Long-circulating bacteriophage as antibacterial agents

Proc.

Natl.

Acad.

Sci.

USA

Vol.

93,

pp.

3188-3192,

April

1996

Microbiology

Long-circulating

bacteriophage

as

antibacterial

agents

(phage/bacteria/reticuloendothelial

system/toxins/antibiotic

resistance)

CARL

R.

MERRIL*,

BISWAJIT

BISWAS*t,

RICHARD

CARLTONt,

NICOLE

C.

JENSEN*t,

G.

JOSEPH

CREED*,

STEVE

ZULLO*,

AND

SANKAR

ADHYAt

*Laboratory

of

Biochemical

Genetics,

National

Institute

of

Mental

Health

Neuroscience

Center

at

Saint

Elizabeths,

Washington,

DC

20032;

tExponential

Biotherapies,

Inc.,

19

West

34th

Street,

Penthouse,

New

York,

NY

10001;

and

tLaboratory

of

Molecular

Biology,

National

Cancer

Institute,

National

Institutes

of

Health,

Bethesda,

MD

20892

Contributed

by

Sankar

Adhya,

December

21,

1995

ABSTRACT

The

increased

prevalence

of

multidrug-

resistant

bacterial

pathogens

motivated

us

to

attempt

to

enhance

the

therapeutic

efficacy

of

bacteriophages.

The

ther-

apeutic

application

of

phages

as

antibacterial

agents

was

impeded

by

several

factors:

(i)

the

failure

to

recognize

the

relatively

narrow

host

range

of

phages;

(ii)

the

presence

of

toxins

in

crude

phage

lysates;

and

(iii)

a

lack of

appreciation

for

the

capacity

of

mammalian

host

defense

systems,

partic-

ularly

the

organs

of

the

reticuloendothelial

system,

to

remove

phage

particles

from

the

circulatory

system.

In

our

studies

involving

bacteremic

mice,

the

problem

of

the

narrow

host

range

of

phage

was

dealt

with

by

using

selected

bacterial

strains

and

virulent

phage

specific

for

them.

Toxin

levels

were

diminished

by

purifying

phage

preparations.

To

reduce

phage

elimination

by

the

host

defense

system,

we

developed

a

serial-

passage

technique

in

mice

to

select

for

phage

mutants

able

to

remain

in

the

circulatory

system

for

longer

periods

of

time.

By

this

approach

we

isolated

long-circulating

mutants

of

Esche-

richia

coli

phage

A

and

of

Salmonella

typhimurium

phage

P22.

We

demonstrated

that

the

long-circulating

A

mutants

also

have

greater

capability

as

antibacterial

agents

than

the

cor-

responding

parental

strain

in

animals

infected

with

lethal

doses

of

bacteria.

Comparison

of

the

parental

and

mutant

A

capsid

proteins

revealed

that

the

relevant

mutation

altered

the

major

phage

head

protein

E.

The

use

of

toxin-free,

bacteria-

specific

phage

strains,

combined

with

the

serial-passage

technique,

may

provide

insights

for

developing

phage

into

therapeutically

effective

antibacterial

agents.

The

discovery

of

viruses

that

can

infect

and

destroy

bacteria

was

greeted

with

considerable

optimism

earlier

in

this

century.

d'Herelle,

one

of

the

discoverers

of

these

viruses

(bacterio-

phages),

promoted

their

use

as

therapeutic

agents

for

the

treatment

of

infectious

diseases

(1).

Despite

the

efforts

of

d'Herelle

and

other

investigators,

the

use

of

bacteriophage

as

an

antibacterial

therapy

was

generally

abandoned

soon

after

the

introduction

of

antibiotics

in

the

1940s.

To

explain

the

relative

failure

of

bacteriophages

as

therapeutic

agents

in

human

infectious

diseases,

Stent

offered

the

following

sugges-

tions:

"..

.the

immediate

antibody

response

of

the

patient

against

the

phage

protein

upon

hypodermic

injection,

the

sensitivity

of

the

phage

to

inactivation

by

gastric

juices

upon

oral

administra-

tion,

and

the

facility

with

which

bacteria

acquire

immunity

or

sport

resistance

against

phages"

(2).

While

these

factors

may

have

been

important,

there

were

also

fundamental

misconceptions

that

may

have

impeded

the

use

of

bacteriophage

therapy.

Perhaps

the

most

serious

was

d'Herelle's

original

belief

that

there

was

"but

one

bacteriophage"

(1).

We

now

know

that

there

are

many

types

of

bacteriophage,

each

of

which

is

specific

for

a

specific

host

range

of

bacteria.

The

earlier

misconception

resulted

in

applications

of

The

publication

costs

of

this

article

were

defrayed

in

part

by

page

charge

payment.

This

article

must

therefore

be

hereby

marked

"advertisement"

in

accordance

with

18

U.S.C.

§1734

solely

to

indicate

this

fact.

3188

phage

capable

of

growing

on

one

bacterial

host

but

with

little,

if

any,

ability

to

influence

clinical

infections

caused

by

other

bac-

terial

strains.

Therapeutic

failures

may

also

have

resulted

from

the

contamination

of

bacteriophage

stocks

with

debris

from

bacterial

lysis,

which

typically

contain

toxins.

Such

contamination

is

known

to

cause

symptoms

ranging

from

mild

fever

to

death.

In

addition,

studies

concerning

the

fate

of

bacteriophage

in

nonimmune

germ-free

mice

suggested

that

even

in

the

absence

of

an

antibody

response,

bacteriophage

tend

to

be

rapidly

eliminated

from

the

circulation

by

the

reticuloendothelium

system

(RES)

(3),

thereby

reducing

the

number

of

phage

available

to

invade

bacteria

infecting

the

patient.

Given

the

increasing

incidence

of

antibiotic-resistant

bac-

teria,

we

undertook

the

current

study

to

determine

whether

it

is

possible

to

increase

the

efficacy

of

bacteriophage

therapy.

This

study

addressed

the

concerns

noted

above

by:

(i)

using

bacteriophages

that

are

specific

for

the

bacterial

strains

used

to

infect

the

experimental

animals,

(ii)

adapting

purification

methods

to

reduce

toxin

concentrations

to

levels

causing

minimal

side

effects,

and

(iii)

developing

methods

to

isolate

phage

mutants

that

have

a

capacity

to

avoid

entrapment

by

the

RES.

The

effects

of

these

methods

of

improving

the

in

vivo

antibacterial

efficacy

of

bacteriophages

were

tested

in

mice

infected

with

lethal

doses

of

bacteria.

In

this

study,

we

have

used

two

bacteriophage

strains,

A

and

P22,

and

their

corre-

sponding

hosts

Escherichia

coli

and

Salmonella

typhimurium.

The

life

cycles

of

these

two

phages

and

their

hosts

have

been

characterized

both

genetically

and

biochemically

in

great

detail

(4-6).

MATERIALS

AND

METHODS

Bacteria

and

Bacteriophage

Strains.

The

strains

of

E.

coli

and

S.

typhimurium

and

the

bacteriophage

strains

used

in

this

study

are

described

in

Table

1.

Preparation

of

Bacteria

and

Bacteriophage

Stocks

and

Detection

of

Toxin

Levels

in

Phage

Lysates.

For

infection

of

animals,

E.

coli

CRM1

or

Sa.

typhimurium

CRM3

were

grown

from

a

single

colony,

in

150

ml

of

LB

medium

in

a

shaking

incubator

at

37°C

until

the

OD600

of

the

culture

reached

1.0

[equivalent

to

109

colony-forming

units

(cfu)].

After

growth,

100

ml

of

the

bacterial

culture

was

harvested

by

centrifugation

at

16,000

x

g

for

10

min

at

4°C.

The

supernatant

was

discarded,

and

the

bacterial

pellet

was

resuspended

in

1

ml

of

LB

at

4°C.

From

this

preparation

the

bacteria

were

serially

diluted

by

a

factor

of 10

in

phosphate-buffered

saline,

from

1010

to

102.

W60,

Argol,

and

Argo2

phage

were

grown

in

CRM1

host

and

R34,

Argo3

and

Argo4

phage

in

CRM3

host,

to

make

high-titer

stocks

using

standard

procedures

(7).

Large-scale

preparations

and

purifications

of

bacteriophage

by

cesium

chloride

density

centrifugation

were

done

according

to

Sambrook

et

al.

(8).

Toxin

levels

in

phage

preparations

were

measured

quantita-

Abbreviations:

RES,

reticuloendothelial

system;

pfu,

plaque-forming

units;

cfu,

colony-forming

units.

Proc.

Natl.

Acad.

Sci.

USA

93

(1996)

3189

Table

1.

Strains

of

E.

coli

and

Sa.

typhimurium

and

bacteriophage

strains

used

Strain

number

Relevant

genotype

Source

Bacteria

CRM1

E.

coli

K-12

btuB::TnlO

NIH

stock

collection

CRM2

E.

coli

K-12

mut

D5

NIH

stock

collection

CRM3

Sa.

typhimurium

LT2

NIH

stock

collection

Phage

W60

AcI60cY17

NIH

stock

collection

Argol

AcI60cY17argl

This

study

Argo2

AcI60cY17arg2

This

study

R34

P22vir3

NIH

stock

collection

Argo3

P22vir

arg3

This

study

Argo4

P22vir

arg4

This

study

NIH,

National

Institutes

of

Health.

tively

by

the

limulus

amebocyte

lysate

assay

(Associates

of

Cape

Cod,

Woods

Hole,

MA).

Animal

Model

of

Bacterial

Infection.

For

each

infection

experiment,

1-week-old

BALB/c

female

mice

in

groups

of

three

were

used.

To

determine

the

lethal

dose

of

bacteria,

serial

dilutions

(102-1011

cfu)

were

injected

i.p.

into

mice,

and

the

animals

were

observed

for

100

hr.

In

addition,

to

determine

whether

bacteria

or

phage

injected

into

the

peritoneal

cavity

can

effectively

enter

the

circulatory

system,

experiments

were

done

in

which

either

bacteria

or

phage

were

injected

i.p.

followed

by

the

collection

of

blood

samples

by

puncturing

the

orbital

plexus

with

a

heparinized

capillary

tube.

Approxi-

mately

1

ml

of

blood

was

collected

into

a

1.5-ml

Eppendorf

tube

containing

10

,ul

of

heparin

(5000

U.S.

Pharmacopeia

units/ml).

The

blood

samples

were

centrifuged

at

10,000

x

g

for

5

min,

and

the

plasma

was

collected

with

sterile

pipettes.

Plasma

from

these

samples

was

titered

for

bacterial

cfu

and

bacteriophage

plaque-forming

units

(pfu).

Blood

samples

were

also

collected

from

these

mice

before

injection

of

phage

or

bacteria

to

assure

that

animals

used

in

these

experiments

were

free

of

contaminating

bacteria

and

phage.

Mice

infected

with

bacteria

were

scored

for

levels

of

illness

by

monitoring

for

the

following

(indicating

progressive

disease

states):

decreased

physical

activity

and

ruffled

fur;

general

lethargy

and

hunch-

back

posture;

exudative

accumulation

around

partially

closed

eyes;

moribund;

and,

finally,

death.

Experiments

to

Study

the

Efficacy

of

Bacteriophage

as

Antibacterial

Agents.

Four

groups

of

mice

were

infected

with

2

x

108

cfu

of

E.

coli

CRM1

by

i.p.

injection.

We

previously

found

that

this

dose

of

CRM1

strain

is

lethal

for

BALB/c

mice

(results

not

shown).

After

infection,

three

of

the

groups

were

injected

i.p.

with

2

x

1010

pfu

of

bacteriophage

A

strain

W60.

The

fourth

group

was

monitored

as

an

untreated

control.

The

mice

in

this

experiment

were

scored

for

degree

of

illness

as

described

above.

Comparison

of

Protein

Profiles

of

Wild-Type

A

and

A

Mutants.

Preparations

of

purified

bacteriophage

were

exam-

ined

by

high-resolution

two-dimensional

protein

electrophore-

sis

(9-11)

to

search

for

virion

protein

differences

between

wild-type

and

mutant

bacteriophage

strains.

Proteins

found

to

vary

by

charge

or

molecular

weight

were

partially

sequenced

by

N-terminal

sequencing

(12,

13).

The

protein

sequence

data

were

used

to

search

the

European

Molecular

Biology

Labo-

ratory

data

bank

to

identify

phage

genes

that

code

for

mutant

proteins.

Genes

coding

for

proteins

of

interest

were

cloned

and

sequenced

(8,

14).

For

PCR

and

cloning,

the

two

primers

selected

were

5'-CCA

GCG

ACG

AGA

CGA

AAA

AAC

G-3'

and

5'-TTC

AGT

TGT

TCA

CCC

AGC

GAG

C-3',

which

yield

a

1545-bp

product

from

114

bp

upstream

to

84

bp

downstream

of

the

A

E

gene.

RESULTS

Experimental

Bacteremia

in

Mice.

We

followed

the

pres-

ence

of

E.

coli

strain

CRM1

in

the

mouse

circulatory

system

after

i.p.

injection

of

108

and

109

bacteria

by

sampling

blood

from

the

orbital

plexus.

Injection

of

108

bacteria

i.p.

resulted

in

the

accumulation

of

6.5

x

106

bacterial

cfu

in

the

blood

by

30

min.

The

cfu

increased

to

4.2

x

107

by

3

hr

and

then

decreased

to

2.5

x

104

cfu

by

24

hr.

Similarly,

30

min

after

i.p.

injection

of

109

bacteria,

blood

levels

of

1.7

x

108

bacterial

cfu

were

observed.

These

levels

rose

to

7

x

108

by

3

hr

and

then

decreased

to

2.5

x

108

by

7

hr.

All

of

the

animals

in

this

latter

group

died

sometime

during

the

next

17

hr.

Use

of

Virulent

Mutants

of

A

and

P22.

As

models

of

bacteriophage

therapy,

we

chose

to

infect

mice

with

a

labo-

ratory

strain

of

E.

coli

CRM1

or

Sa.

typhimurium

CRM3.

The

corresponding

phage

strains

specific

for

these

two

bacterial

species

were

virulent

strains

containing

specific

mutations

to

assure

virulence:

Avir(W60)

and

P22vir(R34).

The

use

of

virulent

bacteriophage

strains

prevented

survival

of

phage-

infected

bacteria

as

lysogens.

Effects

of

Bacterial

Debris

and

Toxins

in

Phage

Prepara-

tions.

Injection

of

109

pfu

of

filter-sterilized

Argol

phage

lysates

made

in

LB

broth

produced

only

a

mild

clinical

reaction

(ruffled

fur)

in

mice,

despite

an

endotoxin

level

of

5

x

104

endotoxin

units

(EU)/ml

in

the

phage

preparation

as

deter-

mined

by

limulus

amebocyte

lysate

assay.

In

contrast,

Argo3

phage

lysates

prepared

in

a

similar

manner

on

Sa.

typhimurium

had

endotoxin

levels

of

5

x

105

EU/ml,

and

all

the

animals

died

within

12

hr

after

i.p.

injection.

However,

cesium

chloride

density-gradient

centrifugation-purified

stocks

of

Argol

and

Argo3,

with

endotoxin

levels

reduced

to

0.3

x

101

EU/ml

and

1

x

103

EU/ml,

respectively,

had

no

adverse

effects

at

all

on

mice

when

injected

i.p.

Development

and

Partial

Characterization

of

Long-

Circulating

Bacteriophage.

Bacteriophage

with

an

enhanced

capacity

to

avoid

entrapment

by

the

RES

were

developed

by

selecting

phage

strains

that

could

remain

in

the

circulatory

system

of

mice

for

progressively

longer

periods.

This

selection

was

started

by

injecting

i.p.

into

each

mouse

1011

pfu

of

W60

grown

on

either

wild-type

CRM1

or

mutator

strain

CRM2,

followed

by

collection

of

blood

samples

from

the

mice

at

7

hr.

The

use

of

the

mutator

E.

coli

CRM2

bacteria

in

one

set

of

serial-passage

experiments

was

intended

to

increase

the

inci-

dence

of

mutation

in

phage

A,

so

as

to

enhance

the

probability

that

one

or

more

of

the

phage

offspring

would

have

properties

that

permit

evasion

of

the

RES.

The

phage

titers

at

7

hr

were

109

and

108

from

the

mutagenized

and

nonmutagenized

phage,

respectively.

The

phage

titers

in

the

circulatory

system

de-

creased

to

102

after

48

hr

and

to

undetectable

levels

after

120

hr.

The

residual

phage

present

at

7

hr

were

grown

to

high

titers

in

bacteria.

These

high-titer

phage

were

purified

by

passing

through

a

0.22-,um

membrane

(Millipore,

Bedford,

MA)

filter.

This

serial

cycling

of

phage,

by

injection

into

animals,

isolation,

and

regrowth

in

bacteria

was

repeated

nine

more

times.

Phage,

at

titers

of

1010

pfu,

were

injected

i.p.

in

the

second

through

the

tenth

cycles.

Of

these,

only

106

pfu

remained

in

the

circulatory

system

at

18

hr

in

the

second

cycle,

but

in

subsequent

cycles

the

titers

gradually

rose,

so

that

109

phage

particles

remained

in

the

circulatory

system

at

18

hr

in

the

fourth

cycle.

The

last

six

selection

cycles

provided

no

significant

increase

in

the

number

of

phage

remaining

in

the

circulatory

system

at

18

hr.

After

the

10th

cycle

of

the

selection

process,

a

single

plaque

from

each

of

the

two

experiments

was

isolated,

puri-

fied,

and

grown

to

high

titers

on

the

CRM1

host

and

designated

as

Argol

(cycled

on

strain

CRM1)

and

Argo2

(cycled

on

strain

CRM2).

Both

Argol

and

Argo2

displayed

similar

enhanced

capacity

to

avoid

RES

entrapment.

For

example,

the

18-hr

survival

after

i.p.

injection

of

Argol

was

16,000-fold

higher

and

that

of

Argo2

was

13,000-fold

higher

than

that

of

the

parental

A

strain

(Fig.

1).

A

similar

selection

process

was

used

to

isolate

long-

circulating

variants

of

the

Salmonella

phage

R34.

After

eight

selection

cycles,

long-circulating

single-phage

plaques

were

Microbiology:

Merril

et

al.

Proc.

Natl.

Acad.

Sci.

USA

93

(1996)

10

9

I

o

O

8

7

6

5

4

0

5

10

15

20

25

Time

(hrs)

FIG.

1.

Capacities

of

the

selected

long-circulating

phage

Argol

(0)

and

Argo2

(I)

phage

to

remain

in

the

mouse

circulatory

system

after

i.p.

injection

of

1010

pfu

into

BALB/c

female

mice

were

compared

to

that

of

the

W60

phage

(A).

Data

are

plotted

in

a

semilogarithmic

graph

in

which

the

logarithm

of

the

phage

titer

is

plotted

against

the

time

blood

was

sampled

from

the

mouse

circulatory

system

in

hours.

The

selected

long-circulating

mutant

Argol

and

Argo2

phage

display

almost

identical

enhanced

capacities

to

avoid

RES

entrapment,

in

comparison'to

their

parental

W60

phage.

Regression

analysis

indicates

that

the

capacity

of

Argol

and

Argo2

phage

to

remain

in

the

circulatory

system

is

significantly

better

(P

=

0.003)

than

that

of

W60

phage.

similarly

isolated

and

purified.

Two

such

isolates,

designated

Argo3

and

Argo4,

were

compared

with

the

parental

phage

for

their

rates

of

clearance

from

the

mouse

circulatory

system.

After

an

i.p.

injection

of

5

x

107

pfu,

there

were

no

detectable

R34

at

24

hr.

In

contrast,

3

x

102

pfu

and

2

x

103

pfu

of

Argo3

and

Argo4,

respectively,

were

detected

under

the

same

con-

ditions.

Comparison

of

Long-Circulating

Argo

Phage

Versus

Wild-

Type

Phage

as

Antibacterial

Agents

in

an

Animal

Model.

Four

groups

of

mice

were

injected

i.p.

with

2

x

108

cfu

of

E.

coli

CRM1.

The

mice

were

scored

for

degree

of

illness

as

de-

scribed.

The

first

group

was

a

control,

with

no

phage

treat-

ment.

Within

5

hr

these

mice

exhibited

ruffled

fur,

lethargy,

and

hunchback

posture.

By

24

hr

they

were

moribund,

and

they

died

within

48

hr.

There

were

three

groups

treated

with

1010

pfu

of

phage.

All

of

the

mice

treated

with

phage

survived.

However,

those

treated

with

W60

(group

2)

had

severe

illness

before

finally

recovering,

whereas

those

treated

with

Argol

(group

3)

and

Argo2

(group

4)

exhibited

only

minor

signs

of

illness

before

complete

recovery

(Fig.

2).

The

following

results

show

that

the

ability

of

phage

to

influence

bacterial

infections

is

dose-dependent.

Mice

were

injected

i.p.

with

strain

CRM1

(2

x

108

cfu)

suspended

in

phosphate-buffered

saline

and

stored

overnight.

In

contrast

to

the

experiment

in

Fig.

2,

the

control

group

that

did

not

receive

any

phage

in

this

experiment

did

not

die

by

48

hr,

an

effect

that

was

likely

due

to

reduced

virulence

of

the

washed

and

refrig-

erated

bacteria

used

in

this

experiment.

However,

they

did

develop

moderately

severe

signs

of

illness.

In

the

phage

treatment

groups,

30

min

after

the

bacterial

injection

the

animals

received

Argol

phage

in

doses

from

102

to

1010

pfu.

As

shown

in

Fig.

3,

at

the

minimum

dose

of

phage

(102

pfu)

at

30

hr

after

infection,

the

animals

showed

symptoms

of

disease

that

were

only

slightly

reduced

from

that

seen

in

the

controls.

5

4

o

a

..

3

o

OB

In

0

r~

0

20

40

60

Time

(Hours)

80

100

FIG.

2.

Comparison

of

the

efficiency

of

wild-type

W60

to

the

selected

long-circulating

strains

of

Argol

and

Argo2

as

therapeutic

agents

for

the

treatment

of

bacteremia

caused

by

the

i.p.

injection

of

E.

coli

CRM1

into

BALB/c

female

mice.

All

of

the

mice

were

injected

i.p.

with

a

lethal

dose

of

E.

coli

CRM1

(2

x

108

cfu).

Thirty

minutes

later

the

mice

in

group

2

(-)

received

an

i.p.

injection

of

W60

and

the

mice

in

group

3

(e)

received

an

i.p.

injection

of

Argol.

Mice

in

group

3

(A)

did

not

receive

any

phage.

A

fourth

group

treated

with

an

Argo2

produced

results

indistinguishable

from

those

observed

with

Argol.

The

mice

were

observed

and

rated

according

to

their

condition

for

a

period

of

100

hr:

0,

normal

mouse;

1,

decreased

activity

and

ruffled

fur;

2,

lethargy,

ruffled

fur,

and hunchback

posture;

3,

lethargy,

ruffled

fur,

hunchback

posture,

and

partially

closed

eyes

with

exudate

around

the

eyes;

4,

moribund;

5,

death.

As

the

observations

are

categorical

condition-stage

observations,

the

actual

distances

between

the

states

is

unknown.

For

this

reason

the

level

of

illness

indicated

is

provided

simply

to

record

the

progression

of the

disease

state.

In

addition,

there

was no

significant

variation

of

symptoms

within

any

of

the

experi-

mental

groups.

With

increased

doses

of

phage,

the

animals

fared

progressively

better.

At

the

maximum

dose

of

phage

used

(1010

pfu),

the

animals

showed

only

minimal

illness

(decreased

physical

ac-

tivity

and

ruffled

fur),

and

these

animals

were

nearly

fully

recovered

at

48

hr.

Characterization

of

Argol

and

Argo2

Phage.

The

capsid

proteins

of

W60

and

its

two

Argo

derivatives

were

analyzed

by

high-resolution

two-dimensional

electrophoresis

as

described.

The

results

revealed

an

alkaline

shift

in

the

38-kDa

major

viral

protein

in

Argol

compared

to

W60

(Fig.

4).

The

same

electrophoretic

protein

shift

was

observed

in

Argo2.

N-

Terminal

amino

acid

sequencing

of

the

38-kDa

protein

by

Edman

degradation

showed

the

following

sequence

for

the

first

15-amino

acid

residues

for

all

three

phage:

S

M

Y

T

T

A

Q

L

L

A

A

N

E

Q

K.

This

sequence

matched

completely

with

corresponding

region

of

the

major

A

capsid

head

protein

E

(15).

Dideoxynucleotide

sequencing

of

the

PCR-amplified

genes

for

the

A

capsid

E

protein

in

Argol

and

Argo2

revealed

a

G

->

A

transition

mutation

at

nt

6606

in

the

A

capsid

E

gene

in

both

Argol

and

Argo2.

This

transition

mutation

resulted

in

the

substitution

of

the

basic

amino

acid

lysine

for

the

acidic

amino

acid

glutamic

acid

at

position

158

of

the

A

capsid

E

protein

in

both

Argo

strains.

Argo2

protein

profiles

displayed

the

presence

of

an

additional

altered

protein,

which

also

had

an

alkaline

shift.

This

second

protein

has

a

molecular

mass

of

11.6

kDa.

N-Terminal

amino

acid

sequence

analysis

of

the

11.6-kDa

protein

in

W60

and

Argo2

phage

showed

the

se-

3190

Microbiology:

Merril

et

al.

Proc.

Natl.

Acad.

Sci.

USA

93

(1996)

3191

A

X

Maior

Head

Protein

(G

_'

^

B

0

10

20

30

40

50

60

Time

(hrs)

FIG.

3.

Dose-response

to

Argol

phage

in

mice

infected

with

E.

coli

CRM1.

The

experimental

design

is

the

same

as

that

used

in

Fig.

2.

In

the

phage

treatment

groups,

30

min

after

bacterial

injection

the

animals

received

Argol

phage

in

the

following

doses:

102

(*),

104

(v),

106

(A),

108

(E),

and

1010

(0).

0,

No

phage

treatment.

Each

experi-

mental

point

represents

the

results

with

five

animals

and

had

an

SEM

of

0.2

or

less.

X

Major

Head

Protein

(GPE)

A

quence

for

the

first

16

residues

to

be

M

T

S

K

E

T

F

T

H

Y

Q

P

Q

G

N

S.

This

sequence

is

identical

with

that

of

the

second

major

capsid

head

protein

D

of

A

(15).

The

DNA

sequence

alteration

in

the

D

gene

of

Argo2

was

not

determined.

DISCUSSION

Despite

the

enthusiasm

with

which

d'Herelle

and

others

promoted

the

use

of

phage

for

treatment

of

infectious

disease,

such

applications

are

now

rare

and

limited

in

scope.

One

of

the

few

remaining,

but

rarely

used,

applications

of

phage

in

treating

infectious

disease

is

based

on

the

use

of

Staphylococcus

aureus

phage

lysates

(16-18).

Bacterial

antigens

liberated

by

lysis

in

such

phage

preparations

have

been

reported

to

have

the

ability

to

stimulate

cell-mediated

immunity

and/or

delayed

hypersensitivity

against

staphylococcal

infections

in

humans.

These

lysates

are

believed

to

function

by

enhancing

production

of

antibody

in

individuals

with

St.

aureus

infections,

rather

than

by

the

direct

bactericidal

effect

of

the

phage.

Although

phage

therapy

by

direct

bactericidal

action

was

hampered

by

the

misconception

that

a

single

strain

of

phage

could

be

effectively

used

against

many

different

bacteria

(1),

some

success

had

been

reported

when

phage

strains

were

used

after

prescreening

for

ability

to

grow

on

the

infectious

bacteria

(19).

In

our

studies,

we

chose

phage

that

were

specific

for

the

host

bacterial

strains

and

that

were

virulent

as

well,

in

order

to

avoid

lysogeny,

for

the

bacteria

used

in

our

models

of

infection.

The

presence

of

bacterial

toxins

in

phage

preparations

may

also

have

limited

the

effectiveness

of

phage

therapy.

Both

endotoxins

and

exotoxins

are

often

released

in

the

lysate

during

the

lytic

growth

of

the

phage.

Because

of

this

problem,

some

investigators

have

administered

phage

orally

to

minimize

the

adverse

effects

of

toxins

(20).

However,

in

our

studies,

no

phage

particles

were

detected

in

the

mouse

circulatory

system

after

the

oral

administration

of

2

x

109

pfu

of

W60

(results

not

shown).

We

have

shown

that

the

toxin

content

of

phage

preparations

can

be

diminished

>100-fold

by

purifying

phage

FIG.

4.

High-resolution

two-dimensional

protein

electrophoreto-

grams

that

demonstrate

an

alkaline

shift

in

charge

of

the

38-kDa

A

capsid

E

protein

found

in

Argol

versus

the

parental

W60

phage.

The

electrophoretic

pattern

of

the

region

of

the

gel

containing

W60

capsid

E

protein

is

illustrated

in

A.

(B)

Same

region

of

the

electrophoreto-

gram

containing

the

Argol

capsid

E

protein,

demonstrating

the

alkaline

shift

in

this

protein.

by

cesium

chloride

density

centrifugation.

While

the

toxin

content

of

our

A

phage

preparations

grown

on

E.

coli

CRM1

do

not

create

severe

problems

and

thus

do

not

require

extensive

purification,

in

contrast,

phage

P22

lysates

grown

on

Sa.

typhimurium

CRM3

are

lethal

in

mice

and

thus

require

extensive

purification

before

parenteral

administration.

The

most

serious

difficulty

that

may

have

limited

the

efficacy

of

phage

therapy

is

the

rapid

elimination

of

>90%

of

the

administered

phage

from

the

circulatory

system,

thereby

de-

creasing

the

effective

dose

available

for

infecting

bacteria

in

vivo.

Although

it

was

postulated

that

antibodies

may

have

served

as

a

major

factor

in

the

elimination

of

phage

(2),

experiments

with

germ-free

animals

lacking

antibodies

against

phage

have

demonstrated

that

the

RES

is

sufficient

for,

and

highly

effective

in,

the

rapid

removal

of

administered

phage

from

the

circulatory

system

(3).

Assuming

that

the

removal

of

phage

by

the

RES

depends

primarily

on

the

nature

of

the

phage

surface

proteins,

we

developed

a

protocol

to

select

for

phage

variants

that

can

evade

capture

by

the

RES.

This

protocol

involves

repeated

serial

passage

of

phage

through

the

mouse

circulatory

system.

The

procedure

selects

for

phage

that

can

remain

in

circulation.

In

our

initial

experiments

we

have

isolated

by

this

protocol,

variants

of

E.

coli

phage

A

and

Sa.

tyhpimurium

phage

P22,

phage

whose

wild

types

have

been

well

characterized

physio-

4

3

Cu

%o

2

I

1

0

Microbiology:

Merril

et

al.

')

;'

.........ff.

:,:

ii-w

.:;.

'Ok

5

4:

k

r

Proc.

Natl.

Acad.

Sci.

USA

93

(1996)

logically

and

whose

genomes

have

been

completely

sequenced

(ref.

15,

and

A.

R.

Poteete,

personal

communication).

In

two

independent

experiments,

by

passing

A

through

10

selection

cycles

we

have

isolated

two

A

variants,

designated

Argol

and

Argo2.

Argol

and

Argo2

displayed,

respectively,

16,000-fold

and

13,000-fold

greater

capacity

to

evade

RES

entrapment

24

hr

after

i.p.

injection,

compared

to

the

parental

A.

While

Argol

was

the

result

of

a

spontaneous

mutation,

and

Argo2

was

isolated

after

mutagenesis,

both

contained

an

identical

muta-

tion

in

the

A

capsid

E

protein,

which shows

the

change

of

glutamic

acid

to

lysine,

presumably

at

the

solvent-exposed

surface

of

the

phage

virion.

While

the

long-circulating

Argo2

phage

had

an

additional

mutation

in

the

capsid

D

protein,

the

presence

of

a

common

alteration

in

the

E

protein

in

both

of

the

independently

isolated

long-circulating

phage

strains

strongly

suggests

that

the

change

in

the

capsid

E

protein

is

the

major

factor

in

diminishing

phage

entrapment

by

the

RES.

The

significance

of

a

double-charge

change,

from

acidic

to

basic,

in

avoiding

entrapment

remains

to

be

determined.

Interestingly,

both

Argol

and

Argo2

displayed

an

almost

identical

ability

to

rescue

mice

infected

with

a

lethal

dose

of

E.

coli

CRM1,

presumably

because

of

their

long-circulating

na-

ture.

The

correlation

between

the

ability

of

the

mutant

phage

to

remain

in

the

circulatory

system

and

their

ability

to

rescue

bacteremic

animals

suggests

that

preventing

the

capture

of

phage

by

the

RES

may

improve

the

capacity

of

phage

to

interact

with

infecting

bacteria.

Because

the

technique

also

succeeded

in

isolating

long-

circulating

mutants

of

Sa.

typhimuruium

phage

P22,

the

results

suggests

that

our

protocol

can

be

used

as

a

general

method

for

obtaining

phage

capable

of

reduced

capture

by

the

RES.

Our

results

also

suggest

that

such

long-circulating

phage

are

useful

as

antibacterial

agents.

Although

we

have

addressed

a

number

of

problems

that

may

have

limited

previous

applications

of

phage

therapy,

phage

candidates

for

therapeutic

applications

must

also

be

screened

against

the

presence

of

undesirable

phage

genes-e.g.,

the

3

toxin

gene

of

Corynebacterium

diph-

theria

(21),

genes

encoding

antibiotic

resistance,

and

genes

that

induce

lysogeny.

In

summary,

we

have

shown

that

phage

can

have

enhanced

therapeutic

efficacy

when

they

are

(i)

virulent

for

the

corre-

sponding

bacterial

host,

(ii)

essentially

free

of

contaminating

bacterial

toxin,

and

(iii)

capable

of

evading

the

RES.

The

development

of

such

phage

may

provide

important

tools

for

the

treatment

of

bacterial

diseases.

Dr.

John

Bartko

provided

expert

assistance

with

statistical

analysis

of

the

data

presented

in

Figs.

1

and

3.

Protein

sequence

analysis

and

composition

were

provided

by

The

Rockefeller

University

Protein/

DNA

Technology

Center,

which

is

supported

in

part

by

National

Institutes

of

Health

shared

instrumentation

grants

and

by

funds

provided

by

the

U.S.

Army

and

Navy

for

purchase

of

equipment.

1.

d'Herelle,

F.

(1922)

in

The

Bacteriophage:

Its

Role

in

Immunity,

trans.

Smith,

G.

H.

(Williams

&

Wilkins,

Baltimore).

2.

Stent,

G.

(1963)

in

Molecular

Biology

of

Bacterial

Viruses

(Free-

man,

San

Francisco),

p.

8.

3.

Geier,

M.

R.,

Trigg,

M.

E.

&

Merril,

C.

R.

(1973)

Nature

(Lon-

don)

246,

221-222.

4.

Hendrix,

R.

W.,

Roberts,

J.

W.,

Stahl,

F.

W.

&

Weisberg,

R.

A.

(1993)

Lambda

II

(Cold

Spring

Harbor

Lab.

Press,

Plainview,

NY).

5.

Poteete,

A.

R.

(1988)

Bacteriophage

P22,

Chapter

11

in

The

Viruses,

Vol.

2:

The

Bacteriophages,

ed.

Calendar,

R.

(Plenum

Press,

NY),

pp.

683-724.

6.

Neidhardt,

F.

C.,

Ingraham,

J.

L.,

Low,

K.

L.,

Magasanik,

B.,

Schaechter,

M.

&

Umbarger,

E.

H.

(1987)

Escherichia

coli

and

Salmonella

typhimurium,

Cellular

And

Molecular

Biology

(Am.

Soc.

Microbiol.,

Washington,

DC).

7.

Silhavy,

T.

J.,

Berman,

M.

L.

&

Enquist,

M.

L.

(1984)

Experi-

ments

with

Gene

Fusion

(Cold

Spring

Harbor

Lab.

Press,

Plain-

view,

NY).

8.

Sambrook,

J.,

Fritch,

E.

F.

&

Maniatis,

T.

(1989)

Molecular

Cloning:

A

Laboratory

Manual

(Cold

Spring

Harbor

Lab.

Press,

Plainview,

NY)

pp.

2.73-2.81.

9.

Bjellqvist,

B.,

Pasquali,

C.,

Ravier,

F.,

Sanchez,

J.-C.

&

Hoch-

strasser,

D.

F.

(1993)

Electrophoresis

14,

1357-1365.

10.

Hughes,

G.

J.,

Frutiger,

S.,

Paquet,

N.,

Ravier,

F.,

Pasquali,

C.,

Sanchez,

J.-C.,

James,

R.,

Tissot,

J.

D.,

Bjellqvist,

B.

&

Hoch-

strasser,

D.

F.

(1992)

Electrophoresis

13,

707-714.

11.

Merril,

C.

R.,

Switzer,

R.

C.

&

Van

Keuren,

M.

L.

(1979)

Proc.

Natl.

Acad.

Sci.

USA

76,

4335-4339.

12.

Hewick,

R.

M.,

Hunkapiller,

M.

W.,

Hood,

L.

E.

&

Dreyer,

W.

J.

(1981)

J.

Biol.

Chem.

256,

7990-7997.

13.

Matsudaira,

P.

(1987)

J.

Biol.

Chem.

262,

10035-10038.

14.

Scharf,

S.

(1990)

PCR

Protocols,

A

Guide

to

Methods

and

Applications,

eds.

Innis,

M.,

Gelfand,

D.,

Sninsky,

J.

&

White,

T.

(Academic

Press,

NY),

pp.

84-91.

15.

Daniels,

D.

L.,

Schroeder,

J.

L.,

Szybalski,

W.,

Sanger,

F.,

Coul-

son,

A.

R.,

Hong,

G.

F.,

Hill,

D.

F.,

Petersen,

G.

V.

&

Blattner,

F.

R.

(1983)

in

Lambda

II,

eds.

Hendrix,

R.

W.,

Roberts,

J.

W.,

Stahl,

F.

W.

&

Weisberg,

R.

A.

(Cold

Spring

Harbor

Lab.

Press,

Plainview,

NY),

pp.

519-676.

16.

Salmon,

G.

&

Symonds,

M.

(1963)

J.

Med.

Soc.

NJ

60,

188-193.

17.

Mudd,

S.

&

Shayegani,

M.

(1974)

Ann.

NY

Acad.

Sci.

236,

244-251.

18.

Dean,

J.,

Silva,

J.,

McCoy,

S.,

Chan,

P.,

Baker,

J.,

Leonard,

C.

&

Herberman,

R.

(1975)

J.

Immunol.

115,

1060-1064.

19.

d'Herelle,

F.

(1930)

The

Bacteriophage

and

Its

Clinical

Applica-

tions

(Thomas,

Baltimore),

p.

167.

20.

Slopek,

S.,

Weber-Dacbrowska,

B.,

Dacbrowski,

M.

&

Kuchare-

wicz-Krukowska,

A.

(1987)

Archiv.

Immunol.

Ther.

Exp.

35,

569-583.

21.

Bishai,

W.

&

Murphy,

J.

(1988)

in

The

Viruses:

The

Bacterio-

phages,

ed.

Calendar,

R.

(Plenum,

New

York),

Vol.

2,

pp.

683-724.

3192

Microbiology:

Merril

et

al.