1

Reversing Bacterial Resistance to Antibiotics by Phage-1

Mediated Delivery of Dominant Sensitive Genes 2

Rotem Edgar,1,3 Nir Friedman,2,3 Shahar Molshanski-Mor, and Udi Qimron* 3

Department of Clinical Microbiology and Immunology, Sackler School of Medicine, Tel 4

Aviv University, Tel Aviv, Israel 5

6

Running title: reversing antibiotic resistance 7

8

Keywords: phage, E. coli, tellurite-resistance, rpsL, gyrA 9

10

*Corresponding author. Mailing address: Department of Clinical Microbiology and 11

Immunology, Sackler School of Medicine, Tel Aviv University, Tel Aviv 69978, Israel. 12

Phone: 972-3-6405191. Fax: 972-3-6409511. E-mail: ehudq@post.tau.ac.il 13

1 Present address: Epidemiology Division, Sourasky Medical Center, Tel Aviv, Israel 14

2 Present address: Department of Ruminant Sciences, Volcani Center, Bet Dagan, Israel 15

3 These authors contributed equally to this study. 16

Copyright 2011, American Society for Microbiology and/or the Listed Authors/Institutions. All Rights Reserved.Appl. Environ. Microbiol. doi:10.1128/AEM.05741-11 AEM Accepts, published online ahead of print on 23 November 2011

2

ABSTRACT 17

Pathogen resistance to antibiotics is a rapidly growing problem, leading to an urgent 18

need for novel antimicrobial agents. Unfortunately, development of new antibiotics 19

faces numerous obstacles, and a method that will resensitize pathogens to approved 20

antibiotics therefore holds key advantages. We present a proof-of-principle for a 21

system that restores antibiotic efficiency by reversing pathogen resistance. This 22

system uses temperate phages to introduce, by lysogenization, genes rpsL and gyrA 23

conferring sensitivity in a dominant fashion to two antibiotics, streptomycin and 24

nalidixic acid, respectively. Unique selective pressure is generated to enrich for 25

bacteria that harbor the phages encoding the sensitizing constructs. This selection 26

pressure is based on a toxic compound, tellurite, and therefore does not forfeit any 27

antibiotic for the sensitization procedure. We further demonstrate a possible way of 28

reducing undesirable recombination events by synthesizing dominant sensitive genes 29

with major barriers to homologous recombination. Such synthesis does not 30

significantly reduce the gene's sensitization ability. Unlike conventional 31

bacteriophage therapy, the system does not rely on the phage's ability to kill 32

pathogens in the infected host, but instead, to deliver genetic constructs into the 33

bacteria, and thus render them sensitive to antibiotics prior to host infection. We 34

believe that transfer of the sensitizing cassette by the constructed phages will 35

significantly enrich for antibiotic-treatable pathogens on hospital surfaces. Broad 36

usage of the proposed system, in contrast to antibiotics and phage therapy, will 37

potentially change the nature of nosocomial infections toward being more 38

susceptible to antibiotics rather than more resistant. 39

3

INTRODUCTION 40

41

Bacteria have evolved to overcome a wide range of antibiotics, and resistance 42

mechanisms against most of the conventional antibiotics have been identified in some 43

bacteria (7). Accelerated development of newer antibiotics is being overrun by the pace 44

of bacterial resistance. In the USA, for example, over 70% of hospital-acquired infections 45

involve bacteria resistant to at least one antibiotic, and in Japan over 50% of the clinical 46

isolates of Staphylococcus aureus are multidrug-resistant (11). 47

This increasing threat has revived research into phage therapy. For example, a 48

clinical phase I and II control trial was recently completed successfully for the treatment 49

of chronic bacterial ear infections (21). Nevertheless, although phage therapy has been 50

practiced for several decades in some of the former Soviet Union countries and Poland, 51

there are still many doubts as to its ability to replace antibiotics. Major concerns over the 52

use of phage therapy include neutralization of phages by the spleen/liver and by the 53

immune system, their narrow host range, bacterial resistance to the phage, and lack of 54

sufficient pharmacokinetic and efficacy studies in humans and animals (1, 11). 55

A recent study used phages as a genetic tool to increase bacterial susceptibility to 56

antibiotics. That study used phage M13, of the Gram-negative Escherichia coli, to 57

genetically target several gene networks, thus rendering the bacteria more sensitive to 58

antibiotics (10). It demonstrated that disrupting the SOS response by M13-mediated 59

gene-targeting renders the bacteria several-fold more sensitive to a variety of antibiotics. 60

It also demonstrated that phage-mediated gene transfer combined with antibiotics 61

increases the survival of mice infected with pathogenic E. coli. Overall, the study showed 62

4

that transferring genes by phage M13 weakens the bacteria, and render them more 63

susceptible to killing by antibiotics. We believed that some aspects of that study required 64

further modification. First, the transferred genes target a beneficial pathway in bacteria, 65

and therefore significantly reduce the fitness of the bacteria harboring this phage. 66

Consequently, negative selection pressure is constantly being applied against transfer of 67

these genes by the M13 phages. Second, no mechanism to facilitate genetic transfer of the 68

M13 genes was used: high multiplicity of infection in the experimental settings 69

compensated for this shortcoming. Nevertheless, such settings cannot be used in field 70

experiments. Third, the phage was experimentally tested in vivo, in mice, but immune 71

responses against it were not examined. Therefore, despite the novelty of that study in 72

terms of unique genetic-targeting by phages, the end result is very similar to conventional 73

phage-therapy practices, in which phages are used to directly kill the pathogen. 74

Different approaches make use of phages as "disinfectants" of pathogens present 75

on edible foods, plants, and farm animals. In addition to increasing the shelf life of these 76

products, the treatment is intended to prevent occasional outbreaks of disease. The US 77

Food and Drug Administration recently approved the use of an anti-Listeria phage 78

cocktail for application on meat and poultry as a preventive measure against Listeria (5). 79

Other phage cocktails have been approved as food additives in Europe, and many are 80

currently being developed by phage biotech companies. These applications demonstrate 81

that phages can be dispersed in the environment and efficiently target pathogens in their 82

surroundings. 83

Here we present a proof-of-principle for genetic delivery of constructs using 84

phages to target pathogens in the environment. In the described system, phages are 85

5

genetically engineered to reverse the pathogens' drug resistance, thereby restoring their 86

sensitivity to antibiotics. The phages transfer, by lysogenization into the pathogens, a 87

drug-sensitizing DNA cassette, which was previously shown to render bacteria sensitive 88

to agents to which they had acquired resistance (6). Pathogens that are lysogenized by the 89

designed phages are selected by tellurite because the phage is engineered to contain a 90

DNA element conferring resistance to this bactericidal agent. Rather than being 91

administered to patients, the phages are intended for dispersion on hospital surfaces, thus 92

gradually reversing the occurrence of drug-resistant pathogens and competing with the 93

resistant pathogens residing in hospitals. This method would thus enable the use of well-94

established antibiotics against which resistance has been acquired. 95

96

MATERIALS AND METHODS 97

98

Bacterial strains. Bacterial strains used in this study are listed in Table S1 in the 99

supplemental material, as well as in Tables 1 and 2. 100

Oligonucleotides. Oligonucleotides used in this study are listed in Table S2 in the 101

supplemental material. 102

Isolation of resistant mutants. Over twenty different overnight cultures of a total 103

of ~1011 E. coli K-12 cells were inoculated on Luria-Bertani (LB)-agar plates containing 104

50 g/ml streptomycin or 50 g/ml nalidixic acid. Resistant mutants emerged, in both 105

cases, at a median frequency of ~1 in 109 CFU, and were picked from different cultures, 106

to reduce the occurrence of sibling mutants. These bacteria were streaked on an agar plate 107

6

containing the appropriate antibiotic. The rpsL or gyrA genes of resistant mutants 108

emerging on the plate were PCR-amplified followed by DNA sequencing. 109

Plasmid construction. Plasmids were constructed using standard molecular 110

biology techniques. DNA segments were amplified by PCR. Standard digestion of the 111

PCR products and vector by restriction enzymes was carried out according to the 112

manufacturer's instructions. Plasmid sequences are provided in the supplemental material. 113

Phages. Genetic engineering of the different phages was carried out using 114

gt11/EcoRI/Gigapack III Gold Cloning Kit (Stratagene) according to the 115

manufacturer's protocols. Briefly, EcoRI-digested arms of phage gt11 were used to 116

construct the lysogenizing phage carrying the different DNA inserts, encoding a 117

chloramphenicol-resistance gene. DNA inserts were PCR amplified from plasmids 118

pRpsL-wt, pRpsL-sil, pRpsL4 (Fig. 1A) using primers 231F/R (Table S2), and digested 119

with MfeI restriction enzyme, which produces ends that are compatible with EcoRI. 120

Ligation was carried out using T4 DNA ligase (New England Biolabs). The ligated 121

products were transformed into E. coli strain Y1088, which supports gt11 growth. 122

Generated plaques were propagated in E. coli Y1088 or E. coli ymel, which were then 123

used to lysogenize the hosts. In several cases, phages were further manipulated in a host 124

which lacks supE, a suppressor gene necessary for phage growth. In such cases, the phage 125

was transferred by P1-mediated transduction to a permissive host and propagated there. 126

Phages carrying tellurite resistance were constructed by homologous recombination-127

based genetic engineering of the tellurite-resistance marker instead of the 128

chloramphenicol-resistance gene. The tellurite-resistance genes tehAB were amplified 129

from the E. coli chromosome using primers N1/N2 (Table S2) for -RpsL4-tell, -130

7

RpsL-wt-tell, -RpsL-sil-tell, -Ctrl-tell, and -GyrA-tell (Fig. 1B). Primers RE22/N2 131

(Table S2) were used for construction of -2xRpsL-tell. The obtained PCR products were 132

used for homologous recombination-based genetic engineering as described below. 133

Sequences of the phage inserts are presented in the supplemental material. 134

Homologous recombination-based genetic engineering. Homologous 135

recombination using short-homology flanking ends was performed as described 136

previously (13). Briefly, an overnight culture of lysogens carrying different DNA inserts 137

encoding the chloramphenicol-resistance gene was diluted 75-fold in 25 ml LB medium 138

with appropriate antibiotics and grown at 32C in a shaking water bath to an OD600 of 139

0.6. Then, half of the culture was heat-induced for recombination function of the 140

prophage at 42C for exactly 4 min in a shaking water bath. The remaining culture was 141

left at 32C as the uninduced control. The induced and uninduced samples were 142

immediately cooled on ice slurry and then pelleted at 3600 g at 0C for 10 min. The pellet 143

was washed twice in ice-cold ddH2O, then resuspended in 200 l ice-cold ddH2O and 144

kept on ice until electroporation with ~500 ng of a gel-purified PCR product encoding the 145

tellurite-resistance genes. A 25-l aliquot of electrocompetent cells was used for each 146

electroporation in a 0.1-cm cuvette at 25 F, 1.75 kV and 200 . After electroporation, 147

the bacteria were recovered in 1 ml LB for 1 h in a 32C shaking water bath and 148

inoculated on selection plates containing 1 to 4 g/ml tellurite. The DNA insertion into 149

the resulting phages, -RpsL4-tell, -RpsL-wt-tell, -RpsL-sil-tell, -Ctrl-tell, -GyrA-150

tell, and -2xRpsL-tell, was confirmed by PCR using primer 233F along with 232F or 151

232R. 152

8

Lysogenization. Overnight culture of the resistant mutants was diluted 1:100 in 153

LB with the appropriate antibiotics, 10 mM MgSO4 and 0.2% (W/V) maltose. When the 154

culture reached an OD600 of 0.60.8, 100 l was mixed with 10 l phage , carrying a 155

resistance gene, in a 1.5-ml tube and incubated at room temperature for 20 min. Cells 156

were inoculated on appropriate selection plates and incubated overnight at 32C. 157

Lysogens emerged on selection plates to which the phage carried a resistance gene. 158

Lysogenization was validated by plating the lysogens at 42C: lysogens cannot grow at 159

this temperature because the prophage is induced to its lytic cycle (see Materials and 160

Methods). 161

Transductions. Transductions were used to transfer antibiotic-resistance markers 162

or complete phages between strains (in cases where the strain did not carry suppressor 163

genes required for growth). P1 lysate was prepared as follows: overnight cultures of 164

donor strain were diluted 1:100 in 2.5 ml LB + 5 mM CaCl2 + 0.2% (W/V) glucose. After 165

1 h shaking at 37C (or 32C for lysogens), 107-108 PFU of phage P1 was added. 166

Cultures were aerated for 1 to 3 h, until lysis occurred. The obtained P1 lysate was used 167

in transduction where 100 l fresh overnight culture was mixed with 1.25 l of 1 M 168

CaCl2 and 0 to 100 l P1 phage lysate. After incubation for 30 min at 30C without 169

shaking, 100 l Na-citrate and 500 l LB were added. Cultures were incubated at 37C or 170

32C for 45 or 60 min, respectively, then 3 ml of warm LB supplemented with 0.7% agar 171

was added and the suspension was poured onto a plate containing the appropriate drug. 172

Transductants obtained on antibiotic plates were streaked several times on selection 173

plates and verified by PCR for the presence of the transduced DNA fragment. 174

9

MIC determinations. MIC determination was carried out by following the 175

procedure described by Wiegand et al. (19). Briefly, bacterial cells were grown overnight 176

at 32C in LB and diluted to 107109 CFU/ml. The obtained suspension was serially 177

diluted 10-fold for different spot concentrations, as indicated. Approximately 1 l of 178

bacterial suspension was then spotted onto selection plates containing different 179

concentrations of either streptomycin or nalidixic acid along with the appropriate 180

selection agent (chloramphenicol or tellurite), as indicated, using a 48-pin replicator. 181

Plates were incubated overnight and photographed using MiniBis Pro (Bio-Imaging 182

Systems). Photographs were digitally manipulated using GIMP2 software to adjust 183

contrast. Liquid-based MIC determination assays were carried out by inoculating serial 184

dilutions of an antibiotic in liquid LB broth with bacterial cultures (OD600 ~0.05) in 96-185

well microtiter plates. Plates were incubated overnight at 32oC and OD600 was then 186

measured. The lowest antibiotic concentration at which the relative growth compared to 187

the "no-drug" control was below 10% was determined as the MIC. 188

189

RESULTS AND DISCUSSION 190

191

Mutations in the target gene, rpsL, constitute a major resistance mechanism 192

to streptomycin. Our overall goal in this study is to provide a proof-of-principle for a 193

genetic system able to restore drug sensitivity to drug-resistant pathogens residing on 194

hospital surfaces. We chose, as a first step, to use streptomycin as the model drug. 195

Streptomycin is highly useful as an effective antibiotic against both Gram-negative and 196

Gram-positive bacteria. For example, streptomycin is a mainstay of tuberculosis therapy. 197

10

However, streptomycin-resistant Mycobacterium tuberculosis emerge during treatment, 198

and 24 to 85.2% of them have mutations in either rpsL or rrs (15). The rpsL gene 199

product, S12, is an essential, highly conserved protein of the 30S small ribosomal 200

subunit. Most of the acquired resistance to streptomycin is due to specific mutations in 201

rpsL that prevent the inhibitory binding of streptomycin to the essential rpsL gene 202

product. We wanted to reproduce these findings in a model bacterium, E. coli, and then to 203

restore its sensitivity to streptomycin. We therefore inoculated E. coli K-12 on LB-agar 204

plates containing 50 g/ml streptomycin and selected for resistant mutants. This 205

procedure fairly simulates the selection of spontaneous drug-resistant-mutant evolution in 206

hospitals following streptomycin treatment. Resistant colonies emerged with a median 207

frequency of 1 in 109 CFU. Mutations in rpsL were found in 21 out of 22 resistant 208

mutants, a frequency that corroborates with that in clinical isolates. As listed in Table 1, 209

10 mutants harbored a K88R substitution in RpsL, 6 had an R86S substitution, and P42S, 210

K43L, K43N, R54S, K88E substitutions were each identified once. These mutation types 211

also corroborate with previous studies, confirming that a major mechanism for 212

streptomycin resistance relies on mutations in rpsL (e.g. (16, 18)). Therefore, targeting 213

this resistance mechanism or reversing its effect should prove highly beneficial in 214

controlling drug-resistant pathogens. 215

Wild-type (wt) rpsL transformed on a plasmid dominantly confers 216

streptomycin sensitivity. Lederberg first reported in 1951 that wt rpsL is a dominant 217

sensitive allele with regard to streptomycin resistance (6). This means that introduction of 218

a sensitive allele of wt rpsL into a streptomycin-resistant bacterial cell will result in 219

sensitivity of the bacterium despite the presence of a resistant rpsL allele. These findings 220

11

were never exploited for clinical practice. To use the dominant sensitivity of rpsL as a 221

platform for the next sets of experiments which provide proof-of-principle for use of the 222

constructs in medical settings in the future, we first established the system with our model 223

strains and plasmids. MICs to streptomycin were determined by agar-plate assay (19). In 224

this assay, ~104 cells are replica-plated on plates with different drug concentrations. The 225

lowest concentration at which there is no visible colony-formation is defined as the MIC. 226

The MICs throughout the study were also measured in a complementary liquid-227

determination assay, giving a similar readout (not shown). Two representatives of the 228

most common streptomycin-resistant strains obtained above were taken for further study: 229

strains Sm6 and Sm13, harboring mutations in rpsL leading to substitutions of R86S and 230

K88R, respectively. Their MICs to streptomycin were 100 g/ml and 200 g/ml, 231

respectively, whereas the MIC of the parental strain was 1.56 g/ml. We transformed 232

these strains with the plasmid pRpsL-wt, encoding the wt rpsL, or a control plasmid, 233

pRpsL4, encoding a mock gene (a defective rpsL with a 4-bp deletion that disrupts the 234

reading frame after amino acid 26 of the RpsL protein; see Fig. 1 as well as the 235

supplemental material for plasmid sequences and maps) under a modified early E. coli 236

promoter from phage T7. Transformed cells were selected on agar plates supplemented 237

with 35 g/ml chloramphenicol, as the plasmid encodes chloramphenicol acetyl 238

transferase, which confers chloramphenicol resistance. The MICs of the transformed 239

strains to streptomycin were then determined. As shown in Fig. 2A, transformation of the 240

plasmid encoding wt rpsL, pRpsL-wt, conferred a dominant sensitive phenotype, 241

restoring the MIC of the resistant mutants Sm6 and Sm13 from 100 g/ml to 12.5 g/ml 242

and from 200 g/ml to 3.125 g/ml, respectively. A control, streptomycin-sensitive E. 243

12

coli transformed with these plasmids (pRpsL4 or pRpsL-wt) retained similar MICs to 244

streptomycin (not shown). These results demonstrate that a wt rpsL allele delivered on a 245

plasmid into a streptomycin-resistant E. coli renders the cell significantly more sensitive 246

to streptomycin. 247

rpsL designed with decreased homology to the wt allele can efficiently restore 248

streptomycin sensitivity. The system we propose as a proof-of-principle is based on the 249

rpsL-containing construct being transferred horizontally between strains by 250

transformation, conjugation or transduction, as described below. Recombination events 251

between the chromosomal resistant rpsL and the delivered wt rpsL may reduce the 252

efficiency of the construct because it may eventually recombine with an rpsL copy that 253

does not confer sensitivity on the transformed strains (nevertheless, there is no danger 254

that it will confer resistance in sensitive strains as the sensitive allele is dominant). In 255

order to reduce the undesired recombination events between the incoming allele 256

conferring sensitivity and the resistant allele in the transformed cell we designed an allele 257

which cannot undergo homologous recombination with the bacterial copy. Efficient 258

homologous recombination requires identity between recombining genes. Reduction of 259

homology from 100% to 90% decreases the frequency of recombination over 40-fold in 260

E. coli (14). In addition, a minimal efficient processing segment of 23 to 27 bp that is 261

identical to the invading strand is required for efficient homologous recombination (14). 262

We synthesized an rpsL gene with silent mutations that maximize the incompatibility of 263

recombination with the sequence of wt rpsL. Silent substitutions were made in every 264

possible case, except where codon usage was less than 10% (see the supplemental 265

material for plasmid sequence). Overall, the genes were identical in only 62% of their 266

13

sequence, and there was no single minimal efficient process segment between the wt rpsL 267

and the new rpsL allele, thus providing efficient barriers against homologous 268

recombination. We designated this allele rpsL-sil, and the plasmid encoding it, pRpsL-sil. 269

The introduced silent mutations might hamper the folding of the encoded protein or its 270

expression levels (3). We therefore tested whether this allele, like the wt rpsL, can 271

dominantly restore sensitivity. As shown in Fig. 2B, dramatic sensitization to 272

streptomycin was observed, with the MIC values decreasing in Sm6 and Sm13 from 100 273

g/ml to 25 g/ml and from 200 g/ml to 6.25 g/ml, respectively. The efficiency of 274

restoration of sensitivity was lower than that observed with the wt rpsL, possibly due to 275

the product's folding efficiency, as already mentioned. Nevertheless, these results indicate 276

that both rpsL and rpsL-sil can efficiently restore sensitivity to streptomycin when 277

expressed from plasmids. 278

A toxic compound, tellurite, efficiently replaces chloramphenicol as a 279

selection marker. In the above experiments, chloramphenicol, under the constitutive bla 280

promoter, was used as a selection and maintenance marker for the rpsL-encoding 281

plasmids. However, chloramphenicol is not a dispensable antibiotic, and by using it in the 282

proposed system, sensitivity to streptomycin is restored by forfeiting sensitivity to 283

chloramphenicol. This outcome is less desirable than one in which drug sensitivity is 284

restored without forfeiting sensitivities to other drugs. We therefore sought to replace 285

chloramphenicol with a dispensable, yet efficient, selection substance. A resistance gene 286

against tellurite (TeO32), a toxic compound, was evaluated. Tellurite is toxic to bacteria 287

as it forms long-lived sulfur complexes, thus disrupting the thiol balance in the bacterial 288

cells. Tellurite was once used as a treatment for microbial infections, especially for 289

14

syphilis, prior to the discovery of antibiotics (22). The tellurite-resistance genes, tehAB, 290

present naturally in the E. coli chromosome, do not confer resistance to E. coli under their 291

endogenous promoter due to low transcription (9). Upon expression from an active 292

promoter (e.g. T7), however, the MIC of tellurite against E. coli increases 50- to 100-293

fold. The observed MIC of E. coli that either possess or lack the chromosomal tehAB 294

genes is equal or less than 2 g/ml, compared to a MIC of 128 g/ml in cells harboring 295

extrachromosomal tehAB with an active promoter (9). Remarkably, all of the known 296

tellurite-resistance genes, including tehAB, are highly specific to tellurite, showing no 297

cross-resistance to other compounds (22), and are therefore a relatively safe choice for 298

our purposes. In addition, the tehAB genes are relatively small (~2.5 kb), making them 299

easy to clone and genetically transfer. Linking these genes to the antibiotic-sensitizing 300

DNA cassette would thus constitute an appropriate selectable sensitizing construct. 301

Plasmids encoding rpsL-sil or the mock gene were constructed, carrying the tellurite-302

resistance genes, tehAB, instead of the gene encoding chloramphenicol acetyl transferase. 303

These plasmids were named pRpsL-sil-tell and pRpsL4-tell. The plasmids were 304

transformed into the streptomycin-resistant strains, Sm6 and Sm13, and the MICs of 305

these transformed cells to streptomycin were determined. Restoration of sensitivity by 306

tellurite-based plasmids was comparable to that observed with the chloramphenicol-based 307

plasmids (Fig. 3). pRpsL-sil-tell sensitized Sm6 from a MIC of 100 g/ml to 1.56 g/ml, 308

and Sm13 from a MIC of 200 g/ml to 12.5 g/ml. These results indicate that tellurite 309

can be used instead of the chloramphenicol-resistance marker. They also demonstrate that 310

tellurite can maintain the plasmids without cross-reactivity with the streptomycin-311

resistance phenotype. 312

15

Streptomycin-resistant bacteria lysogenized with phage encoding rpsL 313

become streptomycin-sensitive. The system is thus shown to restore drug sensitivity 314

using plasmids as a genetic delivery tool without forfeiting other drugs' efficiencies. The 315

system is intended for transfer to resistant pathogens residing on hospital surfaces, thus 316

rendering them treatable by antibiotics. Plasmids are efficiently transferred from host to 317

host mainly via conjugation. However, establishing a conjugation-based system requires 318

dissemination of bacteria harboring a conjugative plasmid, which is not desirable from 319

either regulatory or safety points of view. We therefore evaluated the use of phages as 320

safer delivery vehicles for the designed constructs. We chose , a model phage which has 321

been extensively studied, as a gene-delivery tool. This phage can infect its E. coli host 322

and proceed to the lytic or lysogenic cycle. We used a common phage mutant (gt11, see 323

Materials and Methods) which is directed to a specific cycle type according to the 324

ambient temperature, and has a deletion (nin5) designed to allow stable insertion of up to 325

5 kb of foreign DNA. This phage mutant was engineered to contain wt rpsL, rpsL-sil, or a 326

mock-rpsL, each linked to the tellurite-resistance genes and designated, respectively, -327

RpsL-wt-tell, -RpsL-sil-tell, and -RpsL4-tell (see the supplemental material for maps 328

of the DNA inserts in the phages). One of the streptomycin-resistant strains used above, 329

Sm13, was lysogenized with the recombinant phages and selected on agar plates 330

supplemented with 1.5 g/ml tellurite at 32oC, a temperature at which it favors the 331

lysogenic cycle. The lysogenized bacteria were propagated and their MICs to 332

streptomycin determined. Lysogenization of Sm13 by the phages resulted in sensitization 333

of the resistant mutants (Fig. 4A). The MIC value for the -RpsL4-tell lysogen was 200 334

g/ml, compared to 25 g/ml and 50 g/ml for -RpsL-wt-tell and -RpsL-sil-tell, 335

16

respectively. Although significant, the sensitization was not as efficient as that observed 336

using plasmid delivery. 337

Two copies of the rpsL gene are significantly more efficient than a single copy 338

in reversing resistance. We suspected that the decreased sensitization observed by 339

lysogenization relative to plasmid transformation is due to a lower number of rpsL gene 340

copies introduced by the phage. To test this and improve the sensitization, we cloned 341

the two different rpsL alleles (wt rpsL and rpsL-sil) into the phage, designated -342

2xRpsL-tell, and used it to lysogenize resistant strain Sm13 as above. Introduction of two 343

gene copies dramatically enhanced the sensitization efficiency of the lysogenized strains, 344

resulting in a significant decrease of the MIC from 200 g/ml to 1.56 g/ml, comparable 345

to the MIC observed for the sensitive strain (Fig. 4B). As a whole, these results constitute 346

a proof-of-principle for restoration of sensitivity to streptomycin using a phage that 347

carries sufficient copies of rpsL, at the "genetic cost" of a resistance marker to a toxic 348

compound. 349

Nalidixic acid-resistant bacteria lysogenized with phage encoding gyrA 350

show restored nalidixic acid sensitivity. The above results demonstrate that 351

streptomycin resistance can be reversed by the proposed system. We wished to expand 352

the proof-of-principle to other antibiotics as well, to demonstrate that a "multidrug-353

sensitivity cassette" can theoretically be used. We therefore chose to target quinolone 354

resistance, which also manifests dominant sensitivity by the wt allele (12). The quinolone 355

drug family targets the enzyme gyrase, encoded by gyrA, resulting in DNA-replication 356

arrest. Mutations in gyrA are observed in a specific region termed "quinolone-resistance-357

determining region" (QRDR). The wt gyrA allele is dominant sensitive and may therefore 358

17

reverse resistance (12). Nalidixic acid, the first of the synthetic quinolone family 359

antibiotics, was used here as a representative of the quinolone family. To test whether the 360

system can restore sensitivity to quinolone, we isolated spontaneous nalidixic acid-361

resistant mutants by plating sensitive E. coli on 50 g/ml nalidixic acid. Similar to the 362

isolation of the streptomycin mutants, we obtained mutants with previously reported 363

substitutions in the target gene, gyrA (Table 2). We identified five D87G substitutions 364

and three S83L substitutions in the gyrA gene product. These results corroborate another 365

study on pathogenic E. coli, which showed that 37 out of 38 isolated quinolone-resistant 366

gyrA mutants have substitutions at either S83 or D87 or both. Out of 36 pathogenic E. 367

coli resistant to high levels of nalidixic acid (MIC 256 g/ml), 35 had at least one 368

mutation in gyrA (4). Here, as with the isolation of streptomycin-resistant mutants, the 369

fact that most of the spontaneous mutants are located in the target gene highlights the 370

potential benefit of reversing the effect of these mutations. We next introduced the wt 371

gyrA expressed from its endogenous promoter, or a control construct, both linked to 372

tellurite-resistance genes, into phages, designated -GyrA-tell and -Ctrl-tell 373

respectively. We used these phages to lysogenize a nalidixic acid-resistant strain, Nal2, 374

harboring a S83L substitution in GyrA. The lysogens were selected on 4 g/ml tellurite 375

and tested for sensitization by measuring MICs as described above, using nalidixic acid 376

instead of streptomycin. As shown in Fig. 5, the gyrA construct significantly reversed the 377

mutant's resistance. The MIC of the resistant mutants decreased twofold when 378

lysogenized by a gyrA-encoding phage compared to the control phage. The significance 379

of this sensitization was corroborated by experiments in which we transformed gyrA-380

encoding plasmids into nalidixic acid-resistant mutants and observed a three orders of 381

18

magnitude decrease in the number of CFU on 50 g/ml nalidixic acid compared to 382

resistant cells transformed with a mock plasmid (Fig. S1). Overall, these results indicate 383

that the proposed system can be used to target nalidixic acid resistance as well as 384

streptomycin resistance. 385

Proposed application, safety measures, and advantages of the system. The 386

proof-of-principle presented here is a step toward solving the major threat of emerging 387

drug-resistant pathogens, against which we have limited new emerging antibiotic 388

weapons. It demonstrates that with simple genetic engineering, bacteria can be 389

resensitized to approved and useful antibiotics. It is suggested that the system be applied 390

in a simple treatment of hospital surfaces to reverse the resistance of nosocomial 391

pathogens. Phages against Listeria monocytogenes (ListShield) and E. coli 392

(EcoShield) are used to spray ready-to-eat food and effectively target the 393

contaminating pathogen (5). Phages are also used in the USA as effective pesticides on 394

edible crops, among others (5). Although the above applications used non-genetically-395

altered phages, they support the safety and efficacy of our proposed spraying of phages 396

on hospital surfaces. The proposed uses and advantages of the system are presented in 397

Table 3. Extended transfer of the sensitizing cassette by specifically constructed 398

lysogenizing phages might enrich for antibiotic-treatable pathogens on hospital surfaces. 399

This enriched, sensitive population might then interfere with the establishment of newly 400

introduced resistant pathogens by overtaking their ecological niche. Our approach differs 401

from conventional phage therapy in the sense that it does not use phages to kill the 402

pathogens directly. Consequently, there is no selection against the used phage, but rather 403

selection for pathogens harboring the phage because it contains tellurite resistance. 404

19

Moreover, the approach avoids the use of phages inside the patient's body, thus 405

overcoming toxicity issues and other drawbacks of phage therapy, such as phage 406

neutralization by the spleen and the immune system (11). The presented system makes 407

use of a temperate phage whose lytic cycle is induced at elevated temperatures (8). This 408

feature carries added value because in the environment, as long as the temperature is 409

below 32oC, there is only selective pressure for being lysogenized and taking up the 410

sensitizing genes linked to the tellurite-resistance marker. However, once the pathogens 411

are lysogenized, they are less bound to infect humans or other warm-blooded animals, 412

because such an infection would induce the lytic cycle of the prophage and consequently 413

kill the bacteria. This added benefit is not essential for a future product, but is an optional 414

addition. If successful, this approach will render most of the nosocomial infections 415

treatable by antibiotics, unlike the current situation in which most nosocomial infections 416

are caused by antibiotic-resistant pathogens. The sensitizing cassette may be expanded to 417

include other sensitizing genes such as thyA, conferring sensitivity in a dominant way to 418

trimethoprim (20). Designing several antibiotic-sensitizing genes on a single construct 419

reduces the probability of spontaneous regaining of resistance against all of these 420

antibiotics simultaneously. We believe that the provided proof-of-principle can be applied 421

to different pathogen-phage systems as the extensive co-evolution of phages and bacteria 422

suggests that it is possible to find lysogenizing phages specific to any bacterium. For 423

example, the described system could quite simply be modified to target pathogenic E. coli 424

(e.g. E. coli O104:H4) by carrying out several selection cycles of the described phage 425

on the desired host, until the phage becomes fully adapted to it. In addition, the dominant 426

sensitivity of both rpsL and gyrA alleles has been shown to be broad across bacterial 427

20

species (e.g. (2, 12, 17). Along with the proof-of-principle, we also demonstrate a simple 428

procedure for creating an efficient barrier against homologous recombination. The 429

procedure includes replacement of most of the wobble bases, thus reducing the identity 430

between genes as well as eliminating the minimal efficient processing segment for 431

homologous recombination, without significantly affecting the translated product. This 432

procedure can be carried out with currently available technology for basically any desired 433

gene, as synthetic gene production has become common practice. Further manipulations 434

to reduce the possibility of tellurite resistance being horizontally transferred without the 435

sensitization cassette can be designed in a future product. For example, the sensitization 436

genes can be constructed so that their presence is essential for acquisition of tellurite 437

resistance: the sensitizing genes would be preferentially positioned before a promoterless 438

tellurite-resistance gene, making tellurite resistance dependent on expression of the 439

sensitizing genes. In addition to the system's benefits, which can be improved by 440

implementing the above safety measures, these phages are simple to prepare and to apply, 441

constituting a great advantage. Broad usage of the proposed system, in contrast to 442

antibiotics and phage therapy, will potentially change the nature of nosocomial infections 443

by making the bacteria more susceptible to antibiotics rather than more resistant. 444

445

446

ACKNOWLEDGMENTS 447

448

We thank Nir Osherov for critical reading of the manuscript and Camille 449

Vainstein for professional language editing. 450

21

This research was supported by Israel Science Foundation grant 611/10, 451

Binational Science Foundation grant 2009218, German-Israel Foundation grant 452

2061/2009, and Marie Curie International Reintegration Grants PIRG-GA-2009-256340 453

and GA-2010-266717. 454

455

456

REFERENCES 457

458

1. Brussow, H. 2005. Phage therapy: the Escherichia coli experience. Microbiology 459 151:2133-2140. 460

2. Drecktrah, D., J. M. Douglas, and D. S. Samuels. 2010. Use of rpsL as a 461 Counterselectable Marker in Borrelia burgdorferi. Appl Environ Microbiol 76:985-987. 462

3. Kimchi-Sarfaty, C., J. M. Oh, I. W. Kim, Z. E. Sauna, A. M. Calcagno, S. V. Ambudkar, 463 and M. M. Gottesman. 2007. A "silent" polymorphism in the MDR1 gene changes 464 substrate specificity. Science 315:525-528. 465

4. Komp Lindgren, P., A. Karlsson, and D. Hughes. 2003. Mutation rate and evolution of 466 fluoroquinolone resistance in Escherichia coli isolates from patients with urinary tract 467 infections. Antimicrob Agents Chemother 47:3222-3232. 468

5. Lang, L. H. 2006. FDA approves use of bacteriophages to be added to meat and poultry 469 products. Gastroenterology 131:1370. 470

6. Lederberg, J. 1951. Streptomycin resistance; a genetically recessive mutation. J Bacteriol 471 61:549-550. 472

7. Levy, S. B., and B. Marshall. 2004. Antibacterial resistance worldwide: causes, 473 challenges and responses. Nat Med 10:S122-129. 474

8. Lieb, M. 1964. Ultraviolet Sensitivity of Escherichia Coli Containing Heat-Inducible 475 Lambda Prophages. Science 145:175-176. 476

9. Liu, M., and D. E. Taylor. 1999. Characterization of gram-positive tellurite resistance 477 encoded by the Streptococcus pneumoniae tehB gene. FEMS Microbiol Lett 174:385-478 392. 479

10. Lu, T. K., and J. J. Collins. 2009. Engineered bacteriophage targeting gene networks as 480 adjuvants for antibiotic therapy. Proc Natl Acad Sci U S A 106:4629-4634. 481

11. Merril, C. R., D. Scholl, and S. L. Adhya. 2003. The prospect for bacteriophage therapy in 482 Western medicine. Nat Rev Drug Discov 2:489-497. 483

12. Robillard, N. J. 1990. Broad-host-range gyrase A gene probe. Antimicrob Agents 484 Chemother 34:1889-1894. 485

13. Sharan, S. K., L. C. Thomason, S. G. Kuznetsov, and D. L. Court. 2009. Recombineering: a 486 homologous recombination-based method of genetic engineering. Nat Protoc 4:206-487 223. 488

22

14. Shen, P., and H. V. Huang. 1986. Homologous recombination in Escherichia coli: 489 dependence on substrate length and homology. Genetics 112:441-457. 490

15. Shi, R., J. Zhang, C. Li, Y. Kazumi, and I. Sugawara. 2007. Detection of streptomycin 491 resistance in Mycobacterium tuberculosis clinical isolates from China as determined by 492 denaturing HPLC analysis and DNA sequencing. Microbes Infect 9:1538-1544. 493

16. Sreevatsan, S., X. Pan, K. E. Stockbauer, D. L. Williams, B. N. Kreiswirth, and J. M. 494 Musser. 1996. Characterization of rpsL and rrs mutations in streptomycin-resistant 495 Mycobacterium tuberculosis isolates from diverse geographic localities. Antimicrob 496 Agents Chemother 40:1024-1026. 497

17. Sung, C. K., H. Li, J. P. Claverys, and D. A. Morrison. 2001. An rpsL cassette, janus, for 498 gene replacement through negative selection in Streptococcus pneumoniae. Appl 499 Environ Microbiol 67:5190-5196. 500

18. Timms, A. R., H. Steingrimsdottir, A. R. Lehmann, and B. A. Bridges. 1992. Mutant 501 sequences in the rpsL gene of Escherichia coli B/r: mechanistic implications for 502 spontaneous and ultraviolet light mutagenesis. Mol Gen Genet 232:89-96. 503

19. Wiegand, I., K. Hilpert, and R. E. Hancock. 2008. Agar and broth dilution methods to 504 determine the minimal inhibitory concentration (MIC) of antimicrobial substances. Nat 505 Protoc 3:163-175. 506

20. Wong, Q. N., V. C. Ng, M. C. Lin, H. F. Kung, D. Chan, and J. D. Huang. 2005. Efficient 507 and seamless DNA recombineering using a thymidylate synthase A selection system in 508 Escherichia coli. Nucleic Acids Res 33:e59. 509

21. Wright, A., C. H. Hawkins, E. E. Anggard, and D. R. Harper. 2009. A controlled clinical 510 trial of a therapeutic bacteriophage preparation in chronic otitis due to antibiotic-511 resistant Pseudomonas aeruginosa; a preliminary report of efficacy. Clin Otolaryngol 512 34:349-357. 513

22. Zannoni, D., F. Borsetti, J. J. Harrison, and R. J. Turner. 2008. The bacterial response to 514 the chalcogen metalloids Se and Te. Adv Microb Physiol 53:1-72. 515

516 517

23

TABLES 518

519

TABLE 1. E. coli K-12 streptomycin-resistant mutants, Sm1520

22, isolated on 50 g/ml streptomycin. Substitutions in the 521

RpsL protein are indicated, as well as the MIC to streptomycin. 522

523 Streptomycin-resistant

mutant

Substitution in

RpsL

MIC (g/ml)

Sm1 P42S 50

Sm2 K43L 800

Sm3 K43N 50

Sm4 R54S 50

Sm5 R86S 50

Sm6 R86S 100

Sm7 R86S 50

Sm8 R86S 50

Sm9 R86S 50

Sm10 R86S 50

Sm11 K88E 50

Sm12 K88R 400

Sm13 K88R 400

Sm14 K88R 400

Sm15 K88R 400

Sm16 K88R 400

Sm17 K88R 400

Sm18 K88R 400

Sm19 K88R 400

Sm20 K88R 400

Sm21 K88R 400

Sm22 none 50

24

TABLE 2. E. coli K-12 nalidixic acid-resistant mutants, Nal18, 524

isolated on 50 g/ml nalidixic acid. Substitutions in the GyrA protein 525

are indicated, as well as the MIC to nalidixic acid. 526

527 Nalidixic acid-resistant

mutant

Substitution in

GyrA

MIC (g/ml)

Nal1 S83L 50

Nal2 S83L 50

Nal3 S83L 50

Nal4 D87G 128

Nal5 D87G 128

Nal6 D87G 128

Nal7 D87G 128

Nal8 D87G 128

25

TABLE 3. Proposed procedure for using the lysogenizing phages and the selection agent (tellurite) compared to the current 528 cleaning and disinfecting procedures in hospitals. Advantages of this procedure are indicated. 529

Surface treatment/frequency Current

procedure

Proposed

procedure Advantage of proposed procedure

Floor/furniture

cleansing

Soap V V

Renders bacteria drug-sensitive

Lysogenizing phages X V

Disinfection

Disinfectant V V

Selects for drug-sensitive residual bacteria to occupy

ecological niche

Tellurite X V

Frequency Daily V V Repeated usage results in more sensitive clones

26

FIGURE LEGENDS 530

531

FIG. 1. Plasmid (A) and phage (B) maps. The inserts are drawn to scale, with the 532

relevant genes and genetic elements indicated. CAT - chloramphenicol acetyl transferase, 533

conferring chloramphenicol resistance; Pamp - bla promoter, P A1* - mutated T7-A1 534

promoter of T7 phage. The promoter could not be cloned without a mutation, and 535

therefore we proceeded with the following 1 bp deletion in the promoter sequence 536

(bolded): 537

T7-A1: AAAAGAGTATTGACTTAAAGTCTAACCTATAGGATACT TACAGCCAT, 538

T7-A1*: AAAAGAGTATTGACTTAAAGTCTAA CTATAGGATACT TACAGCCAT; 539

rpsL4 - encodes a truncated RpsL protein, rpsL-wt - encodes the RpsL protein; rpsL-sil 540

- encodes an RpsL protein, harboring numerous silent mutations, with only 62% identity 541

to the wt sequence; tehA/tehB - encode proteins which confer tellurite resistance; gyrA-wt 542

- encodes the Gyrase A protein. 543

544

FIG. 2. rpsL encoded by plasmids efficiently sensitizes streptomycin-resistant 545

mutants. Streptomycin-resistant mutants Sm6 and Sm13, transformed with a plasmid 546

encoding wt rpsL, pRpsL-wt (A) or rpsL having multiple silent mutations, pRpsL-sil (B), 547

become more sensitive to streptomycin as compared to mutants transformed with a 548

plasmid encoding a mock gene, pRpsL4. Serial 10-fold dilutions starting at 105 549

CFU/spot (from top to bottom) of the different mutants were spotted on plates with the 550

indicated streptomycin concentrations. Chloramphenicol was supplemented at 35 g/ml 551

in all plates to maintain the plasmid. Plates were incubated overnight and photographed 552

27

using MiniBis Pro (Bio-Imaging Systems). A representative experiment out of three is 553

presented. 554

555

FIG. 3. Tellurite-resistance genes efficiently replace chloramphenicol acetyl 556

transferase as a selection marker. Streptomycin-resistant mutants transformed with a 557

plasmid encoding wt rpsL as well as a tellurite-resistance gene, pRpsL-sil-tell, become 558

more sensitive to streptomycin as compared to mutants transformed with a plasmid 559

encoding a mock gene, pRpsL4-tell. Serial 10-fold dilutions starting at 105 CFU/spot 560

(from top to bottom) of the different mutants were spotted on plates with the indicated 561

streptomycin concentrations. Tellurite was supplemented at 1.5 g/ml in all plates to 562

maintain the plasmid. Plates were incubated overnight and photographed using MiniBis 563

Pro (Bio-Imaging Systems). A representative experiment out of three is presented. 564

565

FIG. 4. rpsL genes introduced by phage sensitize a streptomycin-resistant 566

mutant. Phage encoding a single copy of either wt rpsL (-RpsL-wt-tell) or rpsL-sil (-567

RpsL-sil-tell) sensitizes a streptomycin-resistant mutant, Sm13, compared to phage 568

encoding a mock gene (-RpsL4-tell) (A). Sensitization is significantly enhanced when 569

the phage carries both copies of rpsL (-2xRpsL-tell) (B). Serial 10-fold dilutions starting 570

at 105 CFU/spot (from top to bottom) of the different lysogens were spotted on plates 571

with the indicated streptomycin concentrations. Tellurite was supplemented at 1.5 g/ml 572

in all plates to maintain the prophage. Plates were incubated overnight and photographed 573

using MiniBis Pro (Bio-Imaging Systems). A representative experiment out of three is 574

presented. 575

28

576

FIG. 5. gyrA introduced by phage sensitizes a nalidixic acid-resistant mutant. 577

Phage encoding a single copy of wt gyrA (-GyrA-tell) sensitizes a nalidixic acid-578

resistant mutant, Nal2, compared to phage encoding a mock gene (-Ctrl-tell). 579

Triplicates of the different lysogens, at 104 CFU/spot, were spotted on plates with the 580

indicated streptomycin concentrations. Tellurite was supplemented at 4 g/ml in all 581

plates. Plates were incubated overnight and photographed using MiniBis Pro (Bio-582

Imaging Systems). A representative experiment out of three is presented. 583

A

Figure 1

A

pRpsL4-tellpRpsL4P amp P amp

pRpsL-wt-tell pRpsL-wt

P A1* P A1*

pRpsL-sil-tellpRpsL-sil

A J cI857 (nin5)

EcoRIgt11

B

-RpsL4-tell tehA tehBP A1* P amp

rpsL4

-Ctrl-tell

p

P A1*tehA tehB

P amp

-RpsL-wt-tell tehA tehBP A1* P amp

rpsL

p

-RpsL-sil-tell tehA tehBP A1* P amp

rpsL-sil

-2xRpsL-tell P amp

rpsL-sil

tehA tehBP A1* P amp

rpsL

-GyrA-tell tehA tehBP A1* P amp

rpsL gyrA

P gyrA

200 streptomycin (g/ml)A 0 1 56 3 125 6 25 12 5 25 50 100

Figure 2

2200 streptomycin (g/ml)A

Sm6

1 2 1 2 1 2 1 2 1 2 1 2 1 2 1 2 10 1.56 3.125 6.25 12.5 25 50 100

B2 - pRpsL-wt

Sm13

0 1.56 3.125 6.25 12.5 25 50 200100 streptomycin (g/ml)

1 pRpsL4Legend:

B

Sm13

Sm6

1 3 1 3 1 3 1 3 1 3 1 3 1 3 1 3 1 3

Sm13

3 - pRpsL-sil1 pRpsL4Legend:

Figure 3

3200 streptomycin (g/ml)

Sm6

1 3 1 3 1 3 1 3 1 3 1 3 1 3 1 30 1.56 3.125 6.25 12.5 25 50 100

1

3 - pRpsL-sil-tell

Sm13

1 pRpsL4-tellLegend:

Figure 4

Sm13

200 streptomycin (g/ml)A 0 1.56 3.125 6.25 12.5 25 50 10021 321 321 321 321 321 321 321 321 3

1 - -RpsL4-tell 2 - -RpsL-wt-tell 3 - -RpsL-sil-tell

B streptomycin (g/ml)1 4 1 4 1 4 1 4 1 4

0 1.56 3.125 6.25 12.5

Legend:

Sm13

Legend: 1 - -RpsL4-tell 4 - -2XRpsL-tell

Figure 5

21 21 21 21 21 21

Nal2

1 - -Ctrl-tell

Nalidixic acid (g/ml)0 8 16 32 64 128

2 - -GyrA-tellLegend: