stewart and costerton 2001

For personal use. Only reproduce with permission from The Lancet Publishing Group.

THE LANCET Vol 358 July 14, 2001 135

Bacteria that adhere to implanted medical devices ordamaged tissue can become the cause of persistentinfections.1,2 These bacteria encase themselves in ahydrated matrix of polysaccharide and protein, forming aslimy layer known as a biofilm. Direct microscopicexamination of colonised surfaces shows dense aggregatesof bacteria held together by diffuse extracellular polymers(figure 1). Biofilm formation is important because thismode of growth is associated with the chronic nature ofthe subsequent infections, and with their inherentresistance to antibiotic chemotherapy.

Periodontitis and chronic lung infection in cysticfibrosis patients are examples of diseases that are generallyacknowledged to be associated with biofilms.3,4 Variousnosocomial infections such as those related to the use ofcentral venous catheters,5 urinary catheters,6 prostheticheart valves,7 and orthopaedic devices8 are clearlyassociated with biofilms that adhere to the biomaterialsurface. These infections share common characteristicseven though the microbial causes and host sites varygreatly. The most important of these characteristics is thatbacteria in biofilms evade host defences and withstandantimicrobial chemotherapy.

Even in individuals with competent innate and adaptiveimmune responses, biofilm-based infections are rarelyresolved. In fact, tissues adjacent to the biofilm mightundergo collateral damage by immune complexes andinvading neutrophils.9 Susceptibility tests with in-vitrobiofilm models have shown the survival of bacterialbiofilms after treatment with antibiotics at concentrationshundreds or even a thousand times the minimuminhibitory concentration of the bacteria measured in asuspension culture.10 In vivo, antibiotics might suppresssymptoms of infection by killing free-floating bacteriashed from the attached population, but fail to eradicatethose bacterial cells still embedded in the biofilm. Whenantimicrobial chemotherapy stops, the biofilm can act as anidus for recurrence of infection. Biofilm infectionsusually persist until the colonised surface is surgicallyremoved from the body.

REVIEW

Lancet 2001; 358: 13538Center for Biofilm Engineering and Department of ChemicalEngineering, Montana State University, Bozeman, MT 59717-3980,USA (P S Stewart, J W Costerton)

Correspondence to: Dr Philip S Stewart(e-mail: [email protected])

As an example of sequelae of biofilms, let us considerthe case of a patient with pacemaker endocarditis.11 A managed 56 years was admitted with a 4-day history ofnausea, vomiting, and shaking chills. Physicalexamination showed a temperature of 392C andtenderness in his upper right quadrant. Staphylococcusaureus grew from blood cultures. He was treatedintravenously with 12 g cloxacillin daily for 4 weeks. 1 week after discharge he developed nausea, vomiting,fever, and sweating. Again, S aureus grew from bloodcultures. For 6 weeks he was treated with 12 g intravenouscloxacillin daily and with 600 mg oral rifampicin daily.There were no signs of endocarditis. He promptlyresponded to antibiotic therapy, but was readmitted athird time 9 days after discharge with the same symptoms.Once again, S aureus grew from blood cultures. The entirepacing system was removed and intravenous cloxacillinwas continued for 4 weeks. He remained well thereafter.Swabbing of the infected pacemaker lead recovered S aureus, and examination by electron microscopy showedlocalised accretions of coccoid bacteria.

Bacteria in biofilms persist in the body by a strategy thatmight be characterised as tenacious survival as opposed toaggressive virulence. Biofilm infections can linger formonths, years, or even a lifetime. Although theycompromise quality of life, these infections are rarely fatal and are often traced to species of bacteria, such as Pseudomonas aeruginosa or S epidermidis, that areubiquitous in water, air, soil, or skin. These areopportunistic pathogens that persist because they areadept at forming biofilms, in which they are protected.

Resistance mechanismsThe familiar mechanisms of antibiotic resistance, such asefflux pumps, modifying enzymes, and target mutations,12

do not seem to be responsible for the protection ofbacteria in a biofilm. Even sensitive bacteria that do nothave a known genetic basis for resistance can haveprofoundly reduced susceptibility when they form abiofilm. For example, a -lactamase-negative strain ofKlebsiella pneumoniae had a minimum inhibitoryconcentration of 2 g/mL ampicillin in aqueoussuspension.13 The same strain, when grown as a biofilm,was scarcely affected (66% survival) by 4 h treatment with5000 g/mL ampicillin, a dose that eradicated free-floating bacteria.13 When bacteria are dispersed from abiofilm they usually rapidly become susceptible toantibiotics,14,15 which suggests that resistance of bacteria inbiofilms is not acquired via mutations or mobile genetic

Antibiotic resistance of bacteria in biofilms

Philip S Stewart, J William Costerton

Review

Bacteria that adhere to implanted medical devices or damaged tissue can encase themselves in a hydrated matrix ofpolysaccharide and protein, and form a slimy layer known as a biofilm. Antibiotic resistance of bacteria in the biofilmmode of growth contributes to the chronicity of infections such as those associated with implanted medical devices.The mechanisms of resistance in biofilms are different from the now familiar plasmids, transposons, and mutationsthat confer innate resistance to individual bacterial cells. In biofilms, resistance seems to depend on multicellularstrategies. We summarise the features of biofilm infections, review emerging mechanisms of resistance, and discusspotential therapies.


REVIEW

136 THE LANCET Vol 358 July 14, 2001

elements. Few studies have investigated the effect of genesencoding multidrug efflux pumps, such as the multipleantibiotic resistance (mar) locus.16 These studies did notshow any important role for these genes in mediatingbiofilm resistance. The mar operon is not induced duringbiofilm growth of Escherichia coli, and mutants withoutmar have a resistance to ciprofloxacin similar to strainswith mar when grown in biofilms.17,18 In biofilms, strains ofP aeruginosa that do not have the MexAB-OprMmultidrug resistance pump also remained resistant tociprofloxacin.19 Preliminary evidence indicates thatconventional antibiotic resistance mechanisms are notsufficient to explain most cases of antibiotic-resistantbiofilm infections. This evidence does not exclude the

possibility that conventional resistance mechanisms, suchas drug pumps, are expressed in biofilms and contributeto antibiotic resistance in the attached mode of growth. However, we should look beyond conventionalmechanisms to understand biofilm resistance. Con-ventional antibiotic resistance can develop in biofilmstreated repeatedly or for a long timestable derepressionof chromosomal -lactamase contributes to thepersistence of P aeruginosa biofilm infections.20

The mechanisms of resistance to antibiotics in bacterialbiofilms are beginning to be elucidated;21 figure 2 showsthree main hypotheses. The first hypothesis is thepossibility of slow or incomplete penetration of theantibiotic into the biofilm. Measurements of antibioticpenetration into biofilms in vitro have shown that someantibiotics readily permeate bacterial biofilms.22 There isno generic barrier to the diffusion of solutes the size ofantibiotics through the biofilm matrix, which is mostlywater.23 However, if the antibiotic is deactivated in thebiofilm, penetration can be profoundly retarded. Forexample, ampicillin can penetrate through a biofilmformed by a -lactamase-negative strain of K pneumoniaebut not a biofilm formed by the -lactamase-positive wild-type strain of the same micro-organism.13 In the wild-strain biofilm, the antibiotic is deactivated in the surfacelayers more rapidly than it diffuses. Antibiotics thatadsorb into the biofilm matrix could also have a retardedpenetration, which might account for the slow penetrationof aminoglycoside antibiotics.24,25 These positively chargedagents bind to negatively charged polymers in the biofilmmatrix.26,27

The second hypothesis depends on an altered chemicalmicroenvironment within the biofilm. Microscalegradients in nutrient concentrations are a well knownfeature of biofilms. Findings from studies with miniature

Figure 1: Pseudomonas aeruginosa biofilm Electron micrograph of a laboratory-grown Pseudomonas aeruginosabiofilm. Bacteria live in multicellular clusters with individual cells in closeproximity. The biofilm was grown on a plastic substratum (bottom). Bar=5 m.

Figure 2: Three hypotheses for mechanisms of antibioticresistance in biofilms The attachment surface is shown at the bottom and the aqueous phasecontaining the antibiotic at the top.


the early stages of biofilm formation. Coordinated byunknown cues, bacteria use flagellar, twitching, andgliding motility mechanisms to grow together in nascentclusters. The further organisation of the biofilm intocomplex structures is regulated by the exchange ofchemical signals between cells in a process known asquorum sensing. Add to these observations the capacityfor bacteria in biofilms to collectively withstandantimicrobial treatments that would kill a lone cell, andthe case for multicellularity in biofilms is compelling. Therecognition of biofilm formation as a multicellulardevelopmental process is important because this insightwill allow new approaches for treatment of the persistentinfections stemming from biofilms.

Potential for new therapiesMore work is needed to fully elucidate antibioticresistance mechanisms in biofilms and develop newtherapeutic strategies, but we have enough evidence tomake some observations and suggestions. Clearly, thereare multiple resistance mechanisms that can act together.Antibiofilm therapies might have to thwart more than onemechanism simultaneously to be clinically effective.Heterogeneity is a common theme of these resistancemechanisms; micro-organisms in a biofilm exist in a broadspectrum of states. First, cells might be exposed todifferent concentrations of antibiotic depending on theirspatial location. Second, gradients in the concentration ofmicrobial nutrients and waste products crisscross thebiofilm and alter the local environment, which leads to abroad range of growth rates of individual microbial cells.Third, a small proportion of cells in a bacterial biofilmmight differentiate into a highly protected phenotypicstate and coexist with neighbours that are antibioticsensitive. The proliferation of states that arises when thesethree types of heterogeneity are crossed means that anygiven antimicrobial agent might be able to kill some of thecells in a biofilm, but is unlikely to effectively target all ofthem. Most or all the antibiotics in current use wereidentified on the basis of their activity against growingcultures of individual cells. New screens of existing andpotential antibiotics that select for activity against non-growing or biofilm cells might yield antimicrobial agentswith clinical efficacy against biofilm infections. As genesthat mediate biofilm resistance to antibiotics are identifiedand their gene products characterised, these will becometargets for chemotherapeutic adjuvants that could be usedto enhance the effectiveness of existing antibiotics againstbiofilm infections.

Because biofilm resistance depends on aggregation ofbacteria in multicellular communities, one strategy mightbe to develop therapies that disrupt the multicellularstructure of the biofilm. If the multicellularity of thebiofilm is defeated, the host defences might be able toresolve the infection, and the efficacy of antibiotics mightbe restored. Potential therapies include enzymes thatdissolve the matrix polymers of the biofilm,38 chemicalreactions that block biofilm matrix synthesis,39 andanalogues of microbial signalling molecules that interferewith cell-to-cell communication, required for normalbiofilm formation.40 As the genetic basis for biofilmdevelopment emerges, the gene products identified asrequired for multicellular colony formation will become apotential target for chemotherapy. In other words, webelieve that treatment strategies will target the formationof multicellular structures rather than essential functionsof individual cells. We will learn to treat the persistentinfections associated with biofilms when the multicellularnature of microbial life is understood.

REVIEW

THE LANCET Vol 358 July 14, 2001 137

electrodes have shown that oxygen can be completelyconsumed in the surface layers of a biofilm, leading toanaerobic niches in the deep layers of the biofilm.28

Concentration gradients in metabolic products mirrorthose of the substrates. Local accumulation of acidicwaste products might lead to pH differences greater than1 between the bulk fluid and the biofilm interior,29 whichcould directly antagonise the action of an antibiotic.Aminoglycoside antibiotics are clearly less effectiveagainst the same micro-organism in anaerobic than inaerobic conditions.30 Alternatively, the depletion of asubstrate or accumulation of an inhibitive waste productmight cause some bacteria to enter a non-growing state, inwhich they are protected from killing. Penicillinantibiotics, which target cell-wall synthesis, kill onlygrowing bacteria.31 This alternative possibility isstrengthened by direct experimental visualisation ofmetabolically inactive zones within continuously fedbiofilms.32 Additionally, the osmotic environment within abiofilm might be altered, leading to induction of anosmotic stress response.33 Such a response couldcontribute to antibiotic resistance by changing the relativeproportions of porins in a way that reduces cell envelopepermeability to antibiotics.

A third and still speculative mechanism of antibioticresistance is that a subpopulation of micro-organisms in abiofilm forms a unique, and highly protected, phenotypicstatea cell differentiation similar to spore formation.This hypothesis is lent support by findings from studiesthat show resistance in newly formed biofilms, eventhough they are too thin to pose a barrier to thepenetration of either an antimicrobial agent or metabolicsubstrates.34,35 Additionally, most bacteria in the biofilm,but not all, are rapidly killed by antibiotics.19,36 Survivors,which might consist of 1% or less of the originalpopulation, persist despite continued exposure to theantibiotic. The hypothesis of a spore-like state enteredinto by some of the bacteria in a biofilm provides apowerful, and generic, explanation for the reducedsusceptibility of biofilms to antibiotics and disinfectants ofwidely different chemistries.

Multicellular nature of biofilm defenceAll three main hypotheses of biofilm resistance toantibiotics depend on the multicellular nature ofbiofilms.37 An antimicrobial agent cannot slowly orincompletely penetrate the biofilm unless the micro-organisms form aggregates that affect its diffusion. Localvariations in the concentrations of microbial substratesand products develop only when a cluster of cells reachesa critical size and the bacteria exert their combinedmetabolic activity. The small population of cells thatdifferentiate into a dormant and protected state dependon their growing neighbours to propagate the genome,and their neighbours depend on them to reseed thecommunity in the event of catastrophic killing. The factthat all these antibiotic resistance mechanisms areinherently multicellular helps to explain why bacteriadispersed from biofilms rapidly revert to a susceptiblephenotype.

Researchers investigating bacterial biofilms arebeginning to discuss biofilm formation in terms ofdevelopmental biology. Recent results lend support to theidea of biofilm formation as a multicellular developmentalprocess. We now know that specific gene products arerequired for the initial association of bacteria with asurface. Dozens of new genes are turned on and others areturned off as bacteria move onto a surface, suggesting apathway of differentiation. Motility seems to be critical in


REVIEW

138 THE LANCET Vol 358 July 14, 2001

Our review was supported by cooperative agreement EEC 8907039between the National Science Foundation and Montana State University,by the industrial partners of the Center for Biofilm Engineering, and byan award from the W M Keck Foundation.

References1 Costerton JW, Stewart PS, Greenberg EP. Bacterial biofilms: a

common cause of persistent infections. Science 1999; 284: 131822.2 Costerton JW, Stewart PS. Biofilms and device-related infections. In:

Nataro JP, Blaser MJ, Cunningham-Rundles S, eds. Persistentbacterial infections. Washington, DC: ASM Press, 2000: 43239.

3 Darveau RP, Tanner A, Page RC. The microbial challenge inperiodontitis. Periodont 2000 1997; 14: 1232.

4 Singh PK, Schaefer AL, Parsek MR, Moninger TO, Welsh MJ,Greenberg EP. Quorum-sensing signals indicate that cystic fibrosislungs are infected with bacterial biofilms. Nature 2000 407: 76264.

5 Passerini L, Lam K, Costerton JW, King EG. Biofilms on indwellingvascular catheters. Crit Care Med 1992; 20: 66573.

6 Morris NS, Stickler DJ, McLean RJ. The development of bacterialbiofilms on indwelling urethral catheters. World J Urol 1999; 17:34550.

7 Hyde JA, Darouiche RO, Costerton JW. Strategies for prophylaxisagainst prosthetic valve endocarditis: a review article. J Heart ValveDis 1998; 7: 31315.

8 Gristina AG, Shibata Y, Giridhar G, Kreger A, Myrvik QN. Theglycocalyx, biofilm, microbes, and resistant infection. SeminArthroplasty 1994; 5: 16070.

9 Hiby N, Fomsgaard A, Jensen ET, et al. The immune response tobacterial biofilms. In: Lappin-Scott HM, Costerton JW, eds.Microbial biofilms. Cambridge: Cambridge University Press, 1995:23350.

10 Ceri H, Olson ME, Stremick C, Read RR, Morck D, Buret A. TheCalgary biofilm device: new technology for rapid determination ofantibiotic susceptibilities of bacterial biofilms. J Clin Microbiol 1999;37: 177176.

11 Marrie TJ, Nelligan J, Costerton JW. A scanning and transmissionelectron microscopic study of an infected endocardial pacemaker lead.Circulation 1982; 66: 133941.

12 Walsh C. Molecular mechanisms that confer antibacterial drugresistance. Nature 2000; 406: 77581.

13 Anderl JN, Franklin MJ, Stewart PS. Role of antibiotic penetrationlimitation in Klebsiella pneumoniae biofilm resistance to ampicillin andciprofloxacin. Antimicrob Agents Chemother 2000; 44: 181824.

14 Anwar H, van Biesen T, Dasgupta M, Lam, K, Costerton JW.Interaction of biofilm bacteria with antibiotics in a novel in vitrochemostat system. Antimicrob Agents Chemother 1989; 33: 182426.

15 Williams I, Venables WA, Lloyd D, Paul F, Critchley I. The effects ofadherence to silicone surfaces on antibiotic susceptibility inStaphylococcus aureus. Microbiology 1997; 143: 240713.

16 Alekshun MN, Levy SB. The mar regulon: multiple resistance toantibiotics and other toxic chemicals. Trends Microbiol 1999; 7:41013.

17 Maira-Litran T, Allison DG, Gilbert P. Expression of the multipleantibiotic resistance operon (mar) during growth of Escherichia coli as abiofilm. J Appl Microbiol 2000; 88: 24347.

18 Maira-Litran T, Allison DG, Gilbert P. An evaluation of the potentialof the multiple antibiotic resistance operon (mar) and the multidrugefflux pump acrAB to moderate resistance towards ciprofloxacin inEscherichia coli biofilms. J Antimicrob Chemother 2000; 45: 78995.

19 Brooun A, Liu S, Lewis K. A dose-response study of antibioticresistance in Pseudomonas aeruginosa biofilms. Antimicrob AgentsChemother 2000; 44: 64046.

20 Bagge N, Ciofu O, Skovgaard LT, Hiby N. Rapid development in

vitro and in vivo of resistance to ceftazidime in biofilm-growingPseudomonas aeruginosa due to chromosomal beta-lactamase. APMIS2000; 108: 589-600.

21 Mah T-F, OToole GA. Mechanisms of biofilm resistance toantimicrobial agents. Trends Microbiol 2001; 9: 3439.

22 Stewart PS. Theoretical aspects of antibiotic diffusion into microbialbiofilms. Antimicrob Agents Chemother 1996; 40: 251722.

23 Stewart PS. A review of experimental measurements of effectivediffusive permeabilities and effective diffusion coefficients in biofilms.Biotechnol Bioeng 1998; 59: 26172.

24 Kumon H, Tomochika K, Matunaga T, Ogawa M, Ohmori H. A sandwich cup method for the penetration assay of antimicrobialagents through Pseudomonas exopolysaccharides. Microbiol Immunol1994; 38: 61519.

25 Shigeta M, Tanaka G, Komatsuzawa H, Sugai M, Suginaka H, Usui T. Permeation of antimicrobial agents through Pseudomonasaeruginosa biofilms: a simple method. Chemotherapy 1997; 43: 34045.

26 Gordon CA, Hodges NA, Marriott C. Antibiotic interaction anddiffusion through alginate and exopolysaccharide of cystic fibrosis-derived Pseudomonas aeruginosa. J Antimicrob Chemother 1988; 22:66774.

27 Nichols WW, Dorrington SM, Slack MPE, Walmsley HL. Inhibitionof tobramycin diffusion by binding to alginate. Antimicrob AgentsChemother 1988; 32: 51823.

28 de Beer D, Stoodley P, Roe F, Lewandowski Z. Effects of biofilmstructure on oxygen distribution and mass transport. Biotechnol Bioeng1994; 43: 113138.

29 Zhang TC, Bishop PL. Evaluation of substrate and pH effects in anitrifying biofilm. Wat Environ Res 1996; 68: 110715.

30 Tack KJ, Sabath LD. Increased minimum inhibitory concentrationswith anaerobiasis for tobramycin, gentamicin, and amikacin,compared to latamoxef, piperacillin, chloramphenicol, andclindamycin. Chemotherapy 1985; 31: 20410.

31 Tuomanen E, Cozens R, Tosch W, Zak O, Tomasz A. The rate ofkilling of Escherichia coli by -lactam antibiotics is strictly proportionalto the rate of bacterial growth. J Gen Microbiol 1986; 132: 1297304.

32 Xu KD, McFeters GA, Stewart PS. Biofilm resistance toantimicrobial agents. Microbiology 2000; 146: 54749.

33 Prigent-Combaret C, Vidal O, Dorel C, Lejeune P. Abiotic surfacesensing and biofilm-dependent regulation of gene expression inEscherichia coli. J Bacteriol 1999; 181: 59936002.

34 Cochran WL, McFeters GA, Stewart PS. Reduced susceptibility ofthin Pseudomonas aeruginosa biofilms to hydrogen peroxide andmonochloramine. J Appl Microbiol 2000; 88: 2230.

35 Das JR, Bhakoo M, Jones MV, Gilbert P. Changes in the biocidesusceptibility of Staphylococcus epidermidis and Escherichia coli cellsassociated with rapid attachment to plastic surfaces. J Appl Microbiol1998; 84: 85258.

36 Goto T, Nakame Y, Nishida M, Ohi Y. In vitro bactericidal activitiesof beta-lactamases, amikacin, and fluoroquinolones againstPseudomonas aeruginosa biofilm in artificial urine. Urology 1999; 53:105862.

37 OToole G, Kaplan HB, Kolter R. Biofilm formation as microbialdevelopment. Annu Rev Microbiol 2000; 54: 4979.

38 Nemoto K, Hirota K, Ono T, et al. Effect of varidase (streptokinase)on biofilm formed by Staphylococcus aureus. Chemotherapy 2000; 46:11115.

39 Yasuda H, Ajiki Y, Koga T, Kawada H, Yokota T. Interactionbetween biofilms formed by Pseudomonas aeruginosa andclarithromycin. Antimicrob Agents Chemother 1993; 37: 174955.

40 Parsek MR, Greenberg EP. Acyl-homoserine lactone quorum sensingin gram-negative bacteria: a signaling mechanism involved inassociations with higher organisms. Proc Natl Acad Sci USA 2000; 97:878993.

stewart and costerton 2001

Documents