Lipid Bilayer Nanodisc Platform for Investigating Polyprenol-Dependent

Home , Polyprenol

Lipid bilayer nanodisc platform for investigating INAUGURAL ARTICLE polyprenol-dependent enzyme interactions and activities

Meredith D. Hartley1, Philipp E. Schneggenburger1, and Barbara Imperiali2

Department of Biology and Department of Chemistry, Massachusetts Institute of Technology, Cambridge, MA 02139

This contribution is part of the special series of Inaugural Articles by members of the National Academy of Sciences elected in 2010.

Contributed by Barbara Imperiali, November 5, 2013 (sent for review September 13, 2013) Membrane-bound polyprenol-dependent pathways are important icance, particularly because extended polyprenols have no other for the assembly of essential glycoconjugates in all domains of life. known role in cells (2, 4, 5). However, despite their prevalence, the functional significance of The discovery (8) and biochemical investigation (9–11) of the extended linear polyprenyl groups in the interactions of the N-linked protein glycosylation in the Gram-negative bacterium glycan substrates, the biosynthetic enzymes that act upon them, Campylobacter jejuni has revealed a canonical polyprenol-dependent and the membrane bilayer in which they are embedded remains glycan assembly pathway. The C. jejuni protein glycosylation (pgl) a mystery. These interactions are investigated simultaneously and pathway is an appealing subject for in-depth analysis because the uniquely through application of the nanodisc membrane technol- component enzymes can be heterologously expressed and purified ogy. The Campylobacter jejuni N-linked glycosylation pathway has in good yields, and can be subjected to protein engineering been chosen as a model pathway in which all of the enzymes and approaches for the introduction of tags and labels. As illustrated in substrates are biochemically accessible. We present the functional Fig. 1A,inthepgl pathway, a heptasaccharide is assembled onto reconstitution of two enzymes responsible for the early mem- undecaprenyl-diphosphate (Und-P-P) by the sequential action of brane-committed steps in glycan assembly. Protein stoichiometry five membrane proteins, designated as PglC, PglA, PglJ, PglH, analysis, fluorescence-based approaches, and biochemical activity and PglI. Of these proteins, PglC and PglI are predicted to be BIOCHEMISTRY assays are used to demonstrate the colocalization of the two integral membrane proteins (12), whereas the other glycan assembly enzymes in nanodiscs. Isotopic labeling of the substrates reveals enzymes (PglA, PglJ, and PglH) lack discrete transmembrane that undecaprenyl-phosphate is coincorporated into discs with the domains (TMDs) and are peripheral membrane proteins. Both two enzymes, and furthermore, that both enzymes are function- the integral and peripheral membrane Pgl proteins form in- ally reconstituted and can sequentially convert the coembedded soluble aggregates in the absence of detergent, which further undecaprenyl-phosphate into undecaprenyl-diphosphate-linked di- supports a model wherein these enzymes are recruited to the saccharide. These studies provide a proof-of-concept demonstrating membrane to collaborate in the sequential glycan assembly process. After biosynthesis of the undecaprenyl-diphosphate hepta- that the nanodisc model membrane system represents a promising fl experimental platform for analyzing the multifaceted interactions saccharide is complete, a ippase (PglK) translocates the assembled among the enzymes involved in polyprenol-dependent glycan as- product to the periplasmic face of the inner membrane (13), where sembly pathways, the membrane-associated substrates, and the the oligosaccharyltransferase (PglB) transfers the assembled glycan lipid bilayer. The stage is now set for exploration of the roles of to asparagine residues in selected acceptor proteins (10, 14). After translocation through the outer membrane, glycosylated proteins the conserved polyprenols in promoting protein–protein interactions among pathway enzymes and processing of substrates through se- fi quential steps in membrane-associated glycan assembly. Signi cance

ellular membranes accommodate abundant biological activ- Linear polyprenols are recurring molecular components in the Cities, including the transport of small molecules and proteins, biosynthetic pathways responsible for the assembly of essen- energy production, and multistep biosynthetic transformations. tial glycoconjugates, including peptidoglycan and N-linked These functions are crucial for cell viability; however, studying glycoproteins. Despite their highly conserved presence in all these processes in biophysical and biochemical detail is chal- domains of life, the role of the extended linear polyprenyl lenging because of the complexity of working with both integral groups in the dynamics of membrane-bound glycan assembly pathways remains a mystery. Here we apply the nanodisc and peripheral membrane proteins in lipid bilayer systems. An model membrane platform to simultaneously assess the inter- important class of membrane-associated pathways involves the actions and activities of the polyprenyl-linked substrates, assembly of complex glycoconjugates, which is dependent on enzymes, and lipid bilayer by investigating initial steps from extended linear polyprenols (1–4). The products of these path- the Campylobacter jejuni N-linked glycosylation pathway. This ways are essential for cellular viability in all domains of life and, work represents a proof-of-concept demonstrating that intriguingly, the polyprenols that are used in glycan assembly nanodiscs can be used for the precise manipulation and study vary between organisms, with considerable differences in the of polyprenol-dependent pathways. overall length and degrees of unsaturation (2, 5). For example, bacteria generate peptidoglycan components on the membrane- Author contributions: M.D.H., P.E.S., and B.I. designed research; M.D.H. and P.E.S. per- anchored undecaprenyl-diphosphate before cell wall assembly formed research; M.D.H. and P.E.S. contributed new reagents/analytic tools; M.D.H., P.E.S., (6), whereas eukaryotes produce glycans for N-linked protein and B.I. analyzed data; and M.D.H., P.E.S., and B.I. wrote the paper. glycosylation via dolichyl-diphosphate–linked intermediates, which The authors declare no conﬂict of interest. feature polyprenols ranging from 14–25 isoprene units (1, 2, 7). See Proﬁle on page 20850. Despite the ubiquitous presence of linear polyprenols as glycan 1M.D.H. and P.E.S. contributed equally to this work. carriers in these important biosynthetic pathways, it remains un- 2To whom correspondence should be addressed. E-mail: [email protected]. clear why these structures have been so faithfully conserved This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10. throughout evolution and what might be their functional signif- 1073/pnas.1320852110/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1320852110 PNAS | December 24, 2013 | vol. 110 | no. 52 | 20863–20870 Downloaded by guest on September 28, 2021 A have also been prepared (26), although the current versions of these discs have been shown to exhibit greater polydispersity. A variety of synthetic lipids have been used for nanodisc preparation, including variants of phosphatidylcholine (PC), phosphati- dylethanolamine (PE), and phosphatidylglycerol (PG), as well as H – J P Escherichia coli lipid extracts (25, 27 29). Diverse membrane P A P C PP PP I proteins have been incorporated into nanodiscs and studies on K these systems have revealed fascinating nuances of protein structure, protein oligomeric state, protein–protein interactions, sub- PP B strate binding, and enzyme activity (30–36). In addition, particularly noteworthy for the current study, the effects of lipid composition on enzymes have been probed in a highly reproducible manner P relative to other membrane mimetic systems (37, 38). To address the fundamental challenge of defining the physical PP and biochemical roles of polyprenols in membrane-associated pathways, we present the development and validation of an B experimental strategy to reconstitute the first two membrane- committed steps of the N-linked glycosylation pathway of C. jejuni (9, 11) in membrane bilayer nanodiscs (28). The approach in this study renders the simultaneous and controlled manipulation of all three determining elements in polyprenol-dependent glycan PglC assembly at the membrane bilayer a tangible objective. Results PglA Incorporation and Characterization of PglC and PglA in Membrane Bilayer Nanodiscs. PglC is a phosphoglycosyltransferase containing one predicted N-terminal TMD (11). PglC catalyzes the Fig. 1. Canonical reaction pathway of bacterial N-linked glycosylation in transfer of phospho-N,N′-diacetylbacillosamine from UDP-N, C. jejuni.(A) Scheme of the membrane-bound enzymatic conversions that N′-diacetylbacillosamine (UDP-diNAcBac) to undecaprenyl- produce the polyprenyl-diphosphate–linked glycan, which is used in protein phosphate (Und-P) (Fig. 1B). For these studies a GB1 (B1 glycosylation. (B) The reactions of the first two enzymes in the C. jejuni domain of Ig binding protein G)–PglC fusion construct was pathway, PglC and PglA, are shown. engineered in which a His6-tag was inserted N-terminal to the GB1 domain (39). The GB1 domain vastly improved the expression and purification of PglC, which was critical for these are displayed on the cellular surface and appear to be involved in studies. The His6-GB1-PglC construct expresses well in E. coli (2– adhesion, colonization, and host-cell invasion (15). 5 mg/L cell culture) and can be solubilized from membrane lipids One of the major hurdles in deciphering the discrete bio- with detergent and purified using Ni-NTA affinity chromatogra- physical and biochemical roles of polyprenols in membrane- phy for reconstitution in the nanodisc system (SI Text). associated pathways is implementing methods to simultaneously PglA is a glycosyltransferase that catalyzes the transfer of GalNAc investigate all three components involved in polyprenol action: from UDP-GalNAc to form Und-P-P-diNAcBac-GalNAc (9) (Fig. the lipid bilayer, membrane-associated proteins, and polyprenyl- 1B). Although there are no predicted TMDs for PglA (12), size- linked substrates. Prior studies on polyprenol-dependent path- exclusion chromatographic (SEC) analysis of the His -tobacco etch ways have principally focused on redacted experimental systems 8 virus (TEV)-PglA construct in buffer without detergent reveals representing two of the three key variables. For example, the that the protein aggregates. Based on this behavior and the fact effect of varying the substrate polyprenol groups has been studied using glycan assembly enzymes that have been solubilized from that PglA interacts with a membrane-bound substrate (Und-PP- the native lipid bilayer with detergent (16–18); and interactions diNAcBac), PglA is designated as a peripheral membrane protein. of polyprenyl-linked compounds with model membrane systems, Well-behaved monodisperse preparations of PglA can be achieved in the presence of β-D-dodecylmaltoside at 0.05% (wt/vol). Typical such as liposomes, in the absence of membrane-bound proteins fi have been investigated using biophysical techniques, including yields of PglA after puri cation and TEV protease treatment for computational modeling and NMR spectroscopy (19–24). His8-tag removal range from 2 to 4 mg/L cell culture (SI Text). To gain insight into the interactions among the C. jejuni glycan Nanodisc preparation was carried out following literature assembly pathway enzymes and the specific roles of the poly- protocols (29) and is illustrated in Fig. 2A for the preparation of prenyl-linked substrates, an ideal model system would allow for nanodiscs involving PglC (PglC-NDs). A polar E. coli lipid ex- reconstitution of the entire biosynthetic pathway in a native mem- tract with a ratio PE/PG/cardiolipin of 67:23:10 (wt %) was used brane environment. Importantly, the system should be flexible because the phospholipid composition of E. coli is expected to be enough to permit structure/activity studies of enzymes with dif- similar to that of C. jejuni, as both are Gram-negative bacteria. ferent membrane components and polyprenyl-linked substrates The MSP1E3 variant used in these studies forms nanodiscs with in a systematic manner. In addition, the system should be ame- a Stokes hydrodynamic diameter of 12.1 nm, as measured pre- nable to detailed biophysical analysis without the constraints viously by SEC, or a diameter of 12.9 nm, as measured by small- imposed by the application of detergent micelles or polydisperse angle X-ray scattering (25). The His6-TEV-MSP1E3 construct liposome structures. was expressed, purified, and subjected to TEV proteolysis to With these parameters in mind, we have turned to nanodiscs, afford high yields of purified protein (60 mg/L cell culture). For a valuable and emerging membrane mimetic technology in- the PglC-ND preparation, a 10:1 MSP:PglC (5:1 ND:PglC) ratio troduced by the Sligar laboratory (25). In nanodiscs, a membrane afforded a discrete population of discs, which could be purified scaffold protein (MSP), derived from the cholesterol transport away from empty discs and the remaining nanodisc assembly protein apolipoprotein A1 (ApoA-1), was engineered to produce components via Ni-NTA chromatography using the unique His6- stable self-assembled structures. Two MSPs form a lariat, which tag on PglC (Fig. 2B). To estimate the number of PglCs per disc, solubilizes the hydrophobic edge of a flat bilayer of lipids to gel densitometry by quantitative Coomassie staining was carried afford soluble assemblies with diameters ranging from 7 nm to out and revealed the presence of one PglC species per nanodisc 12 nm. Larger nanodiscs, with diameters ranging from 16–25 nm, assembly (Fig. 3A and Fig. S1). In addition, SEC analysis of the

20864 | www.pnas.org/cgi/doi/10.1073/pnas.1320852110 Hartley et al. Downloaded by guest on September 28, 2021 PglC-NDs kDa FT W1 W2 W3 E1 E2 E3 E4 ABP 50 INAUGURAL ARTICLE lipids 36 PglC Und-P MSP 22 PglC-NDs incubation Ni-NTA His6 -PglC detergent purification PglC/PglA-NDs kDa withdrawal of PglC-NDs 50 PglA PglA MSP 36 PglC MSP 22 PglC/PglA-NDs Nanodisc assembly mixture purified Nanodiscs

Fig. 2. Protein reconstitution in nanodiscs. (A) Nanodisc self-assembly is initiated by removal of detergent. Nanodiscs are puriﬁed using Ni-NTA chroma-

tography, which binds the unique His6-tag on PglC-NDs. (B) SDS/PAGE analysis of Ni-NTA fractions (E, elution fractions; FT, ﬂow through; W, wash) from a PglC-NDs preparation (Upper) and a PglC/PglA-NDs preparation (Lower).

Ni-NTA–purified nanodiscs showed an elution profile consistent labels (Fig. 4A) to probe the cofacial organization of the two with a PglC-ND assembly (Fig. 3B and Fig. S2). reconstituted proteins in the nanodisc assembly using FRET. Nanodiscs were also prepared in the presence of PglC and PglA Based on estimated distances in the PglC/PglA-ND (Fig. 4B), the at an MSP:PglC:PglA ratio of 10:1:1 (5:1:1 ND/PglC/PglA) and known tetramethylrhodamine (TAMRA) and Cyanine5 (Cy5) fi subjected to af nity chromatography on Ni-NTA. In this case, the FRET-pair (R0 ∼ 6.5 nm) was selected. For fluorophore labeling, only nanodisc species binding to the stationary phase are those PglC lacks native Cys residues and therefore single Cys muta- containing His6-GB1-PglC, because the affinity tags of all other tions were introduced to generate two mutants for site-specific components were removed previously. At this stage, two possible thiol-directed modification using a Cy5-maleimide derivative (SI outcomes were anticipated: Either there would be a statistical Text and Fig. S3). The labeling sites were selected such that the distribution of PglA among all of the discs, and therefore only Cy5 would modify either the globular domain of PglC [PglC(Cy5- 20% of the discs isolated after Ni-NTA purification would contain globular)] or the N terminus [PglC(Cy5-terminal)], which would both PglC and PglA; or alternatively, a positive interaction be- be localized on the distal side of the membrane bilayer (Fig. 4B BIOCHEMISTRY tween PglC and PglA in the discs would bias the distribution. and Fig. S3). The chemical labeling proceeded to ∼50% con- When the purified nanodiscs were analyzed by gel densitometry, version and the precise labeling efficiency was quantified fol- the PglC:PglA ratio was 1.0:0.7 (Fig. 3A). This result implies that lowing reconstitution of the PglC mutants into nanodiscs. With PglA is colocalized with PglC in ∼70% of the population. respect to PglA, the native sequence includes six Cys residues, Dynamic light scattering (DLS) was applied to estimate the and therefore an alternate labeling strategy was adopted. La- hydrodynamic radii of the various nanodisc assemblies (Fig. 3C). beling of PglA (Fig. 4A) with the TAMRA fluorophore was The measurements show particles with average hydrodynamic achieved by sortase-mediated ligation (40); this method guar- particle radii of 6.6 nm for empty nanodiscs, 12.9 nm for PglC- anteed quantitative labeling of the donor species, which is NDs, and 15.4 nm PglC/PglA-NDs, which is consistent with the beneficial in a static ensemble FRET experiment. For this effect of systematically adding proteins to the nanodisc assem- process, a PglA variant with a C-terminal sortase A recognition blies. As a control, PglC and MSP in the absence of phospho- sequence (LPETG) and a TAMRA-labeled, sortase-active pep- lipids did not afford detectable particles, and it is likely that the tide [GGGYK(TAMRA)KG] were reacted to yield the modified protein components precipitated under detergent withdrawal PglA(TAMRA). The C terminus was chosen for labeling because conditions. We note that the measured polydispersities are addition of an N-terminal peptide tag had been found to disrupt somewhat high, which may indicate the presence of a population the association of PglA with PglC-NDs (Fig. S4). Fluorescence of aggregates, which has been documented previously with DLS labeling was carried out as summarized in Fig. 4A and presented measurements on nanodisc assemblies (33). in the SI Text. Fluorescently labeled proteins were reconstituted into nanodiscs Fluorescence Labeling of PglC and PglA and FRET Analysis. PglC and using the approaches developed for the unlabeled species. Fluo- PglA were engineered to incorporate site-specific fluorescent rescent PglC/PglA-NDs were adjusted to the same concentration

MSP (std.) PglC (std.)* * all values in (pmoles) A B 1.6 PglC NDs 3 7 12 16 2 5 8 11 ND

PglC 1.2 MSP ) . u PglC NDs . 0.8 (a MSP (std.)* PglC (std.)* PglA (std.)* . void

Abs Fig. 3. Characterization of nanodiscs. (A)Quan- 9 13.5 18 7 11 15 4 7 10 ND 0.4 PglA tiﬁcation of PglC-NDs and PglC/PglA-NDs by gel ﬁ

/PglA NDs PglC densitometry using quanti ed protein standards 0.0 C MSP of MSP, PglC, and PglA. The estimated ratios were

Pgl 0 5 10 15 20 25 MSP/PglC = 11.4/6.1 pmoles and MSP/PglC/PglA = V (mL) 14.0/8.9/6.1 pmoles. (B) SEC analysis of PglC-NDs. 50 r = 6.6 nm (99.9 % mass, 40 r = 12.9 nm (99.9 % mass, 40 r = 15.4 nm, r = 0.1 nm (99.5 % mass, C 67 % Pd) 47 % Pd) (100 % mass, 80 8 % Pd) 40 The void volume is indicated. (C)DLSanalysisof 30 30 49 % Pd)

s 60 s (left to right): empty nanodiscs, PglC-NDs, PglC/ 30 s 20 mass 20 40 PglA-NDs, and a nanodisc preparation mixture 20 % %ma % mas %mass (PglC-NDs) lacking the lipid components (negative 10 10 20 10 control). The average particle radius (r), the per- 0 0 0 0 0.1 110100 103 0.1 1 10 100 103 0.1 1 10 100 103 0.1 1 10 100 103 centage of the total analyzed mass, and the Diameter (nm) Diameter (nm) Diameter (nm) Diameter (nm) polydispersity (Pd) are displayed in each panel.

Hartley et al. PNAS | December 24, 2013 | vol. 110 | no. 52 | 20865 Downloaded by guest on September 28, 2021 C7 A 1 insert S145C 271

6 TMD Gb1 (7-61) PglC His

TEV (81-103)

Cy5 terminal Cy5 globular

TAMRA TAMRA

H-GGG-YKKG-OH PglA LPET/GG-HHHHHH PglA LPETGGG-YKKG His6 -Sortase A (S. aureus)

CaCl2

B CDdonor only: PglA(TAMRA)/PglC donor only: PglA(TAMRA)/PglC Cy5 globular FRET state: PglA(TAMRA)/PglC(Cy5-term.) FRET state: PglA(TAMRA)/PglC(Cy5-glob.) 4x10 6 5x10 6 2R0 13 nm g TAMRA PlA 4x10 6 PglC 3x10 6 .) u.)

u 6 . 3x10 a. (a 6 (

2x10 m n 15

515nm 6 5 x x 2x10 e 5.5 nm e F F 1x10 6 1x10 6

0 12.5 nm GB1 0 550 600 650 700 750 550 600 650 700 750 Cy5 terminal (nm) (nm)

His6

Fig. 4. Analysis of cofacial localization of PgC and PglA in nanodiscs using FRET. (A) Preparation of ﬂuorescent protein constructs. PglC Cys mutants were labeled with Cy5 maleimide and PglA was labeled with TAMRA using a sortase-mediated transpeptidation with a TAMRA-containing peptide. (B) Graphic of the PglC/PglA-

ND. Nanodisc dimensions and the R0 of the fluorophore pair are indicated. The locations of the fluorophores are based on structure prediction models (SI Text and Fig. S3). (C) Corrected fluorescence emission spectra of PglC(Cy5-terminal)/PglA(TAMRA) nanodiscs (FRET state) and PglC(unlabeled)/PglA(TAMRA) nanodiscs (donor only). (D)Correctedfluorescence emission spectra of PglC(Cy5-globular)/PglA(TAMRA) nanodiscs (FRET state) and donor-only nanodiscs. Data from nanodiscs comprising the PglC species labeled in the globular segment reveal a higher FRET efficiency than nanodiscs containing terminally labeled PglC.

with respect to the PglA(TAMRA) donor using the unique soluble globular domains of PglC and PglA obtain a cofacial ar- absorption signal of the fluorophore at 557 nm. PglC/PglA en- rangement in the nanodiscs, as illustrated in Fig. 4B. sembles contained unlabeled PglC (∼50%),whichledtoadis- proportional overestimation of unquenched donor for the two Functional Reconstitution of PglC and PglA. Radioactivity tracer nanodisc ensembles. To account for this high donor background, studies were used to assess the activities of the enzymes in the the labeled fraction of reconstituted PglC in the nanodisc en- nanodisc assemblies. Specifically, undecaprenyl-[33P]phosphate sembles was quantified by UV spectroscopy to determine the (Und-[33P]P) allowed for quantification of Und-P incorporation number of possible FRET states (see Methods and SI Text and also provided an orthogonal radiolabel that was used in con- for details). cert with [14C]- or [3H]-labeled UDP-sugars (UDP-[14C]diNAcBac To perform the FRET experiment, TAMRA was excited at and UDP-[3H]GalNAc) for precise tracking of the Und-P–linked a low wavelength (515 nm) to minimize cross-excitation of the species. To establish that Und-P incorporated into the nanodiscs, Cy5 acceptor. The fluorescence emission showed defined signals PglC/PglA-NDs were prepared from PE/PG lipids (3:1) and 0.5 mol for the donor emission and excitation of the acceptor attribut- percent Und-[33P]P. The assemblies were purified by Ni-NTA able to FRET (Fig. S5). The resulting spectra of the nanodisc chromatography, and the radioactivity present in the flow-through, ensembles with different PglC mutants were corrected for the wash, and elution fractions was quantified by liquid scintillation number of non-FRET states in the donor fluorescence based on counting, which showed that ∼18% of the total radioactivity was as- UV quantification and for background contributions stemming sociated with nanodiscs (Fig. 5A). Because only 20% of the nanodisc from direct excitation of the Cy5 fluorophore in the acceptor ensembles included PglC, this level of Und-P incorporation was fluorescence (SI Text), as determined by analysis of acceptor-only consistent with a statistical distribution of Und-P. samples of equal concentration (Fig. 4 C and D and Fig. S5). To validate that PglC was functionally reconstituted into the The donor signal was quenched for donor/acceptor nanodisc nanodiscs, PglC-NDs with Und-[33P]P were treated with soluble ensembles with both PglC mutants, revealing a higher FRET ef- UDP-[14C]diNAcBac. After incubation of the reaction for 1 h, an ficiency in the presence of the PglC(Cy5) acceptor labeled within organic-aqueous extraction of the reaction mixture was per- the globular protein segment (E = 0.57) compared with the ter- formed to assess incorporation of [14C], representing diNAcBac, minally labeled PglC(Cy5) (E = 0.43). This trend was also ob- into Und-[33P]P-P-[14C]diNAcBac, which would extract into the served in the intensity of the acceptor signals, revealing a stronger organic phase. The organic extract was analyzed by normal-phase signal for the PglC(Cy5-globular) species (Fig. 4 C and D). FRET HPLC and elution fractions were collected and analyzed by efficiencies were estimated based on donor signal decay upon liquid scintillation counting. The 33P-signal corresponding to quenching, thereby avoiding the use of acceptor standards and Und-[33P]P eluted first, followed by [33P] and [14C] signals that circumventing errors arising from endogenous acceptor quench- coeluted in a single fraction and corresponded to Und-[33P]P-P- ing processes, such as those caused by different probe environ- [14C]diNAcBac (Fig. 5B). ments. Assuming the free rotation of the appended fluorophores, The PglC/PglA-NDs were prepared with Und-[33P]P and ac- the stronger energy transfer for PglC(Cy5-globular) relative to tivity was assessed in a similar fashion, but using first unlabeled PglC(Cy5-terminal) is good evidence for a model in which the UDP-diNAcBac and UDP-[3H]GalNAc (Fig. 5C), and then both

20866 | www.pnas.org/cgi/doi/10.1073/pnas.1320852110 Hartley et al. Downloaded by guest on September 28, 2021 UDP-[14C]diNAcBac and UDP-[3H]GalNAc (Fig. 5D). The [33P]- 33 A

ﬁ INAUGURAL ARTICLE signal corresponding to Und-[ P]P elutes rst in both cases and 60 the disaccharide product is observed with coeluting [33P] and [3H] or coeluting [33P], [14C], and [3H], consistent with the coupled 40 ivity reactions of PglC and PglA. When UDP-diNAcBac is omitted t from the coupled reaction, no ﬁnal product is observed (Fig.

5E), indicating that the sequential reactions of both enzymes radioac 20 are required for production of the Und-P-P-disaccharide. % Because all of the substrates in the assays were discretely and ﬁ 0 quanti ably labeled with radioisotopes, the percent conversion of 1 234 5 6 7 89 33 the nanodisc-associated Und-[ P]P could be determined. After 1 Fractions h we observed that ∼50% of the Und-[33P]P had been converted 33 B Und-[33P]P 33 14 to corresponding Und-[ P]P-P-linked product (Fig. 5 B–D). This 10 Und-[ P]P-P-[ C]Bac ﬁnding is consistent with a model in which the Und-P is statisti- cally distributed in the nanodisc with the phosphate oriented 8 s equally toward both faces of the membrane bilayer. Because it is 6

well established that polyprenyl-phosphates do not undergo pas- pmole sive membrane flipping (41, 42), this finding would correlate with 4 all of the available Und-P being processed through the active 2 14C sites of the enzymes on the same face of the disc. 33P 0 To monitor the time course of the coupled PglC/PglA reaction 15 20 25 30 t (min) and production of the product, a similar method to that described Und-[33P]P-P-Bac-[3H]GalNAc above using unlabeled Und-P, UDP-diNAcBac, and UDP-[3H] C GalNAc, was applied and aliquots of the reaction mixture were 6 Und-[33P]P quenched at multiple time points. The data show a rapid and ef- 5 fi 3 cient production of the labeled Und-P-P-diNAcBac-[ H]GalNAc 4 product (Fig. 6A). To assess for activity because of endogenous 3 Und-P in the native E. coli lipid extract, the nanodiscs were pre- pmoles 2 pared without added Und-P, and indeed a low level of back- BIOCHEMISTRY ground activity was observed (Fig. 6A). When this experiment 1 3H was repeated with nanodiscs prepared with a lipid mixture composed 33P 0 of synthetic PE (67.0%) and PG (23.2%), as well as cardiolipin 15 20 25 30 t (min) (9.8%)—which is comparable to the predominant E. coli membrane Und-[33P]P-P-[14C]Bac-[3H]GalNAc lipid components—background activity was not observed. There- fore, this observation is consistent with the commercial E. coli D Und-[33P]P lipid extract containing a low level of endogenous Und-P. 5 To demonstrate the importance of colocalization in nanodiscs 4

for enzyme activity, the enzymatic reaction rates were determined 3 in nanodiscs and in detergent-disrupted nanodiscs. Two identical nanodisc samples were diluted with either buffer or detergent and pmoles 2 3H

then assayed for activity. The reaction rate of detergent-disrupted 1 14 ∼ C nanodiscs was 12.5-times slower than the reaction rate of the 33P 0 undisrupted sample (Fig. 6B). 15 20 25 30 Finally, evidence for a static colocalization of the PglC and t (min) PglA enzymatic reactions at the nanodisc surface was obtained Und-[33P]P E 6 by evaluating the reaction rates at different PglC/PglA-ND dilutions. Assuming that nanodiscs are stable macromolecular com- 5 plexes containing cofacially oriented PglC, PglA, and Und-P, then 4 es l the observed activity rate should not depend on the concentration o 3 m of PglC, PglA, or Und-P, but rather the absolute amount of these p components. To test this theory, an equimolar amount of nano- 2 discs was assayed at three different dilutions and the concen- 1 3H 33P trations of the soluble UDP-sugars were kept constant. The initial 0 15 20 25 30 reaction rates reflected in the slopes of the reaction turnover are t (min) the same for all dilutions (Fig. 6C). This finding is consistent with Fig. 5. Functional PglC/PglA-NDs demonstrated with triple orthogonal the formation of nanodiscs containing stable cofacially oriented 33 PglC and PglA in tandem with available Und-P substrate. radiolabeling. (A) PglC/PglA-NDs were prepared with Und-[ P]P and purified by Ni-NTA chromatography. The flow-through (fraction 1), washes – – Discussion (fractions 2 5) and elutions (fractions 6 9) were analyzed by liquid scintillation counting and it was determined that ∼18% of the Und-[33P]P We present an experimental platform that exploits membrane coeluted with nanodiscs. (B–E) Functional characterization of PglC/PglA- bilayer nanodiscs for investigating the functional roles of linear ND radiolabeled products by HPLC. (B) PglC-NDs produce Und-[33P]P-P- long-chain polyprenols in glycan assembly pathways. To establish [14C]diNAcBac. (C) PglC/PglA-NDs produce Und-[33P]P-P-diNAcBac-[3H] the potential of the approach, we have focused the current GalNAc. (D) PglC/PglA-NDs produce triple-labeled Und-[33P]P-P-[14C] studies on the first two membrane-committed steps in the C. jejuni diNAcBac-[3H]GalNAc. (E) Negative control for the coupled reaction of Pgl pathway carried out by the phosphoglycosyltransferase PglC PglC/PglA containing Und-[33P]P and UDP-[3H]GalNAc and lacking UDP- and the glycosyltransferase PglA, and have investigated coin- diNAcBac. No double-labeled product is observed. corporation of the enzymes into nanodiscs using multiple biochemical and biophysical methods. (Figs. 2 and 5, and Fig. S1), which is unique in showing the Protein Interactions at the Membrane Interface. We demonstrate oligomeric state of a phosphoglycosyltransferase being probed that PglC incorporates into nanodiscs as a functional monomer in a lipid bilayer. Furthermore, we show that PglC and PglA

Hartley et al. PNAS | December 24, 2013 | vol. 110 | no. 52 | 20867 Downloaded by guest on September 28, 2021 through the application of triple isotopic-labeling approaches A 8 PglC/PglA-Nanodiscs using the uniquely labeled substrates Und-[33P]P, UDP-[14C] control w/o Und-P (E. coli lipids) diNAcBac, and UDP-[3H]GalNAc (Fig. 5D). Importantly, the control w/o Und-P (POPE/POPG/CA) ability to carry out quantitative analysis of activities, for example 6 in the disc disruption and disc dilution assays (Fig. 6 B and C), reveals the impact of enzyme colocalization in a 2D membrane. The results from such studies are crucial for deepening our un- derstanding about how membrane-bound enzymes in multistep 4 pathways function in vivo. Although the current studies have focused on PglC and PglA, (pmoles) the native Pgl pathway comprises five glycan assembly enzymes (PglC, PglA, PglJ, PglH, and PglI, Fig. 1A), and intriguingly only 2 PglC and PglI are predicted to be integral membrane proteins by sequence analysis (12). Based on this established foundation of methods, the nanodisc platform should be readily applicable for investigating the specific roles of PglC and PglI in recruiting the entire 0 cluster of enzymes to the membrane bilayer interface where they 0204060 (min)t function. Additionally, the opportunities for orthogonal isotopic 8 labeling of substrates and intermediates will enable studies to ex- B intact PglC/PglA-Nanodiscs amine whether substrates are transferred directly between sequen- disintegrated Nanodiscs tial enzymes in the pathway. Because substrate processivity in glycan assemblypathways ispoorlyunderstood, this approach promises to 6 provide the experimental system for addressing this key issue. Structure/Function Relationships at the Lipid Bilayer. It is well documented that nanodiscs can be assembled from a variety of fl 4 phospholipids, which may re ect native or nonnative composi- moles) tions. In these studies we demonstrated that disc preparations (p made with native E. coli lipids result in a measureable background activity in the absence of added Und-P, presumably be- 2 cause the native E. coli lipid extract included a small amount of endogenous Und-P (Fig. 6A). This background activity is completely abolished with discs assembled with synthetic phospholipid components. Importantly, addition of dosed quantities 0 of Und-P restores robust activity that can be readily quantified by 0246using isotopically labeled substrates. These studies therefore t (min) show that the nanodisc platform will allow tremendous versatility in quantifying the specific effects of lipid bilayer composition on the C 1.0 3-fold dilution 40 functional efficiency of the sequential membrane enzyme activities. 2-fold dilution The current studies take the versatility of nanodiscs one step undiluted further by showing the opportunity to investigate glycan assembly 0.8 enzymes that require a membrane-bound substrate, in this case negative control (w/o UDP-diNAcBac) 30 a polyprenyl-phosphate derivative. To our knowledge, this de- monstration of enzymes acting on coincorporated membrane- 0.6 %) bound substrates in nanodiscs is unique. Previously, the activity of

on ( a series of diverse polyprenyl-phosphates were previous in-

20 si r vestigated using PglC as a detergent micelle preparation (16). e v (pmoles) The studies revealed that although unsaturation of the α-isoprene 0.4 n

co and a particular combination of E and Z isoprene units were key 10 features determining enzyme activity, the length of the polyprenyl 0.2 chain had signiﬁcantly less impact on protein activity. However, in a detergent-based assay, the effects of polyprenol length on membrane partitioning and dynamics would have been taken out of 0.0 0 consideration. With the nanodisc approach, we are now in a 02468 unique position to quantitatively assess structure/function rela- t (min) tionships of native and nonnative polyprenyl-linked substrates and phospholipids directly in a membrane bilayer, which will en- Fig. 6. Time course of activity in PglC/PglA-NDs as measured by radioactivity. (A) able us to make measurements that are simply inaccessible through Comparison of activity in PglC/PglA-NDs prepared with lipid extracts (E. coli)or any other in vitro or in vivo system. Furthermore, we will be able to synthetic lipid compositions with and without Und-P. (B) Reaction efﬁciency assess the impact of capturing the substrates and enzymes in of PglC/PglA-NDs compared with a detergent-disrupted PglC/PglA-NDs. (C) a 2D bilayer, which is likely to have profound implications for Initial reaction rates of PglC/PglA-NDs in sequentially diluted reactions. substrate availability and propensity of the turnover (42–44).

FRET Analysis of Protein Interactions in Nanodiscs. In contrast to physically and functionally interact in the presence of a lipid vesicular membrane model systems such as liposomes, both faces bilayer. Nonrandom, colocalization of PglC and PglA provides of the nanodisc are accessible for interactions with substrates and clear evidence that these enzymes undergo a specific protein– enzymes. Therefore, we developed orthogonal labeling strategies protein interaction that is dependent on the lipid bilayer in- to prepare fluorescently modified variants of PglC and PglA to terface and does not occur in the presence of detergent (Fig. 6B enable a FRET analysis to assess the topological relationship and Fig. S6). In addition, the sequential activities of the two between the globular domains of the enzymes in the nanodiscs. enzymes in the nanodisc platform are clearly demonstrated The studies clearly reveal a preference for a cofacial relationship

20868 | www.pnas.org/cgi/doi/10.1073/pnas.1320852110 Hartley et al. Downloaded by guest on September 28, 2021 of the globular domains; FRET efficiency is higher when PglC is exploiting PglF, PglE, and PglD of the pgl pathway (47). For the preparation of UDP-[14C]diNAcBac, [14C]acetyl-CoA was applied in the PglD-catalyzed step. labeled on the globular domain compared with N-terminal la- INAUGURAL ARTICLE beling, which places the label on the distal side of the membrane (Fig. 4 C and D). At the current time, further quantitative analysis Nanodisc Preparation and Characterization. Nanodisc assembly methods were of the FRET efficiencies would be premature, because structural based on established protocols (27, 29), and involved incubation of the information is not yet available on either PglC or PglA. The finding protein components (MSP and PglC in the presence and absence of PglA) that the globular domains are cofacially oriented is corroborated with membrane lipids containing Und-P at defined ratios. Self-assembly of by activity analyses and the functional reconstitution of the PglC/ nanodiscs was initiated by detergent removal with hydrophobic adsorbants fi PglA-ND. In particular, in experiments using precisely quantified (e.g., biobeads). Nanodisc assemblies are puri ed using the unique His6-tag Und-[33P]P, all of the available of Und-P is converted to final handle on PglC by using Ni-NTA affinity chromatography. Nanodisc assem- product (Fig. 5 B–D). If the major population of discs did not blies were characterized using SDS/PAGE, SEC, and DLS. Specific quantifica- include PglC and PglA oriented with cofacial active sites, the tion of proteins was carried out using Coomassie blue staining followed by extent of conversion would be lower because assemblies with the gel densitometry. Detailed protocols are described in SI Text. two enzymes oriented on opposite faces would not be able to process two sequential reactions to afford Und-PP-diNAcBac-GalNAc. FRET Analysis. Nanodiscs for FRET analysis were prepared by creating nanodisc Although the FRET analyses in the present studies were ensembles containing the following reconstituted variants of PglC and PglA: (i) carried out with an ensemble of discs including only two protein PglA(TAMRA) and unlabeled PglC; (ii) PglA(TAMRA) and PglC(Cy5-terminal); species, the opportunities for fluorescent labeling of other en- (iii) PglA(TAMRA) and PglC(Cy5-globular); (iv) unlabeled PglA and PglC(Cy5- terminal); (v) unlabeled PglA and PglC(Cy5-globular). The concentrations of the zymes in the glycan assembly phase of the pathway (PglJ, PglH, – and PglI) using similar orthogonal labeling approaches suggests PglA(TAMRA) containing nanodisc ensembles i iii were adjusted to identical concentration on basis of the donor TAMRA UV-absorption signal. Acceptor-only that the nanodisc platform can be extended to the entire path- samples iv and v were adjusted in concentration to match samples ii and iii, way. In this context, it should be noted that although the analysis respectively, applying the Cy5-absorption signals. FRET data were acquired by of nanodisc species as an ensemble was relatively complicated excitation at 515 nm to minimize direct acceptor excitation. Because of the because of incomplete labeling of one of the two enzymes, single- different labeling strategies, the PglA construct was quantitatively labeled, molecule approaches would avoid this complication and render whereas labeling of PglC was incomplete. To minimize errors resulting from analysis of more than two differentially labeled species feasible. different degrees of labeling between the two species, the fractions of la- Single-molecule approaches have already been applied to nano- beled versus nonlabeled PglC were stringently quantified after nanodisc assembly disc systems and, therefore, all of the components are in place for to account for overestimation of the donor fluorescence because of the population taking the current inquiries on the pgl pathway and the roles of of non-FRET PglC/PglA-NDs with unlabeled PglC. These non-FRET states were BIOCHEMISTRY polyprenols in general to the next level of complexity (45). represented by a fraction of 45% in case of the PglC-globular species and 56% in case of PglC-terminal. To account for this, the “donor-only” emission spectrum (i) Conclusion was scaled according to the amount of the non-FRET states in each sample. The two The application of nanodiscs to the well-characterized C. jejuni resulting donor-only curves were used to correct the spectra of the donor/acceptor pgl pathway provides an exciting experimental platform for FRET samples (ii and iii). Subsequently, acceptor signal contributions stemming addressing the evolutionary purpose of linear polyprenols in from direct excitation of the acceptor were subtracted using data from acceptor glycan biosynthetic pathways. The present study presents key only samples (iv and v; full analysis is presented in Fig. S5 and SI Text). steps in establishing the platform and demonstrates that the system delivers robust quantitative measurements of activity that Activity Analysis. Enzyme function in the nanodisc assemblies was assessed will be valuable for investigating the specific roles of polyprenyl- using isotopically labeled substrates (Und-[33P]P, UDP-[14C]diNAcBac, and linked substrates in the integrated function of glycan assembly UDP-[3H]GalNAc). For all concentrations and specific activities, see SI Text. enzymes typified by components of the bacterial pgl pathway. For product analysis by HPLC, four nanodisc preparations containing PglC The study highlights the interaction of PglC and PglA, which are alone or PglC and PglA were prepared with 0.5 mol percent Und-[33P]P. PglC- responsible for the first two membrane-committed steps in the NDs were incubated with UDP-[14C]diNAcBac. PglC/PglA-NDs were incubated pgl pathway. Coincorporation of these enzymes into nanodiscs is with unlabeled UDP-diNAcBac and UDP-[3H]GalNAc, or UDP-[14C]diNAcBac characterized by multiple biophysical methods that support the and UDP-[3H]GalNAc. As a control reaction, PglC/PglA-NDs were incubated cospatial and cofacial localization of the proteins. In tandem with UDP-[3H]GalNAc in the absence of UDP-diNAcBac. with PglC and PglA, the undecaprenyl-phosphate substrate is After 1 h, the reactions were quenched in chloroform:methanol (2:1, 1 mL) also incorporated into nanodiscs and the functional reconstitution and the organic layer was washed with chloroform:methanol: 0.1 M aqueous of a PglC–PglA complex is verified through biochemical assays. potassium chloride (3:48:47, 3× 400 μL). The organic layer was dried and This work represents a proof-of-concept demonstrating that resuspended in chloroform:methanol (4:1, 100 μL) for analysis using NP- nanodiscs can be used for the precise manipulation and study HPLC. Fractions of 1 mL were collected and each fraction was dried under an of polyprenol-dependent pathways. The study is also unique in N2 stream. Radioactivity was determined by liquid scintillation counting. providing a set of evidence supporting the hypothesis that en- Activity analysis for assessing the effects of lipid composition and de- zymes dependent on polyprenyl-phosphates associate into bio- tergent-based nanodisc disruption was carried out using PglC/PglA-NDs synthetic macromolecular complexes. prepared with unlabeled Und-P. Reactions were initiated by the addition of UDP-diNAcBac and UDP-[3H]GalNAc, and aliquots were quenched at various Methods time points and then extracted to isolate organic soluble radiolabeled products (Und-P-P-diNAcBac-[3H]GalNAc). For nanodisc disruption, we used Protein Expression, Purification, and Site-Specific Modification of PglC and PglA Variants. Standard techniques of molecular biology and biochemistry were Triton X-100 as a nondenaturing detergent at a concentration of 0.175% (vol/vol), which is sufficient to completely disrupt the nanodiscs as verified by used to clone, express, and purify His6-GB1-TEV-PglC, the accompanying Ni-NTA chromatographic analysis (Fig. S6 and SI Text). Additionally, mea- single Cys mutants (C7 and S145C), His8-TEV-PglA, and PglA-LPETGG-His6. PglC single Cys mutants were modified using Cy5 maleimide and PglA was surements of the coupled PglC/PglA reaction show that activity is maximal at modified using a sortase-mediated ligation (40) with GGGYK(TAMRA)KG. 0.175% (vol/vol) Triton X-100 (Fig. S6). 33 Detailed protocols are included in SI Text.SeeTables S1 and S2 for primers For the dilution assays, Und-[ P]P was used to enable accurate quantification used in preparation of PglC and PglA DNA constructs. of the disc-loaded substrate and the rates of enzyme-catalyzed conversions were measured by quantifying [3H]GalNAc incorporation into the Und-[33P] 33 Materials. All phospholipids, radioactive compounds, amino acid derivatives P-P-glycan products. Nanodiscs were assayed at 0.05 μM Und-[ P]P, 0.025 μM 33 33 and fluorescent labeling agents were purchased from commercial vendors. Und-[ P]P, or 0.017 μM Und-[ P]P, in volumes of 50 μL, 100 μL, and 150 μL, Undecaprenol was extracted from the leaves of the staghorn sumac (Rhus such that all three reactions contained equimolar amounts of the disc prep- typhina) and phosphorylated using chemical or enzymatic methods (SI Text). aration at different dilutions. The concentrations of UDP-diNAcBac and UDP- Und-[33P]P was prepared by using Streptococcus mutans kinase and [γ33P] [3H]GalNAc remained constant. Aliquots corresponding to 15% of the total ATP (46). UDP-diNAcBac was prepared using a chemoenzymatic approach starting volume were quenched and extracted, as described above.

Hartley et al. PNAS | December 24, 2013 | vol. 110 | no. 52 | 20869 Downloaded by guest on September 28, 2021 ACKNOWLEDGMENTS. The authors thank Prof. Stephen Sligar and Dr. Yelena supported by National Institutes of Health Grant GM039334 (to B.I.); an Grinkova for valuable technical advice and for providing samples of mem- American Chemical Society award of a Fellowship in Medicinal Chemistry brane scaffold protein for our initial nanodisc studies. Also, the authors (to M.D.H.) and a Leopoldina Fellowship Program LPDS 2009-38 (to P.E.S.). are grateful to Dr. Karen Allen, Dr. Angelyn Larkin, and Vinita Lukose The Massachussetts Institute of Technology Biophysical Instrumentation Fa- for their valuable feedback regarding the manuscript. This work was cility is also acknowledged.

1. Larkin A, Imperiali B (2011) The expanding horizons of asparagine-linked glycosyla- 24. van Duijn G, et al. (1986) Dolichyl phosphate induces non-bilayer structures, vesicle tion. Biochemistry 50(21):4411–4426. fusion and transbilayer movement of lipids: A model membrane study. Biochim Bi- 2. Jones MB, Rosenberg JN, Betenbaugh MJ, Krag SS (2009) Structure and synthesis of ophys Acta 861(2):211–223. polyisoprenoids used in N-glycosylation across the three domains of life. Biochim Bi- 25. Denisov IG, Grinkova YV, Lazarides AA, Sligar SG (2004) Directed self-assembly of – ophys Acta 1790(6):485 494. monodisperse phospholipid bilayer Nanodiscs with controlled size. J Am Chem Soc 3. Krag SS (1998) The importance of being dolichol. Biochem Biophys Res Commun 126(11):3477–3487. – 243(1):1 5. 26. Grinkova YV, Denisov IG, Sligar SG (2010) Engineering extended membrane scaffold 4. Hartley MD, Imperiali B (2012) At the membrane frontier: A prospectus on the re- proteins for self-assembly of soluble nanoscale lipid bilayers. Protein Eng Des Sel markable evolutionary conservation of polyprenols and polyprenyl-phosphates. Arch 23(11):843–848. Biochem Biophys 517(2):83–97. 27. Ritchie TK, et al. (2009) Chapter 11—Reconstitution of membrane proteins in phos- 5. Surmacz L, Swiezewska E (2011) Polyisoprenoids—Secondary metabolites or physio- pholipid bilayer nanodiscs. Methods Enzymol 464:211–231. logically important superlipids? Biochem Biophys Res Commun 407(4):627–632. 28. Nath A, Atkins WM, Sligar SG (2007) Applications of phospholipid bilayer nanodiscs in 6. Bouhss A, Trunkfield AE, Bugg TD, Mengin-Lecreulx D (2008) The biosynthesis of – peptidoglycan lipid-linked intermediates. FEMS Microbiol Rev 32(2):208–233. the study of membranes and membrane proteins. Biochemistry 46(8):2059 2069. 7. Weerapana E, Imperiali B (2006) Asparagine-linked protein glycosylation: From eu- 29. Boldog T, Li M, Hazelbauer GL (2007) Using Nanodiscs to create water-soluble karyotic to prokaryotic systems. Glycobiology 16(6):91R–101R. transmembrane chemoreceptors inserted in lipid bilayers. Methods Enzymol 423: 8. Szymanski CM, Yao R, Ewing CP, Trust TJ, Guerry P (1999) Evidence for a system of 317–335. general protein glycosylation in Campylobacter jejuni. Mol Microbiol 32(5):1022–1030. 30. Ritchie TK, Kwon H, Atkins WM (2011) Conformational analysis of human ATP- 9. Glover KJ, Weerapana E, Imperiali B (2005) In vitro assembly of the un- binding cassette transporter ABCB1 in lipid nanodiscs and inhibition by the antibodies decaprenylpyrophosphate-linked heptasaccharide for prokaryotic N-linked glycosyl- MRK16 and UIC2. J Biol Chem 286(45):39489–39496. ation. Proc Natl Acad Sci USA 102(40):14255–14259. 31. Hernández-Rocamora VM, et al. (2012) Dynamic interaction of the Escherichia coli cell 10. Glover KJ, Weerapana E, Numao S, Imperiali B (2005) Chemoenzymatic synthesis of division ZipA and FtsZ proteins evidenced in nanodiscs. J Biol Chem 287(36):30097–30104. glycopeptides with PglB, a bacterial oligosaccharyl transferase from Campylobacter 32. Boldog T, Grimme S, Li M, Sligar SG, Hazelbauer GL (2006) Nanodiscs separate che- jejuni. Chem Biol 12(12):1311–1315. moreceptor oligomeric states and reveal their signaling properties. Proc Natl Acad Sci 11. Glover KJ, Weerapana E, Chen MM, Imperiali B (2006) Direct biochemical evidence for USA 103(31):11509–11514. the utilization of UDP-bacillosamine by PglC, an essential glycosyl-1-phosphate- 33. Inagaki S, et al. (2012) Modulation of the interaction between neurotensin receptor transferase in the C. jejuni N-linked glycosylation pathway. Biochemistry 45(16): NTS1 and Gq protein by lipid. J Mol Biol 417(1-2):95–111. – 5343 5350. 34. Kimmich N, et al. (2007) Electron transfer between cytochrome P450cin and its FMN- 12. Krogh A, Larsson B, von Heijne G, Sonnhammer EL (2001) Predicting transmembrane containing redox partner, cindoxin. J Biol Chem 282(37):27006–27011. protein topology with a hidden Markov model: Application to complete genomes. 35. Nath A, Grinkova YV, Sligar SG, Atkins WM (2007) Ligand binding to cytochrome P450 J Mol Biol 305(3):567–580. 3A4 in phospholipid bilayer nanodiscs: The effect of model membranes. J Biol Chem 13. Alaimo C, et al. (2006) Two distinct but interchangeable mechanisms for flipping of – lipid-linked oligosaccharides. EMBO J 25(5):967–976. 282(39):28309 28320. 14. Scott NE, et al. (2011) Simultaneous glycan-peptide characterization using hydrophilic 36. Denisov IG, Grinkova YV, McLean MA, Sligar SG (2007) The one-electron autoxidation – interaction chromatography and parallel fragmentation by CID, higher energy colli- of human cytochrome P450 3A4. J Biol Chem 282(37):26865 26873. sional dissociation, and electron transfer dissociation MS applied to the N-linked 37. Kawai T, Caaveiro JM, Abe R, Katagiri T, Tsumoto K (2011) Catalytic activity of MsbA glycoproteome of Campylobacter jejuni. Mol Cell Proteomics 10(2):M000031–MCP201. reconstituted in nanodisc particles is modulated by remote interactions with the bi- 15. Karlyshev AV, et al. (2004) The Campylobacter jejuni general glycosylation system is layer. FEBS Lett 585(22):3533–3537. important for attachment to human epithelial cells and in the colonization of chicks. 38. Das A, Sligar SG (2009) Modulation of the cytochrome P450 reductase redox potential Microbiology 150(Pt 6):1957–1964. by the phospholipid bilayer. Biochemistry 48(51):12104–12112. 16. Chen MM, et al. (2007) Polyisoprenol specificity in the Campylobacter jejuni N-linked 39. Cheng Y, Patel DJ (2004) An efficient system for small protein expression and re- glycosylation pathway. Biochemistry 46(50):14342–14348. folding. Biochem Biophys Res Commun 317(2):401–405. 17. Rush JS, Rick PD, Waechter CJ (1997) Polyisoprenyl phosphate specificity of UDP- 40. Popp MW, Antos JM, Ploegh HL (2009) Site-specific protein labeling via sortase-me- GlcNAc:undecaprenyl phosphate N-acetylglucosaminyl 1-P transferase from E. coli. diated transpeptidation. Curr Protoc Protein Sci, Chapter 15: Unit 15.13. Glycobiology 7(2):315–322. 41. Hanover JA, Lennarz WJ (1979) The topological orientation of N,N′-diacety- fi 18. McLachlan KR, Krag SS (1992) Substrate speci city of N-acetylglucosamine 1-phos- lchitobiosylpyrophosphoryldolichol in artificial and natural membranes. JBiol – phate transferase activity in Chinese hamster ovary cells. Glycobiology 2(4):313 319. Chem 254(18):9237–9246. 19. Janas T, Chojnacki T, Swiezewska E, Janas T (1994) The effect of undecaprenol on 42. McCloskey MA, Troy FA (1980) Paramagnetic isoprenoid carrier lipids. 2. Dispersion bilayer lipid membranes. Acta Biochim Pol 41(3):351–358. and dynamics in lipid membranes. Biochemistry 19(10):2061–2066. 20. Zhou GP, Troy FA, 2nd (2005) NMR study of the preferred membrane orientation of 43. de Ropp JS, Troy FA (1985) 2H NMR investigation of the organization and dynamics of polyisoprenols (dolichol) and the impact of their complex with polyisoprenyl recog- polyisoprenols in membranes. J Biol Chem 260(29):15669–15674. nition sequence peptides on membrane structure. Glycobiology 15(4):347–359. 44. Katsikas H, Quinn PJ (1982) The polyisoprenoid chain length influences the in- 21. Zhou GP, Troy FA, 2nd (2003) Characterization by NMR and molecular modeling of teraction of ubiquinones with phospholipid bilayers. Biochim Biophys Acta 689(2): the binding of polyisoprenols and polyisoprenyl recognition sequence peptides: 3D – structure of the complexes reveals sites of specific interactions. Glycobiology 13(2): 363 369. fl 51–71. 45. Nath A, et al. (2010) Single-molecule uorescence spectroscopy using phospholipid – 22. Wang X, Mansourian AR, Quinn PJ (2008) The effect of dolichol on the structure and bilayer nanodiscs. Methods Enzymol 472:89 117. phase behaviour of phospholipid model membranes. Mol Membr Biol 25(6-7): 46. Hartley MD, Larkin A, Imperiali B (2008) Chemoenzymatic synthesis of polyprenyl 547–556. phosphates. Bioorg Med Chem 16(9):5149–5156. 23. Valtersson C, et al. (1985) The influence of dolichol, dolichol esters, and dolichyl 47. Olivier NB, Chen MM, Behr JR, Imperiali B (2006) In vitro biosynthesis of UDP-N,N’- phosphate on phospholipid polymorphism and fluidity in model membranes. J Biol diacetylbacillosamine by enzymes of the Campylobacter jejuni general protein Chem 260(5):2742–2751. glycosylation system. Biochemistry 45(45):13659–13669.

20870 | www.pnas.org/cgi/doi/10.1073/pnas.1320852110 Hartley et al. Downloaded by guest on September 28, 2021