Bioactive cell-like hybrids coassembled from(glyco)dendrimersomes with bacterial membranesQi Xiaoa,1, Srujana S. Yadavallib,1, Shaodong Zhanga, Samuel E. Shermana, Elodie Fiorina, Louise da Silvaa,Daniela A. Wilsona,2, Daniel A. Hammerc,d, Sabine Andrée, Hans-Joachim Gabiuse, Michael L. Kleinf,3, Mark Goulianb,3,and Virgil Perceca,3

aRoy & Diana Vagelos Laboratories, Department of Chemistry, University of Pennsylvania, Philadelphia, PA 19104-6323; bDepartment of Biology, Universityof Pennsylvania, Philadelphia, PA 19104-6313; cDepartment of Bioengineering, University of Pennsylvania, Philadelphia, PA 19104-6321; dDepartment ofChemical and Biomolecular Engineering, University of Pennsylvania, Philadelphia, PA 19104-6391; eInstitute of Physiological Chemistry, Faculty of VeterinaryMedicine, Ludwig Maximilian University, 80539 Munich, Germany; and fInstitute of Computational Molecular Science, Temple University, Philadelphia,PA 19122

Contributed by Michael L. Klein, January 22, 2016 (sent for review December 26, 2015; reviewed by Ling Peng and Donald Tomalia)

A library of amphiphilic Janus dendrimers including two that arefluorescent and one glycodendrimer presenting lactose were usedto construct giant dendrimersomes and glycodendrimersomes.Coassembly with the components of bacterial membrane vesiclesby a dehydration–rehydration process generated giant cell-like hy-brid vesicles, whereas the injection of their ethanol solution intoPBS produced monodisperse nanometer size assemblies. These hy-brid vesicles contain transmembrane proteins including a smallmembrane protein, MgrB, tagged with a red fluorescent protein,lipopolysaccharides, and glycoproteins from the bacterium Escher-ichia coli. Incorporation of two colored fluorescent probes in eachof the components allowed fluorescence microscopy to visualizeand demonstrate coassembly and the incorporation of functionalmembrane channels. Importantly, the hybrid vesicles bind a hu-man galectin, consistent with the display of sugar moieties fromlipopolysaccharides or possibly glycosylated membrane proteins.The present coassembly method is likely to create cell-like hybridsfrom any biological membrane including human cells and thus mayenable practical application in nanomedicine.

E. coli | transmembrane protein | lipopolysaccharides | galectin

Naturally occurring (1), chemically modified (2, 3), and syn-thetic (4, 5) lipids, amphiphilic block copolymers (6, 7),

polypeptides (8), Janus dendrimers (JDs) (9), and Janus glyco-dendrimers (JGDs) (10, 11) self-assemble into vesicles denotedas liposomes, polymersomes, dendrimersomes (DSs), and gly-codendrimersomes (GDSs), respectively. These vesicles providemodels for primitive (12) and contemporary (13, 14) cell mem-branes and drug-delivery devices (15–17). Recently, hybrid ves-icles coassembled from naturally occurring phospholipids andamphiphilic block copolymers (18–20) have been described; thesevesicles eliminated some of the deficiencies of liposomes, such aslimited stability under oxidative conditions and general instabilityover time, and the deficiencies of polymersomes, which possesswide membrane thickness [8–50 nm (20)], exhibit toxicity, and canbe tedious to synthesize. These hybrid vesicles combined thedesirable feature of liposomes—specifically, their biologicallysuitable membrane thickness of 4 nm—with that of polymer-somes, which are known for their stability. In addition, trans-membrane proteins (21–23) could be incorporated into thephospholipid fragments of planar membranes derived from theseassemblies. However, the variability in the extent of miscibilitybetween the hydrophobic fragments of the phospholipid and theblock copolymer (20) generates a complex morphology of thehybrid membrane that requires further characterization toenable practical applications both as drug-delivery devicesand cell membrane models. Here, we report the coassembly ofthe components of DSs and GDSs with those of the bacterialmembrane vesicles (BMVs) to generate functional hybrid vesi-cles. DSs, GDSs, and liposomes have hydrophobic fragmentswith similar chemical structures and similar membrane thickness

(4.5–4.9 nm) (24). Therefore, the bacterial membranes with theirintact native components are expected to be transferred to thehybrid vesicles, providing a new and simple method for thegeneration of bioactive cell-like hybrids of interest as criticalnanoscale design parameters (25).

Results and DiscussionCoassembly of Giant DSs with BMVs.Giant DSs ranging in diameterfrom 2 to 50 μm were prepared by hydration of JD films withultrapure water or PBS (Fig. 1) (9, 24, 26). This protocol waspreviously used for the encapsulation of hydrophilic and hydro-phobic dyes and drugs within giant DSs. Giant DSs prepared bythis method were used for coassembly with BMVs via a de-hydration–rehydration technique (27) (Fig. 1). In addition to alibrary of giant DSs, giant GDSs (10, 11, 28–31) self-assembledfrom sugar-presenting JGDs were also used for coassembly withBMVs. This process was expected to maintain the integrity of thecomponents of bacterial membranes. The envelope of Gram-negative bacteria such as Escherichia coli contains an innerphospholipid bilayer membrane and an outer membrane con-sisting of an asymmetric bilayer with phospholipids in the innerleaflet and lipopolysaccharides (LPSs) in the outer leaflet (32–34). The inner and outer membranes contain various peripheraland integral membrane proteins that perform diverse functions

Significance

Cell surface determinants such as glycans, receptors, and ad-hesion molecules govern cell sociology in a complex manner.By forming cell-like hybrids of chemically programmable(glyco)dendrimersomes with bacterial membrane vesicles, evi-dence is obtained for the feasibility of combining chemical andbiological surface design in one entity. Such tunable cell-likehybrids with custom-made combinations of surface epitopesand active receptors will likely find utility in dissecting thefunctionality of individual entities in complex networks andultimately enable novel biomedical applications.

Author contributions: Q.X., S.S.Y., S.Z., M.G., and V.P. designed research; Q.X., S.S.Y., S.Z.,and S.E.S. performed research; E.F., L.d.S., D.A.W., D.A.H., S.A., and H.-J.G. contributednew reagents/analytic tools; Q.X., S.S.Y., S.Z., M.G., and V.P. analyzed data; and Q.X.,S.S.Y., S.E.S., H.-J.G., M.L.K., M.G., and V.P. wrote the paper.

Reviewers: L.P., Centre Interdisciplinaire de Nanoscience de Marseille (CINaM) CNRS UMR7325; and D.T., NanoSynthons LLC.

The authors declare no conflict of interest.1Q.X. and S.S.Y. contributed equally to this work.2Present address: Institute for Molecules and Materials, Radboud University Nijmegen,6525 AJ, Nijmegen, The Netherlands.

3To whom correspondence may be addressed. Email: mlklein@temple.edu, goulian@sas.upenn.edu, or percec@sas.upenn.edu.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1525589113/-/DCSupplemental.

E1134–E1141 | PNAS | Published online February 16, 2016 www.pnas.org/cgi/doi/10.1073/pnas.1525589113

Dow

nloa

ded

by g

uest

on

Feb

ruar

y 26

, 202

1

such as membrane receptors, transporters, channels, and en-zymes. To label BMVs, a small integral membrane protein,MgrB, which localizes to the inner membrane, was tagged withmCherry and expressed in E. coli (35). These BMVs enrichedwith red fluorescent mCherry–MgrB and provide a simple meansfor following the incorporation of an integral membrane protein byfluorescence microscopy, allowing the self-assembly and coassemblyprocesses to be monitored.

Coassembly of Monodisperse Nanohybrids. The coassembly of JD-3(3,5)12G1-PE-(3,4,5)-3EO-G1-(OCH3)6 (9, 24) with the com-ponents of BMV was investigated by the injection of ethanolsolubilized JD-3 and BMVs into PBS. Control experiments for theself-assembly of JD-3 and BMV components under identical con-ditions were also performed. All assemblies were studied by dy-namic light scattering (DLS). This method produced monodispersehybrid vesicles with dimensions of about 140 nm (Fig. S1). These

values are in the range of DS-3 prepared by injection undersimilar conditions. The initially prepared BMV vesicles exhibit abimodal molecular weight distribution, whereas BMV vesiclesreassembled by ethanol injection into PBS buffer under identicalconditions as DS-3 also exhibited bimodal molar mass distribu-tion and dimensions ranging from 340 and 430 nm. These ex-periments suggest that the size and molar mass distribution ofthe hybrid vesicles coassembled by the injection method is de-termined by the contribution of the JD component. Mono-disperse nanohybrids obtained by the injection method opennumerous potential applications in nanomedicine.

Visualization of BMVs, Giant DSs, and of Their Coassembly. Four JDswere selected from 19 libraries containing 144 compounds (Fig. 2)(9, 24, 26, 36). JD-1, JD-2, and JD-3 are twin–twin molecules(9, 24) containing twin-hydrophobic and twin-hydrophilic den-drons, whereas JD-4 is a single–single compound (26) based on

Fig. 1. Illustration of the preparation and coassembly of hybrid giant vesicles from giant DSs, giant GDSs, and E. coli BMVs enriched with mCherry–MgrB.

Xiao et al. PNAS | Published online February 16, 2016 | E1135

CHEM

ISTR

YBIOCH

EMISTR

YPN

ASPL

US

Dow

nloa

ded

by g

uest

on

Feb

ruar

y 26

, 202

1

single-hydrophobic and single-hydrophilic dendrons. These JDswere characterized by a combination of complementary methodsincluding DLS, cryogenic transmission electron microscopy (cryo-TEM), differential interference contrast (DIC) microscopy, X-raydiffraction, confocal fluorescence microscopy, and micropipetteaspiration (9, 24, 26). JDs possess a bilayer thickness of ∼4.5–4.9 nm (24), which is comparable to that of the phospholipidbilayer of biological membranes (4 nm), unlike most polymer-somes, which are generally thicker (8–50 nm). The giant hybridvesicles coassembled from DSs with BMVs containing mCherry–MgrB by the dehydration–rehydration protocol are comparablein size to giant DSs. These DS vesicles showed a strong red fluo-rescence signal along the boundary, indicating that these vesiclesconsist of both dendrimers and bacterial membrane components.It must be mentioned that BMVs alone do not form giant vesi-cles (Fig. S2) and instead only form small vesicles ranging from20 to 200 nm (37).

Dual-Color Imaging of Coassembled Hybrid DS Vesicles. To demon-strate the coassembly of the DSs with BMVs, DSs labeled with asuitable fluorophore compatible with mCherry were synthesized.Two modified JDs based on the structure of JD-3 were elaborated(Fig. 3). The pentaerythritol (PE) from the center of the JD waschanged to Tris(hydroxymethyl)aminomethane (Tris) and thesmall hydrophobic fluorescent dye coumarin was incorporated togenerate the fluorescent FL-JD-5 and FL-JD-6 (Figs. S3 and S4).These two molecules are constitutional isomers with the sameratio of hydrophobic and hydrophilic dendrons and differ only intheir connectivity. FL-JD-6 is easier to prepare and purify be-cause the hydrophobic fragments were introduced in the laststep of the synthesis, dramatically changing the polarity from thatof their precursors (Fig. S4). Both FL-JD-5 and FL-JD-6 self-assembled into identical giant DSs upon film hydration andshowed intense cyan fluorescence (Fig. 3 A and B and Fig. S5).When the giant fluorescent DSs (FL-DSs) were coassembledwith BMVs enriched with mCherry–MgrB, both cyan and redfluorescence along the boundary of the vesicles were observed,demonstrating that the assemblies were hybrid vesicles (Fig. 3 Cand D). Together with their 3D intensity plots, this dual-colordesign proved that the coassembly of giant FL-DSs and BMVswas achieved.

Bioactive Hybrid Vesicles Containing Functional Channels. E. coli is aGram-negative bacterium with a complex cell envelope con-taining diverse functional components such as channel proteins,transporters, receptors, enzymes, and LPSs in its membranes

(32–34) (Fig. 1). Among the transmembrane proteins, non-specific porins such as OmpF and OmpC located on the outermembrane are particularly abundant (33, 38). These β-barrelproteins allow passive diffusion of small molecules up to 600 Dain size. The coassembly of BMVs with giant DSs was likely toincorporate functional porins in the hybrid vesicles, renderingthe membrane permeable to low molecular weight compounds.To test for membrane permeability, the fluorescent dye rhoda-mine 6G (Rho) (479 Da), which is smaller than the porin sizecutoff, was encapsulated into the giant DS (Fig. 4) to confirmthe membrane impermeability demonstrated previously (9). Asexpected, the giant DS display intense yellow fluorescence in thecenter of the vesicle (Fig. 4C). Because of the wide emissionspectrum of Rho (Fig. S6), a fluorescence signal was also de-tected in the red wavelength region. When Rho was coassembledwith dendrimers and bacterial membranes to make hybrid vesi-cles, on the other hand, the yellow fluorescence was not de-tectable, whereas the red fluorescence of the mCherry–MgrBfrom bacterial membranes was visible at the boundary of thevesicle, as expected. These results are consistent with the low-molecular-weight dye diffusing through porin channels in thehybrid vesicles. In addition to Rho, a green fluorescent dye,calcein (622 Da), which is close in size to the porin diffusion limitwas used. Calcein was encapsulated in both the giant DSs (Fig.5B) and in the hybrid vesicles (Fig. 5C) without any leakagebeing observed, as expected for membrane permeability medi-ated by porins in the hybrid vesicles. The fluorescence of calceinis strongly quenched by cations such as Co2+ at near neutral pH(39). If there are functional channel proteins within the hybridvesicles, Co2+ should quench the fluorescence of calcein. To testthis possibility, Co2+ (1 mM) was added to calcein-loaded giantDS (Fig. 5D) as well as to the hybrid vesicles (Fig. 5E). Sub-sequently, samples were studied under the microscope less than5 min after the addition of Co2+. The fluorescence inside thegiant DSs (Fig. 5D) remained unquenched, whereas the calceinfluorescence from the hybrid vesicles (Fig. 5 A and E) wasquenched. Interestingly, closer examination of a number of hy-brid vesicles in this experiment revealed variable levels of calceinfluorescence quenching (Fig. S7). This result may be attributableto a heterogeneous distribution of porins within each individualhybrid vesicle that depends on the amount of BMVs stochasti-cally incorporated into each hybrid vesicle. It is also possible thatsome of these vesicles are multilamellar, with interior layerspossibly lacking porin channels. These results indicate that smallmolecules like rhodamine 6G (Fig. 4) and cations like Co2+ (Fig.5) can pass through the bacterial membrane porin channels,

Fig. 2. Representative microscopy images of hybrid vesicles coassembled from Giant DS-1 assembled from JD-1 (3,5)12G1-PE-BMPA-G2-(OH)8 (A), Giant DS-2assembled from JD-2 (3,5)12G1-PE-(3,4)-3EO-G1-(OH)4 (B), Giant GS-3 assembled from JD-3 (3,5)12G1-PE-(3,4,5)-3EO-G1-(OCH3)6 (C), and Giant GS-4 assembledfrom JD-4 (3,5)12G1-CH2-PhE- (3,4,5)-3EO-G1-(OCH3)3 (D) with a BMV containing mCherry–MgrB. A phase-contrast image was first acquired, followed by thefluorescence image by successive exposures on the same vesicle.

E1136 | www.pnas.org/cgi/doi/10.1073/pnas.1525589113 Xiao et al.

Dow

nloa

ded

by g

uest

on

Feb

ruar

y 26

, 202

1

whereas larger molecules such as calcein, which are closer to thesize of the porin cutoff (Fig. 5), are not able to pass through.

Human Galectin-8 Binds the LPSs from the Surface of Hybrid Vesicles.LPS is a major component of the outer leaflet of the outermembrane of E. coli and other Gram-negative bacteria. LPS isimportant for membrane stability and defense against antimi-crobial compounds. Galectins are a class of secreted sugar-binding proteins that have specificity for β-galactosides. Galectin-8(Gal-8), a tandem-repeat type galectin (Fig. 6A) (40) regulatescell–matrix interaction and is important for animal cell growth.Gal-8 also plays a significant role in autophagy, enabling mam-malian host defense against bacterial invasion (41). E. coli LPScontains numerous sugar moieties. In addition, some E. colimembrane proteins may be glycosylated (42). To test whetherthe surface of E. coli MG1655 and membrane vesicles derivedfrom this strain can be labeled with Gal-8, a fluorescein iso-thiocyanate (FITC)-labeled Gal-8 (Gal-8–FITC) was used toperform binding experiments with both live E. coli cells (Fig. 6B)and BMVs (Fig. 6C) enriched with mCherry–MgrB. Greenfluorescence from FITC was detected in both samples, con-firming that the lectin binds the E. coli cell surface as well asBMVs. As a control experiment, the giant DS-3 was tested forbinding with Gal-8–FITC. No fluorescence was detected in the

control experiment (Fig. 6D). However, when hybrid vesiclescoassembled from giant DS-3s with BMVs were incubated withGal-8–FITC, both green fluorescence from Gal-8–FITC (al-though a weak signal) and red fluorescence from mCherry–MgrBwere observed along the boundary of the vesicle (Fig. 6E). GiantGDSs (10, 11, 28–31) were assembled from the JGD JGD-7 (Fig.6 F and G) using the protocol used for giant DSs and sub-sequently were tested for Gal-8–FITC binding. As shown in Fig.6E, Gal-8–FITC is strongly targeted to the giant GDSs becauseof the very high surface density of lactose [β-D-galactopyranosyl-(1→4)-D-glucose], which is a β-galactoside sugar (β-galactose +glucose) (11, 30, 43). Hybrid vesicles coassembled from giantGDS-7 and BMVs enriched with mCherry–MgrB also clearlyexhibited both strong green and red fluorescence on the mem-brane (Fig. 6H).

ConclusionsA simple method to create hybrid vesicles from the coassemblyof DSs or GDSs with vesicles generated from bacterial mem-branes is demonstrated. This method transfers many of thecomponents of the bacterial inner and outer membranes, in-cluding transmembrane proteins such as MgrB and porins, andpossibly LPSs, into the structure of the hybrid vesicle. These

Fig. 3. Representative microscopy images of fluorescent (FL) giant DSs Giant FL-GD-5 assembled from of FL-JD-5 (3,5)12-G2-Tris-coumarin-(3,4,5)-3EO-2G1-(OCH3)6 (A) and Giant FL-GD-6 assembled from FL-JD-6 (3,5)12-2G1-coumarin-Tris-(3,4,5)-3EO-G2-(OCH3)6 (B) and hybrid vesicles by Giant FL-GD-5 (C) andGiant FL-GD-6 (D) with a BMV containing mCherry–MgrB. Phase-contrast images and fluorescence images under cyan fluorescence channel and red fluo-rescence channel were taken by successive exposures on the same assembly. Images in the same fluorescence channel were normalized to have the samevalues for the darkest and brightest pixels. The 3D intensity profiles were made using the program ImageJ.

Xiao et al. PNAS | Published online February 16, 2016 | E1137

CHEM

ISTR

YBIOCH

EMISTR

YPN

ASPL

US

Dow

nloa

ded

by g

uest

on

Feb

ruar

y 26

, 202

1

cell-like hybrid structures provide a platform that should be readilyapplicable to the coassembly of other biological membranes—including those from mammalian cells—with DSs and GDSs.Through standard molecular genetic methods, specific mem-brane components of interest can be engineered (such as themembrane protein mCherry–MgrB shown in this work) and in-corporated in the hybrid vesicles. Our attempts to coassemblepolymersomes with BMVs by a similar method did not result inhybrid vesicles, likely because of the large thickness of poly-mersome membranes. Artificial cell-like hybrid assemblies suchas those described here have potential to impact diverse fields,such as targeted drug delivery, detection by fluorescence, vac-cines, and other areas of nanomedicine (44–49). Future experi-ments have the potential to alter the membranes of living cellsdirectly by DSs or GDSs into their membranes without the needfor genetic manipulation.

MethodsGiant DSs and GDSs were prepared by the following general method. Asolution of JD or JGD (10–40 mg·mL–1, 200 μL) in tetrahydrofuran (THF) wasdeposited on the top surface of a roughened Teflon sheet (1 cm2), placedin a flat-bottom vial, and followed by evaporation of the solvent for 2 h.The Teflon sheet was dried in vacuo for additional 12 h. Milli-Q water(2 mL) was added to merge the dendrimer film on Teflon sheet, and thevial was placed in a 60 °C oven for 12 h for hydration. The sample wasthen mixed using a vortex mixer for 30 s with a final concentration of1–4 mg·mL–1.

Strain AML20 (35), which is derived from E. coli K-12 strain MG1655,contains a gene deletion mgrB. AML20 cells were transformed withplasmid pAL22 expressing the mCherry–MgrB fusion protein, which isinserted into the bacterial inner membrane. Strain AML20/pAL22 was

streaked out from the –80 °C freezer stock on LB (Lysogeny broth; FisherScientific) agar (1.5% wt/vol) and incubated overnight at 37 °C to obtainisolated colonies. A single colony was then inoculated in LB medium forovernight growth at 37 °C on a roller drum for aeration. A saturatedculture of AML20/pAL22 was diluted 1:200 in 50 mL of LB medium sup-plemented with ampicillin (100 μg·mL–1) in a 250-mL culture flask. Fol-lowing growth for 1.5 h at 37 °C with aeration, mCherry–MgrB expressionwas induced by the addition of isopropyl β-D-1-thiogalactopyranosideto a final concentration of 0.5 mM. Protein induction was carried out for3.5 h, and then cells were harvested at 3,500 × g for 10 min using aBeckman table-top centrifuge at 4 °C. The cell pellets were washed inTris(hydroxymethyl)aminomethane·HCl (Tris·HCl) buffer (50 mM, pH 8.0)and then frozen at –80 °C.

For membrane preparation, frozen cell pellets were resuspended in 400 μLof solution of sucrose (20%) and Tris·HCl (30 mM, pH 8.0). A fresh stocksolution of lysozyme (Sigma L6876) was prepared by dissolving 10 mg oflysozyme in 1 mL of EDTA (0.1 M, pH 7.3); 50 μL of lysozyme solution wasadded to the resuspended pellet and incubated on ice for 30 min. After thisstep, 5 mL of the solution containing EDTA (3 mM, pH 7.3) and 30 μL ofprotease inhibitor mixture (1:100 dilution final; Sigma P8849) was added toeach tube. Cells were sonicated for 20 s at a low power (30% amplitude) andrepeated for a total of five rounds of 20 s each with 1 min of incubation onice between each round of sonication. Cells were spun at 4 °C for 5 min at7,500 × g and the supernatant was transferred to a Nalgene white screw-cap centrifuge tube to isolate the membranes from the cytoplasmic fractionand cell debris. This supernatant was then spun at 4 °C for 30–45 min at40,000 × g in a Beckman centrifuge fitted with SL-50T rotor. The bacterialmembrane fraction was obtained as vesicles following the high-speed cen-trifugation, and the supernatant was discarded. The membrane vesicleswere resuspended in 500 μL of buffer composed of Tris·HCl (20 mM, pH 8.0)and EDTA (0.1 mM).

The hybrid vesicle preparation protocol was adapted from the literature(27). Bacterial membrane preparation enriched with mCherry–MgrB was

Fig. 4. (A and B) Illustrations of Giant DS encapsulated with rhodamine 6G (Rho) (yellow and red fluorescence inside vesicle) (A) and the hybrid vesiclecoassembled from Giant DS with BMV containing mCherry–MgrB (red fluorescence on the membrane) with Rho (B). (C and D) Microscopy images of GiantDS-2 encapsulated with Rho (C) and hybrid vesicle coassembled from Giant DS-2 with BMV containing mCherry–MgrB, encapsulated with Rho (D). Phase-contrast images and fluorescence images under yellow fluorescence channel and red fluorescence channel were taken by successive exposures on the samevesicle. The brightness of images in the same fluorescence channel was normalized to have the same values for the darkest and brightest pixels.

E1138 | www.pnas.org/cgi/doi/10.1073/pnas.1525589113 Xiao et al.

Dow

nloa

ded

by g

uest

on

Feb

ruar

y 26

, 202

1

added at a final dilution of 1:25–1:100 to a solution of premade giant DSs orGDSs (50 μL) in a flat-bottomed glass vial. The solution was mixed anddehydrated using microcentrifuge adapters in a vacuum centrifugal

evaporator (Savant SpeedVac). Dehydration was performed at roomtemperature for ∼30 min, until all of the liquid was removed and a dry filmwas formed. The glass vials were then loosely capped and transferred to

Fig. 5. (A and B) Illustrations of Giant DS encapsulation with calcein (A), followed by addition of CoCl2 (B). (E and F) Hybrid vesicle coassembled from Giant DS witha BMV containing mCherry–MgrB (red fluorescence on the membrane) with calcein (green fluorescence inside vesicle) (E) and then treated with CoCl2 (F). (C, D, G,and H) Microscopy images of Giant DS-2 encapsulated with calcein (C), Giant DS-2 encapsulated with calcein and treated with CoCl2 (D), hybrid vesicle coassembledfrom Giant DS-2 with BMV containing mCherry–MgrB, encapsulated with calcein (G), and hybrid vesicle coassembled from Giant DS-2 with BMV containingmCherry–MgrB, encapsulated with calcein, and treated with CoCl2 (H). Phase-contrast images and fluorescence images under green fluorescence channel and redfluorescence channel were taken by successive exposures on the same vesicle. The brightness of images in the same fluorescence channel was normalized.

Xiao et al. PNAS | Published online February 16, 2016 | E1139

CHEM

ISTR

YBIOCH

EMISTR

YPN

ASPL

US

Dow

nloa

ded

by g

uest

on

Feb

ruar

y 26

, 202

1

a humid chamber containing excess water to create a water saturatedatmosphere and incubated at 37 °C for 1 h. Sucrose (500 mM, 25 μL) solutionwas added to each vial, and the hybrid vesicles were allowed to swellovernight (∼16 h) at 37 °C. For vesicles encapsulated with dyes, sucrose solution(500 mM, 25 μL) prepared in Tris·HCl (10 mM, pH 7.5) was added duringovernight swelling.

ACKNOWLEDGMENTS. This work was supported by National ScienceFoundation Grants DMR-1066116 and DMR-1120901 (to V.P.), the P. RoyVagelos Chair at the University of Pennsylvania (V.P.), the HumboldtFoundation (V.P.), National Science Foundation Grant DMR-1120901(to M.L.K., M.G., and D.A.H.), National Institutes of Health Grant R01-GM080279 (to M.G.), and the EC Seventh Framework Programme(GLYCOPHARM) (H.-J.G.).

1. Bangham AD, Standish MM, Watkins JC (1965) Diffusion of univalent ions across thelamellae of swollen phospholipids. J Mol Biol 13(1):238–252.

2. Ringsdorf H, Schlarb B, Venzmer J (1988) Molecular architecture and function ofpolymeric oriented systems: Models for the study of organization, surface recogni-tion, and dynamics of biomembranes. Angew Chem Int Ed Engl 27(1):113–158.

3. Thomas JL, Tirrell DA (1992) Polyelectrolyte-sensitized phospholipid vesicles. AccChem Res 25(8):336–342.

4. Kunitake T (1992) Synthetic bilayer membranes: Molecular design, self-organization,and application. Angew Chem Int Ed Engl 31(6):709–726.

5. Guo X, Szoka FC, Jr (2003) Chemical approaches to triggerable lipid vesicles for drugand gene delivery. Acc Chem Res 36(5):335–341.

6. Discher BM, et al. (1999) Polymersomes: Tough vesicles made from diblock copoly-mers. Science 284(5417):1143–1146.

7. Discher DE, Eisenberg A (2002) Polymer vesicles. Science 297(5583):967–973.8. Rodriguez AR, Choe U-J, Kamei DT, Deming TJ (2015) Blending of diblock and triblock

copolypeptide amphiphiles yields cell penetrating vesicles with low toxicity. MacromolBiosci 15(1):90–97.

9. Percec V, et al. (2010) Self-assembly of Janus dendrimers into uniform den-drimersomes and other complex architectures. Science 328(5981):1009–1014.

10. Percec V, et al. (2013) Modular synthesis of amphiphilic Janus glycodendrimers andtheir self-assembly into glycodendrimersomes and other complex architectures withbioactivity to biomedically relevant lectins. J Am Chem Soc 135(24):9055–9077.

11. Zhang S, et al. (2015) Glycodendrimersomes from sequence-defined Janus glyco-dendrimers reveal high activity and sensor capacity for the agglutination by naturalvariants of human lectins. J Am Chem Soc 137(41):13334–13344.

12. Mansy SS, et al. (2008) Template-directed synthesis of a genetic polymer in a modelprotocell. Nature 454(7200):122–125.

13. Blain JC, Szostak JW (2014) Progress toward synthetic cells. Annu Rev Biochem 83(1):615–640.

14. Haluska CK, et al. (2006) Time scales of membrane fusion revealed by direct imagingof vesicle fusion with high temporal resolution. Proc Natl Acad Sci USA 103(43):15841–15846.

15. Kohli AG, Kierstead PH, Venditto VJ, Walsh CL, Szoka FC (2014) Designer lipids fordrug delivery: From heads to tails. J Control Release 190:274–287.

16. Allen TM, Cullis PR (2004) Drug delivery systems: Entering the mainstream. Science303(5665):1818–1822.

17. Pattni BS, Chupin VV, Torchilin VP (2015) New developments in liposomal drug de-livery. Chem Rev 115(19):10938–10966.

18. Ruysschaert T, et al. (2005) Hybrid nanocapsules: Interactions of ABA block copoly-mers with liposomes. J Am Chem Soc 127(17):6242–6247.

19. Le Meins J-F, Schatz C, Lecommandoux S, Sandre O (2013) Hybrid polymer/lipid ves-icles: State of the art and future perspectives. Mater Today 16(10):397–402.

20. Schulz M, Binder WH (2015) Mixed hybrid lipid/polymer vesicles as a novel membraneplatform. Macromol Rapid Commun 36(23):2031–2041.

21. Kumar M, Grzelakowski M, Zilles J, Clark M, Meier W (2007) Highly permeablepolymeric membranes based on the incorporation of the functional water channelprotein Aquaporin Z. Proc Natl Acad Sci USA 104(52):20719–20724.

22. Thoma J, et al. (2012) Membrane protein distribution in composite polymer-lipid thinfilms. Chem Commun (Camb) 48(70):8811–8813.

23. Kowal J, Wu D, Mikhalevich V, Palivan CG, Meier W (2015) Hybrid polymer-lipidfilms as platforms for directed membrane protein insertion. Langmuir 31(17):4868–4877.

24. Peterca M, Percec V, Leowanawat P, Bertin A (2011) Predicting the size and propertiesof dendrimersomes from the lamellar structure of their amphiphilic Janus den-drimers. J Am Chem Soc 133(50):20507–20520.

25. Tomalia DA (2012) Dendritic effects: Dependency of dendritic nano-periodicproperty patterns on critical nanoscale design parameters (CNDPs). New J Chem36(2):264–281.

26. Zhang S, et al. (2014) “Single-single” amphiphilic janus dendrimers self-assemble intouniform dendrimersomes with predictable size. ACS Nano 8(2):1554–1565.

27. Goulian M, et al. (1998) Gramicidin channel kinetics under tension. Biophys J 74(1):328–337.

28. Zhang S, et al. (2014) Mimicking biological membranes with programmable glycanligands self-assembled from amphiphilic Janus glycodendrimers. Angew Chem Int EdEngl 53(41):10899–10903.

29. Zhang S, et al. (2015) Dissecting molecular aspects of cell interactions using glyco-dendrimersomes with programmable glycan presentation and engineered humanlectins. Angew Chem Int Ed Engl 54(13):4036–4040.

Fig. 6. (A) Model of tandem-type human Gal-8. (B–E, G, and H) Representative microscopy images of E. coli K-12 MG1655 cells (B), E. coli BMVs enriched withmCherry–MgrB (C), Giant DS-3 (D), hybrid vesicles coassembled from Giant DS-3 and BMV containing mCherry–MgrB (E), Giant GDS-7 self-assembled fromJGD-7 (G), and hybrid vesicles coassembled from Giant GDS-7 and BMV containing mCherry–MgrB (H). The vesicles were incubated with Gal-8–FITC in PBS. Thechemical structure of JGD-7 is shown (F). The brightness of images in the same fluorescence channel was normalized to have the same values for the darkestand brightest pixels.

E1140 | www.pnas.org/cgi/doi/10.1073/pnas.1525589113 Xiao et al.

Dow

nloa

ded

by g

uest

on

Feb

ruar

y 26

, 202

1

30. Zhang S, et al. (2015) Unraveling functional significance of natural variations of ahuman galectin by glycodendrimersomes with programmable glycan surface. ProcNatl Acad Sci USA 112(18):5585–5590.

31. Xiao Q, et al. (2016) Onion-like glycodendrimersomes from sequence-defined Janusglycodendrimers and influence of architecture on reactivity to a lectin. Proc Natl AcadSci USA 113(5):1162–1167.

32. Beveridge TJ (1999) Structures of gram-negative cell walls and their derived mem-brane vesicles. J Bacteriol 181(16):4725–4733.

33. Silhavy TJ, Kahne D, Walker S (2010) The bacterial cell envelope. Cold Spring HarbPerspect Biol 2(5):a000414.

34. Koebnik R, Locher KP, Van Gelder P (2000) Structure and function of bacterial outermembrane proteins: Barrels in a nutshell. Mol Microbiol 37(2):239–253.

35. Lippa AM, Goulian M (2009) Feedback inhibition in the PhoQ/PhoP signaling systemby a membrane peptide. PLoS Genet 5(12):e1000788.

36. Zhang S, et al. (2014) Self-assembly of amphiphilic Janus dendrimers into uniformonion-like dendrimersomes with predictable size and number of bilayers. Proc NatlAcad Sci USA 111(25):9058–9063.

37. Altendorf KH, Staehelin LA (1974) Orientation of membrane vesicles fromEscherichia coli as detected by freeze-cleave electron microscopy. J Bacteriol117(2):888–899.

38. Cowan SW, et al. (1992) Crystal structures explain functional properties of two E. coliporins. Nature 358(6389):727–733.

39. Bernardi P, Petronilli V, Di Lisa F, Forte M (2001) A mitochondrial perspective on celldeath. Trends Biochem Sci 26(2):112–117.

40. Zick Y, et al. (2004) Role of galectin-8 as a modulator of cell adhesion and cell growth.Glycoconj J 19(7-9):517–526.

41. Thurston TLM, Wandel MP, von Muhlinen N, Foeglein A, Randow F (2012) Galectin 8targets damaged vesicles for autophagy to defend cells against bacterial invasion.Nature 482(7385):414–418.

42. Sherlock O, Dobrindt U, Jensen JB, Munk Vejborg R, Klemm P (2006) Glycosylation ofthe self-recognizing Escherichia coli Ag43 autotransporter protein. J Bacteriol 188(5):1798–1807.

43. Ruiz FM, et al. (2014) Natural single amino acid polymorphism (F19Y) in human ga-lectin-8: Detection of structural alterations and increased growth-regulatory activityon tumor cells. FEBS J 281(5):1446–1464.

44. Menjoge AR, Kannan RM, Tomalia DA (2010) Dendrimer-based drug and imagingconjugates: Design considerations for nanomedical applications. Drug Discov Today15(5-6):171–185.

45. Kannan RM, Nance E, Kannan S, Tomalia DA (2014) Emerging concepts in dendrimer-based nanomedicine: From design principles to clinical applications. J Intern Med276(6):579–617.

46. Stiriba S-E, Frey H, Haag R (2002) Dendritic polymers in biomedical applications: Frompotential to clinical use in diagnostics and therapy. Angew Chem Int Ed Engl 41(8):1329–1334.

47. Haag R, Kratz F (2006) Polymer therapeutics: Concepts and applications. Angew ChemInt Ed Engl 45(8):1198–1215.

48. Wei T, et al. (2015) Anticancer drug nanomicelles formed by self-assembling amphiphilicdendrimer to combat cancer drug resistance. Proc Natl Acad Sci USA 112(10):2978–2983.

49. Farokhzad OC, Langer R (2009) Impact of nanotechnology on drug delivery. ACS Nano3(1):16–20.

50. Moore JS, Stupp SI (1990) Room temperature polyesterification. Macromolecules23(1):65–70.

51. He G, et al. (2009) A color-tunable europium complex emitting three primary colorsand white light. Angew Chem Int Ed Engl 48(33):6132–6135.

52. Ropponen J, Nummelin S, Rissanen K (2004) Bisfunctionalized Janus molecules. OrgLett 6(15):2495–2497.

53. Schneider CA, Rasband WS, Eliceiri KW (2012) NIH Image to ImageJ: 25 years of imageanalysis. Nat Methods 9(7):671–675.

Xiao et al. PNAS | Published online February 16, 2016 | E1141

CHEM

ISTR

YBIOCH

EMISTR

YPN

ASPL

US

Dow

nloa

ded

by g

uest

on

Feb

ruar

y 26

, 202

1