Standardized bacteriophage purification for personalized phage therapy

Abstract

The world is on the cusp of a post-antibiotic era, but researchers and medical doctors have found a way forward—by looking back at how infections were treated before the advent of antibiotics, namely using phage therapy. Although bacteriophages (phages) continue to lack drug approval in Western medicine, an increasing number of patients are being treated on an expanded-access emergency investigational new drug basis. To streamline the production of high-quality and clinically safe phage preparations, we developed a systematic procedure for medicinal phage isolation, liter-scale cultivation, concentration and purification. The 16- to 21-day procedure described in this protocol uses a combination of modified classic techniques, modern membrane filtration processes and no organic solvents to yield on average 23 mL of 10¹¹ plaque-forming units (PFUs) per milliliter for Pseudomonas, Klebsiella, and Serratia phages tested. Thus, a single production run can produce up to 64,000 treatment doses at 10⁹ PFUs, which would be sufficient for most expanded-access phage therapy cases and potentially for clinical phase I/II applications. The protocol focuses on removing endotoxins early by conducting multiple low-speed centrifugations, microfiltration, and cross-flow ultrafiltration, which reduced endotoxins by up to 10⁶-fold in phage preparations. Implementation of a standardized phage cultivation and purification across research laboratories participating in phage production for expanded-access phage therapy might be pivotal to reintroduce phage therapy to Western medicine.

Full text

Standardized bacteriophage purification for
personalized phage therapy
Tiffany Luong
1
, Ann-Charlott Salabarria
1
, Robert A. Edwards
1,2
and Dwayne R. Roach
1,2
✉
The world is on the cusp of a post-antibiotic era, but researchers and medical doctors have found a way forward—by
looking back at how infections were treated before the advent of antibiotics, namely using phage therapy. Although
bacteriophages (phages) continue to lack drug approval in Western medicine, an increasing number of patients are being
treated on an expanded-access emergency investigational new drug basis. To streamline the production of high-quality
and clinically safe phage preparations, we developed a systematic procedure for medicinal phage isolation, liter-scale
cultivation, concentration and purification. The 16- to 21-day procedure described in this protocol uses a combination of
modified classic techniques, modern membrane filtration processes and no organic solvents to yield on average 23 mL of
10
11
plaque-forming units (PFUs) per milliliter for Pseudomonas,Klebsiella, and Serratia phages tested. Thus, a single
production run can produce up to 64,000 treatment doses at 10
9
PFUs, which would be sufficient for most expanded-
access phage therapy cases and potentially for clinical phase I/II applications. The protocol focuses on removing
endotoxins early by conducting multiple low-speed centrifugations, microfiltration, and cross-flow ultrafiltration, which
reduced endotoxins by up to 10
6
-fold in phage preparations. Implementation of a standardized phage cultivation and
purification across research laboratories participating in phage production for expanded-access phage therapy might be
pivotal to reintroduce phage therapy to Western medicine.
Introduction
Academic and military research institutions are being called to immediate action to produce bac-
teriophages—viruses that kill bacteria—for the treatment of antibiotic-resistant infections before drug
approval
1–4
. Sometimes called compassionate use, expanded access provides a patient with an
immediately life-threatening condition or serious disease rapid access to an investigational new drug
(IND) when no satisfactory alternative therapy options are available
5
. Although demand for bac-
teriophages (‘phages’for short) is increasing in several fields, including human and veterinary drugs,
biological products, food supply and cosmetics
4,6,7
, only a limited number of phage products are
produced under the regulations covering Good Manufacturing Practices (GMP)
8,9
. However, an
argument can be made that an effective, consistent and controllable process for phage production,
which also meets safety and efficacy demands for human and animal clinical use, has yet to be
achieved
10,11
. That is, medicinal phage products have generally been of low purity and titer
1,4,12–14
.
According to the Centers for Disease Control and Prevention (CDC) and the World Health
Organization, deaths from antibiotic drug-resistant bacteria are at historic highs
15,16
. In the United
States, more than 2.8 million multidrug-resistant (MDR) infections occur each year, and more than
35,000 people die as a result
15
. As the usefulness of antibiotics is waning, there is an urgent need to
develop new ways to treat infections, or MDR infections will be the leading cause of human death
worldwide by 2050 (ref.
16
). Now, researchers and medical doctors have found a way forward—by
looking back at how infections were treated before the advent of antibiotics
1,2
. Phage therapy started
with its first clinical application in 1919 (ref.
17
) and has seen continuous improvement to this day.
Although phages continue to lack drug approval in Western medicine, an increasing number of
patients have been treated on an emergency IND (eIND) basis under US Food and Drug Admin-
istration (FDA) or European Medicines Agency approval
1–4,14
. With IND clinical approvals taking an
average of 8 years from entry into clinical trials, patients experiencing antibiotic failure and their
family members are themselves advocating for expanded-access phage therapy. Thus, as the antibiotic
1
Department of Biology, San Diego State University, San Diego, CA, USA.
2
Viral Information Institute, San Diego State University, San Diego, CA, USA.
✉e-mail: dwayne.roach@sdsu.edu
NATURE PROTOCOLS | VOL 15 | SEPTEMBER 2020 | 2867–2890 | www.nature.com/nprot 2867
PROTOCOL
https://doi.org/10.1038/s41596-020-0346-0
1234567890():,;
1234567890():,;

resistance crisis continues to rise globally, expanded-access phage therapy is bridging the gap between
current antibiotic failures and future clinical trials.
As a type of virus, phages can reproduce only within a bacterial cell
18,19
. Therefore, a major hurdle
to creating medicinal phages is the separation of phages from bacterial cell debris such as endotoxins
(i.e., lipopolysaccharides (LPS)), peptidoglycan, exotoxins, flagella, nucleic acids, and other com-
pounds. If not adequately removed, these gross impurities could trigger potentially life-threatening
inflammation, sepsis, and septic shock in humans
20–22
. To date, the major safety concern with phage
products for human use has been endotoxin content
1,10,14
. Adequately removing endotoxins from
phage products for human eIND phage therapy has been problematic, often requiring significant
product dilution, which concurrently reduces the active agent of phages
1,3,4
. In addition, GMP
pharmaceutical phage products have reported endotoxin quantity that would not be actionable for
low-dose or topical applications and could be indicative of adverse effects if administered in certain
routes of administration, such as intravenously (i.v.), which is a common route of phage adminis-
tration recommended by the Center for Innovative Phage Applications and Therapeutics
1,2,4
.
Moreover, phage production methods have largely disregarded other unknown gross impurities that
might risk human health. Indeed, phage preparations can be safe in animal models
23–26
. However,
experimental phage therapy studies generally cultivate phages with laboratory-adapted reference
bacterial strains. For expanded-access eIND use, large quantities of a pathogen—often a clinical MDR
bacterial isolate—are required to produce a high number of phages.
Several ad hoc laboratory approaches are currently employed for phage cultivation and
purification for human use, which have been largely developed for small experimental
studies
1–4,14,18,27–34
. However, several pitfalls have materialized, including low phage recovery, high
gross impurities, inadequate endotoxin removal and addition of toxic chemicals (see the ‘Comparison
with other practiced protocols’section). Large-scale GMP pharmaceutical production of large phage
libraries will likely be needed to meet the demands for personalized phage therapy. However, this
approach is currently not time- and cost-effective, thus creating an unmet need for systematic small-
batch phage production to excel current efforts for expanded-access phage therapy.
In this protocol, we outline a systematic, comprehensive and practicable procedure for phage
isolation, selection, liter-scale cultivation, concentration and purification (for schematic, see Fig. 1).
Our protocol uses aspects to demonstrate product identity, purity and quality while protecting
Phage
plaque
isolation
(Steps 1–12)
DNA
extraction
(Steps 21–56)
Liter-scale
cultivation
(Steps 84–90)
Dead-end
filtration
(Step 91)
Phage
titration
(Step 92)
Phage
titration
(Step 105)
CsCI
dialysis
(Steps 112–119)
Cross-flow ultrafiltration
(Steps 93–104) Density
gradient
ultracentrifugation
(Steps 106–111)
Library
preparation
(Steps 57–62)
DNA sequencing
and analysis
(Steps 63–83)
Small
-scale
cultivation
(Steps 13–18)
Phage
titration
(Step 19)
Timing: 5–7 d 3 d3 d 2 d 1 d 2.5 d 1 d 4 d
Phage
titration
(Step 120)
Cell viability
(Steps 157–167)
Phage
titration
(Step 134)
LPS-affinity
chromatography
(Steps 121–133)
Protein
analysis
(Steps 144–156)
Endotoxin
quantification
(Steps 135–143)
START OPTIONAL END
START OPTIONAL END
Fig. 1 | Overview of bacteriophage cultivation and purification. The procedure starts with sourcing and isolating phages with a target bacterial strain.
After multiple rounds of agar plaque isolation, a single plaque is small-scale cultivatated overnight. Next, the newly isolated phage genome is
sequenced, annotated and screened for lysogenic and harmful genes. Phages deemed potentially safe for human use are then liter-scale cultivated.
After overnight culturing, phage lysate is sterilized by pressure-driven double dead-end filtration and cross-flow ultrafiltration (see Fig. 2for filtration
scheme). Cross-flow ultrafiltration also diafiltrates to remove growth medium and concentrates phage particles in buffer. As an option, CsCl density
gradient ultracentrifugation and dialysis can be used to further confirm phage stock homogeneity. LPS-affinity chromatography is used to remove
residual endotoxins. Lastly, the final phage preparation purity and safety is tested by quantifying endotoxin level, protein abundance and cell viability
after phage sample exposure.
PROTOCOL NATURE PROTOCOLS
2868 NATURE PROTOCOLS | VOL 15 | SEPTEMBER 2020 | 2867–2890 | www.nature.com/nprot

scientific data integrity for batch therapeutics. This production process uses typical microbiology
laboratory equipment and no organic solvents. The method yields sufficient phages for most
expanded-access eINDs and potentially for clinical phase I and II applications. We show that phage
preparations obtained using this protocol have high phage purity measured by Limulus amoebocyte
lysate (LAL) chromogenic assays, protein analysis and cell viability analyses (see the ‘Anticipated
results’section). In our hands, all executed runs of the procedure were successful, confirming the
reproducibility of this protocol.
Application of purified phage preparations
In recent years, the demand for small quantities of phages for expanded-access eIND and clinical
phase I and II trials has been growing. Small aliquots of phages are useful for animal experiments and
pilot studies in which high titer and high purity are required. For this purpose, we developed a liter-
scale phage production process. The production method is a scaled-up version of previously described
preparation processes
24,25
and is able to deliver high-titer phages (10
9
–10
12
PFUs) with low endo-
toxins (4.3–24.1 units) per milliliter (Table 1). We are currently in the process of using final patient
formulations in expanded-access eIND phage therapy. Our production method has three major
advantages: (i) it does not require specialized production equipment, such as a bioreactor or a
chromatography system; (ii) by using a cross-flow filtration (CFF) method to concentrate phages
(Steps 93–105) and exchange growth medium, no organic solvents are required; and (iii) filtration
materials are disposable, avoiding any cleaning validation, which saves additional time and costs.
Therefore, this protocol provides a standard of production for medicinal phages, allowing expanded-
access phage therapy to reach more of those in need, starting by increasing the global production of
safe, personalized phage products.
Recently, phages have been shown to play important roles in health through unexpected host
immune interactions
21,25,35,36
. However, immunological response studies have used varying degrees
of purified phages, ranging from sterile filtered lysates to low-endotoxin preparations in saline buf-
fer
25,37,38
. Although there is no evidence of direct toxicity induced by phage particles, study of
mammalian cell–phage interaction requires phage preparations to be free of bacterial cells, toxins and
other compounds to avoid skewing host responses. Owing to the high titer and purity of phage
production, it is suitable for cell culture and animal studies.
We have used this protocol for the production of four Pseudomonas phages, PAK_P1, PAK_P5,
E217 and PYO2; two Klebsiella phages, JG265 and JG266; and one Serratia phage, SM219 (Table 1,
Supplementary Table 1). Although the selection criteria of therapeutic phages are still under inves-
tigation
7,9,25
, and not covered here, the presented production method can be generally used for other
Gram-negative phage and final formulation designs.
Comparison with other practiced protocols
This phage production protocol offers several distinct advantages over currently practiced methods
conducted at laboratory scales. Table 1shows that a 6-L production run can produce between
Table 1 | Bacteriophage product final titer (Step 134), endotoxin level (Step 143) and estimated
number of doses produced
Phage
strain
a
Bacterial
host
PFUs (ml
−1
)
b
Endotoxin
units (ml
−1
)
c
Final
volume (ml)
Estimated
doses at
10
9
PFUs
EU 10
9
PFUs
−1
PAK_P1 P. aeruginosa 5.78 × 10
10
4.49 24 1,387 0.0173
PAK_P5 P. aeruginosa 4.00 × 10
11
4.30 30 12,000 0.00250
E217 P. aeruginosa 1.89 × 10
11
5.26 21 3,970 0.00529
PYO2 P. aeruginosa 4.00 × 10
12
5.05 16 64,000 0.000250
JG265 K. oxytoca 1.30 × 10
10
24.10 28 364 0.07
JG266 K. oxytoca 2.00 × 10
9
17.30 29 58 0.05
SM219 S. marcescens 2.90 × 10
10
18.85 19 552 0.0345
a
GenBank accession nos.: PAK_P1, KC862297.1; PAK_P5, KC862301.1; E217, MF490240; PYO2, MF490236.
b
PFUs determined in Step 20B.
c
EUs
determined by the Pierce LAL Chromogenic Endotoxin Quantitation Kit, Steps 135–143.
NATURE PROTOCOLS PROTOCOL
NATURE PROTOCOLS | VOL 15 | SEPTEMBER 2020 | 2867–2890 | www.nature.com/nprot 2869

6.0 × 10
10
and 6.4 × 10
13
phage virions in total, providing between 58 and 640,000 doses at 10
9
PFUs,
a commonly prescribed intravenous treatment dose
1,2,39
. In addition, endotoxin measured by LAL
chromogenic analysis ranged between 0.0003 and 0.03 endotoxin units (EUs) per 10
9
PFU dose
(Table 1). In comparison, phages prepared under prescription as ‘magistral preparations’for human
treatment in Belgium
40
and for expanded-access eINDs by Chan et al.
39
were reported to yield
~10
7
PFUs·mL
−1
at more than 10 EUs·mL
−1
. Because our phage preparations are of higher titer and
purity than those reported as being administered to patients for expanded-access phage therapy,
implementation of our protocol would conceivably allow for administration of higher phage titer per
treatment dose and maintain an FDA human intravenous limit of less than 5 EUs·kg
−1
h
−1
(ref.
41
).
In contrast to the polyethylene glycol (PEG) precipitation method conducted in most competing
protocols (e.g.., refs.
9,32
), the CFF method employed here (Steps 93–105) is a pressure-driven scalable
membrane filtration process that markedly decreases labor and improves purification reproducibility
(Fig. 2). PEG precipitation methods can produce similarly high phage yields
2,21
. However, high yields
are phage strain dependent
31,32,34
, and PEG precipitation does not effectively remove endotoxins
21
.
Therefore, PEG-precipitated phages typically require substantial dilution or implementation of other
endotoxin removal techniques. CFF is able to reliably fractionate phages from gross bacterial
impurities (e.g., <40 EUs·mL
−1
) and diafilter and concentrate phage particles 4- to 100-fold (Fig. 3).
Notably, this all occurs in a cost-effective, programmed and semi-automated single step (Fig. 1). CFF
was shown to be highly effective at concentrating phages
32
. The use of a CFF molecular weight cutoff
(MWCO) of 100 kDa in combination with several sterile washes appears to be important for the
significant removal of endotoxins from phage lysates because CFF using MWCO <30 kDa has been
shown to concentrate endotoxins along with phages
32
. In addition, CFF with MWCO of 100 kDa
would potentially remove exotoxins, which all have been found to be <100 kDa (ref.
42
). Together, the
inclusion of CFF is a cost-efficient way to concentrate and purify phages. It is also scalable with no
barrier to running several apparatuses with additional hardware (not described in this protocol).
The downstream cesium chloride (CsCl) density gradient ultracentrifugation (Steps 106–111),
dialysis (Steps 112–120) and affinity chromatography (Steps 123–134) processes might not be
required to meet FDA drug product endotoxin safety limits
41
. Without including these steps, our
approach yields ~30 ml of between 2.7 × 10
11
and 1.1 × 10
14
phage virions in total per 6-L batch after
CFF (Supplementary Table 1). This provides an estimated 270–110,000 doses at 10
9
PFUs with
<0.22 EU; the allowable endotoxin exposure for a 70-kg person would be 350 EUs h
−1
. Omission of
lengthy and costly density gradient ultracentrifugation, dialysis and affinity chromatography steps
would reduce phage preparation time to ~5 d. However, if time permits, CsCl density gradient
ultracentrifugation, dialysis and affinity chromatography offer further separation of potential gross
impurities and provide a strategy to avoid unwanted phage contaminants
11,34
. Density gradient
ultracentrifugation can offer visual confirmation that preparations are homogeneous, containing a
single phage strain, because unwanted contaminating phages appear as a second band owing to their
differing density (Fig. 1).
Bacterial cell
Phage virion
Microsolutes
Membrane
Growth medium
0.8 µm
0.45 µm
0.45 µm
0.22 µm
Lysate feed
Dead-end filtration Cross-flow
filtration
Waste
(permeate)
Phage
concentrate
(retentate)
100 kDa
Fig. 2 | Schematic of phage lysate dead-end filtration (Step 91) and cross-flow filtration (Steps 93–104) removal
of impurities. Phage lysates are sterilized by inline 0.8-, 0.45-, 0.45-, and 0.22-µm membrane filtration to remove
whole bacterial cells and cellular debris. Then, CFF is used to remove growth medium and microsolutes <100 kDa
(e.g., endotoxins, peptidoglycans, exotoxins, flagella and nucleic acids), while concentrating the phages in
phosphate buffer.
PROTOCOL NATURE PROTOCOLS
2870 NATURE PROTOCOLS | VOL 15 | SEPTEMBER 2020 | 2867–2890 | www.nature.com/nprot

Our protocol avoids the use of chloroform, denaturing solvents and detergents. Competing
protocols often recommend using chloroform to disrupt the bacterial cell wall to release internal
phage particles in lysates or as standard procedure to remove PEG after precipitating phages
43
. There
are risks to adding chloroform in the purification process, such as denaturing certain phages
18,29
.
Long-term exposure to chloroform by inhalation in humans has resulted in hepatitis, jaundice and
central nervous system effects, and it has been shown to be carcinogenic in animals after oral
exposure, resulting in kidney and liver tumors
44
. In our experience, chloroform also increases the
amount of bacterial cell debris by lysing phage-resistant cells—which harbor no phage particles—that
can emerge with extended bacterial culturing. Organic solvents and detergents can be an effective
method to reduce endotoxins in phage lysates
33,34,45
. However, they can be toxic to users, inhibit
chromogenic-based endotoxin quantification methods and decrease phage stability in storage
34
.
Lastly, we use Oxford Nanopore MinION sequencing over other sequencing technologies because
of several added benefits. The MinION offers rapid generation of sequence information, with the
reads being generated and available for study within minutes after initiation of the run. Importantly,
the latest generation of the MinION sequencer (cat. no. SQK-RAD004), in our recent experience with
phage genomics, allows a few reads to cover the whole genome. Furthermore, the assembly of the
sequences generated are nearly identical to the Illumina assembled sequences, in our case by only a
single base. Finally, the Oxford Nanopore sequencing is significantly less expensive than other
sequencing technologies, with a Flongle generating over 1 GB of sequence data, which is more than
sufficient to accurately assemble a phage genome.
Limitations of this protocol
We have used the production method described in this protocol for the cultivation and purification of
phages that infect Gram-negative bacteria, including Pseudomonas aeruginosa,Klebsiella oxytoca and
Serratia marcescens grown aerobically at 37 °C. Pathogenic strains of P. aeruginosa and Klebsiella,
along with Gram-negative Escherichia coli and Acinetobacter, according to the CDC, are becoming
increasingly resistant to most available antibiotics
15
. Therefore, the case studies presented here are
relevant for current expanded-access eIND phage therapy global needs. Notwithstanding the indi-
vidual culture conditions for each bacterium–phage pair, there should be minor optimization needed
to use this protocol for the production of a variety of phages. For instance, larger volumes of low-titer
phage lysates can be processed with CFF to obtain a sufficiently high-titer sample for downstream
procedures. Nonetheless, further validation of this protocol with other phages, such as those that
infect Gram positives or anaerobes, is needed.
a
1014 PAK_P1 PAK_P5
1012
1010
108
106
Lysate
CFF
CsCI
Chrom
PFU mL–1
b
Lysate
CFF
CsCI
Chrom
Lysate
CFF
CsCI
Chrom
Lysate
CFF
CsCI
Chrom
cd
PYO2 E217 105
103
101
100
10–2
EU 109 PFU–1
efg
JG265 JG266 SM219
Lysate
CFF
CsCI
Chrom
Lysate
CFF
CsCI
Chrom
Lysate
CFF
CsCI
Chrom
1012
1010
108
106
PFU mL–1
105
103
101
100
10–2
EU 109 PFU–1
Titer
Endotoxin
Fig. 3 | Process stepwise phage titer and endotoxin concentration throughout processing. PFUs per mL
(right yaxis; closed circles) and endotoxin units (EUs) normalized to 10
9
PFUs (left xaxis, open circles) after phage
lysate sterilization (Lysate, Step 92), CFF (Step 105), density gradient ultracentrifugation and dialysis (CsCl,
Step 120) and LPS-affinity chromatography (Chrom, Step 134). a–d,P. aeruginosa phages PAK_P1, PAK_P5, PYO2 and
E217. e,f,K. oxytoca phages JG265 and JG266. g,S. marcescens phage SM219. See Step 20B and Steps 121–134 for
phage titration and endotoxin quantification procedures, respectively.
NATURE PROTOCOLS PROTOCOL
NATURE PROTOCOLS | VOL 15 | SEPTEMBER 2020 | 2867–2890 | www.nature.com/nprot 2871

Another drawback, as with most purification techniques, is the loss of phages with increased
number of manipulation techniques. For instance, we found that phage yield decreases significantly
after LPS-affinity chromatography, which appears to be phage strain dependent. For example,
Pseudomonas phage PAK_P1 and PAK_P5 use LPS as their bacterial cell wall binding site. We found
that PAK_P1 and PAK_P5 titers each decreased from 5.19 × 10
11
to 5.78 × 10
10
and from 2.30 × 10
12
to 4.00 × 10
11
PFUs·mL
−1
, respectively (Supplementary Table 1). In contrast, the titers before and
after affinity chromatography remained stable for the other phages tested. We propose that the losses
for some phage strains are dependent on how strongly phages associate with the remaining LPS in
samples before affinity chromatography.
There might be a health risk of using CsCl. Our protocol includes phages retrieved
in 1,500 mg·mL
−1
CsCl, followed by 1:1,000 dialysis against -buffered saline (PBS) and
high-titer aliquots diluted in sterile saline to the required therapeutic phage dose. This suggests that
treatments would contain trace amounts of cesium (<80 ng·mL
−1
). Although radioactive
cesium can cause seizure, syncope, hypokalemia and chronic diarrhea
46
, non-radioactive
cesium is readily found in the human body, mostly derived from consumption of plant and
animal products
47
.
The scalability of the presented phage production process is unclear. The principles of our protocol
are compatible with other existing bioprocess methods used in the manufacture of biotherapeutics.
For example, cultivation of phages in bioreactors would offer scalability over the use of shake flasks
48
,
while still being compatible with presented downstream filtration steps that are easily scalable with
larger filters. Centrifugation steps, however, can have scalability limitations, and affinity chromato-
graphy is not ideal for high-throughput processes.
Experimental design
This protocol outlines a combination of classic and modern viral cultivation and purification tech-
niques, which are experimentally validated for Gram-negative phages (see the ‘Anticipated results’
section below). The protocol design is modular and adaptable, allowing each laboratory to decide how
to best implement the necessary controls, by using scientifically sound design and testing procedures
to achieve higher quality through continual improvement. Implementing this formal system will
ensure the identity, titer, quality and purity of human and animal phage products and prevent
instances of contamination, mix-ups, deviations and errors.
Plaque assay phage isolation
Plaque assays are routinely used to discover new phage strains from a wide range of sources, such as
sewage, bodies of water, liquefied soil and bodily fluids. We use a modified agarose overlay technique,
described previously
18
, where a filtered natural sample is serially diluted and poured over a lawn of
exponentially growing compatible bacteria (Steps 1–12). As the bacterial lawn grows, small zones of
lysis become visible. Each PFU is derived from an initial infection with a single virion followed by
phage-induced lysis of neighboring cells
18
. PFUs displaying different morphologies are selected to
inoculate small-scale liquid cultivation on the compatible bacteria. A critical factor in the isolation of
highly functional phages is the preparation of a pure solution of identical phages. Therefore, the PFU
selection process is repeated several times until all plaques exhibit a similar morphology, suggesting
that a single phage strain has been isolated.
Phage titration
Agar Petri plating techniques are routinely used to quantify phage particle numbers in preparations.
One option is to use the agarose overlay technique as previously described
18
, which consists of mixing
serially diluted phage samples with susceptible bacteria in molten agar and pouring over solid agar in
a Petri plate (Step 20A). A more rapid option is to perform several tenfold serial dilutions of a phage
stock in a microplate and spot these dilutions on a bacterial-seeded agar Petri plate (Step 20B; Fig. 1).
Conventional plaque titering can also be substituted or supplemented by the use of real-time
quantitative PCR (qPCR) if primers for the phage strain are available
49,50
.
Annotation and bioinformatic analysis of phage genomes
The primary goals of genome sequencing and analysis (Steps 57–83) are different when the phages
are required for human therapeutic applications than when they are used for research purposes. For
phages that will be used for expanded-access phage therapy, the primary goal of sequencing and
PROTOCOL NATURE PROTOCOLS
2872 NATURE PROTOCOLS | VOL 15 | SEPTEMBER 2020 | 2867–2890 | www.nature.com/nprot

annotation is to ensure that no harm is likely to be introduced by the phage preparation, and the
primary concern is the transfer of genetic material between bacteria by transduction. Prophages
directly influence the microbes where they reside: they often express genes that are beneficial to the
bacteria (called lysogenic conversion factors)
51
. There are three elements that are the primary concern
for transfer: (i) toxins, (ii) antibiotic resistance genes and (iii) virulence factors. The toxins associated
with cholera
52
, diphtheria
53
, scarlet fever
54
, shigella
55–57
and botulinum
58
are all encoded by phages
integrated into their host bacterial genome
59,60
. Were one of those toxins to be transferred by a phage
used for expanded-access phage therapy, the expression of the toxin might lead to increased mor-
bidity and mortality
57
. Similarly, the introduction of antibiotic resistance genes either already on the
phage genomes
61
or transferred between bacterial hosts by specialized or generalized transduction
62
might render traditional antibiotic therapies ineffective. Although the precise role of phages in the
spread of antibiotic resistance genes between bacteria is somewhat controversial
63
, it has been
demonstrated previously
64,65
. Moreover, the relative ease of antibiotic resistance gene(s) identification
warrants analysis. In addition to antibiotic resistance genes and toxins, phages are also implicated in
the horizontal transfer of virulence genes
66
.
Computational analysis of toxins, antibiotic resistance and bacterial virulence factors present their
own challenges. Novel toxins and mechanisms of virulence are continually being discovered
67
. This
limits understanding of the mechanisms of disease and our ability to predict the detrimental effect of
horizontal gene transfer. The identification of genes that provide direct antibiotic resistance
mechanisms is straightforward; however, the identification of resistant alleles of housekeeping genes
is more complex. For example, it is trivial to identify a β-lactamase encoding gene, but associating
specific point mutations with resistance is more computationally challenging without thousands of
genomes
68
. Multiple databases can be used to identify toxins, antibiotic resistance alleles and viru-
lence factors. For example, Abricate (https://github.com/tseemann/abricate) uses ensemble methods
to compare sequences to the most up-to-date databases, including the Comprehensive Antibiotic
Resistance Database (CARD)
69
, ResFinder
70
, the National Center for Biotechnology Information’s
AMRFinder
71
, Antibiotic Resistance Gene-ANNOTation (ARG-ANNOT)
72
, the virulence factor
database
73
and PlasmidFinder
74
.
In expanded-access phage therapy, care must be taken to avoid temperate phages that are able to
lysogenize their bacterial host. Although there are genetic approaches to inactivate the lysogenic
lifecycle
2
, it is preferable to begin with phages that are unable to lysogenize their host. Phage genome
sequencing allows rapid confirmation of whether a phage is likely to be temperate or virulent. Whole-
genome sequencing might also provide clues to other biological contaminants in the preparation,
depending on the abundance of the contaminants relative to the phage. However, in each of these
considerations, phage genome sequencing will identify only known features. Computational identi-
fication of a toxin, antibiotic resistance gene or virulence factor genes should not infer that other
potentially harmful elements are not present in the phage genome; rather, it only excludes that
currently known elements are not present.
The Oxford Nanopore MinION sequencer is the most convenient ‘long-read’sequencer currently
on the market, and sequencing phages with this device often results in a single read representing the
entire phage
75
. After sequence assembly, the genomes are explored for antibiotic resistance genes and
virulence factors (e.g., with the aforementioned Abricate) and by annotating with PATRIC
76
.
PHACTS is used to determine whether the phages are likely to be virulent or temperate
77
. Phages
without genes of concern and that are predicted to be virulent (i.e., strictly lytic lifecycle) are selected
for downstream liter-scale cultivation.
Finally, the approach described in this protocol is appropriate for the rapid analysis of phage
genomes for expanded-access therapy. However, as we focus on those characteristics of the phage,
further techniques should be applied to the phage and its genome for full characterization
78–80
.
Liter-scaled shake flask cultivation
Test tube and smaller shake flask cultivation in complex medium on a rotary shaker are convenient
for initial phage cultivation for bacterial susceptibility testing and DNA isolation. Large shake batch
cultures in complex medium on a rotary shaker are required for downstream process sterilization of
cultivated phages (Fig. 1). Phage enrichment by shaking flask cultivation can be scaled to produce 6 L
per batch, limited here by centrifuge rotor capacity. Seeding cultures with a multiplicity of infection
(MOI) of 0.1 can produce phage lysates with ~10
9
–10
10
PFUs·mL
−1
. Although mid-log stage of
bacterial growth is the most amenable to infection by most phages
81
, basic parameters for
NATURE PROTOCOLS PROTOCOL
NATURE PROTOCOLS | VOL 15 | SEPTEMBER 2020 | 2867–2890 | www.nature.com/nprot 2873

bacteria–phage interaction should be optimized. For example, selected complex medium and tem-
perature for the bacterium are factors that influence the infectivity of phages and burst size.
Membrane applications in phage production
We use two flow configurations for membrane processes, namely dead-end microfiltration (Steps 91
and 92) and cross-flow ultrafiltration (Steps 93–105; illustrated in Fig. 2). Dead-end microfiltration is
used to retain both bacterial cells and particulates that are collected on the membrane surface forming
afiltration cake. This layer then makes additional filtration effects, improving the separation effi-
ciency. Dead-end filtration is usually a batch-type process, with extensive membrane fouling being its
main disadvantage.
We employ several methods to minimize membrane fouling while filtering phage lysates. First, we
avoid the use of chloroform to sterilize lysates to help reduce bacterial debris that would otherwise be
derived from the final lysis of phage-resistant bacterial cells, which typically develop with prolonged
culturing. Moreover, in our experience, chloroform lysis does not significantly improve phage yield,
and it can harm certain phages
29
. Second, we subject phage lysates to two rounds of centrifugation
with a sterile bottle change in between. Third, we use a pleated cartridge filter design, which provides
a higher surface area. Lastly, two pleated cartridge filters are used for two-step membrane pore sizing.
Next, we use the Vivaflow CFF cassette as an efficient way to ultrafilter lysates with a high
concentration of phages (Steps 93–105). CFF works by introducing dead-end filtered lysate under
pressure across the membrane surface, instead of directly onto the micro-filter. During filtration, any
material smaller than the cross-flow membrane pore of 100 kDa passes through the membrane,
whereas larger suspended phage particles remain in the retentate stream. A membrane pore size of
100 kDa provides an equivalent to a spherical particle with a 3-nm diameter and is therefore sufficient
to retain all known phages
82
. We employ diafiltration with freshly autoclaved buffer to increase the
purity and improve the separation of phages from bacterial debris and the complex growth medium.
Simultaneously, permeate is withdrawn at the same rate that the buffer is added, to flush out
microsolutes from the phage solution. For instance, CFF with a 100-kDa membrane pore size can
remove both endotoxins, which are ~10 kDa in size, and all known exotoxins, which are typically
<30 kDa (ref.
42
). Thus, CFF is used to ‘wash’the phage particles while concentrating up to 2 L by a
factor of 30.
Upscaling the purification process using density gradient ultracentrifugation
Density gradient ultracentrifugation developed by Brakke
83
is a common technique used to isolate
and purify a wide range of phages purely on the basis of their density
28,34
. We perform a self-forming
density gradient isopycnic ultracentrifugal separation technique making use of differences in density
between gross impurities and viruses of a CFF phage preparation (Steps 106–111; Fig. 2). In self-
forming gradients, the phage solution is layered on top of the gradient medium and centrifuged, and,
as the solute molecules sediment to form the gradient, identical phages band at their isopycnic points.
After ultracentrifugation, virus bands may be visualized as a result of their light scattering. We
employ this technique for accomplishing removal of both other unwanted phage strains and gross
impurities. That is, different phages will be separated by their different densities during ultra-
centrifugation
11
. In addition, large endotoxin aggregates will remain near the top of the gradient,
whereas microsolutes will pass to the bottom of the gradient.
We then use dialysis to facilitate the removal of small, unwanted CsCl from phage particles in
solution by selective and passive diffusion through a semi-permeable membrane (Steps 112–120;
Fig. 2). Phage particles that are larger than the 100-kDa membrane pores are retained, but small
molecules and buffer salts pass freely through the membrane, reducing the concentration of those
molecules in the phage preparation. Dialysis also accomplishes both removal of bacterial small
molecules (e.g., 2.5–7,000-kDa LPS molecules) and storage buffer exchange for phage products, but it
dilutes the phage preparation.
Chromatographic removal of endotoxins
Endotoxin removal is critical when producing therapeutic phages in bacterial systems. We further
remove this hydrophobic molecule through commercial purification affinity chromatography (Steps
121–134), which is likely the most reliable and widely applied method used to remove endotoxins. We
then use the LAL assay (Steps 135–143) because of its sensitivity, reproducibility and simplicity for
endotoxin detection, which uses the blood coagulation system of the horseshoe crab and clots upon
exposure to endotoxins
84
.
PROTOCOL NATURE PROTOCOLS
2874 NATURE PROTOCOLS | VOL 15 | SEPTEMBER 2020 | 2867–2890 | www.nature.com/nprot

Phage preparation protein analysis
Bacterial protein contamination is an immunogenic component of phage preparations. We use semi-
quantitative sodium dodecyl SDS–PAGE and Coomassie staining to visualize potential bacterial
protein contamination in the final phage preparations (Steps 144–156). For example, the presence of
protein smears across a variety of sizes indicates high bacterial protein contaminants as seen in the
Klebsiella phage JG265 lysate (see the ‘Anticipated results’section). Phage structural proteins will
dominate clean preparations.
Phage preparation effects on human cell viability
Phages cannot infect human cells. However, phages can indirectly influence mammalian cell activ-
ities
35,36,85
. In addition, bacterial contaminants in phage preparations might elicit cell inflammatory
responses
38,86
. Cell-based assays are often used for screening medicinal products to determine if the
test molecules have effects on cell proliferation or show direct cytotoxic effects that eventually lead to
cell death
87
. In this protocol, we describe a method to test the effects of phage preparations on human
cells that is based on a multi-well format where ATP production is recorded using a plate reader
(Steps 157–167). ATP has been widely accepted as a valid marker of viable cells
87
. When cells lose
membrane integrity, they lose the ability to synthesize ATP, and endogenous ATPases rapidly deplete
any remaining ATP from the cytoplasm. The measurement of ATP using luciferase is the most
commonly applied method for estimating the number of viable cells in high-throughput screening
applications. The ATP assay has a high sensitivity and is less prone to artifacts. The ATP detection
reagent contains a cell lysing detergent, ATPase inhibitors to stabilize released ATP, luciferin as a
substrate and luciferase to catalyze the reaction that generates luminescence
87
.
Materials
Biological materials
Strains
●PAK_P1 phage (GenBank accession no. KC862297.1 | RRID: NCBITaxon_743813)
●PAK_P5 phage (GenBank accession no. KC862301.1 | RRID: NCBITaxon_1327964)
●PYO2 phage (GenBank accession no. MF490236 | RRID: NCBITaxon_2034342)
●E217_A65 phage (GenBank accession no. MF490240 | RRID: NCBITaxon_2034346)
●SM219 phage (this study, available from the corresponding author upon reasonable request)
●JG265 phage (this study, available from the corresponding author upon reasonable request)
●JG266 phage (this study, available from the corresponding author upon reasonable request)
●P. aeruginosa strain PAO1 (ATCC no. 47085 | RRID: NCBITaxon_208964)
●P. aeruginosa strain PAK
88
(RRID: NCBITaxon_1009714)
●Serratia marcescens strain (this study, available from the corresponding author upon reasonable request)
●Klebsiella oxytoca strain (this study, available from the corresponding author upon reasonable request)
Cell lines
! CAUTION The cell lines used in your research should be regularly checked to ensure that they are
authentic and not infected with mycoplasma.
●HeLa cells (ATCC no. CRM-CCL-2 | RRID: CVCL_0030)
●HEK293 cells (ATCC no. CRL-157 | RRID: CVCL_0045)
Reagents
●Tryptone (Fisher Scientific, cat. no. BP1421-2)
●Yeast extract (Fisher Scientific, cat. no. BP1422-2)
●Sodium chloride (NaCl; Fisher Scientific, cat. no. 7647-14-5)
●Agar (CulGenes, cat. no. C6002)
●PBS (BioPioneer, cat. no. MB1001)
●Tris base (Fisher Scientific, cat. no. 77-86-1)
●Glycine (Millipore Sigma, cat. no. G4392)
●SDS (Fisher Scientific, cat. no.BP166-500)
●Ethanol, absolute (Millipore Sigma, cat. no. E7023-500 ml)
●Nanobind CBB Big DNA Kit (Circulomics, cat. no. NB-900-001-01)
●DNase I (Life Technologies, cat. no. 90083)
●RNase A (Life Technologies, cat. no. EN0531)
NATURE PROTOCOLS PROTOCOL
NATURE PROTOCOLS | VOL 15 | SEPTEMBER 2020 | 2867–2890 | www.nature.com/nprot 2875

●PEG-8000 (Fisher Scientific, cat. no. BP233-100)
●Magnesium sulfate heptahydrate (MgSO
4
; Fisher Scientific, cat. no. M63-500)
●Acetic acid (Fisher Scientific, cat. no. A38S 500)
●Pierce High Capacity Endotoxin Removal Spin Columns, 1 mL (ThermoFisher Scientific, cat. no. 88277)
●Pierce LAL Chromogenic Endotoxin Quant Kit (ThermoFisher Scientific, cat. no. 88282)
Equipment
●Biosafety cabinet (ESCO, cat. no. AC2-4S9-NS)
●Microtiter plate reader (BMG, model: Clariostar, cat. no. 0430-100)
●Vortex mixer (VWR, cat. no. 12620-838)
●Spectrophotometer (Denovix, model. no. DS-11 FX+)
●High-speed centrifuge (Beckman Coulter, model: Avanti JXN-26 IVD, cat. no. B38623)
●Fixed-angle centrifuge rotor for 1 L (Beckman Coulter, model: JLA-8.1000, cat. no. 969328)
●Ultracentrifuge (Beckmann Coulter, Optima-L-90K, cat. no. 365672)
●SW41 rotor (Beckman Coulter, cat. no. 331336)
●Microcentrifuge (Sartorius, cat. no. A-14C)
●Mini-PROTEAN Tetra Vertical Electrophoresis System (Bio-Rad, cat. no. 1658028FC)
●Gel imager ChemiDoc XRS+System (Bio-Rad, cat. no. 1708265)
●Bunsen burner (VWR, model no. 89038-530)
●Pipette (1,000, 200, 20 and 10 µL; Sartorius, cat. no. UX-24505-33, -34, -35 and -36, respectively)
●Multichannel pipette (8 × 10 µL; Sartorius, cat. no. UX-24505-44)
●Electronic pipette controller (Argos, model: OmegaZen, cat. no. 25300-96)
●Water bath (10 L; Cole Parmer, cat. no. WE-14576-08)
●Peristaltic pump (Cole Parmer, model: Masterflex, cat. no. EW-07522-20)
●Pump head (Cole Parmer, cat. no. EW-77253-00)
●Tubing (ThermoFisher Scientific, cat. no. 8060-3015; Cole-Parmer, cat. no. ZX-06422-01)
●Microbiological incubator/shaker (Eppendorf, model: S44i, cat. no. S44I200005)
●Cell incubator (Eppendorf, model: C170i, cat. no. 6731010015)
●Incubator (VWR, model: 5420, cat. no. 97005-252)
●ThermoMixer (Eppendorf, cat. no. 2231000574)
●Stir plate (Heathrow, cat. no. HP8885794)
●Hemocytometer (Hauser Scientific, cat. no. 3110V)
●Milli-Q Water Purification System (Millipore Sigma, cat. no. ZRXQ010T0)
●Magnetic Tube Rack (ThermoFisher Scientific, model: DynaMag-2, cat. no. 12321D)
●DNA sequencer (Oxford Nanopore, model: MinION, cat. no. SQK-RAD004)
●Linux or Apple Macintosh computer
Consumables
c
CRITICAL Autoclave glassware and tubing to degrade phage particles before use.
●Petri plates (10 cm; Fisher Scientific, cat. no. 08-757-100D)
●Microcentrifuge tubes (1.7 mL; Sorenson, cat. no. 11500)
●Microcentrifuge tubes (2.0 mL; Sorenson, cat. no. 12030)
●Glass test tubes (VWR, cat. no. 10545-922)
●Glass 250-mL Erlenmeyer flasks with GL45 screw-cap (Kimble-Chase, cat. no. 26720-250)
●Glass 2-L Erlenmeyer flasks with GL45 screw-cap (Kimble-Chase, cat. no. 26720-2000)
●GL45 0.22-µm PTFE membrane vented cap (Corning cat. no. 1395-45LTMC)
●Pipette tips (10–1,000 µL; Sartorius, cat. nos. 37001-162, 53503-294, and 83007-372)
●Pasteur pipettes (VWR, cat. no. 16001-182)
●Centrifuge bottles (1 L; Beckman Coulter, cat. no. A98812)
●Capsule filters (0.8/0.45 µm and 0.45/0.22 µm; Sartorius, model: Sartopore 2, cat. nos, 5441306G5 and
5441307H5)
●Cross-flow ultrafiltration cassette (100 kDa; Sartorius, model: Vivaflow 50R, cat. no. VF20P4)
●Ultraclear tubes (14 × 89 mm; Beckman Coulter, cat. no. 344059)
●Luer-Lok syringe (5 mL; BD, cat. no. 309628)
●Syringe filters (0.45 and 0.22 µm; Sartorius, cat. nos. 16537 and 16541)
●26-gauge needle (BD, cat. no. 305111)
●Conical centrifuge tubes (15 and 50 mL; VWR, cat. nos. 430790 and 430828)
PROTOCOL NATURE PROTOCOLS
2876 NATURE PROTOCOLS | VOL 15 | SEPTEMBER 2020 | 2867–2890 | www.nature.com/nprot

●Glass beaker (2 L; VWR, cat. no. 10754-760)
●Dialysis membrane tubing (100 kDA MWCO; Spectra/Por, cat. no. 131408)
●96-well clear microplate (Corning, cat no. 35177)
●96-well white microplate (Greiner bio-one, cat no. 655088)
●PCR eight-tube strips and caps (Sarstedt, cat. no. 772.991.1002)
●T75 culture flask (Corning, cat. no. 430639)
●MinION flow cell (Oxford Nanopore, cat. no. R9.4.1)
●Kimtech wipes (Fisher Scientific, cat. no. 06-666)
●Gloves (Xceed, cat. no. XC-310-S, M, L)
Software
●MinKNOW MinION Release 19.12.5: https://community.nanoporetech.com/downloads (registration
required)
●Canu v2.0: https://github.com/marbl/canu
●Abricate v1.0: https://github.com/tseemann/abricate
●PHACTS website: https://edwards.sdsu.edu/PHACTS/
●PATRIC website: https://patricbrc.org
Reagent setup
c
CRITICAL Prepare all solutions with sterile double-distilled water (ddH
2
O) and store at room
temperature (RT; 20–25 °C) for up to 3 months, unless otherwise indicated.
Liquid growth medium
Mix 10 g of tryptone, 5 g of yeast extract and 5 g of NaCl in 1 L of ultrapure ddH
2
O. Autoclave.
Solid growth medium
Add 14 g of agar to growth medium and autoclave. Then, cool to 55 °C in a water bath for 30 min–1
h. Sterilely pour solid growth medium into sterile Petri plates. Thickness should be 5–10 mm. Let cool
at RT and store at 4 °C for up to 2 weeks.
Molten soft-agar
Mix 7 g of agar to growth medium and autoclave. Then, cool to 55 °C in a water bath to prevent agar
solidification before plating. Make fresh.
Tris-sodium chloride (TN) buffer
Mix 10 mM Tris (pH 7.0) and 150 mM NaCl. Adjust to pH 7.0. Autoclave and store at 4 °C for up to
3 months.
Tris-EDTA buffer
Mix 10 mM Tris (pH 7.0) and 1 mM EDTA (pH 8.0). Adjust to pH 7.5.
10× SDS–PAGE running buffer
Mix 30.0 g of Tris base, 144.0 g of glycine and 10.0 g of SDS in 1 L of ddH
2
O. Make fresh.
CsCl solutions
Mix 41.2 g (d=1.6), 34.13 g (d=1.5) or 20.49 g (d=1.3) in 50 mL of TN buffer. Sterile filter.
Nuclease solution
Mix 20 mg·mL
−1
of DNase I and 20 mg·mL
−1
of RNase A. Store at −20 °C for up to 30 months.
Precipitant solution (30% (wt/vol) PEG-8000 and 3 M NaCl)
In a sterile bottle, add 110 ml of ddH
2
O and 35 g of NaCl and dissolve completely. Add 60 g of
PEG-8000, cap bottle and shake. Incubate bottle in a 50–60 °C water bath for ~2 h, shaking occa-
sionally (at this point, it is normal for the solution to be turbid and separated into two phases).
Remove and let cool to RT, shaking occasionally. The solution should be clear or slightly turbid.
Dilute to 200 mL with ddH
2
O.
NATURE PROTOCOLS PROTOCOL
NATURE PROTOCOLS | VOL 15 | SEPTEMBER 2020 | 2867–2890 | www.nature.com/nprot 2877

Resuspension buffer (5 mM MgSO
4
)
Add 0.0123 g of MgSO
4
heptahydrate per 10 mL of ddH
2
O.
Sourcing of phages
We outline sourcing phages without enrichment cultures by plating environmental samples directly
onto an isolation host and sample from formed plaques (i.e., zones of lysis). For rarer phage sourcing,
an enrichment step might be required. For details, see ref.
89
.
c
CRITICAL Solid source material such
as soil or feces should be suspended in buffer or liquid growth medium to generate an aqueous sample
that can be plated directly on an isolation host.
Procedure
Phage plaque isolation ●Timing 3–5 d, depending on repetitions needed to obtain a single
plaque morphology for each isolated phage.
! CAUTION Phage isolation (Steps 1–12) should be completed in a Class II Biosafety cabinet.
1 To isolate phage strains de novo, centrifuge aqueous environmental viral sample at 8,000gfor
30 min at 4 °C to remove bulk debris.
c
CRITICAL STEP Volume of aqueous viral sample will be dependent on separation of liquid from
solid debris after centrifugation.
2 Decant the supernatant into a clean tube without disturbing debris pellet.
3 0.2-µm syringe filter the environmental viral sample.
4 In a sterile test tube, mix 100 µL of bacterial host at OD
600
0.2 and 100 µL of the filtered viral sample.
5 Add 3 mL of molten soft agar, mix gently and pour over solidified agar Petri plate.
c
CRITICAL STEP If the bacterial lawn is clear after incubation, dilute the environmental sample to
obtain single plaques.
6 Incubate under appropriate temperature and atmospheric growth conditions for 12–18 h or until
plaques form on a confluent lawn of bacteria.
7 Using a Pasteur pipette, select 1 PFU and re-suspend in 100 µL of PBS in a microcentrifuge tube.
? TROUBLESHOOTING
8 Prepare serial dilutions (e.g., tenfold) in PBS.
9 In sterile test tubes, mix 10 µL of the 10
−5
,10
−6
or 10
−7
dilutions with 500 µL of bacteria grown to
OD
600
0.2.
10 Add 3 mL of molten soft agar to each, mix gently and pour over solidified agar Petri plate.
11 Incubate under appropriate temperature and atmospheric growth conditions for 12–18 h or until
plaques are visible on a confluent lawn of bacteria.
12 Repeat Steps 7–11 three times or until all PFUs exhibit the same observed plaque morphology.
Small-scale cultivation ●Timing 15–21 h
13 Using a Pasteur pipette, select 1 PFU and re-suspend in 100 µL of PBS in a microcentrifuge tube.
14 Warm 50 mL of growth medium in a sterile 250-mL GL45 screw-top flask with a GL45 0.22-µm
PTFE membrane vented cap.
15 Add 500 µL of bacteria grown to OD
600
0.2 and incubate at appropriate temperature and
atmospheric growth conditions for 20 min.
16 Add 50 µL of phage obtained in Step 13 and incubate under appropriate bacterial growth
conditions for 12–18 h.
17 To remove bulk bacterial debris, transfer lysates to 50-mL conical centrifuge tubes and centrifuge at
6,000gfor 30 min at 4 °C.
18 Filter sterilize the supernatant with a 0.2-µm syringe filter into a sterile 50-mL conical tube.
c
CRITICAL STEP Do not disturb the pellet during decanting.
19 Titer the sample as described in Step 20.
j
PAUSE POINT Store phage lysate at 4 °C for up to 12 months.
Phage titration
20 For routine phage titering, we describe two techniques (see the ‘Experimental design’section—‘Phage
titration’): follow ‘option A’foragaroverlaytiteringand‘option B’for spot plaque titering. Alternatively,
phages can be quantified via their genome copy number by qPCR (for details, see refs.
49,50
).
? TROUBLESHOOTING
PROTOCOL NATURE PROTOCOLS
2878 NATURE PROTOCOLS | VOL 15 | SEPTEMBER 2020 | 2867–2890 | www.nature.com/nprot

(A) Agar overlay titering ●Timing 20 h
(i) Add 100 μL of phage sample into a microcentrifuge tube containing 900 μL of PBS,
mix well.
(ii) Pipette 100 μL from the Step20A(i) dilution into a second microcentrifuge tube containing
900 μL of PBS and mix well.
(iii) Repeat Step20A(ii) for the remaining microcentrifuge tubes to create a dilution series of
10
−1
–10
−8
.
(iv) In sterile test tubes, mix 100 µL of each serial dilution with 100 µL of bacteria grown to
OD
600
0.2.
(v) Add 3 mL of molten soft agar, mix gently and pour the mixture over a solidified agar
Petri plate.
(vi) Incubate under appropriate temperature and atmospheric growth conditions for 12–18 h
or until plaques form on a confluent lawn of bacteria.
(vii) Determine phage stock concentration as PFUs·mL
−1
.
(B) Spot plaque titering ●Timing 19 h
(i) Dry a growth agar Petri plate for 30 min in a biosafety cabinet.
(ii) To seed a lawn of bacteria, pour 3 mL of bacteria grown to OD
600
0.2 onto the dry growth
agar Petri plate and quickly remove excess culture.
(iii) Dry the Petri plate in a biosafety cabinet for 15 min.
(iv) While drying, add 90 µL of PBS to eight wells row-by-row in a 96-well microtiter plate.
(v) Add 10 μL of phage sample to the first well and mix well.
(vi) Pipette 10 μL from the first well into the second well and mix well.
(vii) Repeat Step20B(vi) for the remaining wells to create a dilution series (10
−1
–10
−8
).
(viii) Using an eight-channel pipette, spot 4 µL from each well onto a dried seeded lawn of
bacteria prepared in Step20B(i–iii).
c
CRITICAL STEP Completely dry spots before moving the plate to the incubator.
(ix) Incubate the plate at appropriate temperature and atmospheric growth conditions for
12–18 h or until plaques form on a confluent lawn of bacteria.
(x) Determine phage stock concentration as PFUs·mL
−1
.
DNA extraction ●Timing 2h
c
CRITICAL Steps 21–56 describe how to extract genomic DNA from small-scale phage lysates
following a modified version of a previously published protocol
90
. There are many DNA extraction kits,
including Promega Wizard, Qiagen PowerViral Kit and Norgen Biotek. This protocol uses the
Ciculomics Nanobind Tissue Big DNA Kit to generate long DNA fragments suitable for the Oxford
Nanopore MinION sequencer.
21 Place the lysate from Step 18 into a clean 50-mL centrifuge tube. Add 0.5 μL of nuclease solution per mL
of lysate (10 μg·mL
−1
DNase and RNase final). Incubate the lysates at 37 °C for 30 min or at RT for 2 h.
22 Add half volume of the precipitant solution compared to the lysate (10% (wt/vol) PEG-8000, 1 M
NaCl final concentration). Mix gently by inversion. Incubate on ice for at least 60 min; precipitation
works best when incubated at 4 °C overnight. Most phages are stable in this state for several days.
23 Centrifuge the precipitated phage lysate at 10,000gat 4 °C for 10 min.
24 Carefully remove the supernatant either by aspiration or by carefully pouring into a separate tube
and retain the transparent or slightly opaque pellet in the original tube.
25 Gently resuspend the pellet in 0.5 mL of 5 mM MgSO
4
and briefly(5–10 s) centrifuge at 10,000gat
RT to pellet any remaining insoluble particles.
26 Transfer the supernatant to a sterile microcentrifuge tube for the remaining steps.
27 Before proceeding, dilute the supplied CW1 and CW2 buffers with 96–100% (vol/vol) ethanol, as
described by the manufacturer.
28 Add 20 μL of proteinase K solution.
29 Add 20 μL of the supplied CLE3 buffer.
30 Pulse vortex ten times for 1 s each.
31 Incubate in the ThermoMixer at 56 °C and 900 r.p.m. for 30 min.
32 Add 200 μL of the supplied BL3 buffer.
33 Pulse vortex ten times for 1 s each.
34 Incubate the microcentrifuge tube in the ThermoMixer at 56 °C and 900 r.p.m. for 30 min.
35 Add the Nanobind disks to the lysate.
NATURE PROTOCOLS PROTOCOL
NATURE PROTOCOLS | VOL 15 | SEPTEMBER 2020 | 2867–2890 | www.nature.com/nprot 2879

36 Add 300 μL of isopropanol.
37 Mix the microcentrifuge tube by inversion five times.
38 Mix on a rotator at 10 r.p.m. for 10 min.
39 Place the microcentrifuge tube in the Magnetic Tube Rack.
40 Carefully remove the supernatant, taking care not to disturb the Nanobind disks.
41 Add 700 μL of the supplied CW1 buffer.
42 Mix the microcentrifuge tube by vigorous inversion four times.
43 Place the microcentrifuge tube in the Magnetic Tube Rack.
44 Carefully remove the supernatant, taking care not to disturb the Nanobind disks.
45 Add 500 μL of the supplied CW2 buffer.
46 Repeat Steps 42–45.
47 Mix by vigorous inversion four times.
48 Place the microcentrifuge tube in the Magnetic Tube Rack.
49 Carefully remove the supernatant, taking care not to disturb the Nanobind disks.
50 Spin the microcentrifuge tube at 9,000 r.p.m. and RT for 5 s.
51 Remove any remaining liquid without disturbing the Nanobind disks.
52 Repeat Steps 50 and 51.
53 Add 100 μL of the supplied elution buffer and incubate at RT for 10 min.
54 Transfer the DNA eluate to a new, sterile microcentrifuge tube.
55 Spin the microcentrifuge tube containing the Nanobind disks at 9,000 r.p.m. and RT for 5 s.
Remove additional eluate and pipette into the same microcentrifuge tube in Step 54.
56 Leave at RT for ~1 h to let the DNA re-dissolve.
j
PAUSE POINT Phage DNA can be stored at 4 °C for <2 weeks and at −20 °C for <6 months.
Sequencing library preparation ●Timing 10 min
c
CRITICAL Steps 57–62 use Oxford Nanopore MinION for rapid DNA sequencing of phage genomes.
Thaw all components on ice and store on ice until needed. Centrifuge all components at 9,000 r.p.m. at
RT for 5 s before use.
57 In a 0.2-ml PCR tube, mix 7.5 μL of phage DNA (>400 ng) preparation from Step 56 and 2.5 μLof
fragmentation mix (solution FRA).
58 Mix by gently tapping the tube and then centrifuge at 9,000 r.p.m. at RT for 5 s to bring all the
contents to the bottom of the tube.
59 Incubate the tube at 30 °C for 1 min and then at 80 °C for 1 min.
60 Add 1 μL of rapid adapter (solution RAP).
61 Mix by gently tapping the tube and then centrifuge at 9,000 r.p.m. at RT for 5 s to bring all the
contents to the bottom of the tube.
62 Incubate the tube at RT for 5 min.
j
PAUSE POINT The sequencing library can be stored briefly on ice until loaded into the
DNA sequencer.
DNA sequencing ●Timing 1–48 h, depending on desired reads coverage, with first reads
recovered in 1 h. More reads will increase confidence.
63 Open MinKNOW software on the computer.
64 Plugin the MinION device and insert a flow cell (Fig. 2b).
65 Enter the sample ID and flow cell ID on the computer.
66 Run the Platform QC script and confirm that active pores are available.
67 Open the cover of the flow cell and draw back a few microliters of fluid to remove bubbles, taking
care not to remove the buffer from the pores.
68 To prepare the flow cell priming mix, add 30 µL of Flush Tether directly to the tube of mixed flush
buffer and mix by pipetting up and down. Centrifuge at 9,000 r.p.m. at RT for 5 s.
69 Load 800 µL of the priming mix into the flow cell via the priming port. Be careful to avoid
introducing air bubbles into the flow cell. Incubate at RT for 5 min.
70 Mix the sequencing reaction in a new tube:
34 µL sequencing buffer.
25.5 µL loading beads.
4.5 µL nuclease-free water.
11 µL DNA library from Step 62.
PROTOCOL NATURE PROTOCOLS
2880 NATURE PROTOCOLS | VOL 15 | SEPTEMBER 2020 | 2867–2890 | www.nature.com/nprot

71 Mix by pipetting and centrifuge at 9,000 r.p.m. at RT for 5 s.
72 Access the SpotON sample port by lifting the cover.
73 Load 200 µL of the priming mix into the flow cell via the priming port prepared in Step 68. Be
careful to avoid introducing air bubbles into the flow cell.
74 Carefully add the 75 µL of sequencing reaction from Step 71, in a dropwise fashion. Be careful to
ensure that each drop enters the flow cell.
75 Carefully replace the SpotON sample port cover, ensuring the port is filled by the bung, close the
priming port cover and close the MinION lid.
76 On the MinKNOW software and click Start Run.
77 Upon completion of the sequencing, flow cells can be washed and retained for further use. DNA
sequences will be available almost immediately and will continue to accumulate while there are
active pores.
Bioinformatics analysis ●Timing 12 h
78 Upload the fastq files from MinKNOW to a Linux server.
79 Assemble the long reads with canu
91
by running the following code:
a.canu -p phage -d assembly genomeSize=40k -nanopore-raw phage.fastq
The primary options are -p for the name, -d for the location to write the output and the fastq
output from MinKNOW.
80 Compare the phage genomes to antimicrobial and virulence gene databases using Abricate (https://
github.com/tseemann/abricate) by running the following code:
a.for DB in argannot card megares ncbi plasmidfinder resfinder vfdb; do
abricate --threads 10 --db $DB assembly/ done > abricate.out
81 Upload the genome to PHACTS to predict whether it is virulent or temperate virus
77
.
82 Upload the genome to PATRIC (https://patricbrc.org/) to annotate other genes in the genome.
83 Select phages for large-scale production based on the lack of predicted antibiotic and virulence
genes and the likelihood that the isolated phage is virulent (i.e., strictly lytic lifecycle).
Liter-scale shake flask cultivation ●Timing 18–24 h
c
CRITICAL Large phage production batches are limited by maximum centrifuge rotor capacity.
Steps 84–90 describe how to process 1-L cultures; thus, six flask cultures are performed in parallel.
84 In a sterile 2-L GL45 screw-top flask with a GL45 0.22-µm PTFE membrane vented cap, warm 1 L
of complex growth medium (Fig. 1).
85 Add 5 mL of bacteria grown to OD
600
0.2 and incubate under appropriate temperature and
atmospheric growth conditions for 20 min.
86 Add phage from Step 18 at an MOI of 0.1 and incubate under appropriate bacterial growth
conditions for 12–18 h.
87 To remove bulk bacterial debris, transfer lysates to 1-L centrifuge bottles and centrifuge at 8,000g
for 45 min at 4 °C (Fig. 1).
88 Decant the supernatants into fresh, sterile 1-L centrifuge bottles without disturbing the
bacterial pellet.
89 Centrifuge again at 8,000gfor 45 min at 4 °C.
90 Decant the supernatant into a sterile 1-L glass bottle.
Dead-end filtration ●Timing 1h
91 Assemble 0.8/0.45-µm and 0.45/0.2-µm capsule filters inline as shown in Fig. 1. Use the peristaltic
pump to filter sterilized supernatant into sterile glass bottles.
c
CRITICAL STEP Capsule filters can process up to 2 L before clogging but can be cleaned and
reused following the manufacturer’s protocol.
? TROUBLESHOOTING
NATURE PROTOCOLS PROTOCOL
NATURE PROTOCOLS | VOL 15 | SEPTEMBER 2020 | 2867–2890 | www.nature.com/nprot 2881

92 Titer the sample as described in Step 20.
c
CRITICAL STEP The phage lysate titer should be >10
9
PFUs·mL
−1
.
j
PAUSE POINT Phage preparation can be stored at 4 °C for <6 months.
? TROUBLESHOOTING
Ultrafiltration, diafiltration and concentration ●Timing 8 h (2 h per 1.5 L)
93 Assemble the peristaltic pump and CFF cassette as shown in Fig. 1.
94 Circulate 500 mL of sterile ddH
2
O through the cassette and discard.
95 Place the intake and retentate hoses into the 0.2 µm sterile phage supernatant solution from Step 92.
96 Place the filtrate hose into a sink or waste container.
97 Recirculate (up to 1.5 L per batch) the supernatant until ~200 mL remains.
98 Add 400 mL of sterile ddH
2
O.
99 Recirculate the supernatant until ~200 mL remains.
100 Add 400 mL of sterile TN buffer.
101 Recirculate the supernatant until ~200 mL remains.
102 Repeat Steps 100 and 101 until the supernatant becomes clear and colorless.
103 Pause pump, place intake and retentate hoses in a sterile 50-mL conical tube and recirculate while
continuously adding the remaining supernatant until 40 mL of concentrate remains (first fraction).
104 Pause pump and place intake hose with retentate hose into a sterile 50-mL conical tube with 30 mL
of TN buffer. Circulate briefly and remove intake hose. Collect 30 mL of concentrate (second
fraction).
c
CRITICAL STEP Subsequent fractions can be collected if lower-concentration phage stocks are
desired.
105 Titer the sample as described in Step 20. The phage concentrate should be greater than or equal to
tenfold higher in titer than that in Step 92.
j
PAUSE POINT Phage preparation can be stored at 4 °C for <2 weeks.
(Optional) Density gradient ultrapurification ●Timing 6h
c
CRITICAL Density gradient ultracentrifugation (Steps 106–111) and dialysis (Steps 112–120) might
not be required to meet regulatory endotoxin safety limits of phage products
41
. Alternatively, directly
proceed to Step 135 for endotoxin quantification.
106 In an open-top ultraclear (14 × 89 mm) round-bottom tube, prepare a CsCl step density gradient
by layering 2 ml of d=1.6, 3 ml of d=1.5 and 3 ml of d=1.3 from the bottom up.
c
CRITICAL STEP Avoid disturbing the previous density layer during preparation.
107 Fill the remaining tube volume with phage concentrate (~4 ml) from Step 104.
108 Place tubes inside buckets and balance on a 4 °C cooled SW41 rotor.
109 Ultracentrifuge at 28,000gfor 4 h at 4 °C.
110 Carefully extract tubes from buckets using tweezers.
111 With a concentrated light source, extract the visible band containing the phage particles by
puncturing the thin-walled ultraclear tube with a 26-gauge needle and syringe (Fig. 1).
? TROUBLESHOOTING
(Optional) CsCl concentrate dialysis ●Timing 1d
112 Chill up to 3 L of sterile ddH
2
O (4 °C) in a large beaker placed on a stir plate.
113 Place a magnetic stir bar in the beaker.
114 Clamp one end of the 100-kDa MWCO dialysis tubing and add the CsCl phage concentrate from
Step 111.
115 Clamp the other end and secure the clamped dialysis tubing to a float.
116 Dialyze CsCl phage concentrate in the pre-chilled sterile ddH
2
O and stir for 30 min at 4 °C.
117 Exchange ddH
2
O with up to 3 L of pre-chilled sterile storage buffer (e.g., PBS) and dialyze for 1 h.
118 Repeat Step 117 but prolong dialysis to 24 h.
119 Recover the phage concentrate.
120 Titer the sample as described in Step 20. The phage titer should be between 10
9
and 10
12
PFUs·mL
−1
.
j
PAUSE POINT Phage preparation can be stored at 4 °C for <2 weeks.
? TROUBLESHOOTING
PROTOCOL NATURE PROTOCOLS
2882 NATURE PROTOCOLS | VOL 15 | SEPTEMBER 2020 | 2867–2890 | www.nature.com/nprot

LPS-affinity chromatography ●Timing 2h
c
CRITICAL Steps 121–134 are specific to the Pierce High Capacity Endotoxin Removal Spin Column.
121 Equilibrate and regenerate the spin column as per the manufacturer’s instructions.
122 Place the spin column into a collection tube and centrifuge at 500gfor 1 min at RT to discard the
regeneration solution.
123 Remove the cap and insert the bottom plug. Add 8 mL of 2 M NaCl, replace the cap and invert the
column several times.
124 Loosen the cap and remove the bottom plug. Place the column in a collection tube and centrifuge at
500gfor 1 min at RT to discard the solution.
125 Remove the cap and insert the bottom plug. Add 8 mL of supplied endotoxin-free water. Replace
the cap and invert the column several times.
126 Loosen the cap and remove the bottom plug. Place the column in a collection tube and centrifuge at
500gfor 1 min at RT to discard the water.
127 Remove the cap and insert the bottom plug. Add 8 mL of endotoxin-free PBS, replace the cap and
invert the column several times.
128 Loosen the cap and remove the bottom plug. Place the column in a collection tube and centrifuge at
500gfor 1 min at RT to discard the PBS.
129 Repeat Steps 127 and 128 two additional times.
130 Remove the cap and insert the bottom plug. Add 10 mL of dialyzed phage concentrate from Step
119 to the resin, replace the cap and invert the column several times.
131 Incubate the column with gentle end-over-end mixing at 4 °C for 45 min.
132 Loosen the cap and remove the bottom plug. Place the column in a sterile collection tube and
centrifuge at 500gfor 1 min at RT to collect the sample.
133 Repeat Steps 121–132 until all phage concentrate is processed.
134 Titer the sample as described in Step 20. The phage titer should be between 10
9
and 10
12
PFUs·mL
−1
.
j
PAUSE POINT Phage preparation can be stored at 4 °C for <6 months.
? TROUBLESHOOTING
Phage preparation endotoxin quantification ●Timing 2h
c
CRITICAL Steps 135–143 are specific to the Pierce LAL Chromogenic Endotoxin Quantitation Kit.
135 Equilibrate solutions to RT, as per the manufacturer recommendation.
136 Prepare LPS standards provided in the kit using the ‘High Standards’option.
137 Add prepared standards, the blank and samples from Step 133 to a 96-well microtiter plate.
c
CRITICAL STEP Appropriately dilute phage samples to be within the linear range of the
High Standards.
138 Warm the plate to 37 °C and add 50 µL of LAL to each well. Tap the plate lightly ten times to mix.
139 Reconstitute the chromogenic substrate solution and warm it at 37 °C for 5 min before use.
c
CRITICAL STEP Work quickly to prevent inactivation of solutions after reconstitution.
140 Add 100 µL of the chromogenic solution to each well.
141 At 6 min, add 50 µL of 25% (vol/vol) acetic acid to stop the reaction.
142 Immediately measure OD
405nm
in a microplate reader.
143 Extrapolate the endotoxin level from the standard curve. Refer to Supplementary Table 1 for
anticipated endotoxin concentration.
? TROUBLESHOOTING
Phage preparation protein analysis ●Timing 4h
144 Determine the absorbance of phage preparations from Step 133 at 280 nm to measure protein level.
This will vary between phage strains but should be between 1 and 3 mg·mL
−1
.
145 Dilute phages to 15 µg in 20 µL of 1× Laemmli sample buffer.
146 Incubate samples at 90 °C for 5 min.
147 Centrifuge samples at 13,000gfor 60 s at RT.
148 Prepare a 10% (wt/vol) acrylamide gel for SDS–PAGE electrophoresis.
149 Mount the gel into the tank, remove combs and completely fill the inner chamber of the tank and
three-fourths of the outer chamber with 1× SDS–PAGE running buffer.
150 Pipette 3 µl of the standard and 20 µL of the sample into the subsequent wells.
151 Run electrophoresis at 100 V for ~60 min.
NATURE PROTOCOLS PROTOCOL
NATURE PROTOCOLS | VOL 15 | SEPTEMBER 2020 | 2867–2890 | www.nature.com/nprot 2883

152 Wash the gel with ddH
2
O three times for 15 min each.
153 Incubate the gel with 50 mL of Coomassie blue staining solution at RT for 1 h.
154 Decant the solution and add 50 mL of ddH
2
O. Incubate at RT for 30 min on a rocker.
155 Decant and repeat Step 154 three times.
156 Image the gel using a gel dock station or conventional camera.
Phage preparation effects on human cell viability ●Timing 3d
c
CRITICAL In Steps 157–167, we describe how to use the CellTiter-Glo luminescent cell viability assay
(Promega) to determine the number of viable cells in culture after phage preparation exposure based on
quantitation of the ATP present, an indicator of cell proliferation and cytotoxicity.
157 In a 96-well tissue culture plate, seed wells with 100 µL of media containing ~20,000 HeLa or
HEK293 cells.
c
CRITICAL STEP Use a clear-bottom, white-walled microplate to minimize well luminescence
cross-talk.
158 Incubate at 37 °C in 5% CO
2
for 24 h.
159 The next day, check for a confluent monolayer of cell culture.
160 From Step 133, dilute phage stock to 10
8
PFUs·mL
−1
and 10
9
PFUs·mL
−1
.
161 Add 10 µL of 10
8
PFUs·mL
−1
stock for an approximate concentration of 1 cell:100 phages and
10 µL of the 10
9
PFUs·mL
−1
stock for a concentration of 1:1,000 phages.
162 For positive control, add 10 µL of 1% (wt/vol) SDS; for negative control, add 10 µL of PBS.
c
CRITICAL STEP Ensure that two wells are filled with medium only.
163 Incubate at 37 °C in 5% CO
2
for 24 h.
164 Equilibrate the 96-well plate to RT for 30 min before adding 100 µL of CellTiter-Glo to all wells.
165 Mix the plate on an orbital shaker for 2 min to induce cell lysis.
166 Incubate at 22 °C for 10 min.
167 Measure luminescence on a microplate reader (see ‘Anticipated results’below for example data).
Troubleshooting
Troubleshooting advice can be found in Table 2.
Table 2 | Troubleshooting table
Step Problem Possible reason Solution
7 No visible clear zones (i.e., plaques) Bacterial cells poorly lysed Select alternative bacterial host strain
Low viral count in
environmental sample
Concentrate environmental sample by, for example,
CFF (Steps 93–105) or PEG/NaCl precipitation
9,32
20 Not quantifiable owing to the
presence of pinpoint plaques
Actual plaque morphology
on target bacterial
host strain
Use a lighted magnifying glass over a dark backdrop
to count plaques or switch to alternative bacterial
host strain
95
91 Filter clogging Large amounts of cell debris
in supernatant after
centrifugation
Centrifuge phage lysate a third time (Steps 88 and
89) or process 1 L of lysate per batch
92 Phage titer <10
9
PFUs·mL
−1
Poor bacteria growth Optimize bacterial growth conditions and/or reduce
phage MOI for Step 86
Poor phage replication or
low phage burst size
Produce a larger volume of phage lysate (set up
more flasks at Step 84) before proceeding to Step 71
111 Fuzzy band observed at top of
gradient
Gross bacterial debris
present in phage concentrate
0.2-µm syringe filter the CFF fraction before
proceeding to Step 106
Multiple bands observed Unwanted phage
contaminant present in the
phage concentrate
Repeat Steps 106–109 but with 7 mL of d=1.5 in
Step 106 and ultracentrifugation at 38,000gfor 18 h
at 4 °C in Step 109
120 Phage titer drops by >1 log Phages might be damaged
by osmotic shock
Use a 50/50 mix of water and 0.5 M PBS for the
first dialysis in Step 112
134 Phage titer drops by >1 log Phages bound to
column resin
Increase NaCl concentration to 40 mM in the
sample and equilibration buffer before adding
phages to the column in Step 130
143 Endotoxin signal beyond
recommended standard range
Endotoxin >1 EUs·mL
−1
Dilute phage sample 10- to 1,000-fold in sterile PBS
before adding to the microtiter plate in Step 137
PROTOCOL NATURE PROTOCOLS
2884 NATURE PROTOCOLS | VOL 15 | SEPTEMBER 2020 | 2867–2890 | www.nature.com/nprot

Timing
The timing information is an estimate for phages for aerobic Gram-negative bacteria.
Steps 1–12, phage plaque isolation: 3–5 d, depending on repetitions needed to obtain a single plaque
morphology
Steps 13–19, small-scale cultivation: 15–21 h
Step 20A, phage titration using agar overlay titering: 20 h
Step 20B, phage titration using spot plaque titering: 19 h
Steps 21–56, DNA extraction: 2 h
Steps 57–62, sequencing library preparation: 10 min
Steps 63–77, DNA sequencing: 1–48 h, depending on desired reads coverage
Steps 78–83, bioinformatics analysis: 12 h
Steps 84–90, large-scale cultivation: 24–30 h
Steps 91 and 92, dead-end filtration: 1h
Steps 93–105, ultrafiltration, diafiltration and concentration: 8 h (2 h per 1.5 L)
Steps 106–111, density gradient ultrapurification: 4 h
Steps 112–120, CsCl concentrate dialysis: 24 h
Steps 121–134, LPS-affinity chromatography: 2 h
Steps 135–143, phage preparation endotoxin quantification: 2 h
Steps 144–156, phage preparation protein analysis: 4 h
Steps 157–167, phage preparation effects on human cell viability: 3 d
Anticipated results
Process optimization at liter-scale production
This protocol for purifying phages employs a combination of modified classic techniques, modern
membrane filtration processes and omission of certain common practices (for an overview, see
Fig. 1). A typical 6-L shake flask cultivation yields between 16 and 30 ml of final product, providing
up to 64,000 treatment doses at 10
9
PFUs—a commonly prescribed i.v. dose
1
—depending on the
phage–bacteria pair (Table 1and Supplementary Table 1). Because final phage products contain very
low endotoxin levels, theoretically it would be possible that treatments of up to 10
12
PFUs could be
prescribed to humans and be within regulatory limits of 5 EUs·kg
−1
h
−1
(ref.
41
). In comparison,
Belgium is the only Western country that routinely produces phages in a laboratory under pre-
scription, as a ‘magistral’preparation (i.e., drug compounding)
40
. To our knowledge, magistral phage
products are generally purified by high-speed lysate centrifugation and subsequent affinity chro-
matography endotoxin removal
40
. In practice, this approach yields ~10
7
PFUs·mL
−1
and
12.5 EUs·mL
−1
(ref.
39
). A recent extended-access phage therapy reported that PEG/NaCl, density
gradient ultracentrifugation and dialysis purification of Mycobacterium phages yielded a similarly
high 10
11
PFUs·mL
−1
and undetectable endotoxins
2
.Mycobacteria, however, do not produce LPS.
PEG-CsCl dialysis can effectively reduce LPS to <0.05 EUs·mL
−1
experimentally; it does so by adding
organic solvents and generally at the expense of phage yield
21
.
In Fig. 3, we show that a relatively high phage titer is maintained during downstream processes,
while endotoxins are being reduced to a safe level for human i.v. use. In addition, we show that
high titer and low endotoxin phage preparation are independent of phage strain and viral structure.
For example, myophages PAK_P1 and PAK_P5 were isolated from French wastewater with the
P. aeruginosa strain PAK
92
, whereas myophage E217 and podophage PYO2 were isolated from Italian
wastewater with the P. aeruginosa strain PAO1 (ref.
24
). Final production runs produced titers of
6×10
10
,4×10
11
,2×10
11
, and 4 × 10
12
PFUs·mL
−1
, respectively (Table 1). High-titer final stocks are
dependent on the initial phage cultivation performance. Six-liter lysates of Pseudomonas phages
produced titers of 10
10
PFUs·mL
−1
that could be concentrated up to 10
12
PFUs·mL
−1
, whereas lysates
of K. oxytoca and S. marcescens phages produced lower titers of 10
9
PFUs·mL
−1
that could be
concentrated to 10
10
PFUs·mL
−1
(Fig. 3, Supplementary Table 1). Because CFF was found to be the
main phage-concentrating technique, a larger volume of lysate from phages that do not produce high
titers would be required to achieve higher final product concentrations.
Other practiced phage cultivation and purification approaches, such as PEG precipitation, cen-
trifugal ultrafiltration, organic solvent extraction, enzymatic inactivation and anion-exchange chro-
matography were previously shown to produce equally high phage concentrations but can retain high
amounts of endotoxins
11,18,27–34,45
. Our protocol focuses on early removal of endotoxins by
NATURE PROTOCOLS PROTOCOL
NATURE PROTOCOLS | VOL 15 | SEPTEMBER 2020 | 2867–2890 | www.nature.com/nprot 2885

conducting multiple low-speed centrifugation steps, microfiltration and cross-flow ultrafiltration with
phage particle washing steps, which reduce most LPS in phage samples (Fig. 3). By contrast, we found
that subsequent CsCl density gradient ultracentrifugation increases, in some preparations, LPS
quantity by up to 30% (Supplementary Table 1). Our results conflict with the recent study by
Van Belleghem et al. that compared several endotoxin removal methods and concluded that CsCl
density gradient ultracentrifugation was the most effective at removing endotoxins
34
. The study also
showed that effective endotoxin removal came at the expense of a phage yield reduction. In this study,
we observed that the phage titer generally increased after CsCl density gradient ultracentrifugation
and dialysis, except for phage JG266 (Fig. 3).
Our results suggest that CFF with an MWCO 100-kDa pore size, in combination with several
buffer washes, produces a phage product with endotoxin quantities within FDA regulatory limits for
human intravenous use. Notably, most exotoxins are smaller than 100 kDa (ref.
42
). The semi-
automated CFF used in our approach is able to concentrate phages between 10- and 100-fold, while
removing endotoxins by >4,000-fold and rich bacterial growth medium in a single step (Fig. 3). CFF is
an effective phage concentration technique
32,33
. However, using lower MWCO pore sizes appears to
trap endotoxins along with the phage particles
32
. The recent ‘Phage On Tap’purification protocol
showed that centrifugal ultrafiltration was a more rapid technique to concentrate phage particles but
also trapped endotoxins
33
. Nevertheless, we show that endotoxins in phage samples can be further
reduced by commercially available LPS-affinity chromatography (Fig. 3and Supplementary Table 1).
Phage product safety testing
Throughout the phage cultivation and purification process, we maintain phage strain homogeneity by
monitoring for changes in plaque morphology while phage titering. After each processing step, phage
stocks should maintain a distinct single plaque phenotype. Homogeneity is further confirmed by
obtaining a distinct single band after density gradient, as would be expected with a sample containing
phage particles with the same density and shape
83
.
As mentioned above, at a therapeutic dose of 10
9
PFUs, endotoxins in phage lysates are reduced at
least 10
6
-fold for all seven phages tested, without the use of organic solvents (Fig. 3and Table 1).
Simultaneously, <100 kDa of microsolute permeate is withdrawn from phage preparations during
CFF, suggesting that small molecules, such as exotoxins that are typically smaller than 30 kDa
a
kDa
JG265 lysate
PAK_P1
PAK_P5
E217
PYO2
150
75
50
37
25
Marker
JG265
JG266
SM219
b
100
PAK_P1
PAK_P5
PYO2
E217
HeLa viability (%)
50
0Lysate 1:100 1:1,000
c
100
50
0
HEK293 viability (%)
Lysate 1:100 1:1,000
PAK_P1
PAK_P5
PYO2
E217
d
100
JG265
JG266
HeLa viability (%)
50
0
Lysate
1:100
1:1,000
e
100
50
0
Lysate
1:100
1:1,000
Lysate
1:100
1:1,000
Lysate
1:100
1:1,000
HEK293 viability (%)
JG265
JG266
f
100
HeLa viability (%)
50
0
g
100
HEK293 viability (%)
50
0
Fig. 4 | Purity and safety analyses of final phage preparations. a, SDS–PAGE analysis of protein content in final phage samples (Step 133) at 10
9
PFUs
compared to an exemplified sterilized phage lysate (produced by Steps 84–92) and Precision Plus Protein ladder (Bio-Rad). b–g, Human HeLa cell line
(b,d,f) and HEK293 cell line (c,e,g) viability after 24 h of co-incubation with final preparations of P. aeruginosa phages (b,c), K. oxytoca phages (d,e) and
S. marcescens phages (f,g) at cell:phages ratios of 1:100 and 1:1,000 quantified by the CellTiter-Glo assay (Steps 157–167). Viability was normalized to
untreated cells. Pseudomonas and Serratia sterilized phage lysates were significantly lower than corresponding final phage preparations (P< 0.006),
whereas there was no significant difference in cell viability between Klebsiella sterilized lysate and phage treated. One-way ANOVA; error bars
represent s.e.m.; n=3–6).
PROTOCOL NATURE PROTOCOLS
2886 NATURE PROTOCOLS | VOL 15 | SEPTEMBER 2020 | 2867–2890 | www.nature.com/nprot

(ref.
42
), are also removed from final phage products. However, none of our phages or bacterial strains
contained a known exotoxin to validate this claim.
Figure 4a shows that our example phage preparations are free of gross bacterial proteins, with
SDS–PAGE staining not showing significant protein smearing. This becomes especially clear when
comparing the 0.2-µm filtered lysate of Klebsiella phage JG265 with its purified preparation (Fig. 4a).
The lysate displays many bands, whereas the purified phage preparation displays a few distinct bands
unique for each denatured phage, indicating differences in their structural proteins. PAK_P1 and
PAK_P5 share a major protein band at ~50 kDa and E217, PYO2, JG265 and JG266 at 75 kDa.
Between phage strains, however, other proteins are not conserved with a variety of bands ranging
from 20 to 150 kDa. To identify these specific phage proteins, further mass spectrometry analysis
would need to be conducted
93
.
Because SDS–PAGE is not sufficient to guarantee absence of harmful toxins, we further conducted
cell viability testing
94
. In vitro models to test for phage preparation effects on human cells and exclude
potentially harmful products is an important consideration, but there is no standard model of phage
pharmaceutical standard available. As a rapid test for eIND request, quantitation of intracellular ATP
can be used as a measurement of the metabolic activity of human cells. With two human cell lines,
HeLa and HEK293, we show that cell viability is not significantly different from untreated cell
controls after exposure to each of the phage strains purified after cultivation with Pseudomonas,
Klebsiella or Serratia (Fig. 4b–g). By contrast, centrifuged phage lysate causes a significant decrease in
cell metabolic activity. Therefore, our results suggest that downstream purification of phage lysates
can produce a product free of harmful bacterial toxins and other components.
Together, these results suggest that our protocol for phage purification is reliable and reproducible,
helping to ensure the safety and efficacy of phage products for human use.
Reporting Summary
Further information on research design is available in the Nature Research Reporting Summary
linked to this article.
Data availability
The data that support the findings of this study are available from the corresponding author upon
reasonable request.
References
1. Schooley, R. T. et al. Development and use of personalized bacteriophage-based therapeutic cocktails to treat
a patient with a disseminated resistant Acinetobacter baumannii infection. Antimicrob. Agents Chemother.
61, e00954–17 (2017).
2. Dedrick, R. M. et al. Engineered bacteriophages for treatment of a patient with a disseminated drug-resistant
Mycobacterium abscessus.Nat. Med. 25, 730–733 (2019).
3. Onsea, J. et al. Bacteriophage application for difficult-to-treat musculoskeletal infections: development of a
standardized multidisciplinary treatment protocol. Viruses 11, 891 (2019).
4. Nir-Paz, R. et al. Successful treatment of antibiotic resistant poly-microbial bone infection with bacter-
iophages and antibiotics combination. Clin. Infect. Dis.69, 2015–2018 (2019).
5. Van Norman, G. A. Expanding patient access to investigational drugs: single patient investigational new drug
and the “right to try”.JACC Basic Transl. Sci. 3, 280–293 (2018).
6. Svircev, A., Roach, D. & Castle, A. Framing the future with bacteriophages in agriculture. Viruses 10,
218 (2018).
7. Segall, A. M., Roach, D. R. & Strathdee, S. A. Stronger together? Perspectives on phage–antibiotic synergy in
clinical applications of phage therapy. Curr. Opin. Microbiol. 51,46–50 (2019).
8. Moye, Z. D., Woolston, J. & Sulakvelidze, A. Bacteriophage applications for food production and processing.
Viruses 10, 205 (2018).
9. Lehman, S. M. et al. Design and preclinical development of a phage product for the treatment of antibiotic-
resistant Staphylococcus aureus infections. Viruses 11, 88 (2019).
10. Pirnay, J.-P. et al. Quality and safety requirements for sustainable phage therapy products. Pharm. Res. 32,
2173–2179 (2015).
11. Mutti, M. & Corsini, L. Robust approaches for the production of active ingredient and drug product for
human phage therapy. Front. Microbiol.10, 2289 (2019).
12. Kutateladze, M. & Adamia, R. Phage therapy experience at the Eliava Institute. Med. Mal. Infect. 38,
426–430 (2008).
13. Letkiewicz, S. et al. The perspectives of the application of phage therapy in chronic bacterial prostatitis. FEMS
Immunol. Med. Microbiol. 60,99–112 (2010).
NATURE PROTOCOLS PROTOCOL
NATURE PROTOCOLS | VOL 15 | SEPTEMBER 2020 | 2867–2890 | www.nature.com/nprot 2887

14. Jault, P. et al. Efficacy and tolerability of a cocktail of bacteriophages to treat burn wounds infected by
Pseudomonas aeruginosa (PhagoBurn): a randomised, controlled, double-blind phase 1/2 trial. Lancet Infect.
Dis. 19,35–45 (2019).
15. US Department of Health and Human Services (Antibiotic Resistance Coordination and Strategy Unit within
the Division of Healthcare Quality Promotion, Centers for Disease Control and Prevention, 2019).
16. O’Neill, J. Tackling Drug-Resistant Infections Globally: Final Report and Recommendations (Department of
Health, United Kingdom, 2016).
17. d’Hérelle, F. L'étude d’une maladie: le choléra, maladie à paradoxes (Rouge, 1946).
18. Adams, M. Enumeration of bacteriophage particles. In: Bacteriophages (ed Adams MH) 27–34 (Interscience
Publishers, 1959).
19. Roach, D. R. & Debarbieux, L. Phage therapy: awakening a sleeping giant. Emerg. Top. Life Sci. 1,93–103 (2017).
20. Dalpke, A., Frank, J., Peter, M. & Heeg, K. Activation of toll-like receptor 9 by DNA from different bacterial
species. Infect. Immun. 74, 940–946 (2006).
21. Sweere, J. M. et al. Bacteriophage trigger antiviral immunity and prevent clearance of bacterial infection.
Science 363, eaat9691 (2019).
22. Opal, S. M. Endotoxins and other sepsis triggers. Contrib. Nephrol. 167,14–24 (2010).
23. Capparelli, R., Parlato, M., Borriello, G., Salvatore, P. & Iannelli, D. Experimental phage therapy against
Staphylococcus aureus in mice. Antimicrob. Agents Chemother. 51, 2765 (2007).
24. Forti, F. et al. Design of a broad-range bacteriophage cocktail that reduces Pseudomonas aeruginosa biofilms
and treats acute infections in two animal models. Antimicrob. Agents Chemother. 62, e02573–17 (2018).
25. Roach, D. R. et al. Synergy between the host immune system and bacteriophage is essential for successful
phage therapy against an acute respiratory pathogen. Cell Host Microbe 22,38–47 (2017).
26. Chibani-Chennoufi, S. et al. In vitro and in vivo bacteriolytic activities of Escherichia coli phages: implications
for phage therapy. Antimicrob. Agents Chemother. 48, 2558 (2004).
27. Yamamoto, K. R., Alberts, B. M., Benzinger, R., Lawhorne, L. & Treiber, G. Rapid bacteriophage sedi-
mentation in the presence of polyethylene glycol and its application to large-scale virus purification. Virology
40, 734–744 (1970).
28. Bachrach, U. & Friedmann, A. Practical procedures for the purification of bacterial viruses. Appl. Microbiol.
22, 706–715 (1971).
29. Ackermann, H. W. et al. Guidelines for bacteriophage characterization. Adv. Virus Res. 23,1–24 (1978).
30. Boratynski, J. et al. Preparation of endotoxin-free bacteriophages. Cell Mol. Biol. Lett. 9, 253–259 (2004).
31. Gill, J. & Hyman, P. Phage choice, isolation, and preparation for phage therapy. Curr. Pharm. Biotechnol. 11,
2–14 (2010).
32. Bourdin, G. et al. Amplification and purification of T4-like Escherichia coli phages for phage therapy: from
laboratory to pilot scale. Appl. Environ. Microbiol. 80, 1469–1476 (2014).
33. Bonilla, N. et al. Phage on tap-a quick and efficient protocol for the preparation of bacteriophage laboratory
stocks. PeerJ 4, e2261 (2016).
34. Van Belleghem, J. D., Merabishvili, M., Vergauwen, B., Lavigne, R. & Vaneechoutte, M. A comparative study
of different strategies for removal of endotoxins from bacteriophage preparations. J. Microbiol. Methods 132,
153–159 (2017).
35. Van Belleghem, J. D., Dabrowska, K., Vaneechoutte, M., Barr, J. J. & Bollyky, P. L. Interactions between
bacteriophage, bacteria, and the mammalian immune system. Viruses 11, 10 (2018).
36. Gogokhia, L. et al. Expansion of bacteriophages is linked to aggravated intestinal inflammation and colitis.
Cell Host Microbe 25, 285–299 (2019).
37. An, T. W., Kim, S. J., Lee, Y. D., Park, J. H. & Chang, H. I. The immune-enhancing effect of the Cronobacter
sakazakii ES2 phage results in the activation of nuclear factor-kappaB and dendritic cell maturation via the
activation of IL-12p40 in the mouse bone marrow. Immunol. Lett. 157,1–8 (2014).
38. Van Belleghem, J. D., Clement, F., Merabishvili, M., Lavigne, R. & Vaneechoutte, M. Pro- and anti-
inflammatory responses of peripheral blood mononuclear cells induced by Staphylococcus aureus and
Pseudomonas aeruginosa phages. Sci. Rep. 7, 8004 (2017).
39. Chan, B. K. et al. Phage treatment of an aortic graft infected with Pseudomonas aeruginosa.Evol. Med. Public
Health 2018,60–66 (2018).
40. Pirnay, J. P. et al. The magistral phage. Viruses 10, 64 (2018).
41. US Department of Health and Human Services Food and Drug Administration (Office of Communications,
Division of Drug Information (US Food and Drug Administration, 2012).
42. Spaulding, A. R. et al. Staphylococcal and streptococcal superantigen exotoxins. Clin. Microbiol. Rev. 26,
422–447 (2013).
43. Carlson, K. Working with Bacteriophages: Common Techniques and Methodological Approaches, Vol 1 (CRC
Press, 2005).
44. Davidson, I. W., Sumner, D. D. & Parker, J. C. Chloroform: a review of its metabolism, teratogenic,
mutagenic, and carcinogenic potential. Drug Chem. Toxicol. 5,1–87 (1982).
45. Szermer-Olearnik, B. & Boratyński, J. Removal of endotoxins from bacteriophage preparations by extraction
with organic solvents. PLoS ONE 10, e0122672 (2015).
46. Melnikov, P. & Zanoni, L. Z. Clinical effects of cesium intake. Biol. Trace Elem. Res. 135,1–9 (2010).
47. Centeno, J. A. et al. Blood and tissue concentration of cesium after exposure to cesium chloride: a report of
two cases. Biol. Trace Elem. Res. 94,97–104 (2003).
PROTOCOL NATURE PROTOCOLS
2888 NATURE PROTOCOLS | VOL 15 | SEPTEMBER 2020 | 2867–2890 | www.nature.com/nprot

48. Mancuso, F., Shi, J. & Malik, D. J. High throughput manufacturing of bacteriophages using continuous stirred
tank bioreactors connected in series to ensure optimum host bacteria physiology for phage production.
Viruses 10, 537 (2018).
49. Edelman, D. C. & Barletta, J. Real-time PCR provides improved detection and titer determination of bac-
teriophage. Biotechniques 35, 368–375 (2003).
50. Anderson, B. et al. Enumeration of bacteriophage particles: comparative analysis of the traditional plaque
assay and real-time QPCR- and nanosight-based assays. Bacteriophage 1,86–93 (2011).
51. Freeman, V. J. Studies on the virulence of bacteriophage-infected strains of Corynebacterium diphtheriae.
J. Bacteriol. 61, 675–688 (1951).
52. Waldor, M. K. & Mekalanos, J. J. Lysogenic conversion by a filamentous phage encoding cholera toxin.
Science 272, 1910–1914 (1996).
53. Hatano, M., Nakamura, K. & Kurokawa, M. Isolation of a new temperature phage causing the lysogenic
conversion in Corynebacterium diphtheriae.Jpn. J. Microbiol. 3, 301–311 (1959).
54. Weeks, C. R. & Ferretti, J. J. The gene for type-A streptococcal exotoxin (erythrogenic toxin) is located in
bacteriophage-T12. Infect. Immun. 46, 531–536 (1984).
55. Schmidt, H. Shiga-toxin-converting bacteriophages. Res. Microbiol. 152, 687–695 (2001).
56. Plunkett, G., Rose, D. J., Durfee, T. J. & Blattner, F. R. Sequence of shiga toxin 2 phage 933W from
Escherichia coli O157:H7: shiga toxin as a phage late-gene product. J. Bacteriol. 181, 1767–1778 (1999).
57. Cobian Guemes, A. G. et al. Cystic fibrosis rapid response: translating multi-omics data into clinically
relevant information. mBio 10, e00431-19 (2019).
58. Eklund, M. W., Poysky, F. T., Reed, S. M. & Smith, C. A. Bacteriophage and the toxigenicity of Clostridium
botulinum type C. Science 172, 480–482 (1971).
59. Brussow, H., Canchaya, C. & Hardt, W. D. Phages and the evolution of bacterial pathogens: from genomic
rearrangements to lysogenic conversion. Microbiol. Mol. Biol. Rev. 68, 560–602 (2004).
60. Wagner, P. L. & Waldor, M. K. Bacteriophage control of bacterial virulence. Infect. Immun. 70,
3985–3993 (2002).
61. Brown-Jaque, M. et al. Antibiotic resistance genes in phage particles isolated from human faeces and induced
from clinical bacterial isolates. Int. J. Antimicrob. Agents 51, 434–442 (2018).
62. Purdy, A., Rohwer, F., Edwards, R., Azam, F. & Bartlett, D. H. A glimpse into the expanded genome content
of Vibrio cholerae through identification of genes present in environmental strains. J. Bacteriol. 187,
2992–3001 (2005).
63. Enault, F. et al. Phages rarely encode antibiotic resistance genes: a cautionary tale for virome analyses. ISME J.
11, 237–247 (2017).
64. Billard-Pomares, T. et al. Characterization of a P1-like bacteriophage carrying an SHV-2 extended-spectrum
beta-lactamase from an Escherichia coli strain. Antimicrob. Agents Chemother. 58, 6550–6557 (2014).
65. Zhou, W., Liu, L., Feng, Y. & Zong, Z. A P7 phage-like plasmid carrying mcr-1 in an ST15 Klebsiella
pneumoniae clinical isolate. Front. Microbiol. 9, 11 (2018).
66. Aziz, R. K. et al. Mosaic prophages with horizontally acquired genes account for the emergence and
diversification of the globally disseminated M1T1 clone of Streptococcus pyogenes.J. Bacteriol. 187,
3311–3318 (2005).
67. Lacey, J. A., Johanesen, P. A., Lyras, D. & Moore, R. J. In silico identification of novel toxin homologs and
associated mobile genetic elements in Clostridium perfringens.Pathogens 8(2019).
68. Nguyen, M. et al. Using machine learning to predict antimicrobial MICs and associated genomic features for
nontyphoidal Salmonella.J. Clin. Microbiol.57, e1260-18 (2019).
69. Jia, B. et al. CARD 2017: expansion and model-centric curation of the comprehensive antibiotic resistance
database. Nucleic Acids Res. 45, D566–D573 (2017).
70. Zankari, E. et al. Identification of acquired antimicrobial resistance genes. J. Antimicrob. Chemother. 67,
2640–2644 (2012).
71. Feldgarden, M. et al. Validating the AMRFinder Tool and Resistance Gene Database by using antimicrobial
resistance genotype-phenotype correlations in a collection of isolates. Antimicrob. Agents Chemother. 63,
e00483–19 (2019).
72. Gupta, S. K. et al. ARG-ANNOT, a new bioinformatic tool to discover antibiotic resistance genes in bacterial
genomes. Antimicrob. Agents Chemother. 58, 212–220 (2014).
73. Chen, L., Zheng, D., Liu, B., Yang, J. & Jin, Q. VFDB 2016: hierarchical and refined dataset for big data
analysis-10 years on. Nucleic Acids Res. 44, D694–D697 (2016).
74. Carattoli, A. et al. In silico detection and typing of plasmids using PlasmidFinder and plasmid multilocus
sequence typing. Antimicrob. Agents Chemother. 58, 3895–3903 (2014).
75. Beaulaurier, J. et al. Assembly-free single-molecule sequencing recovers complete virus genomes from natural
microbial communities. Genome Res. 30, 437–446 (2020).
76. Wattam, A. R. et al. Improvements to PATRIC, the all-bacterial bioinformatics database and analysis
resource center. Nucleic Acids Res. 45, D535–D542 (2017).
77. McNair, K., Bailey, B. A. & Edwards, R. A. PHACTS, a computational approach to classifying the lifestyle of
phages. Bioinformatics 28, 614–618 (2012).
78. Edwards, R. A., McNair, K., Faust, K., Raes, J. & Dutilh, B. E. Computational approaches to predict
bacteriophage–host relationships. FEMS Microbiol. Rev. 40, 258–272 (2016).
NATURE PROTOCOLS PROTOCOL
NATURE PROTOCOLS | VOL 15 | SEPTEMBER 2020 | 2867–2890 | www.nature.com/nprot 2889

79. McNair, K. et al. in Bacteriophages: Methods and Protocols Vol 3 (eds Clokie, M.R.J., Kropinski, A.M. &
Lavigne, R.) 231–238 (Springer, 2018).
80. Philipson, C. W. et al. Characterizing phage genomes for therapeutic applications. Viruses 10, 188 (2018).
81. Kutter, E. in Bacteriophages (eds Clokie, M.R.J. & Kropinski A.M.) 141–149 (Springer, 2009).
82. Erickson, H. P. Size and shape of protein molecules at the nanometer level determined by sedimentation, gel
filtration, and electron microscopy. Biol. Proced. Online 11, 32 (2009).
83. Brakke, M. K. Density gradient centrifugation - a new separation technique. J. Am. Chem. Soc. 73,
1847–1848 (1951).
84. Petsch, D. & Anspach, F. B. Endotoxin removal from protein solutions. J. Biotechnol. 76,97–119 (2000).
85. Barr, J. J. et al. Bacteriophage adhering to mucus provide a non–host-derived immunity. Proc. Natl Acad. Sci.
USA 110, 10771 (2013).
86. Bocian, K. et al. The effects of T4 and A3/R bacteriophages on differentiation of human myeloid dendritic
cells. Front. Microbiol.7(2016).
87. Riss, T. L. et al. in Assay Guidance Manual (eds Sittampalam, G. S. et al.) (Eli Lilly & Company and the
National Center for Advancing Translational Sciences, 2004).
88. Moir, D. T., Ming, D., Opperman, T., Schweizer, H. P. & Bowlin, T. L. A high-throughput, homogeneous,
bioluminescent assay for Pseudomonas aeruginosa gyrase inhibitors and other DNA-damaging agents.
J. Biomol. Screen. 12, 855–864 (2007).
89. Van Twest, R. & Kropinski, A. M. Bacteriophage enrichment from water and soil. Methods Mol. Biol. 501,
15–21 (2009).
90. Summer, E. J. Preparation of a phage DNA fragment library for whole genome shotgun sequencing. Methods
Mol. Biol. 502,27–46 (2009).
91. Koren, S. et al. Canu: scalable and accurate long-read assembly via adaptive k-mer weighting and repeat
separation. Genome Res. 27, 722–736 (2017).
92. Henry, M., Lavigne, R. & Debarbieux, L. Predicting in vivo efficacy of therapeutic bacteriophages used to treat
pulmonary infections. Antimicrob. Agents Chemother. 57, 5961–5968 (2013).
93. Boulanger, P. in Bacteriophages: Methods and Protocols Vol 2 (eds Clokie, M.R.J. & Kropinski A.M.) 227–238
(Humana Press, 2009).
94. Trend, S. et al. Use of a primary epithelial cell screening tool to investigate phage therapy in cystic fibrosis.
Front. Pharmacol. 9, 1330 (2018).
95. Roach, D. R., Sjaarda, D. R., Castle, A. J. & Svircev, A. M. Host exopolysaccharide quantity and composition
impacts bacteriophage pathogenesis of Erwinia amylovora.Appl. Environ. Microbiol. 79, 3249–3256 (2013).
Acknowledgements
We thank J. Grose at Brigham Young University for gifting us phages JG265, JG266 and SM219 and the Serratia and Klebsiella strains.
We thank R. Schooley at the University of California, San Diego for providing feedback. This research was supported in part by National
Institutes of Health grant RC2DK116713 to R.A.E. and San Diego State University startup funds to D.R.R.
Author contributions
D.R.R. and R.A.E. conceived the concepts and supervised the research and development. T.L. and A.C.S. conducted the in vitro
experiments. R.A.E. performed the genomic analyses. All authors wrote and commented on the manuscript.
Competing interests
The authors declare no competing interests.
Additional information
Supplementary information is available for this paper at https://doi.org/10.1038/s41596-020-0346-0.
Correspondence and requests for materials should be addressed to D.R.R.
Reprints and permissions information is available at www.nature.com/reprints.
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Received: 29 October 2019; Accepted: 24 April 2020;
Published online: 24 July 2020
Related links
Key references using this protocol
Forti, F. et al. Antimicrob. Agents Chemother.62, e02573-17 (2018): https://aac.asm.org/content/62/6/e02573-17
Roach, D. R. et al. Cell Host Microbe 22,38–47 (2017): https://doi.org /10.1016/j.chom.2017.06.018
PROTOCOL NATURE PROTOCOLS
2890 NATURE PROTOCOLS | VOL 15 | SEPTEMBER 2020 | 2867–2890 | www.nature.com/nprot

1
nature research | reporting summary October 2018
Corresponding author(s): Dwayne Roach
Last updated by author(s): Apr 19, 2020
Reporting Summary
Nature Research wishes to improve the reproducibility of the work that we publish. This form provides structure for consistency and transparency
in reporting. For further information on Nature Research policies, see Authors & Referees and the Editorial Policy Checklist.
Statistics
For all statistical analyses, confirm that the following items are present in the figure legend, table legend, main text, or Methods section.
n/a Confirmed
The exact sample size (n) for each experimental group/condition, given as a discrete number and unit of measurement
A statement on whether measurements were taken from distinct samples or whether the same sample was measured repeatedly
The statistical test(s) used AND whether they are one- or two-sided
Only common tests should be described solely by name; describe more complex techniques in the Methods section.
A description of all covariates tested
A description of any assumptions or corrections, such as tests of normality and adjustment for multiple comparisons
A full description of the statistical parameters including central tendency (e.g. means) or other basic estimates (e.g. regression coefficient)
AND variation (e.g. standard deviation) or associated estimates of uncertainty (e.g. confidence intervals)
For null hypothesis testing, the test statistic (e.g. F, t, r) with confidence intervals, effect sizes, degrees of freedom and P value noted
Give P values as exact values whenever suitable.
For Bayesian analysis, information on the choice of priors and Markov chain Monte Carlo settings
For hierarchical and complex designs, identification of the appropriate level for tests and full reporting of outcomes
Estimates of effect sizes (e.g. Cohen's d, Pearson's r), indicating how they were calculated
Our web collection on statistics for biologists contains articles on many of the points above.
Software and code
Policy information about availability of computer code
Data collection Scrips are provided in main text.
Data analysis Provide
a
description
of
all
commercial,
open
source
and
custom
code
used
to
analyse
the
data
in
this
study,
specifying
the
version
used
OR state that no software was used.
For manuscripts utilizing custom algorithms or software that are central to the research but not yet described in published literature, software must be made available to editors/reviewers.
We strongly encourage code deposition in a community repository (e.g. GitHub). See the Nature Research guidelines for submitting code & software for further information.
Data
Policy information about availability of data
All manuscripts must include a data availability statement. This statement should provide the following information, where applicable:
- Accession codes, unique identifiers, or web links for publicly available datasets
- A list of figures that have associated raw data
- A description of any restrictions on data availability
The data that support the findings of this study are available on request from the corresponding author DRR.
Field-specific reporting
Please select the one below that is the best fit for your research. If you are not sure, read the appropriate sections before making your selection.
Life sciences Behavioural & social sciences Ecological, evolutionary & environmental sciences

2
nature research | reporting summary October 2018
For a reference copy of the document with all sections, see nature.com/documents/nr-reporting-summary-flat.pdf
Life sciences study design
All studies must disclose on these points even when the disclosure is negative.
Sample size No same size was pre-calculated for experiments. Cell viability data consisted of 3-6 biological replicates with 2 technical replicates each. Data
showed reproducible results of no effects of phage preparations on cell viability compared to buffer treated controls (p>0.05).
Data exclusions No data was excluded.
Replication Cell viability tests were performed at least 3 times as individual experiments and yielded reproducible results.
Randomization Cells were randomly allocated to the wells and then phage preparations were added. Treatments within an experiment (n) was added at the
same time.
Blinding Blinding was not done. Cell viability read-out was done by BMG ClarioStar and was therefore not subject to human bias. Therefore blinding
was not relevant in our study.
Reporting for specific materials, systems and methods
We require information from authors about some types of materials, experimental systems and methods used in many studies. Here, indicate whether each material,
system or method listed is relevant to your study. If you are not sure if a list item applies to your research, read the appropriate section before selecting a response.
Materials & experimental systems
n/a Involved in the study
Antibodies
Eukaryotic cell lines
Palaeontology
Animals and other organisms
Human research participants
Clinical data
Methods
n/a Involved in the study
ChIP-seq
Flow cytometry
MRI-based neuroimaging
Antibodies
Antibodies used Describe
all
antibodies
used
in
the
study;
as
applicable,
provide
supplier
name,
catalog
number,
clone
name,
and
lot
number.
Validation Describe
the
validation
of
each
primary
antibody
for
the
species
and
application,
noting
any
validation
statements
on
the
manufacturer’s website, relevant citations, antibody profiles in online databases, or data provided in the manuscript.
Eukaryotic cell lines
Policy information about cell lines
Cell line source(s) ATCC
Authentication Cells were authenticated by ATCC prior to receipt.
Mycoplasma contamination All cell lines were tested for Mycoplasma contamination using PCR with a protocol derived from Uphoff CC & Drexler HG.
(http://www.ncbi.nlm.nih.gov/pubmed/21516400 & http://www.ncbi.nlm.nih.gov/pubmed/11928999). No contamination
was found.
Commonly misidentified lines
(See ICLAC register)
According to the ICLAC register the cell lines used in this study are not commonly misidentified or cross-contaminated.
Palaeontology
Specimen provenance Provide
provenance
information
for
specimens
and
describe
permits
that
were
obtained
for
the
work
(including
the
name
of
the
issuing authority, the date of issue, and any identifying information).
Specimen deposition Indicate
where
the
specimens
have
been
deposited
to
permit
free
access
by
other
researchers.
Dating methods If
new
dates
are
provided,
describe
how
they
were
obtained
(e.g.
collection,
storage,
sample
pretreatment
and
measurement),

3
nature research | reporting summary October 2018
Dating methods where
they
were
obtained
(i.e.
lab
name),
the
calibration
program
and
the
protocol
for
quality
assurance
OR
state
that
no
new
dates are provided.
Tick this box to confirm that the raw and calibrated dates are available in the paper or in Supplementary Information.
Animals and other organisms
Policy information about studies involving animals; ARRIVE guidelines recommended for reporting animal research
Laboratory animals For
laboratory
animals,
report
species,
strain,
sex
and
age
OR
state
that
the
study
did
not
involve
laboratory
animals.
Wild animals Provide
details
on
animals
observed
in
or
captured
in
the
field;
report
species,
sex
and
age
where
possible.
Describe
how
animals
were caught and transported and what happened to captive animals after the study (if killed, explain why and describe method;
if
released, say where and when) OR state that the study did not involve wild animals.
Field-collected samples For
laboratory
work
with
field-collected
samples,
describe
all
relevant
parameters
such
as
housing,
maintenance,
temperature,
photoperiod and end-of-experiment protocol OR state that the study did not involve samples collected from the field.
Ethics oversight Identify
the
organization(s)
that
approved
or
provided
guidance
on
the
study
protocol,
OR
state
that
no
ethical
approval
or
guidance was required and explain why not.
Note that full information on the approval of the study protocol must also be provided in the manuscript.
Human research participants
Policy information about studies involving human research participants
Population characteristics Describe
the
covariate-relevant
population
characteristics
of
the
human
research
participants
(e.g.
age,
gender,
genotypic
information, past and current diagnosis and treatment categories). If you filled out the behavioural & social sciences study design
questions and have nothing to add here, write "See above."
Recruitment Describe
how
participants
were
recruited.
Outline
any
potential
self-selection
bias
or
other
biases
that
may
be
present
and
how
these are likely to impact results.
Ethics oversight Identify
the
organization(s)
that
approved
the
study
protocol.
Note that full information on the approval of the study protocol must also be provided in the manuscript.
Clinical data
Policy information about clinical studies
All manuscripts should comply with the ICMJE guidelines for publication of clinical research and a completed CONSORT checklist must be included with all submissions.
Clinical trial registration Provide
the
trial
registration
number
from
ClinicalTrials.gov
or
an
equivalent
agency.
Study protocol Note
where
the
full
trial
protocol
can
be
accessed
OR
if
not
available,
explain
why.
Data collection Describe
the
settings
and
locales
of
data
collection,
noting
the
time
periods
of
recruitment
and
data
collection.
Outcomes Describe
how
you
pre-defined
primary
and
secondary
outcome
measures
and
how
you
assessed
these
measures.
ChIP-seq
Data deposition
Confirm that both raw and final processed data have been deposited in a public database such as GEO.
Confirm that you have deposited or provided access to graph files (e.g. BED files) for the called peaks.
Data access links
May remain private before publication.
For
"Initial
submission"
or
"Revised
version"
documents,
provide
reviewer
access
links.
For
your
"Final
submission"
document,
provide a link to the deposited data.
Files in database submission Provide
a
list
of
all
files
available
in
the
database
submission.
Genome browser session
(e.g. UCSC)
Provide
a
link
to
an
anonymized
genome
browser
session
for
"Initial
submission"
and
"Revised
version"
documents
only,
to
enable peer review. Write "no longer applicable" for "Final submission" documents.
Methodology
Replicates Describe
the
experimental
replicates,
specifying
number,
type
and
replicate
agreement.

4
nature research | reporting summary October 2018
Sequencing depth Describe
the
sequencing
depth
for
each
experiment,
providing
the
total
number
of
reads,
uniquely
mapped
reads,
length
of
reads and whether they were paired- or single-end.
Antibodies Describe
the
antibodies
used
for
the
ChIP-seq
experiments;
as
applicable,
provide
supplier
name,
catalog
number,
clone
name, and lot number.
Peak calling parameters Specify
the
command
line
program
and
parameters
used
for
read
mapping
and
peak
calling,
including
the
ChIP,
control
and
index files used.
Data quality Describe
the
methods
used
to
ensure
data
quality
in
full
detail,
including
how
many
peaks
are
at
FDR
5%
and
above
5-fold
enrichment.
Software Describe
the
software
used
to
collect
and
analyze
the
ChIP-seq
data.
For
custom
code
that
has
been
deposited
into
a
community repository, provide accession details.
Flow Cytometry
Plots
Confirm that:
The axis labels state the marker and fluorochrome used (e.g. CD4-FITC).
The axis scales are clearly visible. Include numbers along axes only for bottom left plot of group (a 'group' is an analysis of identical markers).
All plots are contour plots with outliers or pseudocolor plots.
A numerical value for number of cells or percentage (with statistics) is provided.
Methodology
Sample preparation Describe
the
sample
preparation,
detailing
the
biological
source
of
the
cells
and
any
tissue
processing
steps
used.
Instrument Identify
the
instrument
used
for
data
collection,
specifying
make
and
model
number.
Software Describe
the
software
used
to
collect
and
analyze
the
flow
cytometry
data.
For
custom
code
that
has
been
deposited
into
a
community repository, provide accession details.
Cell population abundance Describe
the
abundance
of
the
relevant
cell
populations
within
post-sort
fractions,
providing
details
on
the
purity
of
the
samples
and how it was determined.
Gating strategy Describe
the
gating
strategy
used
for
all
relevant
experiments,
specifying
the
preliminary
FSC/SSC
gates
of
the
starting
cell
population, indicating where boundaries between "positive" and "negative" staining cell populations are defined.
Tick this box to confirm that a figure exemplifying the gating strategy is provided in the Supplementary Information.
Magnetic resonance imaging
Experimental design
Design type Indicate
task
or
resting
state;
event-related
or
block
design.
Design specifications Specify
the
number
of
blocks,
trials
or
experimental
units
per
session
and/or
subject,
and
specify
the
length
of
each
trial
or block (if trials are blocked) and interval between trials.
Behavioral performance measures State
number
and/or
type
of
variables
recorded
(e.g.
correct
button
press,
response
time)
and
what
statistics
were
used
to establish that the subjects were performing the task as expected (e.g. mean, range, and/or standard deviation across
subjects).
Acquisition
Imaging type(s) Specify:
functional,
structural,
diffusion,
perfusion.
Field strength Specify
in
Tesla
Sequence & imaging parameters Specify
the
pulse
sequence
type
(gradient
echo,
spin
echo,
etc.),
imaging
type
(EPI,
spiral,
etc.),
field
of
view,
matrix
size,
slice thickness, orientation and TE/TR/flip angle.
Area of acquisition State
whether
a
whole
brain
scan
was
used
OR
define
the
area
of
acquisition,
describing
how
the
region
was
determined.
Diffusion MRI Used Not used

5
nature research | reporting summary October 2018
Preprocessing
Preprocessing software Provide
detail
on
software
version
and
revision
number
and
on
specific
parameters
(model/functions,
brain
extraction,
segmentation, smoothing kernel size, etc.).
Normalization If
data
were
normalized/standardized,
describe
the
approach(es):
specify
linear
or
non-linear
and
define
image
types
used for transformation OR indicate that data were not normalized and explain rationale for lack of normalization.
Normalization template Describe
the
template
used
for
normalization/transformation,
specifying
subject
space
or
group
standardized
space
(e.g.
original Talairach, MNI305, ICBM152) OR indicate that the data were not normalized.
Noise and artifact removal Describe
your
procedure(s)
for
artifact
and
structured
noise
removal,
specifying
motion
parameters,
tissue
signals
and
physiological signals (heart rate, respiration).
Volume censoring Define
your
software
and/or
method
and
criteria
for
volume
censoring,
and
state
the
extent
of
such
censoring.
Statistical modeling & inference
Model type and settings Specify
type
(mass
univariate,
multivariate,
RSA,
predictive,
etc.)
and
describe
essential
details
of
the
model
at
the
first
and second levels (e.g. fixed, random or mixed effects; drift or auto-correlation).
Effect(s) tested Define
precise
effect
in
terms
of
the
task
or
stimulus
conditions
instead
of
psychological
concepts
and
indicate
whether
ANOVA or factorial designs were used.
Specify type of analysis: Whole brain ROI-based Both
Statistic type for inference
(See Eklund et al. 2016)
Specify
voxel-wise
or
cluster-wise
and
report
all
relevant
parameters
for
cluster-wise
methods.
Correction Describe
the
type
of
correction
and
how
it
is
obtained
for
multiple
comparisons
(e.g.
FWE,
FDR,
permutation
or
Monte
Carlo).
Models & analysis
n/a Involved in the study
Functional and/or effective connectivity
Graph analysis
Multivariate modeling or predictive analysis
Functional and/or effective connectivity Report
the
measures
of
dependence
used
and
the
model
details
(e.g.
Pearson
correlation,
partial
correlation, mutual information).
Graph analysis Report
the
dependent
variable
and
connectivity
measure,
specifying
weighted
graph
or
binarized
graph,
subject- or group-level, and the global and/or node summaries used (e.g. clustering coefficient, efficiency,
etc.).
Multivariate modeling and predictive analysis Specify
independent
variables,
features
extraction
and
dimension
reduction,
model,
training
and
evaluation
metrics.