A polycationic antimicrobial and biocompatible hydrogel with microbe membrane suctioning ability

ARTICLESPUBLISHED ONLINE: 12 DECEMBER 2010 | DOI: 10.1038/NMAT2915

A polycationic antimicrobial and biocompatiblehydrogel with microbe membranesuctioning abilityPeng Li1†, Yin Fun Poon1†, Weifeng Li2, Hong-Yuan Zhu3, Siew Hooi Yeap1, Ye Cao1, Xiaobao Qi1,Chuncai Zhou1, Mouad Lamrani4, Roger W. Beuerman3,5, En-Tang Kang6, Yuguang Mu7,Chang Ming Li1, Matthew W. Chang1, Susanna Su Jan Leong1 and Mary B. Chan-Park1*

Despite advanced sterilization and aseptic techniques, infections associated with medical implants have not been eradicated.Most present coatings cannot simultaneously fulfil the requirements of antibacterial and antifungal activity as well asbiocompatibility and reusability. Here, we report an antimicrobial hydrogel based on dimethyldecylammonium chitosan (withhigh quaternization)-graft-poly(ethylene glycol) methacrylate (DMDC-Q-g-EM) and poly(ethylene glycol) diacrylate, whichhas excellent antimicrobial efficacy against Pseudomonas aeruginosa, Escherichia coli, Staphylococcus aureus and Fusariumsolani. The proposed mechanism of the antimicrobial activity of the polycationic hydrogel is by attraction of sectionsof anionic microbial membrane into the internal nanopores of the hydrogel, like an ‘anion sponge’, leading to microbialmembrane disruption and then microbe death. We have also demonstrated a thin uniform adherent coating of the hydrogelby simple ultraviolet immobilization. An animal study shows that DMDC-Q-g-EM hydrogel coating is biocompatible with rabbitconjunctiva and has no toxicity to the epithelial cells or the underlying stroma.

Infections arising in association with medical implants anddevices are a significant problem and, when these happen,surgical removal or other surgical intervention, with attendant

medical risks and complications, is often inevitable1. To minimizealteration of bulk properties (for example mechanical strengthor transparency) of an implant, coating with an effective an-timicrobial agent seems to be an attractive approach to combatinfection1–3. Earlier generations of antimicrobial coatings basedon drug/chemical elution have only short-term antimicrobial ef-fect, and cause cumulative toxicity and/or microbe resistance4,5.A contact-active coating with immobilized antimicrobial agentis generally less likely to lead to the development of microberesistance. This class of coating disrupts the microbes’ membraneswithout targeting their metabolic activity, which is associated withthe emergence of resistance6.

Antimicrobial polymers have been applied as contact-active coatings2. Cationic polymers such as derivatives ofpolyethylenimine7, poly(vinyl-N -hexylpyridinum)8, polynor-bornene9, polymethacrylates10, poly(phenylene ethynylene)11 andso on, in solution form, have been reported to disrupt the pathogencytoplasmic membrane, and some have impressive selectivity forbacterial over mammalian cells11–13. However, when these polymersare immobilized, their antimicrobial activities may be greatlyreduced because their diffusion into cellmembranes is impeded14,15.Polymers that retain their antimicrobial activities even afterimmobilization typically contain dangling hydrophobic polycations

1School of Chemical and Biomedical Engineering, Nanyang Technological University, 62 Nanyang Drive, Singapore 637459, Singapore, 2School of Physicaland Mathematical Sciences, Nanyang Technological University, 21 Nanyang Link, Singapore 637371, Singapore, 3Singapore Eye Research Institute, 11 ThirdHospital Avenue, Singapore 168751, Singapore, 4Menicon Co. Ltd., Immeuble Espace Cordelier, 2 Président Carnot, Lyon 69002, France, 5Duke-NUS, SRPNeuroscience and Behavioral Disorders, 8 College Road, Singapore 169857, Singapore, 6Department of Chemical and Biomolecular Engineering, NationalUniversity of Singapore, 10 Kent Ridge, Singapore 119260, Singapore, 7School of Biological Sciences, Nanyang Technological University, 60 Nanyang Drive,Singapore 637551, Singapore. †These two authors contributed equally to the paper. *e-mail: [email protected].

but these often have high toxicity tomammalian cells16,17. At presentthere are few reports of coatings that are broadly antimicrobialto fungi and bacteria (both Gram-negative and Gram-positive)and that are also non-haemolytic and biocompatible18. Also, mostreported methods of surface immobilization of polymers aremultistep and post-synthesis3,8,19, involve organic solvents20,21 anddo not result in permanent coatings.

In this work we demonstrate highly antimicrobial surfaces basedon in situ ultraviolet immobilization of a protein-/cell-repelling andcontact-active hydrogel layer made from quaternized ammoniumchitosan-graft -poly(ethylene glycol) methacrylate (qC-g -EM;Fig. 1a). We chose chitosan, an inherently biocompatible andantimicrobial material, for further derivation to add modalities to(1) increase the antibacterial and antifungal activity, (2) achieveexcellent biocompatibility and (3) enable easy in situ coating(that is, surface grafting carried out concurrently with hydrogelcrosslinking). Accordingly, we have modified chitosan to add(1) a hydrophobic alkyl side chain and cationic charge throughquaternization of the amino group, (2) hydrophilic poly(ethyleneglycol) with six ethylene glycol repeats (PEG6) and (3) methacrylatefunctionality (Fig. 1a). Others have shown that quaternizedchitosan derivatives are water soluble and more antimicrobialthan pristine chitosan22. PEGylation of chitosan derivativeshas been shown to decrease cytotoxicity and haemolysis23,24.However, PEGylated quaternized chitosan derivatives have notbeen reported for use as antimicrobial agents. Also, hydrogels

NATURE MATERIALS | VOL 10 | FEBRUARY 2011 | www.nature.com/naturematerials 149© 2011 Macmillan Publishers Limited. All rights reserved.

http://www.nature.com/doifinder/10.1038/nmat2915

mailto:[email protected]

http://www.nature.com/naturematerials

ARTICLES NATURE MATERIALS DOI: 10.1038/NMAT2915

t = 0 ns t = 50 ns

DMDC-Qhydrogel

LPS

Lipid bilayer

Substrate

Disrupted microbe outer membrane

qC-g-EM

Pore

Cytoplasmic side

Cationic antimicrobialhydrogel

__ _ _ _ _ _ _ _ _ __ _ _ _ __

Anioniclipids

+ + +++ + + +++

+ + ++++ + +++

+ + +++ + + +++

OH

OH

OH OH

NRR'

O OO O

m n

O OOO

OH

m n

NH2

NR2

m + n

O

OHO

OH OH

O OO OHO

OH

m n p

OH OH

OO O

O OO

OO

HO HO HO

OH OH

NR'2

OO

OH

HO HOO

OHO

m n p

HO

NH (R) NH2m n

HO

HO HO

HO

N+R3

N+(R)

N+(CH3)3

N+(CH3)3

(CH3)2

NR'R"N+R3

(I) N-alkylation

(I) QuaternizationNaOH, Nal, RX with x = Br or I50 °C¬70 °C, 24 h

(III) PEG methacrylation(II) PEG methacrylationCl¬CH2COO¬(CH2CHO2O)6¬CO¬C(CH3)=CH2r.t., 3 h

RCHO, NaBH4r.t., 1.5 h

Series II

(II) QuaternizationNaOH, Nal, CH3l, 50 °C, 24 h

Cl¬CH2COO¬(CH2CH2O)6¬CO¬C(CH3)=CH2r.t., 3 h

CH2(CH2)4CH3

CH2(CH2)8CH3

CH3 or H

R =

R =

R' =

for DMHC-g-EM

for DMDC-g-EM DMDC-Q-g-EM

C=CH2

=

OCO)6CH2 CR" = R' or

=O

O (CH2CH2

CH3

N+(R)

(CH3)2

CH3 or H for TMC-g-EM TMC-Q-g-EM

R =

Series I

O

=

C=CH2O)6 CO (CH2CH2

O

=

R' = R or CH2C

CH3

a

b c d

Figure 1 | qC-g-EM polymers and the antimicrobial killing mechanism of their hydrogels. a, Synthetic scheme for preparation of TMC-g-EM andTMC-Q-g-EM (series I, left side) and DMHC-g-EM, DMDC-g-EM and DMDC-Q-g-EM (series II, right side). b, Schematic diagram of the ‘anion sponge’model—parts of the negatively charged bacterial membrane are ‘suctioned’ into the pores of the qC-g-Em hydrogel. c,d, Computer simulation of the killingmechanism showing the ‘suctioning’ of P. aeruginosa bacterial membrane (lipid bilayer) LPS molecules into the DMDC-Q hydrogel after 50 ns.

based on qC-g -EM have not been reported or explored asantimicrobial coatings. Coulombic attraction would exist betweenthe cationic quaternized ammonium group of qC-g -EM and thebacterial or fungal cell membrane, which is typically anionic12,25.However, unlike other coatings2,3, hydrogels are porous26, sothey have interior ‘space’ to receive parts of the microbemembrane that may be ‘attracted into’ the hydrogel. Whenthe attraction between anionic membrane components and thecationic hydrogel is strong enough, energy minimization and

entropy maximization will drive anionic outer microbe membranecomponents into the nanopores of the hydrogel, leading tomembrane disruption (Fig. 1b).

Three qC-g -EMs, specifically trimethylammonium chitosan-g -EM (TMC-g -EM), dimethylhexylammonium chitosan-g -EM(DMHC-g -EM) and dimethyldecylammonium chitosan-g -EM(DMDC-g -EM), with different alkyl chains, were synthesizedand tested (Fig. 1a and entries 2–4 in Table 1). Two differentdegrees of quaternization, that is 20–27% and 51–56% (Table 1),

150 NATURE MATERIALS | VOL 10 | FEBRUARY 2011 | www.nature.com/naturematerials

© 2011 Macmillan Publishers Limited. All rights reserved.



NATURE MATERIALS DOI: 10.1038/NMAT2915 ARTICLES

Table 1 |Characteristics of the qC-g-EM polymers.

No Name Series Degree ofquaternization (%)

Double bondcontent (mmol g−1)

Mn (×105 g mol−1) Mw (×105 g mol−1) PID

1 C-g-EM - 0.76 0.74 0.86 1.152 TMC-g-EM I 24 0.84 0.75 1.22 1.633 DMHC-g-EM II 27 1.02 0.49 0.67 1.384 DMDC-g-EM II 20 1.00 0.36 0.52 1.465 TMC-Q-g-EM I 51 0.80 0.35 0.41 1.176 DMDC-Q-g-EM II 56 0.80 0.24 0.38 1.567 DMDC-M II 20 1.13 0.33 0.36 1.06

Mn : number-average molecular weight. Mw : weight-average molecular weight. PID: polydispersity index.

were explored; ‘Q’ is added to the polymer name, for exampleTMC-Q-g -EM, to denote the higher degree of quaternization(entries 5 and 6). In addition, the effect of a PEG6 side chainwas also explored: DMDC-M (entry 7, Table 1), which doesnot have the PEG6 side chain but is methacrylated, was alsoprepared by reacting DMDC with glycidyl methacrylate in thepresence of triethylamine27 rather than with α-terminated PEG6methacrylate (Cl-PEG6M). The solutions and hydrogels basedon the various qC-g -EMs were tested for efficacy in killingfour clinically significant pathogens: the Gram-negative bacteriaPseudomonas aeruginosa (P. aeruginosa, ATCC9027 isolated fromear infection) and Escherichia coli (E. coli, ATCC8739 isolatedfrom faeces), the Gram-positive bacterium Staphylococcus aureus(S. aureus, ATCC6538 isolated from human lesion) and thefungus Fusarium solani (F. solani, ATCC36031 isolated fromhuman corneal ulcer).

NMR analyses confirm the successful syntheses of thesepolymers (Supplementary Fig. S1 in Supplementary Section S1).After quaternization and PEG6 methacrylate (PEG6M) grafting,all chitosan derivatives are soluble in neutral water. Minimuminhibitory concentration (MIC) determinationwas carried out withthe chitosan derivative solutions to evaluate their antimicrobialactivities (Supplementary Section S2). The results (Table 2) confirmthat the quaternized chitosan derivatives (entries 2–7) have betterantimicrobial activities than unquaternized chitosan (entry 1). TheMICs of qC-g -EMs against fungi are generally low (≤200 µgml−1).Increasing the alkyl chain length from TMC-g -EM to DMDC-g -EM (entries 2–4, Table 2) alone decreases the MICs againstGram-positive S. aureus but not against the Gram-negativeE. coli and P. aeruginosa. However, the more highly quaternizedTMC-Q-g -EM and DMDC-Q-g -EM (entries 5 and 6, Table 2)have dramatically lower MICs against both bacteria and fungi(24–200 µgml−1), which are comparable to those of commonantimicrobial peptides that we also tested (Table 2). Further,TMC-Q-g -EM and DMDC-Q-g -EM are fairly non-haemolytic,with 0% haemolytic activity at 3,100 µgml−1 and 1,600 µgml−1respectively (Supplementary Fig. S3), which are at least 15 timestheir bacterium/fungus MICs (Table 2).

To form free-standing hydrogel films, the chitosan derivatives,poly(ethylene glycol) diacrylate (with 13 ethylene glycol repeats,denoted hereafter as PEG13DA) andwater weremixed in the ratio of1:1:8 (w/w), together with 0.1 wt% photoinitiator (Irgacure 2959),and then ultraviolet irradiated. The hydrogels were challenged withthe four pathogens at a concentration of about 2.5×106 CFU cm−2and the cell-count reductions after one hour were recorded.Figure 2a shows the antimicrobial activity of the hydrogel films.The hydrogel based on the non-quaternized C-g -EM (entry 1)exhibited low activity against these microbes. All the qC-g -EMhydrogels (entries 2–6, Fig. 2a) show outstanding antifungal activityagainst F. solani but differing antibacterial activities. Increasing

the alkyl-side-chain length from TMC to DMHC to DMDC(entries 2–4, Fig. 2a) increases the killing efficacy against Gram-positive S. aureus. More interestingly, the two highly quaternizedTMC-Q-g -EM and DMDC-Q-g -EM hydrogels (entries 5 and 6)have excellent log reductions of above 2.0 (>99% kill) for allfour microbes, which are superior to their counterparts withlower degrees of quaternization (entries 2 and 4). (More detailedantimicrobial activity results of the hydrogels can be found inSupplementary S3.)

The LIVE/DEAD bacterial viability assay was carried out onDMDC-Q-g -EM hydrogel with S. aureus, using PEG13DA hydrogelas control (Fig. 2c,d). Using this assay, bacterial cells that look greenare live cells with intactmembranes whereas bacterial cells that stainred are dead cells that have damagedmembranes. Figure 2c,d showsthat S. aureus cells are alive on the PEG13 DA control but deadon the DMDC-Q-g -EM hydrogel. Field emission scanning electronmicroscopy was also carried out to observe the morphologies ofthe variousmicrobes onDMDC-Q-g -EMhydrogel after 1 h contact(Fig. 2e–h). Distorted and wrinkled membranes were observed inE. coli and F. solani on the hydrogel surfaces (Fig. 2f,h). Lesions andholes were observed in P. aeruginosa and S. aureus (Fig. 2e,g) aftercontact with the hydrogel. Many of the small pores in P. aeruginosacells (arrows in Fig. 2e) were near the edge where the microbemembrane contacted the hydrogel.

We postulate that our cationic nanoporous hydrogels actlike a molecular ‘anion sponge’, suctioning out parts of theanionic microbe membrane into the gel interior voids to causemicrobe membrane disruption (Fig. 1b). This interpretation issuggested by computer molecular dynamics simulations conductedto probe the interaction between DMDC-Q hydrogel and bacterialmembranes. (See Supplementary Section S4 on charges ofmicrobialand mammalian cell membranes.) Figure 1c,d shows a modelof Gram-negative P. aeruginosa membrane, consisting of 243zwitterionic 1,2-dioleoyl-sn-glycero-3-phosphocholine and nineanionic lipopolysaccharide (LPS) molecules. The molecular modelof LPS of P. aeruginosa PAO1 was generated on the basis ofexperimentally sequenced data28. When the bacterial lipid bilayerwas placed near the DMDC-Q chains for 50 ns (Fig. 1d), itwas significantly disturbed; some of the anionic LPSs were evencompletely pulled out of the bilayer and drawn into the poresbetween the DMDC-Q chains. The negatively charged regions ofthe remaining LPS molecules moved into close proximity to thehydrogel surface. The suctioning of cell-membrane componentsinto the porous hydrogel distorts and wrinkles the membrane,or even produces holes in it, corroborating our field emissionscanning electron microscopy results (Fig. 2e). This membranedisruption eventually kills the microbe. Similar results werealso found in the simulation of the interaction between aGram-positive bacterial-membrane model and DMDC-Q hydrogel(Supplementary Section S4.2).





Table 2 |MICs of the qC-g-EM solutions against various pathogens.

No Name MIC (µg ml−1)

Gram negative Gram positive Fungi

P. aeruginosa E. coli S. aureus F. solani

1 C-g-EM 13,000 13,000 6,300 3,1002 TMC-g-EM 780 780 1,600 2003 DMHC-g-EM 780 3,100 200 2004 DMDC-g-EM 780 1,600 200 2005 TMC-Q-g-EM 98 200 98 246 DMDC-Q-g-EM 98 98 49 247 DMDC-M 1,600 1,600 390 200

Melittin 63 63 8 -LL-37 >250 >250 >250 -Indolicidin 130 >250 130 -Magainin I >250 130 130 -Defensin (HNP-1) >130 >130 >130 -

Our ‘anion sponge’ model of these cationic nanoporoushydrogels differs from previous models of cell-membranedisruption based on the penetration of antimicrobial polymer intothemicrobemembrane or ion exchange between the cell membraneand antimicrobial polymer6,8. Previous antimicrobial surfaces havetypically been based on brushes or pendant chains of hydrophobiccationic polymers emanating from a monolayer or thin coating.These coatings do not contain pores needed to ‘receive’ disruptedmembrane parts6,8 and the antimicrobial polymers may showgreatly reduced microbe activities after immobilization14,15. OurqC-g -EM hydrogel, however, retains its antimicrobial efficacy afterultraviolet immobilization because of the presence of the pores.The high antifungal activity of all the highly hydrated hydrogels(entries 2–6, Fig. 2a) may possibly be due to the compatibilityof the chitosan derivatives with fungal cell wall, which containsa significant amount of chitin, enabling close approach of thehydrogel to the cell membrane.

Increasing the positive charge density and pore size of ourcationic hydrogels would significantly increase their killing efficacy.The measured surface zeta potentials (Supplementary Section S3.7)confirm that the highly quaternized TMC-Q-g -EM and DMDC-Q-g -EM hydrogels have higher charge densities: entries 5 and 6have zeta potentials of 11.9±0.4mVand 13.7±0.9mVrespectively,higher than the values (10.4±0.3mV and 11.4±0.2mV) for theircounterparts with lower degrees of quaternization (entries 2 and 4).The higher antibacterial efficacies of TMC-Q-g -EM and DMDC-Q-g -EM hydrogels compared with TMC-g -EM and DMDC-g -EM(Fig. 2a) illustrate the significance of hydrogel charge density ineffective microbe killing by these materials. The effect of chargeon killing efficacy was corroborated by computer simulation ofless-charged DMDC hydrogel, which seems not to interact asstrongly with the microbe membranes (Supplementary SectionS4.2) compared withDMDC-Qhydrogel (Fig. 1c,d).

To demonstrate the importance of the pores of the cationicnetworks for achieving effective microbe killing, we comparedDMDC-g -EM hydrogel (entry 4) with two other smaller-pore-size hydrogels formulated using DMDC-M (entries 7 and 8,Fig. 2a). Specifically, we replaced (1) DMDC-g -EMwith DMDC-M(without the PEG6 side chain) in entry 7 (Fig. 2a) and (2) PEG13DAcrosslinker with a shorter diethylene glycol diacrylate (DEG2DA)in DMDC-M-LS (low swelling) (entry 8 in Fig. 2a). The killingefficacies of DMDC-M (entry 7) and, especially, DMDC-M-LS(entry 8) are lower than that of DMDC-g -EM hydrogel (entry 4).The water uptakes of hydrogels 4, 7 and 8 were measured to be1,040±30%, 830±8% and 110±3% respectively (SupplementarySection S3.5). On the basis of the Peppas–Merill theory26, we

calculated the pore sizes of hydrogels 4, 7 and 8 to be 16.5 nm,10.4 nm and 1.6 nm respectively (Supplementary Section S3.6).The zeta potentials of hydrogels 7 and 8 (10.9 ± 0.4mV and11.9± 0.1mV, Supplementary Section S3.7) are similar to that ofhydrogel 4 (11.4± 0.2mV), ruling out the effect of differences incharge density in this series. Decreasing the water swelling andhence pore size in hydrogel 8 significantly decreases its killingefficacy (Fig. 2a). Conversely, hydrogel 4 has the best killing efficacybecause of its largest pore size of about 16.5 nm. Each folded LPSmolecule is estimated to be about 3.5 nm–4.5 nm in size and soit seems that hydrogel 8 does not have large enough pores toreceive these molecules. Computer simulation also confirmed thatdecreased hydrogel pore size causes fewer LPS molecules to bepulled into the hydrogel (Supplementary Section S4.4).

The DMDC-Q-g -EM hydrogel was repeatedly challenged withS. aureus up to four times. The tested hydrogel films were washedwith PBS and vortexed before the next use. Figure 2b showsthat, even after three previous challenges with high concentrationsof microbes, the dead cells easily washed from the hydrogel,preserving its antimicrobial activity in the next (fourth) challenge.One problem of present contact-active antimicrobial surfacesin application is that they can easily be masked by adsorbedconditioning films from organic compounds such as proteins orremnants of dead cells, resulting in loss of effectiveness29. Thisconditioning film undoubtedly blocks contact-active antimicrobialsurfaces from contact with microbes, causing loss of antimicrobialactivity. In contrast, hydrated PEGylated hydrogels are knownto be inherently cell/protein resistant30, accounting partly forthe observed durable antimicrobial activity. Further, diffusion ofdisrupted membrane molecules into the interior of the gel layeris expected to prevent the gel surface from becoming fouledafter repeated challenges. Large pores are needed to admit themicrobe membrane fragments. Also, a high total pore volumefraction guarantees plenty of interior volume for membranefragments to diffuse into, so that the contact-active surfacepossibly cleans itself if it becomes momentarily fouled by absorbedmembrane components after contactwith amicrobe.Our hydrogels(entries 1–7, not entry 8) have high pore volume, as can beinferred from their high water volume fractions of greater than 90%(Supplementary Fig. S6).

As a demonstration of application of our material and coating, ahydrogel layer of DMDC-Q-g -EMwas in situ immobilized on a flu-oropolymer substrate (contact lens Z from Menicon). The surfaceto be coated was first surface activated with peroxides using argonplasma followed by air ageing, and the substrate was then placed inthe hydrogel precursor solution and photopolymerized without a





NATURE MATERIALS DOI: 10.1038/NMAT2915 ARTICLESDMDC-Q-g-EM hydrogelControl

0

1

2

3

4

5

6a

b c

d

e

f

g

h

HydrogelDMDC-M-LSDMDC-M

P. aeruginosaE. coliS. aureusF. solani

Log

redu

ctio

n

90.000

99.000

99.900

99.990

99.999

%kill

1 3 4 5 6 97 8Coating

(DMDC-Q)DMHCC TMC DMDC TMC-Q DMDC-Q

0

1

2

3

4

5

6

No. of repeated challengeson DMDC-Q hydrogel (6)

99.99

99.99

Log

redu

ctio

n

S. aureus

2 3 41

90.00

99.00

99.90

%kill

P. aeruginosa

E. coli

S. aureus

F. solani

20 µm

2

20 µm

1 µm1 µm

1 µm 1 µm

1 µm 1 µm

10 µm 10 µm

1 µm 1 µm

1 µm 1 µm

10 µm 10 µm

1 µm

Figure 2 |Antimicrobial activities of qC-g-EM hydrogels against various bacteria and fungi. a, Log reduction and %kill of four pathogens on variousqC-g-EM hydrogels. b, Multiple-use antimicrobial activities of DMDC-Q-g-EM hydrogel (entry 6) against S. aureus. Error bars represent mean± standarddeviation of mean for n= 3. c,d, LIVE/DEAD bacterial viability assay of S. aureus on PEG13DAcontrol (c) and DMDC-Q-g-EM hydrogel (entry 6) (d) after1 h incubation at 37 ◦C and washing (scalebar= 20 µm). e–h, Morphology of various pathogens in contact with DMDC-Q-g-EM hydrogel (right, entry 6)and control (left). Arrows indicate lesions and holes on the cell membrane after contact with DMDC-Q-g-EM hydrogel.

a b

dc

10 µm 10 µm

10 µm 10 µm

Figure 3 | Coating of DMDC-Q-g-EM hydrogel on fluoropolymer substrate. a,b, Photographic image of uncoated (a) and fluorescein-stainedDMDC-Q-g-EM hydrogel-coated (b) substrate. c,d, Scanning electron microscopy images of top view and cross-section (inset) of uncoated (c) andfreeze-dried DMDC-Q-g-EM hydrogel- (entry 6) coated (d) substrate surface.

mould. Under ultraviolet irradiation, a crosslinked hydrogel layeremanates from the surface as the acrylate/methacrylate function-ality of the precursor solution reacts with the peroxide-decoratedsurface and with itself. The coating is concurrently covalentlyattached to the surface as the hydrogel is crosslinked from the

precursor solution. Our method is simpler than typical proceduresfor forming antimicrobial coatings.

Figure 3 shows photographs and scanning electron microscopyimages of the hydrogel-coated fluoropolymer substrate, togetherwith the control uncoated disc. Fluorescein, a dye that binds to





0

1.2a

b c

d e

f g

1.0

0.8

0.6

0.4

0.2

TCPS control

DMDC-Q hydrogel (6)

50 µm 50 µm

1 day 3 days 5 days 7 days

MT

T a

bsor

banc

e

100 µm

100 µm100 µm

100 µm

Figure 4 | In vitro and in vivo biocompatibility studies. a, MTT activities(absorbance at 490 nm) of primary epidermal keratinocytes on TCPScontrol and DMDC-Q-g-EM hydrogel (entry 6). Error bars represent mean± standard deviation of mean for n= 3. b,c, LIVE/DEAD analysis of primaryepidermal keratinocytes on TCPS control (b) and DMDC-Q-g-EM hydrogel(entry 6) (c) after 7 days of culture. d–g, Microscopic observations ofhaematoxylin- and eosin-stained frozen sections of conjunctiva. d, Normalconjunctiva epithelium showing normal epithelium and stromal bloodvessels. e, PO day 5 positive control, tissue overlying the surgically createdpocket without a lens implant. f, PO day 5, tissue overlying the pocket withan uncoated lens. g, PO day 5, tissue overlying the pocket containing aDMDC-Q-g-EM hydrogel-coated lens (entry 9). White arrows indicate theconjunctival epithelium and black arrows indicate blood vessels.

the N+ present on the quaternized chitosan, was used to stain thehydrogel coating. Figure 3a,b shows the gross visual appearanceof the uncoated and fluorescein-stained hydrogel-coated surfaces.Figure 3b shows that the hydrogel layer was fairly uniformlycoated on the substrate. Figure 3d shows that the freeze-driedhydrogel layer was a few micrometres thick and had good adhesionto the fluoropolymer substrate, although the hydrogel has veryhigh water content. The thin layer of hydrogel grafted on thesurface showed good antimicrobial action, with log reductions of2.1–4.2, which are comparable to those of free DMDC-Q-g -EMhydrogel films (Fig. 2a).

The biocompatibility of DMDC-Q-g -EM hydrogel was demon-strated by both in vitro and in vivo tests. The in vitro test contacted

human primary epidermal keratinocytes with DMDC-Q-g -EMhydrogel films for 7 days, with tissue culture polystyrene dish(TCPS) as control (Supplementary Section S3.8). Methyl tetra-zolium (MTT) and LIVE/DEAD cell viability assays (Fig. 4a–c)show that the human keratinocytes proliferate well in contact withDMDC-Q-g -EM hydrogel. The MTT assay measures mitochon-drial activity and cell viability and proliferation. Figure 4a showsthat the MTT absorbance increases with culture time, indicatingthat the cells proliferate even in the presence of the hydrogel. InFig. 4b,c, the keratinocytes are mostly stained green, indicating thatthey are alive on the hydrogels, suggesting that our hydrogel isin vitro biocompatible.

The in vivo biocompatibility of DMDC-Q-g -EM hydrogelcoating with the epithelial cells of the rabbit ocular surface wasexamined. Uncoated and coated lenses were each implanted intoa pocket underneath the rabbit conjunctival epithelium. Theresponse of the conjunctival tissue was examined clinically byslit-lamp and in vivo confocal microscopy over 5 days. The bulbarconjunctiva overlying the control and DMDC-Q-g -EM hydrogel-coated lenses remained healthy as determined by examination within vivo confocal microscopy on post-operative (PO) days 3 and5 (Supplementary Fig. S12). Representative histological imagesof rabbit conjunctiva collected at PO day 5 and stained byhaematoxylin and eosin are shown in Fig. 4d–g. Importantly, therewas no indication of epithelial erosion, or of unusual neutrophilinfiltration other than what may be expected after surgery. A fewneutrophils appeared in the conjunctival tissue associated withthe lens (Fig. 4f,g). No erosions of the epithelium, or vascularchanges, were found to be associated with the presence of thelens. The in vivo results show that the coating has no effecton the epithelial cells and does not lead to any pathologicalchanges. The coated lens is compatible with conjunctiva andthe hydrogel coating has no toxicity to the epithelial cells orthe underlying stroma.

Our hydrogel has good in vitro and in vivo biocompatibility.Unlike microbial membranes, the outer leaflet of the mammaliancell membrane lacks anionic lipids12,25. There is therefore muchless Coulombic attractive ‘suctioning’ force to disrupt mammaliancells when they are in contact with the ‘anion sponge’, whichaccounts for its excellent biocompatibility. The lack of Coulom-bic attraction between DMDC-Q-g -EM and zwitterionic outermammalian membrane is also shown by computer simulation(Supplementary Section S4.3).

We have demonstrated simultaneous polymerization and cova-lent grafting of a qC-g -EM-based hydrogel onto a fluoropolymersubstrate surface by means of a simple two-step process to create auniform covalent adherent thin coating that effectively reduces theviability of four tested pathogens by two tomore than four orders ofmagnitude. Specifically, DMDC-Q-g -EM hydrogel exhibits supe-rior antimicrobial activity with an inhibition above 99% for all fourclinically significant pathogens tested, includingGram-negative andGram-positive bacteria and fungi. Our hydrogel is contact-activeantimicrobial and also in vitro and in vivo biocompatible. Thecoating has very good adhesion and is re-useable. We proposethat the polycationic hydrogel acts like an ‘anion sponge’ to drawanionic phospholipids out of the bacterial cell membrane into thegel pores, and that the hydrogel positive charge density and poresize determine the killing efficacy of the coating. We believe thatthis PEGylated quaternized chitosan-based hydrogel coating, whichis easily ultraviolet immobilized on any surface, will be widely appli-cable for combating infection in many classes of implant/prosthesisand in other medical devices.

MethodsSynthesis of qC-g -EMs. Quaternized chitosan (qC) was first synthesized by eitherroute I or route II (shown in Fig. 1a). For quaternized TMC (route I), chitosan





NATURE MATERIALS DOI: 10.1038/NMAT2915 ARTICLES(1 g, 6.21mmol) in N -methyl-2-pyrrolidone (50ml) was added to NaOH solution(1.5M, 15ml). Sodium iodide (1.08 g, 7.23 mmol) and methyl iodide (11.2 g,78.7mmol) were then added to the chitosan/ N -methyl-2-pyrrolidone/NaOHmixture and then reacted for 24 h at 50 ◦C. For DMDC (route II), chitosan (1 g,6.21mmol) was first predissolved in acetic acid (1%, 100ml) before addingdecanal (0.97 g, 6.2mmol). After an hour of stirring at room temperature, thesolution pH was increased to 4.5 followed by addition of sodium borohydride(9.32mmol). The product was precipitated by adding NaOH solution (1M). ForTMC-Q and DMDC-Q, the reaction procedure to obtain TMC and DMDC wascarried out followed by three further repetitions of the methylation reaction. Toobtain qC-g -EM, NaOH solution (0.38M, 0.30ml) was added to quaternizedchitosan solution (0.2 g in 0.45ml water), followed by addition of Cl-PEG6M(0.50 g) pre-dissolved in isopropanol (1.50ml). The solution was stirred atroom temperature for 3 h and qC-g -EM was obtained through precipitation andcentrifugation followed by dialysis. Other details of the synthesis can be found inSupplementary Section S1.

Preparation of hydrogel and hydrogel coating. A typical hydrogel was preparedby blending qC-g -EM (10wt%) and PEG13 DA (10wt%) in deionized water(80wt%). The mixed solution was added with the photoinitiator Irgacure 2959(0.1 wt%) and ultraviolet irradiated for 15min (at wavelength 365 nm and intensityof 10mWcm−2). Other details can be found in Supplementary Section S3. Thehydrogel coating of the contact lens (contact lens Z, Menicon) was carried outsimilarly but without using the photoinitiator and by first surface activating thecontact lens with plasma (Supplementary Section S5.1).

Antimicrobial assay for hydrogels. The hydrogel films were soaked and rinsed insterilized PBS for 3 days and then cut into discs of 1.5 cm diameter. 10 µl bacterialsuspension (5×108 CFUml−1) was spread onto each hydrogel film in a tissueculture plate. The inoculated hydrogel films were incubated for 1 h at 37 ◦C (28 ◦Cfor F. solani) and a relative humidity of not less than 90%. 1ml of neutralizing brothwas then added to each well to recover any microbial survivors. A series of tenfolddilutions was prepared, and plated out in Luria–Bertani agar (yeast–malt agar forF. solani). The plates were incubated for 36–48 h at 35 ◦C (28 ◦C for F. solani),and counted for colony-forming units. A hydrogel of PEG13 DA (20wt%) withoutqC-g -EMwas used as a control. The results are expressed as

log reduction

= log (cell count of control)− log(survivor count on qC-g -EM hydrogel)

%kill=cell count of control− survivor count on qC-g -EM hydrogel

cell count of control×100

LIVE/DEAD assay to examine bacterial viability. 10 µl bacterial suspension(5×108 CFUml−1) in PBS was spread onto the hydrogel films, which were thenincubated for 1 h at 37 ◦C and a relative humidity of not less than 90%. Thefilms were then stained with a LIVE/DEAD Kit (L13152, Invitrogen) for 30minat room temperature. After rinsing with PBS, they were imaged with a Zeissinverted optical microscope.

Computer simulation of microbe killing mechanism. The simulations werecarried out with the GROMACS package, using an all-atom GLYCAM/AMBERforce field. The hydrogel coating was modelled by three layers of quaternizedchitosan derivative chains in which each layer consisted of eight quaternizedchitosan derivative chains with an inter-chain separation of about 2.5 nm.Each quaternized chitosan derivative chain contained ten monosaccharideunits. Each chitosan chain in DMDC-Q hydrogel was modelled to contain fivepositive charges. Position restraints were applied to tip-linkage oxygen atomsof the chitosan chains to simulate the crosslinking effect of crosslinkers inthe hydrogel. The outer layer of the P. aeruginosa membrane was simulatedto be composed of 1,2-dioleoyl-sn-glycero-3-phosphocholine and LPS. Therough LPS structure of P. aeruginosa PAO1 was modelled on the basis ofexperimentally sequenced data28.

In vivo biocompatibility study. The DMDC-Q-g -EM hydrogel-coated anduncoated contact lenses were inserted in a manually created sub-conjunctivalpocket in the upper bulbar region of the eye (female white New Zealand rabbits,n= 3, 3.0–3.5 kg). In vivo confocal microscopy (Heidelberg HRT3, HeidelbergEngineering, Germany) was carried out on PO days 3 and 5 to evaluate theappearance of the surface conjunctival epithelial cells overlying the pocket.Conjunctiva overlying the pocket were collected on PO day 5 and embedded inoptimal cutting temperature compound (Leica, Nussloch, Germany). Cryosections(Microm, Walldorf, Germany) of 10 µm thickness from each animal were stainedwith haematoxylin and eosin (Sigma-Aldrich, Missouri, USA), then imaged witha Zeiss Axioplan microscope (Zeiss, Oberkochen, Germany). (More details are inSupplementary Section S5.3.)

Received 23 May 2010; accepted 2 November 2010;published online 12 December 2010

References1. Hetrick, E. M. & Schoenfisch, M. H. Reducing implant-related infections:

Active release strategies. Chem. Soc. Rev. 35, 780–789 (2006).2. Ferreira, L. & Zumbuehl, A. Non-leaching surfaces capable of killing

microorganisms on contact. J. Mater. Chem. 19, 7796–7806 (2009).3. Klibanov, A. M. Permanently microbicidal materials coatings. J. Mater. Chem.

17, 2479–2482 (2007).4. Kristinsson, K. G. et al. Antimicrobial activity of polymers coated with

iodine-complexed polyvinylpyrrolidone. J. Biomater. Appl. 5, 173–184 (1991).5. Smith, A. W. Biofilms and antibiotic therapy: Is there a role for combating

bacterial resistance by the use of novel drug delivery systems. Adv. DrugDelivery Rev. 57, 1539–1550 (2005).

6. Milovic, N. M., Wang, J., Lewis, K. & Klibanov, A. M. Immobilized N-alkylatedpolyethylenimine avidly kills bacteria by rupturing cell membranes with noresistance developed. Biotechnol. Bioeng. 90, 715–722 (2005).

7. Lin, J., Qiu, S. Y., Lewis, K. & Klibanov, A. M. Bactericidal properties offlat surfaces and nanoparticles derivatized with alkylated polyethylenimines.Biotechnol. Prog. 18, 1082–1086 (2002).

8. Tiller, J. C., Liao, C. J., Lewis, K. & Klibanov, A. M. Designing surfaces that killbacteria on contact. Proc. Natl Acad. Sci. USA 98, 5981–5985 (2001).

9. Ilker, M. F., Nusslein, K., Tew, G. N. & Coughlin, E. B. Tuning the hemolyticand antibacterial activities of amphiphilic polynorbornene derivatives.J. Am. Chem. Soc. 126, 15870–15875 (2004).

10. Kuroda, K. & DeGrado, W. F. Amphiphilic polymethacrylate derivatives asantimicrobial agents. J. Am. Chem. Soc. 127, 4128–4129 (2005).

11. Tew, G. N., Clements, D., Tang, H., Arnt, L. & Scott, R. W. Antimicrobialactivity of an abiotic host defense peptide mimic. Biochim. Biophys. Acta 1758,1387–1392 (2006).

12. Gabriel, G. J., Som, A., Madkour, A. E., Eren, T. & Tew, G. N. Infectiousdisease: Connecting innate immunity to biocidal polymers.Mater. Sci. Eng. R57, 28–64 (2007).

13. Kenawy, E. R., Worley, S. D. & Broughton, R. The chemistry and applicationsof antimicrobial polymers: A state-of-the-art review. Biomacromolecules 8,1359–1384 (2007).

14. Bagheri, M., Beyermann, M. & Dathe, M. Immobilization reduces the activityof surface-bound cationic antimicrobial peptides with no influence upon theactivity spectrum. Antimicrob. Agents Chemother. 53, 1132–1141 (2009).

15. Imazato, S., Russell, R. R. B. & McCabe, J. F. Antibacterial activity of MDPBpolymer incorporated in dental resin. J. Dent. 23, 177–181 (1995).

16. Sambhy, V., Peterson, B. R. & Sen, A. Antibacterial and hemolytic activitiesof pyridinium polymers as a function of the spatial relationship betweenthe positive charge and the pendant alkyl tail. Angew. Chem. Int. Ed. 47,1250–1254 (2008).

17. Stratton, T. R., Rickus, J. L. & Youngblood, J. In vitro biocompatibilitystudies of antibacterial quaternary polymers. Biomacromolecules 10,2550–2555 (2009).

18. Zumbuehl, A. et al. Antifungal hydrogels. Proc. Natl Acad. Sci. USA 104,12994–12998 (2007).

19. Fuchs, A. D. & Tiller, J. C. Contact-active antimicrobial coatings derived fromaqueous suspensions. Angew. Chem. Int. Ed. 45, 6759–6762 (2006).

20. Nurdin, N., Helary, G. & Sauvet, G. Biocidal polymers active by contact.2. Biological evaluation of polyurethane coating with pendant quaternaryammonium-salts. J. Appl. Polym. Sci. 50, 663–670 (1993).

21. Madkour, A. E., Dabkowski, J. A., Nusslein, K. & Tew, G. N. Fast disinfectingantimicrobial surfaces. Langmuir 25, 1060–1067 (2009).

22. Jia, Z. S., Shen, D. F. & Xu, W. L. Synthesis and antibacterial activities ofquaternary ammonium salt of chitosan. Carbohydr. Res. 333, 1–6 (2001).

23. Mao, S. R. et al. Synthesis, characterization and cytotoxicity ofpoly(ethyleneglycol)-graft -trimethyl chitosan block copolymers. Biomaterials26, 6343–6356 (2005).

24. Zhu, S. Y., Qian, F., Zhang, Y., Tang, C. & Yin, C. H. Synthesis andcharacterization of PEG modified N-trimethylaminoethylmethacrylatechitosan nanoparticles. Eur. Polym. J. 43, 2244–2253 (2007).

25. Theis, T. & Stahl, U. Antifungal proteins: Targets, mechanisms and prospectiveapplications. Cell. Mol. Life Sci. 61, 437–455 (2004).

26. Peppas, N. A., Hilt, J. Z., Khademhosseini, A. & Langer, R. Hydrogels in biologyand medicine: From molecular principles to bionanotechnology. Adv. Mater.18, 1345–1360 (2006).

27. Li, Q., Wang, D. A. & Elisseeff, J. H. Heterogeneous-phase reactionof glycidyl methacrylate and chondroitin sulfate: Mechanism ofring-opening-transesterification competition. Macromolecules 36,2556–2562 (2003).

28. Sadovskaya, I., Brisson, J. R., Lam, J. S., Richards, J. C. & Altman, E. A.Structural elucidation of the lipopolysaccharide core regions of the wild-typestrain PAO1 and O-chain-deficient mutant strains AK1401 and AK1012 fromPseudomonas aeruginosa serotype O5. Eur. J. Biochem. 255, 673–684 (1998).





29. Cheng, G., Xue, H., Zhang, Z., Chen, S. & Jiang, S. A switchable biocompatiblepolymer surface with self-sterilizing and nonfouling capabilities. Angew. Chem.Int. Ed. 47, 8831–8834 (2008).

30. Ostuni, E., Chapman, R. G., Holmlin, R. E., Takayama, S. & Whitesides, G. M.A survey of structure–property relationships of surfaces that resist theadsorption of protein. Langmuir 17, 5605–5620 (2001).

AcknowledgementsThis work was funded and supported by Menicon Holdings (Japan), a SingaporeMinistry of Education Tier 2 grant (M45120007), Nanyang Technological University(Singapore) and a Singapore SingHealth Foundation grant (SHF/09/GMC(1)/012(R)(R705)). R.W.B. and H-Y.Z. were supported by NMRC/TCR/002-SERI/2008 R618. Y.C.was supported by SingHealth Foundation SHF/09/GMC(1)/012(R) (R705). W.L. andY.M. were supported by a Singapore Ministry of Education Tier 2 grant (T206B3210RS).We acknowledge the Singapore General Hospital (Pathology Department) for carryingout some of the early antimicrobial tests. We thank Y. Shucong, W. Xiujuan and F. Ningfor their help in using field emission scanning electron microscopy, scanning electron

microscopy and atomic force microscopy. The provision of computation time from theNTUHPC centre is gratefully acknowledged.

Author contributionsP.L. carried out the testing and coating experiments. Y.F.P., P.L. and S.H.Y. didthe syntheses and characterization of all the polymers. Y.C. carried out the in vitrobiocompatibility studies. X.Q. carried out some early antimicrobial testing. W.L. andY.M. did the computer simulation and related writing. H-Y.Z. and R.W.B. did the animalstudy and related writing. C.Z., E-T.K., M.L., M.W.C., S.S.J.L., C.M.L. and M.B.C-P.advised on the design and interpretation of the experiments.M.B.C-P. directed the overallproject. P.L., Y.F.P. andM.B.C-P. did themain writing of themanuscript.

Additional informationThe authors declare competing financial interests: details accompany the paper atwww.nature.com/naturematerials. Supplementary information accompanies this paperon www.nature.com/naturematerials. Reprints and permissions information is availableonline at http://npg.nature.com/reprintsandpermissions. Correspondence and requestsfor materials should be addressed to M.B.C-P.




www.nature.com/naturematerials


http://npg.nature.com/reprintsandpermissions


Documents

A polycationic antimicrobial and biocompatible hydrogel with microbe membrane suctioning ability