• We are sorry, but NCBI web applications do not support your browser and may not function properly. More information
Logo of aacPermissionsJournals.ASM.orgJournalAAC ArticleJournal InfoAuthorsReviewers
Antimicrob Agents Chemother. Jun 2007; 51(6): 2156–2163.
Published online Apr 2, 2007. doi:  10.1128/AAC.00163-07
PMCID: PMC1891362

Targeted Drug-Carrying Bacteriophages as Antibacterial Nanomedicines[down-pointing small open triangle]


While the resistance of bacteria to traditional antibiotics is a major public health concern, the use of extremely potent antibacterial agents is limited by their lack of selectivity. As in cancer therapy, antibacterial targeted therapy could provide an opportunity to reintroduce toxic substances to the antibacterial arsenal. A desirable targeted antibacterial agent should combine binding specificity, a large drug payload per binding event, and a programmed drug release mechanism. Recently, we presented a novel application of filamentous bacteriophages as targeted drug carriers that could partially inhibit the growth of Staphylococcus aureus bacteria. This partial success was due to limitations of drug-loading capacity that resulted from the hydrophobicity of the drug. Here we present a novel drug conjugation chemistry which is based on connecting hydrophobic drugs to the phage via aminoglycoside antibiotics that serve as solubility-enhancing branched linkers. This new formulation allowed a significantly larger drug-carrying capacity of the phages, resulting in a drastic improvement in their performance as targeted drug-carrying nanoparticles. As an example for a potential systemic use for potent agents that are limited for topical use, we present antibody-targeted phage nanoparticles that carry a large payload of the hemolytic antibiotic chloramphenicol connected through the aminoglycoside neomycin. We demonstrate complete growth inhibition toward the pathogens Staphylococcus aureus, Streptococcus pyogenes, and Escherichia coli with an improvement in potency by a factor of ~20,000 compared to the free drug.

The increasing development of bacterial resistance to traditional antibiotics has reached alarming levels (2, 3), spurring a strong need to develop new antimicrobial agents. Classical short-term approaches include chemical modification of existing agents to improve potency or spectrum. Long-term approaches rely on bacterial and phage genomics to discover new antibiotics that attack new protein targets which are essential to bacterial survival and therefore with no known resistance (1, 8). In both traditional and newly developed antibiotics, the target selectivity lies in the drug itself, in its ability to affect a mechanism that is unique to the target microorganism and absent in its host. As a result, a vast number of potent drugs have been excluded from use as therapeutics due to low selectivity. This brings to mind the limited selectivity of anticancer drugs and recent efforts to overcome it by developing targeted therapeutic strategies. Antibody-based targeted drug delivery approaches have been developed since the advent of monoclonal antibodies (6). Since then, monoclonal antibodies and derived single-chain antibodies were used to deliver potent cytotoxic components to cancer cells that, once bound, internalize and kill the target cell (7, 12). A similar immunotargeting of bacteria is not feasible due to the lack of a bacterial internalization process, making the use of an extracellular release mechanism necessary for a targeted antibacterial approach. Moreover, in comparison to cancer internalized targeting devices such as immunotoxins and immunoconjugates, common antibiotics are less-potent drugs in which a threshold number of several thousands of molecules are needed to inhibit or kill a single bacterium. Thus, a targeted antibacterial platform should have a significantly larger drug-carrying capacity than an anticancer one.

Filamentous bacteriophages (phages) are the workhorse of antibody engineering and are gaining increasing importance in nanobiotechnology (9). Here we present targeted, drug-carrying phages as a platform for targeting pathogenic bacteria. Due to genetic and chemical modifications, these phages represent a modular targeted drug-carrying platform of nanometric dimensions where targeting moieties and conjugated drugs may be exchanged at will.

Recently, we have shown the feasibility of using phages as targeted antibacterial drug carriers (13). In our system, chloramphenicol (which is rarely used to treat patients systemically due to toxicity) was attached as a prodrug to p8 coat protein molecules on the surface of filamentous phage. The phages were then targeted to bind to pathogenic bacteria and, upon release of active chloramphenicol, retarded bacterial growth. The reported system had a limited capacity for inhibition of bacterial growth due to a limited arming capacity of less than 3,000 drug molecules/phage. We have now overcome this limitation by designing a unique drug conjugation chemistry, comprising the use of (hydrophilic) aminoglycoside antibiotics as branched, solubility-enhancing linkers. By changing the arming chemistry and a modification of the antibody-phage conjugation method, our system, as illustrated in Fig. Fig.1,1, was transformed into a viable and versatile tool for the targeting of a broad range of pathogenic bacteria.

FIG. 1.
Schematic representation of drug-carrying bacteriophages. (a) Drawing of a single fUSE5 ZZ-displaying bacteriophage. Small turquoise spheres represent major coat protein p8 monomers. Purple sphere and sticks represent the 5 copies of minor coat protein ...


All the chemicals used were of analytical grade and were purchased from Sigma (Israel). Unless stated otherwise, reactions were carried out at room temperature (about 22°C).

Synthesis and evaluation of chloramphenicol prodrug.

A chloramphenicol prodrug, where chloramphenicol is linked through a labile ester bond to an N-hydroxysuccinimide (NHS)-ester for conjugation to amine groups was prepared and evaluated as described previously (13). The drug release rate when serum esterases are used is about 15% of the conjugated chloramphenicol released from the carrier after 1 h of incubation in serum at 37°C with linear kinetics (13).

Preparation of ZZ domain-displaying phages.

Phage fUSE5-ZZ, which displays the Fc-binding ZZ domain of protein A on all copies of the phage p3 minor coat protein, was constructed as described previously (13).

Preparation of phages for drug conjugation.

fUSE5-ZZ filamentous phages (13) were routinely propagated in DH5α F′ cells using standard phage techniques as described previously (5). Phages were usually recovered from overnight 1-liter cultures of carrying bacteria. The bacteria were removed by centrifugation, and the phage-containing supernatant was filtered through a 0.22-μm filter. The phages were precipitated by the addition of 20% (wt/vol) polyethylene glycol 8000-2.5 M NaCl, followed by centrifugation as described previously (5). The phage pellet was suspended in sterile Milli-Q double-distilled water at a concentration of 1013 PFU/ml and stored at 4°C.

Conjugation of aminoglycosides to chloramphenicol or to FITC.

Solid neomycin or hygromycin and a stock solution of 100 μM chloramphenicol prodrug in dimethyl sulfoxide or of fluorescein isothiocyanate (FITC) in dimethyl sulfoxide were used in all conjugations. They were mixed within 0.1 M NaHCO3, pH 8.5, at a molar ratio of 1:2 for the chloramphenicol prodrug-neomycin or at a molar ratio of 1:10 for FITC-hygromycin. The reaction was stirred overnight. Next, the prepared neomycin-chloramphenicol adduct was purified by reverse-phase high-performance liquid chromatography (HPLC). A reverse-phase C18 column was used on a Waters machine with a gradient 0% to 100% of acetonitrile (stock solution of 80% [wt/wt] in water) and water (100% water to 0%) in the mobile phase, at a 1-ml/min flow rate. Under these conditions, the neomycin-chloramphenicol adduct eluted 18 min after sample injection, while the intact chloramphenicol-prodrug eluted 24 min after sample injection (Fig. 2a and b). Following validation by matrix-assisted laser desorption ionization-time of flight mass spectrometry (MS) (Fig. (Fig.2c),2c), the purified neomycin-chloramphenicol adduct was lyophilized and conjugated to antibody-complexed phage nanoparticles by the 1-ethyl-3-[3-dimethylaminopropyl] carbodiimide (EDC) procedure.

FIG. 2.
Reverse-phase HPLC purification and MS analysis of the chloramphenicol-neomycin adduct. (a) HPLC analysis of chloramphenicol-NHS prior to conjugation to neomycin. The chloramphenicol-NHS prodrug was separated using a gradient of acetonitrile in water ...

EDC chemistry.

The phage major coat protein p8 contains 3 carboxylic amino acids (glu2, asp4, asp5) that can be conjugated by application of EDC chemistry, a rapid reaction performed at a mildly acidic pH (4.5 to 5.5) (11). Here, all conjugations were done within a total volume of 1 ml of 0.1 M Na-citrate buffer, pH 5, 0.75 M NaCl, 2.5 × 10−6 mol of the aminoglycoside, 1012 phages conjugated to an immunoglobulin G (IgG). The reaction was initiated by the addition of 2.5 × 10−6 mol of EDC, which was repeated two more times at time intervals of 30 min. Reactions were carried out at room temperature with gentle stirring (10 rpm) in 2-ml Eppendorf tubes for a total of 2 h. The targeted drug-carrying phage nanoparticles were separated from the reactants by two dialysis steps of 16 h each against 1,000 volumes of sterile phosphate-buffered saline (PBS).

Quantifying linked chloramphenicol molecules-phage.

Linked chloramphenicol molecules-phage were quantified essentially as described previously (13). Briefly, conjugated chloramphenicol was released from the phages by incubation in rabbit serum (as a source of esterases) at 37°C for 48 h. The free chloramphenicol was separated from the phages by using 10-kDa-cutoff ultrafiltration cartridges (Millipore), and the number of released chloramphenicol molecules was calculated by using a calibration curve of free chloramphenicol absorbance that was recorded at 280 nm.

Evaluation of serum titers by ELISA using whole bacteria as antigens.

Enzyme-linked immunosorbent assay (ELISA) plates (flat bottom; Nunc, Sweden) were coated with bacteria as follows. Cells from a fresh overnight culture were collected by centrifugation and suspended in PBS at about 108 cells/ml. An aliquot of 100 μl of the cell suspension was applied to each well of the plate that was spun in a centrifuge at 4,000 rpm for 5 min at 4°C. The supernatant was carefully removed, and 100 μl of 3% glutaraldehyde in PBS was added to each well and left to fix the cells for 1 h. Next, the plate was blocked with 50% bovine serum. Tested sera were added in serial dilutions, the plates were incubated for 1 h at room temperature (about 22°C), washed three times with PBS and incubated with horseradish peroxidase-conjugated goat anti-human or anti-rabbit antibodies (Jackson Immunoresearch Laboratories) for an additional 1 h. Plates were then washed four times and developed with 3,3′,5,5′-tetramethyl benzidine (Dako). Reactions were terminated with 1 M H2SO4. The plate was read in an ELISA reader at 450 nm.

Growth inhibition experiments.

A 100-μl aliquot of overnight bacterial culture was collected by centrifugation and washed in 1 ml of cold PBS. The cells were collected again and resuspended in 100 μl cold PBS. Ten microliters of washed bacteria (107 cells) was incubated with 100 to 500 μl of targeted neomycin-chloramphenicol carrying phage nanoparticles (1010 to 1011 particles) for 1 h on ice. Next, an equal volume of rabbit serum was added (100 to 500 μl) and incubated for 3 h at 37°C. One hundred to five hundred microliters of this mixture was diluted in 3 ml growth medium (Staphylococcus aureus, tryptic soy broth; Streptococcus pyogenes, Todd-Hewitt broth; Escherichia coli, 2× YT [16 g/liter Bacto-tryptone, 10 g/liter Bacto-yeast extract, 5 g/liter NaCl, double-distilled water complete to 1 liter]) containing 50% rabbit serum in 13-ml tubes with shaking at 250 rpm at 37°C. Growth was recorded by monitoring the absorbance at 600 nm.

Calculation of potency improvement factor of targeted versus free drug.

The calculation of the potency improvement factor is based on 1010 chloramphenicol-carrying phages, each carrying 104 drug molecules, which inhibit the growth of 107 cells as effectively as 15 μg of free chloramphenicol. The percentage of relevant phages is based on the ~5% target-specific IgG within the polyclonal serum. The estimate that 30% of the drug is released during the experiment is based on reference 13.


Conjugation of chloramphenicol to phages through aminoglycoside linkers.

As a solution to the limited arming efficiency that was mainly due to drug hydrophobicity (13), we applied aminoglycoside antibiotics as branched solubility-enhancing linkers. This resulted in overcoming the hydrophobic barrier in aqueous solutions on one hand and significantly increased the loading potential of drug molecules per phage particle on the other. The first example was conjugation of chloramphenicol to the aminoglycoside neomycin. Each neomycin molecule has 6 primary amines, 1 of which was linked to a chloramphenicol molecule, while the other amines were left for further conjugation to carboxyl residues of the phage coat proteins by EDC chemistry (11) (Fig. (Fig.3a).3a). The use of phage coat carboxyl residues for drug conjugation instead of the amine residues multiplied the direct drug-carrying capacity potential to ~11,400 molecules per phage (3,800 p8 coat protein copies on our fUSE5-ZZ phage with a genome size of 9,200 bases × 3 accessible carboxyl residues on each p8 monomer).

FIG. 3.
Schematic representation the chemical reactions used to prepare drug for conjugation. (a) Preparation of a neomycin-chloramphenicol adduct. (1) Two chemical steps were used to modify chloramphenicol for conjugation to amine groups. In the first step, ...

Quantification of payload.

To indirectly quantify the reactive carboxyl residues on the phage surface, we prepared a FITC-hygromycin conjugate (Fig. (Fig.3b).3b). The aminoglycoside hygromycin is similar to neomycin in structure but differs in the number of amines groups. A hygromycin molecule contains 2 primary amines; one may be attached to the fluorescent dye FITC, while the second is left for further conjugation to free carboxyl residues on the phage coat by EDC chemistry. Since hygromycin serves as a direct linker, it allows a true estimation of conjugation-to-phage events. By a linear calibration curve of fluorescence intensity as a function of FITC concentration, we deduced the number of FITC molecules and, hence, that of hygromycin molecules that were linked to the phages at ~10,000, which is comparable to the number of reactive carboxyl residues per phage as calculated above.

To directly quantify the chloramphenicol that was conjugated through the neomycin linker, we released the conjugated chloramphenicol molecules by incubating drug-carrying phages with serum as described previously (13). We calculated about 10,000 chloramphenicol molecules per phage. Because of the additional primary amines available on neomycin, an even larger drug payload could be obtained by manipulating reactant concentrations and incubation times, and we could obtain over 40,000 chloramphenicol molecules/phage without compromising the phage integrity.

Improving targeting efficiency.

We complexed the phages with polyclonal antibodies that served as the targeting moieties in our targeted drug-carrying platform. We used human sera against staphylococci and streptococci and a rabbit, protein A-purified IgG against E. coli O78. All serum titers, as determined by ELISA on immobilized whole bacteria, were in the range of 1:10,000 to 1:100,000 (Fig. (Fig.4).4). To facilitate accurate calculation of the improvement in potency, we measured the fraction of target-specific IgGs within the sera. We found that the total IgG concentration in the sera was ~15 mg/ml, of which 4 to 5% was target specific.

FIG. 4.
Titers of the polyclonal sera. The titers of human anti-Staphylococcus aureus (SA) and -Streptococcus pyogenes (SP) sera and of rabbit anti-E. coli IgG were analyzed by ELISA with fixed bacteria as antigens, titers above 1:50,000 were recorded for all ...

In our previous study, we complexed the phages with the targeting antibodies following drug conjugation (13). Here, we complexed the phages with the targeting antibodies prior to drug conjugation. As a result, in addition to linking the drug to the phage coat, the applied EDC chemistry cross-linked the targeting antibodies to the phages (data not shown). This is important when considering in vivo applications, where resident nonspecific antibodies may compete out the targeting antibody from the ZZ domain.

Growth inhibition of target bacteria by drug-carrying phages.

We tested the ability of the targeted drug-carrying phage nanoparticles to inhibit the growth of three different strains of common pathogenic bacteria. Two were gram positive: the methicillin-resistant Staphylococcus aureus COL and a clinical isolate of Streptococcus pyogenes. The third was a gram-negative, avian pathogenic E. coli O78 (781) (14).

The growth inhibition experiments with staphylococci were done with the minimal amount of phages that gave total growth inhibition, 1010 phage particles per 107 bacteria. Negative controls were nontreated bacteria as well as bacteria treated with nonimmune human IgG or human Fc complexed bacteriophages conjugated to the same amount of antibiotic. The results were compared to growth inhibition resulting from various concentrations of free chloramphenicol. We found that 1010 targeted drug-carrying phages inhibited bacterial growth, as do 15 μg of free chloramphenicol (Fig. 5a and b). Similar growth inhibition experiments were carried out with the other bacterial targets where we could also observe growth inhibition (Fig. 5c and d). The growth inhibition profile of SP and E. coli O78 by free chloramphenicol was similar to that of the SA cells (not shown). The partial growth inhibition observed for E. coli O78 that was treated with nontargeted drug-carrying phages (Fig. (Fig.5d)5d) was expected due to the low level of nonspecific binding of filamentous phages to this E. coli strain that we reported previously (13).

FIG. 5.
Growth inhibition curves. Three strains of common pathogenic bacteria were tested: Staphylococcus aureus (SA) COL, Streptococcus pyogenes (SP), and E. coli O78. (a) Growth curve of SA treated with 1011 (filled triangles) or 1010 (filled squares) targeted ...

Calculation of potency improvement in comparison to free drug.

We calculated the potency improvement factor by assuming that the fraction of relevant phages (that bind the target bacteria) is equal to the fraction of target-relevant IgGs within the sera. As was shown in Fig. Fig.5,5, 1010 targeted chloramphenicol-carrying phages inhibited bacterial growth as effectively as did 15 μg of free chloramphenicol. Considering that, based on the fraction of target-specific IgG in the serum, ~5% of the phages are targeted, which for 107 bacteria yields 50 drug-carrying phages that actually bind each target bacterium. Each phage carries ~104 chloramphenicol molecules, of which 30% (3,000) are released during the time course of the experiment (based on release kinetics reported in reference 13). This yields 150,000 drug molecules released for each target bacterium.

For 107 target bacteria, 15 μg of free chloramphenicol corresponds to ~3 × 109 molecules/bacterium. Hence, the potency improvement factor in comparison to the free drug is about 20,000 [3 × 109free/150,000targeted].


The emergence of bacterial drug resistance calls for creative measures to overcome it. The use of some extremely potent antibacterial agents is limited by their lack of selectivity. A solution to such a problem may be provided in the form of targeted therapy (12). An efficient targeted drug delivery platform should satisfy criteria of target binding selectivity, large drug-carrying capacity, and timely drug release at the target.

We offer targeted, drug-carrying phage nanoparticles as a versatile way to meet these criteria, as shown here with growth inhibition of both gram-positive and gram-negative pathogenic bacteria. As a model drug, we used the bacteriostatic antibiotic chloramphenicol which is usually limited to topical application due to toxicity to blood cells (10). The phage represents here a nanometric size particle that, due to the modular assemblage of its coat, offers excellent drug-carrying capacity for its size. The arrangement of drug that is conjugated on the exterior of the targeted particle is unique in comparison to particulate drug-carrying devices such as liposomes or virus-like particles. An additional innovation was introduced by the use of the amino sugar-based aminoglycosides as branched, hydrophilic linkers, providing for the solvation of hydrophobic materials such as chloramphenicol, FITC, or Z-Phe (not shown). This had solved a major obstacle in allowing conjugation to a biological entity (phage) in aqueous solutions, enabling us to conjugate a fairly large amount of hydrophobic molecules to each phage. Indeed, we could conjugate over 40,000 chloramphenicol molecules/phage without compromising the phage integrity. However, working at a conjugation level of 10,000 molecules/phage was sufficient to obtain complete growth inhibition while saving on precious reagents.

The neomycin we used as the aminoglycoside linker is itself an antibiotic to which the bacteria we tested are sensitive. However, since it was linked to the phage coat by a nonlabile bond, it could not be released and contribute to bacterial growth inhibition. One can imagine a more elegant conjugation design where the aminoglycosides are linked to the phage by a labile bond subject to controlled release, in which case an additive or synergistic drug effect could have been obtained.

Our study demonstrated an improvement factor of 20,000 in comparison to the free drug. In our previous study, we could show a limited growth inhibition with a much lower potentiation factor. This drastic improvement can probably not be explained on the basis of increasing drug payload from 3,000 to 10,000 alone. Rather, it must be a synergistic effect resulting from the improved chemistry which probably affected the overall solubility of the entire drug-carrying platform, and the new targeting approach in which the antibodies were complexed to the phages prior to drug conjugation with concomitant cross-linking of the antibodies to the drug-carrying phages.

Our results of growth inhibition were obtained within an artificial closed system and surely do not reflect an in vivo application, which will offer additional challenges to our approach, as those facing other potential nanomedicines (4). For one, the usage of polyclonal serum as we did in the presented model system will not be suitable for treatment without a prior affinity purification of bacterium-specific IgGs because we “waste” ~95% of our drug-carrying phages that are not targeted. In fact, antibodies may not be the ideal targeting molecules for drug-carrying phages because the target bacteria may be already opsonized by patient antibodies. In such a case, one may consider other targeting approaches, as we did with peptide-displaying phages in our previous study (13). We hope that this work would lead to more creative and versatile methods to fight our most ancient and intimate enemies, the pathogenic bacteria.


We thank Ehud Gazit (Tel-Aviv University) for critical reading of the manuscript. We thank Marina Shamis and Doron Shabat for preparing the chloramphenicol-NHS compound.

I.Y. was supported by a Ph.D. scholarship from the Gertner Research Institute for Nanomedical Systems, Tel-Aviv University, and by the Dan David scholarship for young scholars in future dimension.


[down-pointing small open triangle]Published ahead of print on 2 April 2007.


1. Barrett, J. F. 2005. Can biotech deliver new antibiotics? Curr. Opin. Microbiol. 8:498-503. [PubMed]
2. Bax, R. P., and N. Mullan. 1999. Response of the pharmaceutical industry to antimicrobial resistance. Balliere Clin. Infect. Dis. 5:289-304.
3. Coates, A., Y. Hu, R. Bax, and C. Page. 2002. The future challenges facing the development of new antimicrobial drugs. Nat. Rev. Drug Discov. 1:895-910. [PubMed]
4. Duncan, R. 2006. Polymer conjugates as anticancer nanomedicines. Nat. Rev. Cancer 6:688-701. [PubMed]
5. Enshell-Seijffers, D., and J. M. Gershoni. 2002. Phage display selection and analysis of Ab-binding epitopes, p. 9.8.1-9.8.27. In J. C. Colligan (ed.), Current protocols in immunology. John Wiley & Sons, Inc., Hoboken, NJ.
6. Köhler, G., and C. Milstein. 1975. Continuous cultures of fused cells secreting antibody of predefined specificity. Nature 256:495-497. [PubMed]
7. Kreitman, R. J. 1999. Immunotoxins in cancer therapy. Curr. Opin. Immunol. 11:570-578. [PubMed]
8. Payne, D., and A. Tomasz. 2004. The challenge of antibiotic resistant bacterial pathogens: the medical need, the market and prospects for new antimicrobial agents. Curr. Opin. Microbiol. 7:435-438. [PubMed]
9. Sarikaya, M., C. Tamerler, A. K. Jen, K. Schulten, and F. Baneyx. 2003. Molecular biomimetics: nanotechnology through biology. Nat. Mater. 2:577-585. [PubMed]
10. Skolimowski, I. M., D. A. Rowley, R. C. Knight, and D. I. Edwards. 1981. Reduced chloramphenicol-induced damage to DNA. J. Antimicrob. Chemother. 7:593-597. [PubMed]
11. Staros, J. V., R. W. Wright, and D. M. Swingle. 1986. Enhancement by N-hydroxysulfosuccinimide of water-soluble carbodiimide-mediated coupling reactions. Anal. Biochem. 156:220-222. [PubMed]
12. Trail, P. A., H. D. King, and G. M. Dubowchik. 2003. Monoclonal antibody drug immunoconjugates for targeted treatment of cancer. Cancer Immunol. Immunother. 52:328-337. [PubMed]
13. Yacoby, I., M. Shamis, H. Bar, D. Shabat, and I. Benhar. 2006. Targeting antibacterial agents by using drug-carrying filamentous bacteriophages. Antimicrob. Agents Chemother. 50:2087-2097. [PMC free article] [PubMed]
14. Yerushalmi, Z., N. I. Smorodinsky, M. W. Naveh, and E. Z. Ron. 1990. Adherence pili of avian strains of Escherichia coli O78. Infect. Immun. 58:1129-1131. [PMC free article] [PubMed]

Articles from Antimicrobial Agents and Chemotherapy are provided here courtesy of American Society for Microbiology (ASM)
PubReader format: click here to try


Related citations in PubMed

See reviews...See all...

Cited by other articles in PMC

See all...


Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...