WoundRepairandRegeneration Development of a novel, highly quantitative in vivo model for the study of biofilm-impaired cutaneous wound healing AnandevN.Gurjala,MD,MS1;MatthewR.Geringer,BS1;AkhilK.Seth,MD1;SeokJ.Hong,PhD1; MarkS.Smeltzer,PhD2;RobertD.Galiano,MD1;KaiP.Leung,PhD3;ThomasA.Mustoe,MD1 1.DivisionofPlasticSurgery,FeinbergSchoolofMedicine,NorthwesternUniversity,Chicago,Illinois, 2.DepartmentofMicrobiologyandImmunology,UniversityofArkansasMedicalSciences,LittleRock,Arkansas,and 3.MicrobiologyBranch,USArmyDentalandTraumaResearchDetachment,InstituteofSurgicalResearch,FortSamHouston,Texas Reprintrequests: ABSTRACT ThomasA.Mustoe,DivisionofPlastic Surgery,FeinbergSchoolofMedicine, A growing body of evidence suggests that in addition to hypoxia, ischemia- NorthwesternUniversity,675N.St.Clair, reperfusioninjury,andintrinsichostfactors,bacterialbiofilmsrepresentafourth Suite19250,Chicago,IL60611. majorpillarinchronicwoundpathogenesis.Giventhatmoststudiestodaterely Tel:113126956022; oninvitroorobservationalclinicaldata,ouraimwastodevelopanovel,quan- Fax:113126955672; titativeanimalmodelenablingfurtherinvestigationofthebiofilmhypothesisin Email:[email protected] vivo.DermalpunchwoundswerecreatedinNewZealandrabbitears,andused as uninfected controls, or inoculated with green fluorescent protein-labeled Manuscriptreceived:September24,2010 Staphylococcusaureustoformwoundswithbacteriapredominantlyintheplank- Acceptedinfinalform:January25,2011 tonic or biofilm phase. Epifluorescence and scanning electron microscopy re- vealedthatS.aureusrapidlyformsmaturebiofilminwoundswithin24hoursof DOI:10.1111/j.1524475X.2011.00690.x inoculation, with persistence of biofilm viability over time seen through serial bacterial count measurement and laser scanning confocal imaging at different time points postwounding and inoculation. Inflammatory markers confirmed thatthebiofilmphenotypecreatesacharacteristic,sustained,low-gradeinflam- matory response, and that over time biofilm impairs epithelial migration and granulation tissue in-growth, as shown histologically. We have established and validated a highly quantitative, reproducible in vivo biofilm model, while pro- viding evidence that the biofilm phenotype specifically contributes to profound cutaneous wound healing impairment. Our model highlights the importance of bacterialbiofilmsinchronicwoundpathogenesis,providinganinvivoplatform forfurtherinquiryintothebasicbiologyofbacterialbiofilm–hostinteractionand high-throughputtestingofantibiofilmtherapeutics. Bacterialbiofilms,sessilecommunitiesofbacteriaencased woundhealingprocess.PhilStewartfromMontanaState in a protective polysaccharide matrix, can find harbor in Universityinparticularhasputforthaconsiderableseries manyparts of thebody, having beenreportedonbiolog- ofpapersarticulatingthishypothesis.Inaclinicalstudyof icalsurfacesasdisparateasmyocardium,nasalepithelium, humanwounds,hisgroupconvincinglydemonstratedus- anddentalenamel.1Oneachofthesesurfacestherelation- ing light and scanning electron microscopy (SEM) that ship between bacteria and host may range from purely bacteriainbiofilm-phasearepresentinalltypesofchronic commensal (as in the intestinal gut) to highly destructive wounds,establishingakeylinkbetweenbiofilmandnon- when established on microbiologically naıve tissues (such healing.3Studiesbyothergroupsusingsophisticatedmo- asinendocarditisorcysticfibrosis).Whetherondamaged lecular techniques for bacterial recovery have further heartvalvesorpoorlyfunctioningrespiratoryepithelium, established that a diverse array of bacterial species exists the pathophysiology of bacterial biofilms remains con- inhumanchronicwounds,andthatthesebacteriapersist served:hostdefensesarerenderedunabletoclearbiofilm over time.4,5 Moreover, a variety of in vitro models have duetoitsadaptivesurvivalmechanisms,andyetarecon- made essential contributions, including demonstration of stantly stimulated by its physical presence, shedding of robustbiofilmresistancetoantimicrobials,6importantge- planktonic bacteria, and elaboration of virulence factors. netic pathways for bacterial communication including The result is a chronic inflammatory cycle causing injury quorum sensing,7 and even inhibitory effects of biofilms tohosttissuesovertime.2 againstculturedhumankeratinocytes.8 Inrecentyearsthewoundbedofcompromisedskinhas While the aforementioned studies have been crucial to been thought to offer yet another biological surface onto developinganunderstandingofbiofilminwounds,equiv- which bacterial biofilms can take hold. The susceptibility alentstrideshavenotbeenmadeintheareaofinvivoex- ofopenwoundstobacterialseedingandthemoist,nutri- perimentation,principallyduetothelackofanestablished tionally supportive microenvironment of the wound ma- animalmodel.Giventhecomplexityofwoundhealing,ex- trixcreateanidealsettingforbiofilmtointerferewiththe trapolation from in vitro biofilm studies to the clinic has 400 WoundRepReg(2011)19400 410 (cid:2)c 2011bytheWoundHealingSociety Report Documentation Page Form Approved OMB No. 0704-0188 Public reporting burden for the collection of information is estimated to average 1 hour per response, including the time for reviewing instructions, searching existing data sources, gathering and maintaining the data needed, and completing and reviewing the collection of information Send comments regarding this burden estimate or any other aspect of this collection of information, including suggestions for reducing this burden, to Washington Headquarters Services, Directorate for Information Operations and Reports, 1215 Jefferson Davis Highway, Suite 1204, Arlington VA 22202-4302 Respondents should be aware that notwithstanding any other provision of law, no person shall be subject to a penalty for failing to comply with a collection of information if it does not display a currently valid OMB control number 1. REPORT DATE 2. REPORT TYPE 3. DATES COVERED 01 MAY 2011 N/A - 4. TITLE AND SUBTITLE 5a. CONTRACT NUMBER Development of a novel, highly quantitative in vivo model for the study of 5b. GRANT NUMBER biofilm-impaired cutaneous wound healing. 5c. PROGRAM ELEMENT NUMBER 6. AUTHOR(S) 5d. PROJECT NUMBER Gurjala A., Geringer M. R., Hong S.-J. Smeltzer M. S., Galiano R. D., 5e. TASK NUMBER Leung K. P., Mustoe T. A., 5f. WORK UNIT NUMBER 7. PERFORMING ORGANIZATION NAME(S) AND ADDRESS(ES) 8. PERFORMING ORGANIZATION United States Army Institute of Surgical Research, JBSA Fort Sam REPORT NUMBER Houston, TX 9. SPONSORING/MONITORING AGENCY NAME(S) AND ADDRESS(ES) 10. SPONSOR/MONITOR’S ACRONYM(S) 11. SPONSOR/MONITOR’S REPORT NUMBER(S) 12. DISTRIBUTION/AVAILABILITY STATEMENT Approved for public release, distribution unlimited 13. SUPPLEMENTARY NOTES 14. ABSTRACT 15. SUBJECT TERMS 16. SECURITY CLASSIFICATION OF: 17. LIMITATION OF 18. NUMBER 19a. NAME OF ABSTRACT OF PAGES RESPONSIBLE PERSON a REPORT b ABSTRACT c THIS PAGE UU 11 unclassified unclassified unclassified Standard Form 298 (Rev. 8-98) Prescribed by ANSI Std Z39-18 Gurjalaetal. Animalmodelforbiofilmimpairedwoundhealing been challenging, because only animal models can take plasmid, pCN57, was introduced to UAMS-1 by phage intoaccountsuchcriticalinvivofactorsastheinterplayof transduction as described by Charpentier et al.11 with the host immune response and wound bed components modifications.Briefly,strainRN9623(recentlydesignated withbacteria.Secondly,humanstudieshavebeenonlyob- as NARSA strain ID# NRS623, which contains pCN57) servational and correlative, lacking the experimental and wasculturedtoexponentialgrowthphaseinTSBcontain- causative data necessary to fully establish the biofilm hy- ing 5mM CaCl (Sigma-Aldrich, St. Louis, MO) and 2 pothesis.Thereisthusawideconsensusamongthebiofilm 10mg/mL erythromycin (Sigma-Aldrich). To prepare research community on a need for more in vivo biofilm phage lysates, strain RN9623 was infected with phage 11 models,andthattheywillbenecessarytoolsforfurthering at a multiplicity of infection (MOI) 0.5 and cultured at ourunderstandingofbiofilm–hostbiology. 371Cuntilcelllysisoccurred.Lysateswerecentrifugedto Inthepresentstudyweaimto buildonpreviouswork removecelldebrisandfilteredthrougha0.45mmfilter.For by developing a novel, in vivo biofilm model in rabbit, transduction, phage lysates at a MOI 0.2 were incubated capableofpreciselydefiningtheimpactofbiofilmsoncu- withUAMS-1thathasbeenresuspendedinTSBcontain- taneouswoundhealing.Throughrigoroustesting,weval- ing5mMCaCl .After20minutesincubationat371C,the 2 idate that our model allows for biofilm formation within reactionmixturewaswashedwithice-cold20mMsodium woundsandmaintenanceofitsstructureandviabilityover citrate and plated in TSA containing 10mg/mL erythro- time.Althoughwoundinfectionsrepresentaspectrumof mycin.ThepresenceofpCN57inUAMS-1wasconfirmed bacterialphenotypes,involvingbacteriainboththeplank- byplasmidisolation,restrictionanalysis,andfluorescence tonicandbiofilmphases,weuseourmodeltostudydiffer- microscopy. ences in planktonic- and biofilm-dominant infections, which are classically associated with acute and chronic Woundprotocolandbacteriabiofilmmodel wound pathogenesis, respectively. In particular, we hy- pothesize that both planktonic-dominant, or active, and Forcreationofthedermalwounds,therabbitswereanes- biofilm-dominant wound infections will show severe thetized with an intramuscular injection of ketamine wound healing deficits when compared with uninfected (22.5mg/kg)andxylazine(3.5mg/kg)mixturebeforesur- wound beds. Furthermore, we propose that phenotypic gery.Earswereshavedandsterilizedwithabetadinescrub differencesbetweenthesetwotypesofinfectionwillleadto and70%ethanol.Followingintradermalinjectionof1% distinct host inflammatory responses even under circum- lidocaine with epinephrine local anesthetic, six 6-mm- stances where both infections have similar bacterial bur- diameterfull-thicknessdermalwoundswerecreateddown dens. Through these experiments, we hope to further totheperichondriumontheventralsurfaceoftheearand definetheintricaciesofwoundbacterialbiofilms,whilein- dressed with semiocclusive transparent film (Tegaderms, troducing and establishing an in vivo biofilm model that 3MHealthCare,St.Paul,MN).Forthecontrolarmofthe willserveasafoundationforfurthermechanisticandclin- experiments, wounds were redressed with sterile Tega- icallyrelevantinvestigation. derms on postoperative days (PODs) 3, 5, 6, 8, and 10. For the infected wound groups, bacteria grown on agar plates were harvested and proliferated in culturemedium MATERIALSANDMETHODS broth, followed by suspension in phosphate buffer solu- tion (PBS), from which a concentration was measured. Animals Approximately106colony-formingunits(CFUs)ofplank- tonicbacteriawereinoculatedfromsuspensionontoeach Young, adult New Zealand white rabbits (3–6 months, woundbedonPOD3andallowedtoproliferateunderthe (cid:3)3kg) were acclimated to standard housing and fed ad Tegadermsdressing.Fortheactiveinfectionarm,plank- libitum under a protocol approved by the Animal Care tonic bacteria were allowed to continue active prolifera- and Use Committee at Northwestern University. All ani- tion, with Tegaderms dressing changes on PODs 5, 6, 8 malswerehousedinindividualcagesunderconstanttem- and10.Forthebiofilmarm,topicalmupirocin(2%)anti- perature (221C) and humidity with a 12-hour light–dark biotic ointment (Teva Pharmaceuticals, Sellersville, PA) cycle.Atotalof82rabbitswereusedinthedevelopmentof wasappliedonPOD4toeliminateplanktonic-phasebac- this model after a series of pilot studies manipulating the teria and create a predominantly biofilm-phase infection. followingvariablesindifferentpermutations:woundsize, In order to prevent seroma formation and maintain woundlocation,numberofwoundsperear,wounddepth biofilm-phase infection by preventing regrowth of plank- (partialvs.fullthickness),timingofinoculation,lengthof tonic bacteria, an antimicrobial absorbent dressing antibiotictreatment,typeofabsorptivedressing,timingof containing polyhexamethylene biguanide (Telfas AMD, dressingchanges,andlengthofwoundhealingperiodbe- Tyco Healthcare Group, Mansfield, MA) was applied to fore harvest. Following exhaustive characterization of biofilmwoundsonPODs5,6,8,and10.Alldressingswere thesepermutationsthefinalmodelprotocoldescribedbe- checkeddailythroughouttheentireprotocol. lowwaschosenbasedonrecapitulationofessentialaspects ofthehumanbiofilm-infectedchronicwound. Harvestingofwounds Aftereuthanizingtheanimalsbyanintracardiaceuthasol Bacterialstrains,plasmid,andphagetransduction injection,woundswereharvestedforvariousanalyses.For Staphylococcus aureus strain UAMS-19,10 was grown on the biofilm time course experiment, wounds were har- tryptic soy agar (TSA) and in tryptic soy broth (TSB) at vested at 6-hour intervals following inoculation until 24 371C. The green fluorescent protein (GFP)-expressing hours,andanalyzedbyepifluorescenceandSEM.Totest WoundRepReg(2011)19400 410 (cid:2)c 2011bytheWoundHealingSociety 401 Animalmodelforbiofilmimpairedwoundhealing Gurjalaetal. for the presence of bacteria in the wound bed, bacterial werealsoanalyzedusinglaserscanningconfocalmicroscopy countsandlaserscanningconfocalmicroscopywerecon- (Zeiss LSM 510 META, Carl Zeiss Microimaging, LLC, ductedonPODs4,6,8,10,and12.Inaddition,SEMwas Thornwood,NY)tovisualizeGFP-labeledS.aureus. performed POD 12 to visually confirm biofilm formation within the wound bed. For the wound healing experi- SEM ments,woundswereharvestedatPOD12forhistological analysisusinghematoxylin&eosin(H&E)staining.RNA Todeterminethestructureofbiofilmsformedinrabbitear extractionandreversetranscriptionquantitativepolymer- wounds, wound samples were mounted by double-sided asechainreaction(qPCR)wascarriedoutonPOD6and tapetospecimenstubs,andexamined‘‘insitu’’byvariable POD 12 on wounds to determine the cytokine profiles as pressureSEM.Theprocedurebypassesthestepsforsam- theresultofthehostinflammatoryresponse. ple preparation used in conventional SEM and therefore minimizes the introduction of artifacts, providing an ex- cellent preservation of sample integrity. Imaging of the Viablebacterialcountsusingdropplatemethod sampleswasaccomplishedbytheuseofaCarlZeiss1450 For viable bacterial counts, wound beds were biopsied VP scanning electron microscope (Carl Zeiss Microima- with 10mm punches and using No. 15 scalpels (Becton ging,LLC)operatedatthegaschamberpressureof70Pa Dickinson AcuteCare, Franklin Lakes, NJ) the wound with the specimen cooling stage set at (cid:4)251C and the beds were excised and tissue samples collected into sepa- scanningvoltagesetat10kV. rate MagNA Lyser Green Beads tubes (Roche Diagnos- tics, Indianapolis, IN) each containing 1mL of PBS. All TotalmRNAextractionandreverse-transcriptionqPCR samples were homogenized at 5,000rpm for 30 seconds (MagNA Lyser, Roche Diagnostics), and then sonicated WoundswereharvestedformRNAextractionandsubse- (MicrosonUltrasonicCellDisrupter,HeatSystems-Ultra- quent cDNA conversion as part of reverse transcription sonics Inc., Farmingdale, NY) for 2 minutes at 5–8W to qPCR.Thedermallayerontheventralsideoftheearwas break up any bacterial aggregates that were present in removed and the wound bed and circumferential areas woundsamples.Timeintervalandpowersettingusedfor (rings) extending from the wound bed were punched out the sonication were preempirically determined to ensure and immediately snap-frozen in liquid nitrogen. Wound thatbacterialviabilitywasmaintainedintheprocess. samples were homogenized using a Mini-bead beater-8 Theresultingsolutionswereseriallydilutedandplated equipment(BiospecProductsInc.,Bartlesville,OK)using on both Blood Agar (Hardy Diagnostics, Santa Maria, zirconiabeads(2.0mmdiameter,BiospecProductsInc.)in CA) and Lipovitellin Salt Mannitol Agar (selective me- the presence of Trizol Reagent (Sigma-Aldrich). Total dium for S. aureus) (Hardy Diagnostics) and incubated RNAwasisolatedaccordingtothemanufacturer’sproto- overnightat371C.CFUsweredeterminedbythestandard col.ContaminatinggenomicDNAduringRNAprepara- colonycountingmethod. tionwasremovedusingtheTurboDNA-freekit(Ambion, Austin, TX). Five mg of total RNA was used to prepare cDNA using superscript II (Invitrogen) with 100ng of Histologicalanalysis randomprimers(Invitrogen). Wounds were excised with a 10mm punch and bisected at For quantitative analysis of the expression level of theirlargestdiameterforH&Estaining.Tissueswerefixed mRNAs,realtimeqPCRanalysesusingSYBRgreen1were in formalin, embedded in paraffin, cut into 4mm sections, performed utilizing an ABI prism 7000 sequence detection andstainedtobeanalyzedunderalightmicroscope.Slides system(AppliedBiosystems,FosterCity,CA).PCRprimers wereexaminedforquantificationofepithelialandgranula- weredesignedusingthePrimer3program(http://frodo.wi. tiongap,andtotalgranulationareausingadigitalanalysis mit.edu/). Expression of each gene was normalized to the system (NIS-Elements Basic Research, Nikon Instech Co., levelofglyceraldehyde3-phosphatedehydrogenase(GAD- Kanagawa, Japan) as described previously.12 Two blinded PH),thehousekeepinggene,togetDC.The2DDCt method t and independent observers evaluated all histological sec- wasusedtocalculategeneexpressionofinterleukin(IL)-1b tions. The results of both examiners were averaged. Slides andtumornecrosisfactor-a(TNF-a)ofcontrol,biofilm-in- wereomittedifresultsdiffered > 30%amongexaminers. fectedwounds,andwoundsunderactiveinfection.Expres- sion of genes was detected by PCR with the following Fluorescentstainingandlaserscanningconfocal oligonucleotides:IL-1b(50-CCACAGTGGCAATGAAAA TG-30and50-AGAAAGTTCTCAGGCCGTCA-30),TNF- microscopy a(50-CCAGATGGTCACCCTCAGAT-30 and50-TGTTC Woundsamplesforfluorescencestainingwereembeddedin TGAGAGGCGTGATTG-30), GADPH (50-AGGTCATC O.C.T. Compound (Sakura Finetek, Torrance, CA), snap- CACGACCACTTC-30 and 50-GTGAGTTTCCCGTTCA frozeninliquidnitrogen,andstoredat(cid:4)801Cuntiltimeof GCTC-30). cryosectioning. Ten micrometer sections of biofilm-infected woundtissueswerepreparedwiththeuseofacryostat.Sec- Statisticalanalysis tions for the biofilm time course experiment were stained with concanavalin-A conjugated to Texas Red (Invitrogen, Dataarepresentedingraphicalformasmean(cid:5)standard Carlsbad,CA)toidentifythepresenceofexopolysaccharides errors and were analyzed using the Student’s t-test (two- richinmannoseandglucosethatarecommonlythebasisof tailedandunpaired)tocomparecontrols,activeinfection, biofilm extracellular matrix (ECM), and with 40-6-diami- andbiofilm woundsamples.Thelevel ofsignificancewas dino-2-phenylindole to visualize host cells. Section slides setatp < 0.05. 402 WoundRepReg(2011)19400 410 (cid:2)c 2011bytheWoundHealingSociety Gurjalaetal. Animalmodelforbiofilmimpairedwoundhealing Figure1. Time-courseofbiofilmdevelopmentin6-hourintervalsfrominoculationtotheformationofmaturebiofilm.Staphylococ- cusaureusestablishesamaturebiofilminrabbitdermalulcerswithin24hoursasreviewedbyscanningelectronmicroscopy(SEM) (A–D)andfluorescencelightmicroscopy(E–H).(A)and(E)showthatat6hourspostinfectionbacteria(green)sparselycoverthe woundsurface(blue).At12hourspostinfectionS.aureusspreadsmoreevenlyoverthewoundsurface(BandF).(C)and(G)show thatby18hourspostinfectionS.aureusassumesearlybiofilmmorphology.Notethechangeofarchitectureofthebiofilmasre- vealedbySEM.By24hourspostinfection,amaturebiofilmofS.aureusisformedwithbacteriaencasedwithinexopolysaccharide matrix(red)(DandH)(magnification:A–D,(cid:6)200;E–H,(cid:6)1,000). RESULTS confocal microscopy was also performed at these time points,showingthepresenceofGFP-labeledS.aureusat To determine the kinetics of biofilm formation in vivo, a POD 6 (Figure 3A) and POD 12 (Figure 3B) within the time-course study using the rabbit biofilm wound model wound bed, clearly separated from the underlying, intact wasperformed.TheresultsshowedthatS.aureusformeda earcartilage.Inaddition,SEMwasperformedatPOD12, mature biofilm within 24 hours in wounds after inocula- showing the presence of individual bacteria within a well tion (Figure 1). A spatially organized three-dimensional biofilm structure began to emerge 18 hours postinocula- tionasrevealedbySEM(Figure1C).At24hourspostin- oculation the bacteria on the wound became more organized showing a complex architecture with dense ECM(Figure1D).Stainingofmannoseandglucosemoi- etieswithinthewoundbyTexasRed-labeledconcanavalin Aalsorevealedincreasingexopolysaccharidesaroundthe bacteriasuchthatby24hourstheyarecompletelyencased byacarbohydrate-richECMasshownbyepifluorescence microscopy(Figure1GandH). As part of model testing and validation, biofilm devel- opmentandviabilitywasverifiedthroughseveraldifferent techniques. Viable bacterial counts, measured at multiple timepointsfollowinginoculation,showedthepersistence oflivebacteriawithinthewoundbedsofbiofilmanimals Figure2. Viable bacterial count measurements from biofilm- (Figure2).Inparticular,followinginoculationonPOD3 infected wounds at successive time points following inocula- with approximately 106CFUs of planktonic bacteria, an tiononPOD3.Inoculationwithapproximately106colony-form- initial proliferation on POD 4 up to 107 is followed by ing units (CFUs) of Staphylococcus aureus is followed by an topical antibiotic and absorptive, antimicrobial dressing initialproliferationofbacteriabyPOD4,followedbytopicalan- placement, resulting in a significant (p < 0.05) decreased tibiotic and antimicrobial, absorbent gauze placement to de- numberofviablebacteriaatPOD6.Thisdecreasecanbe velop and maintain a biofilm-dominant wound. Note an initial attributed to the elimination of free-floating planktonic significantdecreaseinviablebacteriabetweenPOD4andPOD bacteria, creating a biofilm-dominant wound. However, 6, most likely representing the elimination of free-floating betweenPOD6andPOD12,biofilmwoundsmaintaina planktonicbacteriabytheaforementionedtreatmentmethods. consistent level of viable bacteria averaging between 105 SuccessivebacterialcountmeasurementatPODs6,8,10,and and106CFUsperinfectedwound,verifyingthatwounds 12showviablebacteriathataremaintainedataconsistentlevel maintainlivebiofilmbacteriathroughoutthetime-course over time, averaging between 105 and 106CFUs per wound, of our animal model. This consistency models the representative of the ‘‘steady-state’’ that is often seen be- presumed ‘‘steady-state’’ that develops between chronic tween wound biofilms and their respective hosts (np<0.05) wound biofilms and their respective hosts. To correlate (n58–15wounds/time-point). POD, postoperative day. Error with bacterial count measurements, laser scanning barssignifystandarderrorofmeans. WoundRepReg(2011)19400 410 (cid:2)c 2011bytheWoundHealingSociety 403 Animalmodelforbiofilmimpairedwoundhealing Gurjalaetal. Figure3. Laser scanning confocal microscopy of biofilm wounds in- fected with green fluorescent pro- tein-labeled Staphylococcus aureus. ImagingdoneatPOD6andPOD12 validatesthepresenceofviablebac- teriawithinbiofilmwoundsoverthe course of time used in our model. Notetheclusteringofbacteriawithin thewoundbed(green)withunderly- ingintactcartilageoftheear(arrow) (magnification(cid:6)20). POD, postoper- ativeday. developed, encasing biofilm matrix architecture that cov- eredthewoundbedsurface(Figure4) Despite the spectrum of infection that exists between planktonic- and biofilm-phase bacteria, our model has been developed to create a distinct infection phenotype characterized by a predominance of biofilm-phase bacte- ria.Toshowthismodelcapability,viablebacterialcounts were performed from biofilm-dominant and planktonic- dominant,i.e.,activeinfection,woundsatPOD6(Figure 5A).Thisanalysisrevealednosignificantdifferenceinvi- able bacteria despite differences in how the wounds were treated. At this same time-point, host inflammatory response to these two types of infections was quantified using real-time qPCR (Figure 5B and C). Biofilm infec- tion elicited a significantly lower-grade host response (p < 0.05), measured through IL-1b and TNF-a expres- sion, than active infection wounds, indicatinga true phe- notypic difference in the bacteria–host response between thesetwotypesofinfectiondespiteasimilarbacterialbur- den. In addition, when the expression of these inflamma- tory mediators in biofilm wounds was measured at POD 12, there is evidence of a maintained low-grade host re- sponse, with minimal decrease in the level of TNF-a ex- pression and a continued high level of IL-1b expression. (Figure6)Toensurethatourmodelcreatedonlylocalized wound infections, thus preventing the activation and in- fluence of systemic inflammatory mediators, successively largercircumferentialringsoftissuewereanalyzedoutside of the wound bed. As seen in Figure 7, expression of in- flammatorymediatorsquicklytaperstosignificantlylower levels in both biofilm and active infection wounds (p < 0.05),verifyingthatourmodelmaintainsalocalized Figure4. Scanning electron microscopy of Staphylococcus woundinfection.Furthermore,noanimalsineithergroup aureusbiofilm-infectedwoundsonPOD12showingdeveloped showedsignsofsystemicinfectionandmaintainednormal biofilm architecture. (A) Intricate extracellular matrix encases body weights throughout the experimental period (data individualbacterialcells (arrow)in dense,lattice-like structure not shown). These data not only distinguish and charac- (magnification (cid:3)(cid:6)1000).(B)Highermagnificationviewofindi- terize the biofilm wound phenotype, which triggers and vidualS.aureuswithsurroundingextracellularmatrixmaterial maintains a low-grade host inflammatory response, but adjacent to bare wound surface (arrow) (magnification (cid:3)(cid:6) also validate the sensitivity of our model to distinguish 4000).POD,postoperativeday. 404 WoundRepReg(2011)19400 410 (cid:2)c 2011bytheWoundHealingSociety Gurjalaetal. Animalmodelforbiofilmimpairedwoundhealing Figure5. Demonstration of pheno- typic differences between biofilm- and planktonic-dominant (i.e., active infection) wounds. (A) At POD 6, thereisnosignificantdifferenceinvi- able bacterial counts between un- treated active infection (AI) wounds and treated biofilm-dominant (BF) wounds (n515wounds/group). (B) and (C) despite similar bacterial bur- dens,biofilmwoundstriggerasignifi- cantlylower-gradehostinflammatory response,measuredthroughinterleu- kin(IL)-1bandtumornecrosisfactor-a (TNF-a)woundbed(WB)expression, as compared with active infection wounds. This validates a phenotypic differencebetweenthetwoinfection groups in their interaction with the host (np<0.05 as compared with CTRL; nnp<0.05 as compared with BF)(n56wounds/group).POD, post- operativeday.Errorbarssignifystan- darderrorofmeans. betweenthesedifferentwoundphenotypesdespitesimilar planktonic bacteria damage the cartilage, while biofilm levelsofinitialbacterialburden. histology shows cartilage integrity. Large amounts of de- Having established and validated our wound biofilm brisintheactiveinfectionwerepresentrepresentinganin- model,weaimedtoassesstheeffectsofS.aureusbiofilm creasedinfluxofwhitebloodcells(neutrophils)abovethe on quantitative wound healing parameters histologically. woundbed.Thedebriswasalsofoundinthebiofilm,but WoundswereharvestedonPOD12,withgrossevidenceof significantly reduced, confirming a decreased inflamma- minimal wound healing in both biofilm- and planktonic- tory response histologically. Through quantitative image dominant(activeinfection)woundsascomparedwithcon- analysis, measurement of epithelial gap (Figure 10A), trol (Figure 8). Also note the increased level of purulent granulationtissuegap(Figure10B),andnewgranulation exudateoveractiveinfectionwounds(Figure8B)ascom- tissue (Figure 10C) showed significant differences across pared with the so-called ‘‘film’’ overlying biofilm wounds all analyzed parameters between control and infected (Figure 8C), consistent with the differences seen in host groups (p < 0.05), verifying that our model is capable of inflammatory response. Looking histologically, S. aureus simulating the impairment of wound-healing seen in bio- in the form of active or biofilm infections impaired cuta- film-infectedwounds. neouswoundhealing(Figure9).Infectedwoundsshowed delayedreepithelializationoverthewoundbedandthick- DISCUSSION eningoftheepidermallayerattheleadingedge(Figure9B and C). Differences between active infection and biofilm Predating any bacterial biofilm theories, chronic wounds arealsoseeninthecartilageanddebrisaroundthewound ofalltypeshavesharedacommoncharacteristic:apersis- bed. In the active infection, highly virulent and invasive tent inflammatory state, marked by a destructive milieu Figure6. Maintenance of a low-grade inflammatory response secondary to biofilm wound infection over time. (A) Expression of tumor necrosis factor-a (TNF-a) is maintained between POD 6 andPOD12atverysimilarlevels.(B)In- terleukin(IL)-1bexpressionisdecreased butremainsmarkedlyelevatedatPOD 12, validating the idea that biofilm trig- gersalow-grade,chronicinflammatory response, which can be simulated through this model (n56wounds/ group). POD, postoperative day. Error barssignifystandarderrorofmeans. WoundRepReg(2011)19400 410 (cid:2)c 2011bytheWoundHealingSociety 405 Animalmodelforbiofilmimpairedwoundhealing Gurjalaetal. Figure7. Comparison of inflamma- tory profiles from control and in- fected wounds, showing highly localized infections at POD 6. Note significantlyhigherlevelofinflamma- torymediatorexpressionforbothtu- mornecrosisfactor-a(TNF-a)(A)and interleukin (IL)-1b (B) within wound beds of infected wounds as com- pared with successive outer rings (ring 1: concentric ring of tissue 2mmdistaltothewoundedge;ring 2:concentricringoftissue2mmdistaltotheedgeofring1)(np<0.05ascomparedwithwoundbed)(n56wounds/group).POD, postoperativeday.Errorbarssignifystandarderrorofmeans. and failure of the wounds to progress normally into the wound microbiology was conventionally conceptualized proliferative and remodeling phases of wound healing.13 as contamination, colonization, and infection by plank- Clinically, chronic wounds are observed to languish in a tonic bacteria, it is now clear that biofilm phase bacteria friable, highly exudative, and oftentimes ‘‘slimy’’ condi- playaprominentroleinthespectrumofclinicalwounds.16 tion,mostdisplayingnoevidenceofplanktonicorfrankly This spectrum includes both infections that are predomi- purulent bacterial infection, some responding tofrequent nantlyintheplanktonic-orbiofilm-phases,althoughitis debridementandwoundcare,andothersremainingrecal- clearthatamixtureofbacterialphenotypescan,andtend citranttowoundhealinginterventions.14 to, exist within a given wound. There is likely a balance Bacterial biofilms offer a powerful and parsimonious betweenhostvulneraryfactors/defensesandbacterialpro- explanation for both the cause and persistence of the in- liferationinitsvariousforms,thepredominanceofwhich flammatoryarrestofchronicwounds.Thelogicalconnec- influences wound outcome. Bacteria in a wound may be tion between bacteria and chronic wounds is intuitively either overcome or cleared by host defenses; if bacteria clear,butwaslikelydelayedbymedicine’shistoricreliance persist asbiofilms,theymaybe heldincheckbythehost on planktonic culture techniques, which fail to recover defenses,yetcreateastateofchroniclow-gradeinflamma- biofilm phase bacteria, especially from chronic wound tion.Giventhisexpandedviewofwoundmicrobiology,it samples.Indeed,asbiofilmshaverecentlybeenfoundus- iscriticalthatanyinvivomodelofbiofilmsimulateboth ingmoresophisticatedtechniquestobepresentinpressure itsabilitytoimpaircutaneouswoundhealinganditsmain- sores,venousstasisulcers,anddiabeticfootulcers—tradi- tenance of a low-grade, chronic inflammatory host re- tionally regarded as having distinct etiologies—all three sponse that is distinct from the response to a more chronicwoundarchetypesmayinfactbelinkedbyasingle ‘‘active’’planktonic-dominantinfection. contributing factor. This factor has high potential as a Recentworkhastakenimportantstepstowardachiev- therapeutic target, as elimination of bacterial biofilms inganinvivobiofilmmodel,withRashidetal.17showing alone could be instrumental in shifting the balance be- thatcutaneousPseudomonasaeruginosabiofilmcanbeex- tween inhibitory and vulnerary factors toward a healing perimentally grown in thermally induced mouse wounds. phenotype. In addition to the classic tenets of hypoxia, Using their porcine model, Davis and colleagues inocu- ischemia-reperfusion,andintrinsichostdisease,then,bac- lated S. aureus onto partial thickness cutaneous wounds, terialbiofilmsholdconsiderablepromiseasafourthmajor andassessedforpresenceofbiofilmafter48hours.Multi- pillarofimpairedwoundhealing.15 plemodesofimagingwereabletoconfirmpresenceofliv- Theadditionofbiofilmtheorytowoundshasincreased ing biofilm in these wounds.18 In 2009, our own group our understanding of bacteria–host interaction. Whereas published the infected splinted mouse model, in which Figure8. Photographs of control (uninfected)andinfectedwoundsat POD12.Control(A),activeinfection (B), and biofilm infection (C). Note theabsenceofpurulenceanddebris inbiofilmwoundsascomparedwith active infection wounds, instead showinga‘‘film’’ofbacteriaspread across the wound. Also note the clear difference in extent of wound epithelializationbetweencontroland infected wounds. POD, postopera- tiveday. 406 WoundRepReg(2011)19400 410 (cid:2)c 2011bytheWoundHealingSociety Gurjalaetal. Animalmodelforbiofilmimpairedwoundhealing wounds, this removal of dermis more closely models the dermal-lossseeninhumanchronicwounds.Additionally, themajorityofhumanwoundshealthroughepithelializa- tionandgranulation,incontrasttothecontracture-based healingseeninmice.19Theunderlyingcartilageoftherab- bit ear serves as a natural splint, preventing healing by contracture,andthusallowingforaccuratequantification ofepithelialandgranulationtissueformationfromthepe- riphery of the wound. Previous work with the rabbit ear dermal ulcer model has also showed that hypertrophic scarringwithintherabbitearissimilartothatseeninhu- mans grossly, histologically, and in its response to scar treatmentssuchassteroidinjection.31,32Moreover,multi- pleidenticalwoundscanbemadeinoneanimalwithcon- tralateralcontrols,creatingbothastandardizedandhigh- throughputwoundmodel.Inoculationofwoundsisdone onpostwoundingday3usingculturemedium-grownbac- teria with a measured size of inoculum of approximately 106CFUsofbacteria.Incontrasttootherpublishedmod- els where preformed in vitro biofilm is directly applied to wounds, we believe that inoculation with planktonic, free-floating bacteria more closely represents the seeding mechanismofhumanchronicwounds,withthewoundbed itselfplayingacriticalroleinthetransformationofbacteria into the biofilm state.33,34 Furthermore, as suggested by our presented data,inoculationin this manner triggers the formationofbiofilm-phasebacteriawithin24hours. In another significant departure from other published animalmodels,followinginvivoproliferationoftheinoc- ulated bacteria, including formation of bacterial biofilm, our wounds are treated with topical antibiotic. As ex- Figure9. Woundhealinghistology.Hematoxylin&eosinstain- pected, this reduces the presence of active, planktonic- ingofPOD12rabbitearwoundsfromcontrol(A),activeinfec- phase bacteria, but also by definition leaves behind what tion(B),andbiofilminfection(C)groups.Thewoundedgewas should be predominantly biofilm-phase bacteria, which determinedbytheappearanceofanickinthecartilagemade are more resistant to antimicrobial challenge due to a during wounding. Cartilage nick (arrow), epidermal thickening protective ECM. We then utilize a combination of an (arrowhead),cartilagedegradation(D),anddebrisfromwhite occlusive dressing (Tegaderms) with an underlying anti- blood cell influx (W) are labeled (magnification, (cid:6)20 [A– microbial (polyhexamethylene biguanide) absorptive C]).POD,postoperativeday. gauze pad (AMD Telfas). This form of wound coverage helps maintain the predominance of biofilm-phase bacte- openwoundswereinoculatedwitheitherbiofilm-ornon- ria in two ways. First the antimicrobial impregnation of biofilm–forming bacteria, and treated with a biofilm in- thegauzehelpslimit proliferationofplanktonicbacteria. hibiting peptide; data from this paper suggested delayed Second,itisclearthatactivelyinfectedwoundsarepredis- epithelializationofthesewounds,andthatthisdelaywasa posed to the formation of purulent exudates. Previously biofilm-dependent effect.19 Most recently Zhao et al.20 published models have utilized fluid impermeable, occlu- have developed an infected wound model in db/db mice, sivedressingsfortheirinfectedwoundswithoutanunder- applyingpreformedP.aeruginosabiofilmtofull-thickness lying absorptive gauze pad.19,20 Clinical observations, as punch wounds, and found a marked reduction in a gross wellasourownpilotstudies,demonstratethatplacement measure (percent healing) of infected vs. noninfected ofocclusivedressingsoverthisdevelopingfluidcollection wounds. createsaseromawithinthedeadspacebeneaththedress- Despitetheadvancesinvivoanimalmodeling,currently ing,representinganidealculturemediumforproliferation availablewoundbiofilmmodelspossesscertainlimitations of planktonic-phase bacteria. In this setting a mixed and shortcomings. In trying to address these deficiencies, planktonic-biofilminfectioncanbecomeapredominantly we have developed the aforementioned rabbit ear wound planktonic, purulent infection, more similar to a superfi- biofilmmodel,whichwebelieveisthemostaccuraterep- cial abscess than the wound surface biofilms seen in resentation of biofilm-infected human chronic wounds to chronic wounds. Therefore, the use of absorptive gauze date. Our model utilizes an adaptation of the rabbit der- helps to minimize seroma formation, which when com- mal ulcer model,21 an FDA-recognized model of wound bined with previously administered antibiotics, creates a healingthathasbeenutilizedbyourlabandothersfor20 ‘‘steady-state’’ predominantly biofilm-phase infection by years.22–30 In this model, full-thickness, circular punch- POD6.Finally,wealsoperformfrequentdressingchanges woundsaremade inthe earsofNew ZealandWhite rab- beforewoundharvest,modelingthecommonclinicalman- bits down to cartilage, affording a number of important agement of chronic wounds. In summary, our in vivo advantages. For example, in contrast to partial-thickness wound biofilm model is the most accurate and consistent WoundRepReg(2011)19400 410 (cid:2)c 2011bytheWoundHealingSociety 407 Animalmodelforbiofilmimpairedwoundhealing Gurjalaetal. Figure10. Quantification of healing parameters for uninfected (control), active-,andbiofilm-infectedwounds. (A)Epithelialgap,(B)granulationgap, and (C) granulation tissue area. In- fected wounds have significantly lesshealingacrossallmeasuredpa- rameters when measure to control wounds, verifying in particular that biofilm does impair cutaneous woundhealingandthatthisresultis effectivelysimulatedwiththismodel (np<0.05ascomparedwithcontrol wounds) (n58–12wounds/group). Error bars signify standard error of means. simulation of biofilm-infected human chronic wounds to islocalizedtothewoundbed,similartoahumanchronic date,whichwevalidatethroughthedatapresentedinthis wound that elicits little systemic inflammatory effects. manuscript. Taking the clinical applicability of our model further, we Development of a model requires systemic testing and investigated the impairment of wound healing related to verificationbeforeitcanbeusedforexperimentationand biofilm, an increasingly recognized phenomenon in non- analysis.Wehavepresentedanextensivesetofdatadem- healingwounds.Grossly,thereisacleardifferenceinboth onstrating this validation. Using epifluorescence and the hostinflammatoryresponseand extentof healingbe- SEM,wehaveshownthatinoculationwithplanktonicS. tween infected and control wounds, whichis further sup- aureus establishes mature biofilm in cutaneous wounds, ported by histological analysis. In fact, quantitative with marked speed and under no special conditions. To analysisyieldsstatisticallysignificantdifferencesinmulti- ensurethatourmodelmaintainsviablebiofilm-phasebac- ple histological wound healing parameters between unin- teriaoverthecourseofourprotocoltimeline,wemeasured fected and biofilm-infected wounds. Interestingly, vs. viablebacterialcountsatvarioustimepointspostwound- granulation tissue formation, we found maximal impair- ingand-inoculation.Althoughourdevelopmentofabio- ment to be in reepithelialization of biofilm-infected film-predominant infection through antimicrobial wounds. This observation agrees well with the frequent techniquesdoesdecreaseviablebacterialcountsfromtheir clinicalobservationofrecedingofepitheliumorcomplete postinoculationvalues,theconsistentlevelofbacteriaseen failure ofreepithelialization(even intheface ofadequate between POD 6 and POD 12 reveals a trend that models granulationtissue)inhumanwounds.Thesefindingssug- the ‘‘steady-state’’ or chronic nature of biofilm, and di- gestthattheprimarymechanismofbiofilmimpairmentof rectly correlates to the static nature of human chronic woundhealingmaybethroughinhibitionofepithelializa- wounds. Corroborating this bacterial count analysis, we tion. Although our data only represents part of the con- usedlaserscanningconfocalmicroscopyandSEMtover- tinued investigation into biofilm, we submit that our ify the physical presence of viable bacteria within the model provides a powerful in vivo methodology for woundbedandtheexistenceofcomplexbiofilmstructure woundbiofilmstudythatiscapableofquantitative,repro- morphologicallyatPOD12.Theseexperimentsclearlyin- ducible,andsensitivedataanalysisthroughbothimaging dicate that our model is capable of developing consistent andnonimagingmodalities. woundbiofilmsthatprovideafoundationforfurtherbio- Although we have ensured biofilm phase infection and filmcharacterizationandexperimentation. mirrored key inflammatory aspects of chronic wounds in Wethenaimedtoassessthecapabilitiesandsensitivity thismodel,ourassessmentofwoundhealingisstilloccur- of our model, utilizing quantitative analysis of host in- ringwithinanacutetimeframe.Furtherworkwillbenec- flammatory gene expression to distinguish between our essary to investigate the long-term effects of our biofilm developed biofilm wounds and untreated, planktonic- protocol.Moreover,althoughmorphologybasedonSEM dominant,activeinfectionwounds.Correlatingwithclin- is the current gold standard for identifying bacterial bio- icalsuspicion,wehaveshownthatthereisadistinctphe- films, it would have been ideal to confirm biofilm pheno- notypic difference between biofilm and active infection typeinmorespecificways.Unfortunately,visualizationby wounds. In particular, despite similar bacterial burdens, fluorescence is currently limited to using nonspecific car- thehostwoundbedshowsauniqueinflammatorymedia- bohydrate markers as concanavalin A, as no specific torprofileinresponsetothebiofilmphenotype.Addition- markers for biofilm yet exist. In addition, our work thus ally,weverifiedthatthisuniqueprofile,characterizedbya farhasbeenlimitedtoasinglespeciesofbacteria,namely low-gradeinflammatoryresponse,ismaintainedovertime S. aureus. Organisms such as P. aeruginosa and various withinbiofilm-infectedwoundsandthatusingourmodelit anaerobic bacteria are also very common within chronic 408 WoundRepReg(2011)19400 410 (cid:2)c 2011bytheWoundHealingSociety