary structure during the rapid association and dissociation to prevent aggregation. (4) The protein is attracted to the membrane surface by electrostatic interactions between the strong negative charges of the pyrophosphate and the positive charges of the basic amino acids of the substrate. Lack of integration activity of ECA suggested that proteins captured at the tips of long glycans would not be able to approach the membrane surface without the strong charge of the pyrophosphate. (5) The hydrophobic region of the protein is immediately thrust into the loosened membrane through hydrophobic interactions with acyl chains of membrane lipids. (6) The protein is delivered to other factors, such as YidC, for complete integration if necessary. MPIase is regenerated to accept a new substrate, which enables the catalytic mechanism. The accumulation state of MPIase on the membrane and its localization in vivo remain to be elucidated. Given the inclusion of the translation process in our assay system, conducting kinetic analysis of membrane integration is challenging. Nonetheless, we are actively working on developing an SPR system to immobilize a membrane containing MPIase and introduce proteins as analytes, aiming to explore the cooperativity between MPIase and membrane lipids.

12. Conclusion and outlook

In conclusion, our studies have revealed a unique mechanism of Sec-independent membrane protein integration mediated by a glycolipid. The amino acid sequences of membrane proteins are diverse, in contrast to the highly specific interactions between lectins and glycans, so the interactions between MPIase and a substrate protein are weak, multiple, and relatively unspecific, leading to repeated binding and unbinding. MPIase is also involved in more intricate protein transport pathways, such as Sec-dependent integration and secretion; therefore, MPIase may direct various proteins to the membrane surface and deliver them to translocons or related factors. Further studies will provide a comprehensive understanding of the diverse roles of MPIase in each system. Although MPIase is not a proteinaceous molecule, it acts like a chaperone to prevent protein aggregation and transports proteins to the membrane surface. These findings provide deep insights into novel biological functions of glycolipids. Finally, whether glycolipids function similarly in other organisms, especially eukaryotes, remains to be addressed.

Acknowledgements

We are grateful to Dr. Toshiyuki Yamaguchi, Dr. Masahide Maeda, Dr. Ryohei Nagase (Suntory Foundation for Life Sciences), Prof. Masafumi Shionyu, Prof. Tsuyoshi Shirai, Prof. Takao Yoda (Nagahama Institute of Bio-Science and Technology), Dr. Taku Yoshiya, Dr. Hideki Nishio, Mr. Shun Masuda, Dr. Kumiko Yoshizawa-Kumagaye, Dr. Shugo Tsuda (Peptide Institute, Inc.), Prof. Kenichi Morigaki (Kobe University), and Prof. Toshinori Shimanouchi (Okayama University) for their contribution to the research on MPIase.

This work was supported by JSPS KAKENHI (grant numbers JP19H02843 and JP22H02213 to K.S.; JP18K06143 and JP22K05323 to K.N.; JP20K05738 and JP23K04955 to K.F.; JP22H02567, JP22K19262, JP22H05392 and JP23H04536 to K-i.N.).

Notes

Edited by Takao SEKIYA, M.J.A.

Correspondence should be addressed to: K. Shimamoto, 8-1-1 Seikadai, Seika-cho, Soraku-gun, Kyoto 619--0284, Japan (e-mail: [e-mail]).

References

• 1)  O'Keefe, S., Pool, M.R. and High, S. (2022) Membrane protein biogenesis at the ER: the highways and byways. FEBS J. 289, 6835-6862.
• 2)  Tsirigotaki, A., De Geyter, J., Šoštaric, N., Economou, A. and Karamanou, S. (2017) Protein export through the bacterial Sec pathway. Nat. Rev. Microbiol. 15, 21-36.
• 3)  Denks, K., Vogt, A., Sachelaru, I., Petriman, N.A., Kudva, R. and Koch, H.G. (2014) The Sec-translocon mediated protein transport in prokaryotes and eukaryotes. Mol. Membr. Biol. 31, 58-84.
• 4)  Nishiyama, K., Maeda, M., Abe, M., Kanamori, T., Shimamoto, K., Kusumoto, S. et al. (2010) A novel complete reconstitution system for membrane integration of the simplest membrane protein. Biochem. Biophys. Res. Commun. 394, 733-736.
• 5)  Nishiyama, K., Maeda, M., Yanagisawa, K., Nagase, R., Komura, H., Iwashita, T. et al. (2012) MPIase is a glycolipozyme essential for membrane protein integration. Nat. Commun. 3, 1260.
• 6)  Sawasato, K., Sato, R., Nishikawa, H., Iimura, N., Kamemoto, Y., Fujikawa, K. et al. (2019) CdsA is involved in biosynthesis of glycolipid MPIase essential for membrane protein integration in vivo. Sci. Rep. 9, 1372.
• 7)  Nishikawa, H., Sawasato, K., Mori, S., Fujikawa, K., Nomura, K., Shimamoto, K. et al. (2022) Interaction between glycolipid MPIase and proteinaceous factors during protein integration into the cytoplasmic membrane of E. coli. Front. Mol. Biosci. 9, 986602.
• 8)  Osawa, T., Fujikawa, K. and Shimamoto, K. (2024) Structures, functions, and syntheses of glycero-glycophospholipids. Front. Chem. 12, 1353688.
• 9)  Fujikawa, K., Suzuki, S., Nagase, R., Ikeda, S., Mori, S., Nomura, K. et al. (2018) Syntheses and activities of the functional structures of a glycolipid essential for membrane protein integration. ACS Chem. Biol. 13, 2719-2727.
• 10)  Fujikawa, K., Han, Y., Osawa, T., Mori, S., Nomura, K., Muramoto, M. et al. (2023) Structural requirements of a glycolipid MPIase for membrane protein integration. Chemistry 29, e202300437.
• 11)  Nomura, K., Yamaguchi, T., Mori, S., Fujikawa, K., Nishiyama, K.I., Shimanouchi, T., et al. (2019) Alteration of membrane physicochemical properties by two factors for membrane protein integration. Biophys. J. 117, 99-110.
• 12)  Nomura, K., Mori, S., Fujikawa, K., Osawa, T., Tsuda, S., Yoshizawa-Kumagaye, K. et al. (2022) Role of a bacterial glycolipid in Sec-independent membrane protein insertion. Sci. Rep. 12, 12231.
• 13)  Mori, S., Nomura, K., Fujikawa, K., Osawa, T., Shionyu, M., Yoda, T. et al. (2022) Intermolecular interactionsbetween a membraneprotein and a glycolipid essential for membrane proteinintegration. ACS Chem. Biol. 17, 609-618.
• 14)  Fujikawa, K., Mori, S., Nishiyama, K.I. and Shimamoto, K. (2022) A bacterial glycolipid essential for membrane protein integration. Adv. Carbohydr. Chem. Biochem. 81, 95-129.
• 15)  Opekarová, M. and Tanner, W. (2003) Specific lipid requirements of membrane proteins-a putative bottleneck in heterologous expression. Biochim. Biophys. Acta 1610, 11-22.
• 16)  Schneiter, R. and Toulmay, A. (2007) The role of lipids in the biogenesis of integral membrane proteins. Appl. Microbiol. Biotechnol. 73, 1224-1232.
• 17)  Kawashima, Y., Miyazaki, E., Müller, M., Tokuda, H. and Nishiyama, K. (2008) Diacylglycerol specifically blocks spontaneous integration of membrane proteins and allows detection of a factor-assisted integration. J. Biol. Chem. 283, 24489-24496.
• 18)  Blobel, G. (2000) Protein targeting (Nobel lecture). ChemBioChem 1, 86-102.
• 19)  Rapoport, T.A., Goder, V., Heinrich, S.U. and Matlack, K.E. (2004) Membrane-protein integration and the role of the translocation channel. Trends Cell Biol. 14, 568-575.
• 20)  Smets, D., Loos, M.S., Karamanou, S. and Economou, A. (2019) Protein transport across the bacterial plasma membraneby theSec pathway. Protein J. 38, 262-273.
• 21)  Kaushik, S., He, H. and Dalbey, R.E. (2022) Bacterial signal peptides-Navigating the journey of proteins. Front. Physiol. 13, 933153.
• 22)  Blaimschein, N., Parameswaran, H., Nagler, G., Manioglu, S., Helenius, J., Ardelean, C. et al. (2023) The insertase YidC chaperones the polytopic membrane protein MelB inserting and folding simultaneously from both termini. Structure 31, 1419-1430.
• 23)  Tsukazaki, T. (2019) Structural basis of the Sec-translocon and YidC revealed through X-ray crystallography. Protein J. 38, 249-261.
• 24)  Mateja, A. and Keenan, R.J. (2018) A structural perspective on tail-anchored protein biogenesis by the GET pathway. Curr. Opin. Struct. Biol. 51, 195-202.
• 25)  Chio, U.S., Cho, H. and Shan, S.O. (2017) Mechanisms of tail-anchored membrane protein targeting and insertion. Annu. Rev. Cell. Dev. Biol. 33, 417-438.
• 26)  Kiefer, D. and Kuhn, A. (1999) Hydrophobic forces drive spontaneous membrane insertion of the bacteriophage Pf3 coat protein without topological control. EMBO J. 18, 6299-6306.
• 27)  Chen, M., Samuelson, J.C., Jiang, F., Muller, M., Kuhn, A. and Dalbey, R.E. (2002) Direct interaction of YidC with the Sec-independent Pf3 coat protein during its membrane protein insertion. J. Biol. Chem. 277, 7670-7675.
• 28)  Kiefer, D. and Kuhn, A. (2018) YidC-mediated membrane insertion. FEMS Microbiol. Lett. 365, fny106.
• 29)  Hennon, S.W., Soman, R., Zhu, L. and Dalbey, R.E. (2015) YidC/Alb3/Oxa1 family of insertases.J. Biol. Chem. 290, 14866-14874.
• 30)  Dalbey, R.E., Kuhn, A., Zhu, L. and Kiefer, D. (2014) The membrane insertase YidC. Biochim. Biophys. Acta 1843, 1489-1496.
• 31)  Samuelson, J.C., Chen, M., Jiang, F., Mӧller, I., Wiedmann, M., Kuhn, A. et al. (2000) YidC mediates membrane protein insertion in bacteria. Nature 406, 637-641.
• 32)  Wickles, S., Singharoy, A., Andreani, J., Seemayer, S., Bischoff, L., Berninghausen, O. et al.. (2014) A structural model of the active ribosome-bound membrane protein insertase YidC. eLife 3, e03035.
• 33)  Kuhn, A. and Kiefer, D. (2017) Membrane protein insertase YidC in bacteria and archaea. Mol. Microbiol. 103, 590-594.
• 34)  White, S.H. and von Heijne, G. (2004) The machinery of membrane protein assembly. Curr. Opin. Struct. Biol. 14, 397-404.
• 35)  Lee, H. and Kim, H. (2014) Membrane topology of transmembrane proteins: determinants and experimental tools. Biochem. Biophys. Res. Commun. 453, 268-276.
• 36)  Bogdanov, M., Xie, J. and Dowhan, W. (2009) Lipid-protein interactions drive membrane protein topogenesis in accordance with the positive inside rule. J. Biol. Chem. 284, 9637-9641.
• 37)  van Bloois, E., Nagamori, S., Koningstein, G., Ullers, R.S., Preuss, M., Oudega, B. et al. (2005) The Sec-independent function of Escherichia coli YidC is evolutionary-conserved and essential. J. Biol. Chem. 280, 12996-13003.
• 38)  Nishiyama, K., Ikegami, A., Moser, M., Schiltz, E., Tokuda, H. and Muller, M. (2006) A derivative of lipid A is involved in signal recognition particle/SecYEG-dependent and -independent membrane integrations. J. Biol. Chem. 281, 35667-35676.
• 39)  Ding, B., Soblosky, L., Nguyen, K., Geng, J., Yu, X., Ramamoorthy, A. et al. (2013) Physiologically-relevant modes of membrane interactions by the human antimicrobial peptide, LL-37, revealed by SFG experiments. Sci. Rep. 3, 1854.
• 40)  Nakamura, S., Suzuki, S., Saito, H. and Nishiyama, K.I. (2018) Cholesterol blocks spontaneous insertion of membrane proteins into liposomes of phosphatidylcholine. J. Biochem. 163, 313-319.
• 41)  Nishikawa, H., Sasaki, M. and Nishiyama, K.I. (2020) In vitro assay for bacterialmembrane protein integration into proteoliposomes. Bio Protoc. 10, e3626.
• 42)  Shimizu, Y., Inoue, A., Tomari, Y., Suzuki, T., Yokogawa, T., Nishikawa, K. et al. (2001) Cell-free translation reconstituted with purified components. Nat. Biotechnol. 19, 751-755.
• 43)  Mäkelä, P.H. and Mayer, H. (1976) Enterobacterial common antigen. Bacteriol. Rev. 40, 591-632.
• 44)  Kuhn, H.M., Meier-Dieter, U. and Mayer, H. (1988) ECA, the enterobacterial common antigen. FEMS Microbiol. Rev. 4, 195-222.
• 45)  Rai, A.K. and Mitchell, A.M. (2020) Enterobacterial common antigen: synthesisand function of an enigmatic molecule. mBio 11 e01914-20.
• 46)  Rick, P.D., Hubbard, G.L., Kitaoka, M., Nagaki, H., Kinoshita, T., Dowd, S. et al. (1998) Characterization of the lipid-carrier involved in the synthesis of enterobacterial common antigen (ECA) and identification of a novel phosphoglyceride in a mutant of Salmonella typhimurium defective in ECA synthesis. Glycobiology 8, 557-567.
• 47)  Meier-Dieter, U., Starman, R., Barr, K., Mayer, H. and Rick, P.D. (1990) Biosynthesis of enterobacterial common antigen in Escherichia coli. Biochemical characterization of Tn10 insertion mutants defective in enterobacterial common antigen synthesis. J. Biol. Chem. 265, 13490-13497.
• 48)  Barr, K. and Rick, P.D. (1987) Biosynthesis of enterobacterial common antigen in Escherichia coli. In vitro synthesis of lipid-linked intermediates. J. Biol. Chem. 262, 7142-7150.
• 49)  Barr, K., Ward, S., Meier-Dieter, U., Mayer, H. and Rick, P.D. (1988) Characterization of an Escherichia coli rff mutant defective in transfer of N-acetylmannosaminuronic acid (ManNAcA) from UDP-ManNAcA to a lipid-linked intermediate involved in enterobacterial common antigen synthesis. J. Bacteriol. 170, 228-233.
• 50)  Barr, K., Nunes-Edwards, P. and Rick, P.D. (1989) In vitro synthesis of a lipid-linked trisaccharide involved in synthesis of enterobacterial common antigen. J. Bacteriol. 171, 1326-1332.
• 51)  Rick, P.D., Barr, K., Sankaran, K., Kajimura, J., Rush, J.S. and Waechter, C.J. (2003) Evidence that the wzxE gene of Escherichia coli K-12 encodes a protein involved in the transbilayer movement of a trisaccharide-lipid intermediate in the assembly of enterobacterial common antigen. J. Biol. Chem. 278, 16534-16542.
• 52)  Barr, K., Klena, J. and Rick, P.D. (1999) The modality of enterobacterial common antigen polysaccharide chain lengths is regulated by o349 of the wec gene cluster of Escherichia coli K-12. J. Bacteriol. 181, 6564-6568.
• 53)  Kajimura, J., Rahman, A. and Rick, P.D. (2005) Assembly of cyclic enterobacterial common antigen in Escherichia coli K-12. J. Bacteriol. 187, 6917-6927.
• 54)  Morris, K.N. and Mitchell, A.M. (2023) Phosphatidylglycerol is the lipid donor for synthesis of phospholipid-linked enterobacterial common antigen. J. Bacteriol. 205, e0040322.
• 55)  Kajimura, J., Rahman, A., Hsu, J., Evans, M.R., Gardner, K.H. and Rick, P.D. (2006) O acetylation of the enterobacterial common antigen polysaccharide is catalyzed by the product of the yiaH gene of Escherichia coli K-12. J. Bacteriol. 188, 7542-7550.
• 56)  Jorgenson, M.A., Kannan, S., Laubacher, M.E. and Young, K.D. (2016) Dead-end intermediates in the enterobacterial common antigen pathway induce morphological defects in Escherichia coli by competing for undecaprenyl phosphate. Mol. Microbiol. 100, 1-14.
• 57)  Kamemoto, Y., Funaba, N., Kawakami, M., Sawasato, K., Kanno, K., Suzuki, S. et al. (2020) Biosynthesis of glycolipid MPIase (membrane protein integrase) is independent of the genes for ECA (enterobacterial common antigen). J. Gen. Appl. Microbiol. 66, 169-174.
• 58)  Sato, R., Sawasato, K. and Nishiyama, K.I. (2019) YnbB is a CdsA paralogue dedicated to biosynthesis of glycolipid MPIase involved in membrane protein integration. Biochem. Biophys. Res. Commun. 510, 636-642.
• 59)  Hikage, R., Sekiya, Y., Sawasato, K. and Nishiyama, K.I. (2024) CdsA, a CDP-diacylglycerol synthase involved in phospholipid and glycolipid MPIase biosynthesis, possesses multiple initiation codons. Genes Cells 29, 347-355.
• 60)  Tamura, Y., Harada, Y., Nishikawa, S., Yamano, K., Kamiya, M., Shiota, T. et al. (2013) Tam41 is a CDP-diacylglycerol synthase required for cardiolipin biosynthesis in mitochondria. Cell Metab. 17, 709-718.
• 61)  Sawasato, K., Suzuki, S. and Nishiyama, K.I. (2019) Increased expression of the bacterial glycolipid MPIase is required for efficient protein translocation across membranes in cold conditions. J. Biol. Chem. 294, 8403-8411.
• 62)  Sawasato, K., Sekiya, Y. and Nishiyama, K.I. (2019) Two-step induction of cdsA promoters leads to upregulation of the glycolipid MPIase at cold temperature. FEBS Lett. 593, 1711-1723.
• 63)  Kamemoto, Y., Hikage, R., Han, Y., Sekiya, Y., Sawasato, K. and Nishiyama, K.I. (2023) Coordinated upregulation of two CDP-diacylglycerol synthases, YnbB and CdsA, is essential for cell growth and membrane protein export in the cold. FEMS Microbiol. Lett. 370.
• 64)  Liu, L., Zha, J., DiGiandomenico, A., McAllister, D., Stover, C.K., Wang, Q. et al. (2015) Synthetic enterobacterial common antigen (ECA) for the development of a universal immunotherapy for drug-resistant enterobacteriaceae. Angew. Chem. Int. Ed. Engl. 54, 10953-10957.
• 65)  Paulsen, H. and Lorentzen, J.P. (1986) [Synthesis of trisaccharide determinants of enterobacterial common antigens]. Carbohydr. Res. 150, 63-90.
• 66)  Bartra, M., Romea, P., Urpi, F. and Vilarrasa, J. (1990) A fast procedure for the reduction of azides and nitro compounds based on the reducing ability of Sn(SR)₃-species. Tetrahedron 46, 587-594.
• 67)  Zulueta, M.M.L., Lin, S.-Y., Lin, Y.-T., Huang, C.-J., Wang, C.-C., Ku, C.-C. et al. (2012) α-Glycosylation by d-glucosamine-derived donors: Synthesis of heparosan and heparin analogues that interactwith mycobacterial heparin-binding hemagglutinin. J. Am. Chem. Soc. 134, 8988-8995.
• 68)  Mori, S., Shionyu, M., Shimamoto, K. and Nomura, K. (2024) Bacterial glycolipid acting on protein transport across membranes. ChemBioChem 25, e202300808.
• 69)  Markgren, P.-O., Schaal, W., Hämäläinen, M., Karlén, A., Hallberg, A., Samuelsson, B. et al. (2002) Relationships between structure and Iinteraction kinetics for HIV-1 protease inhibitors. J. Med. Chem. 45, 5430-5439.
• 70)  Molecular Operating Environment (MOE). (2022) 2022.02 ed. Chemical Computing Group ULC, Montreal, QC, Canada.
• 71)  Strandberg, E. and Ulrich, A.S. (2015) AMPs and OMPs: Is the folding and bilayer insertion of β-stranded outer membrane proteins governed by the same biophysical principles as for α-helical antimicrobial peptides? Biochim. Biophys. Acta 1848, 1944-1954.
• 72)  Nomura, K., Mori, S. and Shimamoto, K. (2024) Roles of a glycolipid MPIase in Sec-independent membrane protein insertion. Membranes (Basel) 14, 48.
• 73)  Wang, Y. and Jardetzky, O. (2002) Probability-based protein secondary structure identification using combined NMR chemical-shift data. Protein Sci. 11, 852-861.
• 74)  Wang, Y. and Jardetzky, O. (2002) Probability-based protein secondary structure identification using combined NMR chemical-shift data. Protein Sci. 11, 852-861.
• 75)  Lentz, B.R. (1993) Use of fluorescent probes to monitor molecular order and motions within liposome bilayers. Chem. Phys. Lipids 64, 99-116.
• 76)  Nishikawa, H., Kanno, K., Endo, Y. and Nishiyama, K.I. (2021) Ring assembly of c subunits of F₀F₁-ATP synthase in Propionigenium modestum requires YidC and UncI following MPIase-dependent membrane insertion. FEBS Lett. 595, 647-654.
• 77)  Nishikawa, H., Sasaki, M. and Nishiyama, K.I. (2017) Membrane insertion of F₀ c subunit of F₀F₁ ATPase depends on glycolipozyme MPIase and is stimulated by YidC. Biochem. Biophys. Res. Commun. 487, 477-482.
• 78)  Sasaki, M., Nishikawa, H., Suzuki, S., Moser, M., Huber, M., Sawasato, K. et al. (2019) The bacterial protein YidC accelerates MPIase-dependent integration of membrane proteins. J. Biol. Chem. 294, 18898-18908.
• 79)  Endo, Y., Shimizu, Y., Nishikawa, H., Sawasato, K. and Nishiyama, K.I. (2022) Interplay between MPIase, YidC, and PMF during Sec-independent insertion of membrane proteins. Life Sci. Alliance 5, e202101162.
• 80)  Kumazaki, K., Kishimoto, T., Furukawa, A., Mori, H., Tanaka, Y., Dohmae, N. et al. (2014) Crystal structure of Escherichia coli YidC, a membrane protein chaperone and insertase. Sci. Rep. 4, 7299.
• 81)  Kumazaki, K., Chiba, S., Takemoto, M., Furukawa, A., Nishiyama, K., Sugano, Y. et al. (2014) Structural basis of Sec-independent membrane protein insertion by YidC. Nature 509, 516-520.
• 82)  Moser, M., Nagamori, S., Huber, M., Tokuda, H. and Nishiyama, K. (2013) Glycolipozyme MPIase is essential for topology inversion of SecG during preprotein translocation. Proc. Natl. Acad. Sci. U.S.A. 110, 9734-9739.
• 83)  Van den Berg, B., Clemons, W.M., Jr., Collinson, I., Modis, Y., Hartmann, E., Harrison, S.C. et al. (2004) X-ray structure of a protein-conducting channel. Nature 427, 36-44.
• 84)  Breyton, C., Haase, W., Rapoport, T.A., Kühlbrandt, W. and Collinson, I. (2002) Three-dimensional structure of the bacterial protein-translocation complex SecYEG. Nature 418, 662-665.
• 85)  Nishiyama, K., Suzuki, T. and Tokuda, H. (1996) Inversion of the membrane topology of SecG coupled with SecA-dependent preprotein translocation. Cell 85, 71-81.

Non-standard abbreviation list

2D

two-dimensional

Ac

acetyl

Bn

benzyl

Bz

benzoyl

CD

circular dichroism

CDP

cytidine diphosphate

DAG

diacylglycerol

DGP

diacylglycerol phosphate

DHPC

1,2-diheptanoyl- sn-glycero-3-phosphocholine

DPH

1,6-diphenyl-1,3,5-hexatriene

DS

docking simulation

ECA

enterobacterial common antigen

E. coli

Escherichia coli

EPL

E. coli phospholipid

ER

endoplasmic reticulum

Fuc4NAc

N-acetyl-4-aminofucose

GC

gas chromatography

GlcNAc

N-acetyl-glucosamine

IM

inner membrane

INV

inverted inner membrane vesicle

K _D

dissociation constant

LPS

lipopolysaccharide

MALDI

matrix-assisted laser desorption/ionization

ManNAcA

N-acetyl-mannosaminuronate

MPIase

membrane protein integrase

MtlA

mannitol permease

NMR

nuclear magnetic resonance

NOESY

nuclear Overhauser effect spectroscopy

PC

phosphatidylcholine

PE

phosphatidylethanolamine

PG

phosphatidylglycerol

PMF

proton motive force

PS

phosphatidylserine

Q-TOF-MS

quadrupole time-of-flight mass spectrometry

RNC

ribosome-nascent chain complex

SPR

surface plasmon resonance

SR

signal recognition particle receptor

SRP

signal recognition particle

ssNMR

solid-state nuclear magnetic resonance

STD

saturation transfer difference

TA

tail-anchored

TBDPS

tert-butyldiphenylsilyl

TLC

thin-layer chromatography

TM

transmembrane

UDP

uridine diphosphate

Und-P

undecaprenyl phosphate

Und-PP

undecaprenyl pyrophosphate

WT

wild-type

Profile

Keiko Shimamoto studied organic synthesis at Osaka University and received her MS in 1986. She joined Suntory Institute for Bioorganic Research (SUNBOR) in 1986. She received her PhD degree from Osaka University in 1991. She is working at Suntory Foundation for Life Sciences, Bioorganic Research Institute as an executive researcher and is a specially associated professor at Osaka University Graduate School of Science. Her research interests lie in the investigation of the functions of bioactive amino acids and glycolipids based on the syntheses of novel molecular probes.

Kohki Fujikawa graduated from Gifu University in 2005. He received his PhD degree from Gifu University in 2010. After 2 years of postdoctoral fellow at the University of Missouri St. Louis, USA, he joined the JST ERATO Ito glycotrilogy project as a postdoctoral researcher from 2012 to 2015. Currently, he is a researcher at Suntory Foundation for Life Sciences, Bioorganic Research Institute (SUNBOR). His research interests include the synthesis of bioactive glycans and elucidation of their function.

Tsukiho Osawa studied organic synthesis at Tokushima University and received her MS in 2017. After working at Nippon Soda Co., Ltd. (2017-2019), she joined Suntory Foundation for Life Sciences, Bioorganic Research Institute (SUNBOR) in 2019. She engages in research on glycolipids based on the synthesis of novel molecular probes.

Shoko Mori studied organic chemistry at Osaka University and received her MS in 2013. She joined Suntory Foundation for Life Sciences, Bioorganic Research Institute (SUNBOR) in 2013. She received her PhD degree from Osaka University in 2022. She engages in research on the intermolecular interaction of biomolecules and the structural analysis of natural products based on physicochemical techniques such as NMR.

Kaoru Nomura is a senior researcher at Suntory Foundation for Life Science, Bioorganic Research Institute (SUNBOR). She received her PhD from Kyoto University in 1999. After working as a Japan Society for the Promotion of Science (JSPS) postdoctoral fellow at Tokyo Metropolitan University (1999-2002), she joined SUNBOR in 2002. Her research focuses on elucidating the roles and functional mechanisms of biomembrane-active molecules, such as glycolipids, GPI-anchored proteins, and antimicrobial peptides, using solid-state NMR studies.

Ken-ichi Nishiyama studied biochemistry at the University of Tokyo. After receiving his PhD degree in 1994, he continued his work as a JSPS postdoctoral fellow, assistant professor, and associate professor at the University of Tokyo until 2009. From 2002 to 2004, he worked with Professor Matthias Müller at the University of Freiburg, Germany, as an EMBO long-term fellow. He moved to Iwate University as a professor in 2010. His research interest is to elucidate the molecular mechanisms underlying protein export such as membrane protein integration and preprotein translocation.