Cell-free production of personalized therapeutic phages targeting multidrug-resistant bacteria

Bacteriophages are potent therapeutics against biohazardous bacteria, which rapidly develop multidrug resistance. However, routine administration of phage therapy is hampered by a lack of rapid production, safe bioengineering, and detailed characterization of phages. Thus, we demonstrate a comprehensive cell-free platform for personalized production, transient engineering, and proteomic characterization of a broad spectrum of phages. Using mass spectrometry, we validated hypothetical and non-structural proteins and could also monitor the protein expression during phage assembly. Notably, a few microliters of a one-pot reaction produced effective doses of phages against enteroaggregative Escherichia coli (EAEC), Yersinia pestis, and Klebsiella pneumoniae. By co-expressing suitable host factors, we could extend the range of cell-free production to phages targeting gram-positive bacteria. We further introduce a non-genomic phage engineering method, which adds functionalities for only one replication cycle. In summary, we expect this cell-free methodology to foster reverse and forward phage engineering and customized production of clinical-grade bacteriophages.

Here, we introduce the phage production pipeline “phactory,” which enables on-demand and safe generation of a broad spectrum of phages in a host-independent manner, including transiently engineered and therapeutic phages against antibiotic-resistant bacteria. This approach has also enabled us to validate the existence of dozens of highly conserved hypothetical proteins, including non-structural proteins. It also allowed us to systematically monitor phage assembly in correlation with its expression profile using time-resolved mass spectrometry. These innovations can significantly foster our understanding of molecular phage biology, accelerate future phage engineering, and improve therapeutic phage production.

So far, neither the cell-free generation of clinically relevant bacteriophages nor a comprehensive characterization of phage proteomes has been achieved.

We thus reasoned that cell-free production systems could address all of the challenges mentioned above, as shown in Figure 1 : they could (1) produce phages efficiently and safely at high titers, (2) facilitate novel methodologies for phage engineering, and (3) enable a better characterization of phage proteomes and function, which is essential to improve phage therapies further. Cell-free expression systems have recently been employed for the biosynthesis of glycoproteins (), the incorporation of non-canonical amino acids (), and as components of minimal cells (). However, for phage production, cell-free systems made from E. coli extracts were limited to the generation of model phages such as the E. coli phages T7, phiX174, and MS2 ().

An optimized transcription-translation system with additional host factors and crowding reagents can produce phages that target a broad range of Gram-positive and Gram-negative bacteria. Non-genomic phage engineering can safely add transient functionalities to phages. Cell-free produced phages can be analyzed by mass spectrometry without purification, facilitating the identification of non-structural phage proteins. Time-resolved proteomic analyses can further be correlated with the phage assembly process. At the microliter scale, a one-pot reaction enables the production of therapeutic phages against multidrug-resistant bacteria in a matter of hours at clinically relevant titers.

Apart from the elaborate production process, the limited understanding of the molecular phage-host interaction can lead to erratic treatment outcomes (). This is partly due to the insufficient information on phage protein expression and function. Even though phages have been described for over a century (), the characterization of phage proteomes is still largely obscure. Based on the analysis of several hundred phage genomes, recent studies determined over 60% of the annotated gene products as “hypothetical proteins” () without any experimental validation.

In addition, the production of engineered phages is limited by the necessity of modifying phage genomes within living bacteria, e.g., by CRISPR-Cas or by using yeast artificial chromosomes, which reduces efficiency ().

Because of the increasing need for such alternatives to antibiotics, Belgium and France have recently allowed the use of naturally occurring, well-characterized bacteriophages for personalized bacteriophage (short “phage”) therapy under the legal framework of magistral preparation (). Unfortunately, the current production processes for natural and engineered phages pose considerable biosafety concerns and are often relatively inefficient and unreliable. In both cases, a pure culture of the pathogenic host is required for phage amplification and engineering (). This process poses a high risk of prophage contamination (), which can induce virulence factors in bacteria (). Moreover, the standard cultivation of phages with bacteria is time-consuming, generates relatively low titers, and requires the appropriate lab security level for the respective pathogen. These factors make it challenging to parallelize the production of personalized phage cocktails that typically contain up to 16 phages (). Recent double-blind patient trials have also revealed the instability and correspondingly low shelf half-life of prepared phage cocktails, emphasizing the importance of a fast and reliable production process ().

The growing number of multidrug-resistant (MDR) bacteria has been classified as one of the most critical global health threats by the World Health Organization (). It is estimated that costs of $2.2 billion in the United States in 2014 alone () and 33,000 deaths per year in Europe are caused by MDR bacteria (). MDR is particularly severe in gram-negative bacteria, with most cases reported for Escherichia coli. However, classic hospital pathogens are also increasingly resistant and frequently found in bacterial superinfections in COVID-19 patients, especially the ESKAPE pathogen Klebsiella pneumoniae (). In addition, pathogens such as Yersinia pestis, a potential biological warfare agent, still contribute to naturally occurring outbreaks of epidemics in Africa with high morbidity rates ().

Attributable deaths and disability-adjusted life-years caused by infections with antibiotic-resistant bacteria in the EU and the European Economic Area in 2015: a population-level modelling analysis.

Notably, the bioluminescence signal disappeared after one phage replication cycle, showing that the phage modification at the protein level was only transient ( Figure 4 E).

Upon co-expression of the His-tagged capsid protein, phages could be purified by affinity chromatography at nearly two log-fold higher titers, indicating that the G10B-His was integrated into the outer phage structure and modified phages were still effective against the host bacteria ( Figure 4 D). We also co-expressed HiBit with a His-tag fused to G10B, purified the resulting phages, and incubated them with E. coli. After washing the bacteria, a substantial luminescence signal was detected compared with phages produced by co-expression of his-tagged HiBit without the capsid protein G10B or wild-type T7 ( Figure 4 E). These data show that the engineered phages could attach to the host bacteria and potentially serve phage biotyping.

Given the ease with which proteins can be co-expressed in the cell-free system, we reasoned that it should be possible to engineer phages at the protein level without altering the phage genome. To demonstrate this approach, we generated variants of the T7 phage with the modified minor capsid protein Gp10B, including a polyhistidine-tag (His-tag) and the split luciferase HiBit part (). The HiBit luciferase consists of two parts. The minor C-terminal domain was genetically fused to the capsid protein structure, while the major N-terminal domain was part of the assay buffer. The complementation of both parts generates a bioluminescent signal. This approach thus enables the affinity purification and detection of the modified phage with a relatively small protein modification as compared with fluorescent proteins or other luciferase systems.

Capitalizing on the direct genetic access in the cell-free system, we thus co-expressed the corresponding SigA host factor. This modification indeed enabled the assembly of the B. subtilis-targeting phages, shown by spot assay ( Figure 4 B). Only the combination of SigA, phage DNA, and the cell-free system led to the production of functional phages ( Figure S2 A), with a plaque size identical to phages amplified in their bacterial host ( Figures S2 B and S2C). To exclude lytic proteins generated in the cell-free system rather than by complete infectious phages as a possible cause for plaque formation, the plaques were picked, the contained phages amplified with the appropriate host, and their identity validated by PCR with phage-specific primers ( Figure S2 C). This analysis confirmed that the adaptation of the RNA polymerase in the cell-free system was sufficient for generating fully functional phages. These data show cell-free phage production can be extended to the important class of phages targeting gram-positive bacteria.

The early genes of a phage genome are usually transcribed by the host’s RNA polymerase. Therefore, we reasoned that modifying the E. coli RNA polymerase by incorporating the B. subtilis housekeeping sigma factor SigA should enable the production of these phages ( Figures 4 A and 4C). The RNA polymerase holoenzymes of E. coli and B. subtilis show a 96.5% sequence similarity ( Table S3 ), and the sigma factors were shown to be interchangeable ().

To demonstrate the generalizability of phactory beyond phages targeting gram-negative bacteria, we sought to produce the phages Phi29 (NCBI ID: NC_011048.1 ) and Goe1 (vB_BsuP_Goe1, NCBI ID NC_049975.1 ) with the cell-free system, targeting the gram-positive bacterium Bacillus subtilis. ( Figure 3 C).

Collectively our results demonstrate a complete pipeline for personalized phage therapy (phactory) that includes the isolation of a multidrug-resistant pathogen from a patient, the selection of a specific therapeutic phage, its cell-free production, and successful application.

Encouraged by this result, we assessed whether we could produce therapeutic phages on-demand that could target an MDR ESKAPE pathogen in a realistic patient-dependent scenario. We thus isolated a carbapenem-resistant K. pneumoniae strain (Bw1) with wzi capsule type K17 from colonized patient skin, which is untreatable with common antibiotics. We then isolated a bacteriophage, which we designated vB_KpS_Muc5 (MUC-5), from a wastewater plant via amplification in K. pneumoniae strain Bw1 followed by isolation of phage DNA. When we subsequently added MUC-5 phages produced in the cell-free system to the pathogenic bacteria, we observed bacterial lysis within approximately 30 min with sufficient phage titers produced in just a few microliters of the one-pot reaction ( Figure 4 A).

As shown in Figure 4 A , it was indeed possible to produce clinically relevant titers (on the order of 10PFU/mL) of specific phages within just 4 h.

(E) The modification with a luciferase is present for only one replication cycle of the phage. Bioluminescent signals before and after the replication cycle of the phage by infection of the host bacteria. Host bacteria were incubated with modified phages, pelleted, and washed (before replication). The same experiment was repeated with the resulting phages, which lost the transient modification after replication. Phages produced with co-expression of Gp10B-HiBit-His, or controls (HiBit-His without Gp10B or WT). The mean luminescence signal is shown here ±SEM (n = 3). ∗∗∗p < 0.001 for an ordinary two-way ANOVA with Tukey correction.

(D) Titers of T7 phages with co-expression of a modified capsid protein (Gp10B) fused to split luciferase (HiBit) and a polyhistidine-tag (Gp10B-HiBit-His) before and after affinity purification (elution) compared to wildtype T7 phages (WT). Titers were adjusted to 10 6 pfu/mL for both conditions. The mean phage titer is shown here ±SEM (n = 3). ∗p < 0.05 for an unpaired two-tailed t-test with Welch’s correction.

(C) Schematic of phage rebooting via co-expression of an appropriate host factor (HF, blue arrows) to produce phages targeting bacteria distant from E. coli. Non-genomic phage engineering is achieved by the co-expression of proteins of interest (POI, red) that co-assemble into the phage structure to produce phages with transiently modified properties.

(A and B) Quantification of the titers of the phages produced by the cell-free system. Titers sufficient for therapy (>10 6 pfu/mL) were obtained for CLB-P3 targeting EAEC, PhiA1122 targeting Y. pestis, and vB_KpS_Muc5 (MUC-5) phages targeting K pneumoniae (K.pneum.). The mean phage titer is shown here ±SEM (n = 3). (B) Spot assay results from some of the phages shown in (a). The assay plates include the phage stock as a positive control (+) as well as the cell-free system with (+) phage DNA and without (−).

We next tested whether we could also produce biomedically relevant phages targeting bacteria from a different genus ( Table S2 ). We thus chose PhiA1122 (NCBI ID: NC_004777.1 ) phage targeting Y. pestis, a possible biological warfare agent.

Combined with the plaque assay results ( Figure 3 B), which show the first assembled phages between 60 and 90 min, these proteins appear to be limiting factors for the assembly of fully functional phages or the release of the phages from the bacteria. On this basis, we aligned the protein expression patterns to the position of their coding sequence in the T7 genome. The trajectory obtained by time-resolved proteomics of the cell-free system is consistent with the current consecutive expression model proposed for T7 phage ( Figure 3 C). It is even possible to analyze the kinetics of cell-free expression with single protein precision ( Figure 3 D), exemplarily shown for T7 RNA polymerase (RPOL), T7 DNA polymerase (DPOL), as well as the late genes, Tail tubular gp12 (TUBE2) and Terminase S (TER S).

Interestingly, only four proteins were expressed particularly late after 40 min. These four proteins entail the two subunits of the terminase (TerL and TerS, necessary for the loading of DNA into the phage capsid []), the cell lysis protein holin (necessary for the timed release of the phage from the bacteria (), and the yet uncharacterized Y195.

Encouraged by the high proteome coverage, we reasoned that it should be possible to correlate protein expression to phage assembly by time-resolved mass spectrometry and plaque assay to analyze the tightly orchestrated protein expression with molecular precision. We thus focused on the T7 phage, for which a clustering into early-, middle-, and late-expressing genes has been known. Here, all proteins are produced in under 20 min, making it challenging to track the expression of every single protein in a time-resolved manner in bacteria (). Leveraging the cell-free system, we monitored the protein expression and T7 phage titer from a 70 μL cell-free reaction over 10 time points (0–240 min). Specifically, half of the removed volume at each time point (3 μL) was prepared for LC-MS/MS analysis by immediately adding lithium dodecyl sulfate (LDS) to stop the expression and assembly process ( Figure 3 A ). From the remaining volume, the phage titer was determined via plaque assay ( Figure 3 B). The mass spectrometric results show that most phage proteins are present after 40 min ( Figure 3 A), but fully assembled and functional phages first appear after 60 min ( Figure 3 B).

(D) Protein expression patterns can be differentiated with molecular precision. Early genes, e.g., DNA/RNA polymerase (P00581 and P00573), are already expressed after 10 min, whereas structurally relevant proteins, e.g., TUBE2 and late proteins such as TerS are expressed after 20 and 30 min. Data were obtained from three biological replicates. Error bars in b and d represent the SD of the mean (n = 3).

(C) Mapping of protein expression onto the phage genome accurately recapitulates early, middle, and late gene products in the synthetic system. Time points for which a protein was detected with a 50% max. MS intensities were aligned to the position in the T7 genome.

(A) The protein expression of the T7 proteome was analyzed by LC-MS/MS from 0 to 240 min within the cell-free system. The heatmap shows the relative abundance of proteins at different time points, with the dendrogram indicating the protein similarity derived from the measured peptides.

Protein expression and phage assembly from a cell-free reaction were monitored with respect to proteome composition and titer over different time points.

As shown in Figure 2 B, we could identify most of the CLB-P3 phage proteins in both samples, but the cell-free system produced more stable results. In total, 68 of 87 predicted proteins were identified by LC-MS/MS. Only 38 of these 68 proteins could be directly assigned functional names based on genome annotation. The remaining 30 are labeled “hypothetical.” Importantly, the corresponding heatmap ( Figure 2 exhibits a similar distribution of the precursor signal as the heatmap obtained by analyzing the T7 phage. This comparison enables data-driven functionalization based on protein intensity and coverage. Similar to the heatmaps of the T7 phage, the intensities of the soluble factors of the CLB-P3 are lower in the phage stock sample than in the cell-free system. In contrast, structural proteins were present with high abundance in both sample types, as phage stocks (e.g., by polyethylene glycol [PEG] precipitation or ultracentrifugation) enrich completely assembled virion particles. Importantly, it was possible to track these proteins in sequencing data of several hundred phages targeting gram-negative and gram-positive bacteria. These include potentially therapeutic phages such as the Shigella phage psf-2 (NC_026,010.1), the Salmonella phage Jersey (NC_021,777.1), the Klebsiella phage Sushi (NC_028,774.1), and the Cronobacter phage phiES15 (NC_018,454.1). Moreover, the identified proteins are also encoded in phages potentially relevant for the food industry, such as Lactococcus phage TP901-1 (NC_002,747.1) and Pseudomonas phage phiPSA1 (NC_024,365.1) ( Figure S1 ).

Motivated by the success in validating hypothetical proteins based on the well-characterized genome of T7, we next tested whether the cell-free approach could be applied to yet uncharacterized phages. We chose phage CLB-P3 (), which is directed against a strain of pathogenic enteroaggregative E. coli (EAEC). To assemble a protein FASTA file for LC-MS/MS analysis, we de novo sequenced CLB-P3 phages using next-generation sequencing (using both Nanopore and Illumina Sequencing) and performed a hybrid genome assembly from MinION and MiSeq reads. We then analyzed the proteins from cell-free expression by MS/MS.

The unbiased clustering of proteins (rows in the heatmap) suggests overall two groups of proteins according to their abundance with a general trend toward the more abundant proteins corresponding to structural proteins, whereas the less abundant proteins were non-structural ( Figure 2 A). Non-structural proteins (based on the protein uniprot ID and their correlated function) exhibited a much lower abundance in the phage stock samples than in the cell-free products. We also found that several of the newly validated proteins from phage T7 ( Table S1 ) are conserved in other clinically relevant phages targeting K. pneumoniae and Y. pestis, e.g., Yersinia phage PhiA1122 (NCBI ID: NP_848312.1 ) and Klebsiella phage (vB_KpnP_KpV767 GenBank ID: AOZ65500), suggesting that these proteins play a universal role in phage biology.

To this end, the cell-free system, based on E. coli cell extract, buffers, and metabolites, was mixed with purified phage DNA, which resulted in the assembly of functional phages. The phages from classic bacterial cultures and cell-free produced phages were subjected to in-gel digestion and analyzed by high-resolution liquid chromatography coupled to tandem mass spectrometry (LC-MS/MS). Remarkably, the proteome coverage of phages generated by the cell-free system was 72% (41 of 57), with 28% (16 proteins) of the proteome represented by hypothetical proteins ( Figure 2 A ). Thus, the cell-free system enabled the detection of almost all previously characterized proteins and validated the existence of at least half of the hypothetical proteins.

(A and B). Phage proteomes expressed in a cell-free system were analyzed by LC-MS/MS. The heatmaps from (A) T7 and (B) CLB-P3 proteomes were obtained from three independent replicates (S-1, S-2. S-3) for phage stock (stock) and cell-free assembled phages (cell-free). The logarithm transformed (base 10) iBAQ intensity was used to represent the protein abundance. Missing values were replaced with the lowest measured iBAQ intensity. The clustering was based on the Ward linkage method using the euclidean distance. Dendrograms display similarity between rows (or columns). Less abundant proteins in the phage stock proteome clustered mostly as non-structural proteins, e.g., not contained in the phage capsid, base-plate, and tail fiber. Stars indicate all hypothetical proteins validated in this study.

To date, phage proteomes, particularly proteins involved in phage assembly, are not yet characterized in detail. To assess if the incomplete phage proteome could be analyzed in-depth, we first compared the protein composition of the model T7 phage (NCBI ID: NC_001604.1 ) derived from classic bacteria-dependent amplification (phage stock) with phages synthesized in the cell-free reaction.

Discussion

We have introduced phactory, a cell-free phage production and analysis pipeline, which facilitates the host-independent one-pot synthesis of potentially clinically relevant phages enabling the systematic identification of phage proteins and phage engineering.

However, the success of cell-free production of other phages is still limited by the knowledge of required interaction with other host proteins, the presence of membrane components, and DNA packing mechanisms.

Haldenwang, 1995 Haldenwang W.G. The sigma factors of Bacillus subtilis. Davison et al., 1980 Davison B.L.

Murray C.L.

Rabinowitz J.C. Specificity of promoter site utilization in vitro by bacterial RNA polymerases on Bacillus phage phi 29 DNA. Transcription mapping with exonuclease III. Whiteley and Ernest Hemphill, 1970 Whiteley H.R.

Ernest Hemphill H. The interchangeability of stimulatory factors isolated from three microbial RNA polymerases. Phactory increases the diversity of bacteriophages produced by cell-free expression beyond E. coli phages. We were able to produce phages (Phi29 and Goe1) against gram-positive bacteria by co-expressing the primary B. subtilis RNA polymerase sigma factor SigA, which mediates the interaction of RNA polymerase with B. subtilis-specific promoter sequences () ( Figure 4 B). These results are in line with the finding by Whiteley and colleagues (), demonstrating that the B. subtilis SigA factor can direct the E. coli RNA-polymerase to transcribe foreign promoters, even if the natural sigma70 cofactor is still active in the system.

Vellanoweth and Rabinowitz, 1992 Vellanoweth R.L.

Rabinowitz J.C. The influence of ribosome-binding-site elements on translational efficiency in Bacillus subtilis and Escherichia coli in vivo. Roberts and Rabinowitz, 1989 Roberts M.W.

Rabinowitz J.C. The effect of Escherichia coli ribosomal protein S1 on the translational specificity of bacterial ribosomes. Our results suggest that the E. coli and B. subtilis RNA-Polymerase core enzymes are sufficiently similar to allow some promiscuity in integrating the host factor. In addition, our data also make it appear likely that additional phages for hosts distinct from E. coli could be generated by supplementing appropriate host factors ( Figures 4 A–4C). Interestingly, the E. coli ribosomes can non-specifically translate other bacterial mRNAs (e.g., from B. subtilis), while this does not seem to occur vice versa (). The promiscuity of E. coli translation factors is thus a powerful feature of the phactory production process.

Vogele et al., 2021 Vogele K.

Falgenhauer E.

von Schönberg S.

Simmel F.C.

Pirzer T. Small antisense DNA-based gene Silencing enables cell-free bacteriophage manipulation and genome replication. At the present stage, phactory relies on isolating phage DNA from purified phage stocks. With such phage genomes, approximately 0.126 (MS2), 0.05 (CLP-P3), 0.00002 (PhiA1122), 0.0025 (Phi29), and 0.0014 (Goe1) infectious phage particles per genome were obtained. In comparison, the T7 phage can replicate within the cell-free system, resulting in an infectious particle/genome ratio of 2.7 ().

Kodumal et al., 2004 Kodumal S.J.

Patel K.G.

Reid R.

Menzella H.G.

Welch M.

Santi D.V. Total synthesis of long DNA sequences: synthesis of a contiguous 32-kb polyketide synthase gene cluster. Richardson et al., 2017 Richardson S.M.

Mitchell L.A.

Stracquadanio G.

Yang K.

Dymond J.S.

DiCarlo J.E.

Lee D.

Huang C.L.V.

Chandrasegaran S.

Cai Y.

et al. Design of a synthetic yeast genome. Cassini et al., 2019 Cassini A.

Högberg L.D.

Plachouras D.

Quattrocchi A.

Hoxha A.

Simonsen G.S.

Colomb-Cotinat M.

Kretzschmar M.E.

Devleesschauwer B.

Cecchini M.

et al. Attributable deaths and disability-adjusted life-years caused by infections with antibiotic-resistant bacteria in the EU and the European Economic Area in 2015: a population-level modelling analysis. Tacconelli et al., 2018 Tacconelli E.

Carrara E.

Savoldi A.

Harbarth S.

Mendelson M.

Monnet D.L.

Pulcini C.

Kahlmeter G.

Kluytmans J.

Carmeli Y.

et al. Discovery, research, and development of new antibiotics: the WHO priority list of antibiotic-resistant bacteria and tuberculosis. Given that it is already possible to generate synthetic genomes between 5 and 1,500 kb (), we anticipate that phages can soon be de novo designed and produced from synthetic genomes. With this approach, it should be possible to generate a variety of phages that tackle different MDR bacteria for which the development of antibiotic resistance has become a major global health problem ().

Yamamoto et al., 1970 Yamamoto K.R.

Alberts B.M.

Benzinger R.

Lawhorne L.

Treiber G. Rapid bacteriophage sedimentation in the presence of polyethylene glycol and its application to large-scale virus purification. Phactory also enables the identification of non-structural phage proteins, particularly by comparing cell-free produced phages to phage stocks. It is known that phage stock samples enrich functioning phages, which include the structural phage proteins during PEG purification, e.g., tail fiber proteins, stem proteins, and base plate proteins (). On the other hand, soluble factors necessary for intracellular phage replication, e.g., polymerases, nuclease inhibitors, are probably lost during buffer exchange, as they are not physically connected to the phage structure.

) . Indeed, the proteomics data generated from phactory were highly reproducible and enabled the identification of over 44 proteins with almost complete proteome coverage that had been previously categorized only as hypothetical ( Figure 2 ). In contrast, samples obtained via bacteria-dependent production displayed fewer identified proteins with a high variability of the obtained intensities. In these samples, many of the proteins could not be identified reproducibly ( Figure 2

We found that the set of proteins in both the cell-free and bacteria-derived samples of phage T7 could be assigned mainly to structural proteins instead of soluble factors not attached to the phage structure. By contrast, proteins exhibiting lower MS intensities were mainly related to proteins not part of the phage structure ( Figure 2 A). As the majority of the hypothetical proteins clustered together with already characterized non-structural T7 proteins, we assumed that these proteins also belong to the class of non-structural proteins. A similar intensity distribution was found for the previously uncharacterized and de-novo sequenced CLB-P3 phage ( Figure 2 B). We thus expected the proteins found in both types of preparations to be structural, as observed for T7 proteins of the "core proteome." Accordingly, we assumed that proteins that are exclusively well-represented in samples from the cell-free system are mainly non-structural. It is likely that this identification method for phage proteins, exemplarily demonstrated for CLB-P3, will facilitate the further characterization of other phages. It can be seen that a range of CLB-P3 proteins is well conserved in many bacteriophages, while others appear to be host-specific (Enterobacteria/Escherichia phage).

Garenne et al., 2019 Garenne D.

Beisel C.L.

Noireaux V. Characterization of the all- E. coli transcription-translation system myTXTL by mass spectrometry. It is also interesting to see that CLB-P3 shares a large portion of proteins with Escherichia phage ADB-2 (JX912252.1) and Shigella phage pSf-2 (NC_026,010.1), highlighting its clinical potential ( Figure S1 ). In addition, we observed that the proteins from the cell-free expression system itself (i.e., from the E. coli proteome) are ten times less concentrated than the actual bacterial cytosol, allowing the detection of also less abundant phage proteins by MS analysis. Given that all of these proteins are conserved in other phages, these results may pave the way for consecutive functional and structural analysis enabling a deeper understanding of phage biology or even interaction with the host. Notably, during their life cycle, bacteriophages interact with several host proteins. Some of these known interaction partners have been identified in the cell-free system, e.g., GroEL/ES ().

McAllister and Wu, 1978 McAllister W.T.

Wu H.L. Regulation of transcription of the late genes of bacteriophage T7. Heineman and Bull, 2007 Heineman R.H.

Bull J.J. Testing optimality with experimental evolution: lysis time in a bacteriophage. Phactory furthermore allows for longitudinal monitoring of phage protein expression over time. When examining the time course of phage protein expression, the natural protein expression patterns for phage assembly is closely recapitulated in vitro. The E. coli RNA polymerase-dependent proteins cluster separately from the T7 RNA polymerase-dependent proteins ( Figure 4 A), and the T7 typical gene expression sequence from early via middle to late genes is confirmed ( Figure 4 C) (). However, the overall expression trajectory appears to be decelerated by about a factor of two to three compared to the in vivo case, where T7 phage replicates typically within 20–30 min (). This may be partially explained by the dilution of the bacterial cell extract compared to the bacterial cytosol (by a factor of ∼10), which may lead to a decrease in reaction rates in the extract. At the same time, this deceleration enables the expression analysis in detail while probably still maintaining the natural sequence.

Carter et al., 1981 Carter A.D.

Morris C.E.

McAllister W.T. Revised transcription map of the late region of bacteriophage T7 DNA. The time-resolved analysis allowed us to analyze the expression sequence between class II and class III genes () at the protein level with total proteome coverage. Interestingly our data indicates a fourth cluster of late-expressing genes, including TerS, TerL, holin, and Y195 ( Figure 3 A).

Daudén et al., 2013 Daudén M.I.

Martín-Benito J.

Sánchez-Ferrero J.C.

Pulido-Cid M.

Valpuesta J.M.

Carrascosa J.L. Large terminase conformational change induced by connector binding in bacteriophage T7. Wang et al., 2000 Wang I.N.

Smith D.L.

Young R. Holins: the protein clocks of bacteriophage infections. Previously, it was suspected that TerS and TerL were responsible for incorporating DNA into the phage capsid () and holin for the disruption of the bacterial cell wall (). Consistently with the phage titer ( Figure 3 B), we determined that the first infectious phages were generated within a time frame of 60–90 min.

The position at the end of the linear genome and their late expression, coincident with the increase in phage titer ( Figures 3 A and 3C, may indicate a function of these proteins, including the newly validated Y195, in delaying cell escape until phage assembly is complete ( Figures 3 A and 3C).

Rustad et al., 2017 Rustad M.

Eastlund A.

Marshall R.

Jardine P.

Noireaux V. Synthesis of infectious bacteriophages in an E. Coli-based cell-free expression system. Vogele et al., 2021 Vogele K.

Falgenhauer E.

von Schönberg S.

Simmel F.C.

Pirzer T. Small antisense DNA-based gene Silencing enables cell-free bacteriophage manipulation and genome replication. Garneau et al., 2010 Garneau J.E.

Dupuis M.È.

Villion M.

Romero D.A.

Barrangou R.

Boyaval P.

Fremaux C.

Horvath P.

Magadán A.H.

Moineau S. The CRISPR/Cas bacterial immune system cleaves bacteriophage and plasmid DNA. Kutter, 2009 Kutter E. Phage host range and efficiency of plating. It may thus be possible that these type IV genes depend not only on the T7 RNA polymerase but might exploit additional promoter-independent regulatory elements, e.g., RNAs amenable to conformational changes. Besides the identical sequence of protein expression in the cell-free system, the complete DNA replication was also shown, including cleavage of the concatemers at the specific pac sites for the T7 phage (). With respect to producing therapeutic phages, the titers obtained by phactory for PhiA1122, CLB-P3, and MUC-5 targeting Y. pestis, EAEC, and the MDR K. pneumoniae, respectively, were substantially higher (5-fold) than those achieved with amplification in bacterial liquid culture ( Figure 4 A). This may be explained by the fact that phactory provides a stably buffered solution with high metabolite concentration and is unaffected by phage-induced bacterial defense mechanisms such as CRISPR, cell death (), or differences in the relative efficiency of plating of host bacteria ().

Furthermore, we demonstrated an approach for transient phage engineering by temporarily integrating a luciferase and a polyhistidine tag in the capsid structure of the T7 phage. This form of phage engineering is modular and convenient, requiring only the addition of plasmids without altering the phage genome. The method also offers a high degree of biosafety because the non-genomic modification cannot be passed on to the offspring. This transient phage bioengineering at the protein level can thus become a rapid and adaptive method to systematically modify, e.g., tail-fibers to respond to MDR bacteria without risking the spread of genetically modified phages to the environment.

In summary, we have developed and deployed phactory, a cell-free production and analysis pipeline for bacteriophages, which enables the rapid, reliable, and host-independent production of a wide range of phages, including therapeutic phages. This versatile platform allows for the systematic analysis of phage proteomes and bioengineering of phages at the protein level, which will accelerate the biological development of tailor-made phage therapies against the growing number of pathogenic MDR bacteria.