Recombinant Reporter Phage rTUN1::nLuc Enables Rapid Detection and Real-Time Antibiotic Susceptibility Testing of Klebsiella pneumoniae K64 Strains

The emergence of multi-drug-resistant Klebsiella pneumoniae (Kp) strains constitutes an enormous threat to global health as multi-drug resistance-associated treatment failure causes high mortality rates in nosocomial infections. Rapid pathogen detection and antibiotic resistance screening are therefore crucial for successful therapy and thus patient survival. Reporter phage-based diagnostics offer a way to speed up pathogen identification and resistance testing as integration of reporter genes into highly specific phages allows real-time detection of phage replication and thus living host cells. Kp-specific phages use the host’s capsule, a major virulence factor of Kp, as a receptor for adsorption. To date, 80 different Kp capsule types (K-serotypes) have been described with predominant capsule types varying between different countries and continents. Therefore, reporter phages need to be customized according to the locally prevailing variants. Recently, we described the autographivirus vB_KpP_TUN1 (TUN1), which specifically infects Kp K64 strains, the most predominant capsule type at the military hospital in Tunis (MHT) that is also associated with high mortality rates. In this work, we developed the highly specific recombinant reporter phage rTUN1::nLuc, which produces nanoluciferase (nLuc) upon host infection and thus enables rapid detection of Kp K64 cells in clinical matrices such as blood and urine. At the same time, rTUN1::nLuc allows for rapid antibiotic susceptibility testing and therefore identification of suitable antibiotic treatment in less than 3 h.

T he ESKAPEE bacteria (Enterococcus faecium, Staphylococcus aureus, Klebsiella pneumoniae, Acinetobacter baumanni, Pseudomonas aeruginosa, Escherichia coli, and Enterobacter species) are responsible for the majority of nosocomial infections and typically possess multi-drug resistance (MDR). Among these, the opportunistic pathogen K. pneumoniae (Kp) poses a particular threat for public health. Ubiquitously occurring as a commensal, Kp can be found among gastro-intestinal microbiota, in the respiratory tract, and on the skin. As a facultative pathogen, Kp can cause a variety of severe nosocomial diseases such as pneumonia, sepsis, wound, and urinary tract infections (UTIs). In recent years, the number of community-acquired cases of Kp-caused pneumonia and meningitis has dramatically increased. 1 Together with the emergence of acquired MDR against broad-spectrum antibiotics (ABs) and the natural resistance mechanisms of the bacterium, treatment of Kp infections has become a global healthcare challenge. 2−4 Infections with carbapenem-resistant Kp are associated with serious symptoms and high mortality rates. 5,6 Therefore, rapid diagnosis of the infection and detection of resistance markers are crucial for patient survival. Classical approaches to identify carbapenem-resistant Kp rely on bacterial cell culture with subsequent biochemical profiling and therefore provide results only after a few days. While nucleic acid-based tools such as diagnostic real-time PCR or loop-mediated isothermal amplification enable fast and specific detection of the pathogen, the established assays mainly target carbapenemresistance genes and therefore cannot distinguish between living cells and DNA residues present in the sample. 7−9 Diagnostic bacteriophages (phages) represent a promising alternative approach. In addition to their therapeutic potential, phages are suited for the identification of bacteria due to the viruses' high specificity. Phage amplification assays have been in practice for centuries for a variety of pathogens such as Bacillus anthracis 10 or Yersinia pestis. 11 Such techniques include antibody-based detection of phages using ELISA 12 or detection of amplified phage DNA. 13 Another option is to exploit the bacteriolytic activity of phages and detect released cellular components such as ATP. 14 The most promising methodology, however, is the use of reporter phages, which have a reporter molecule either directly attached to the virion surface or produced during phage replication. 15−18 While most phages of Gram-negative bacteria utilize surface structures such as lipopolysaccharides, teichoic acids, or proteins as host receptors for adsorption, 19 the specificity of Kp phages is mostly determined by the different Kp capsule polysaccharides (CPSs). CPSs represent a major virulence factor of Kp as the thick layer protects the bacterial cell from phagocytosis and prevents complement binding during host response. In addition, the CPS forms a physical barrier toward ABs as it hampers diffusion into the bacterial cell. 20−22 To date, about 80 different Kp capsule serotypes (K-types) have been described. 23 Since Kp-specific phages are usually capable of infecting only one or a few capsule types, the selection of a suitable reporter phage is crucial for reliable diagnostics. In Tunisia, for example, the majority of nosocomial infections at the Military Hospital of Instructions in Tunis (MHT) are caused by a Kp strain featuring capsule type 64 (K64). These Kp K64 infections are associated with high case-fatality rates, especially among intensive care patients. 24  In Vitro DNA Assembly of Synthetic Phage Genomes. For isolation of genomic DNA (gDNA) of phages, a MasterPure Complete DNA and RNA Purification Kit (Lucigen, Wisconsin, USA) was used according to the manufacturer's protocol. Prior to assembly of the synthetic TUN1 WT genome, TUN1 WT gDNA was used as PCR template to amplify five fragments. This was achieved by using the primer pairs TUN1 F1 + R1 (= fragment 1), TUN1 F2 + R2 (= fragment 2), TUN1 F3 + R3 (= fragment 3), TUN1 F4 + R4 (= fragment 4), and TUN1 F5 + R5 (= fragment 5). Primers were designed to generate 15−25 bp overlaps between adjacent fragments to ensure efficient assembly. The circular approach was done in the same way, with the difference of replacing fragment 1 and fragment 5 by Fragment 5_1, amplified using the primers TUN1 F5 + R1 ( Figure  S1A).
If not stated differently, the four TUN1 fragments of the circular approach described above were used for all further TUN1 DNA assemblies.
To generate the triple mutant TUN1 Δhpgc123, fragment 3 and fragment 4 were amplified from TUN1 WT gDNA using the primer pairs mentioned above. Amplification of modified fragment 1 was achieved by using the primers TUN1 F5 and gp6 R and gDNA of TUN1 Δhpgc1 as a PCR template. Furthermore, fragment 2 was amplified from TUN1 Δhpgc2 gDNA by using the primer pair TUN1 gp9 F + R2 ( Figure S1E).
Phage Rebooting. Phage assembly and rebooting were achieved in a non-replicative host by transforming chemically competent E. coli 10-beta cells (New England Biolabs) with 5 μL of the assembly mix according to the protocol provided by the manufacturer. After transformation, 950 μL of LB medium was added and the cells were grown for 2 h at 37°C with horizontal shaking at 120 rpm. Subsequently, the cells were harvested for 5 min at 3000g and the supernatant (lysate), containing rebooted phages, was stored at 4°C until further use.
Plaque-Assay. To check for functionality of (recombinant) TUN1, a fresh culture of Kp was grown to OD 600 = 0.4−0.6 and then 350 μL of the culture was mixed with 2.5 mL of hand-warm soft agar [LB + 0.6% agar (w/v)] together with 100 μL of phage stock/ transformation supernatant. Then, the mixture was evenly distributed on an LB-agar plate and, after solidification, the plates were incubated at 37°C for 16 h. The next day, plaque forming units (PFUs) were calculated and the plaque morphology was analyzed.

Bacterial Growth.
To determine the bacteriolytic activity of TUN1 phages on Kp strains, bacterial growth was measured comparing samples with and without phages. For this purpose, 100 μL of the respective Kp culture (10 7 CFUs/mL) was mixed with either 100 μL of TUN1 phage lysate (1 × 10 3 PFUs/mL) or 100 μL of LB medium as a control in a 96-well plate with clear bottom. Plates were incubated at 37°C with shaking, and bacterial growth (OD 600nm ) was measured every 30 min. rTUN1::nLuc Functionality. To measure luciferase activity on plaque containing LB-agar plates, 5 μL of the diluted (1:50, in PBS) nanoluciferase substrate 2-furyl methyl-deoxy-coelenterazine (Furimazine; Promega, Fitchburg, USA) was dripped onto the respective plaques. After incubation at RT for 3 min, the plates were analyzed for luminescence signals using a ChemiDoc Imaging System and Image Lab 6.0.1 software (Bio-Rad Laboratories Inc., Hercules, USA). The exposure time was set to 5 s. The obtained luminescence signals were then merged with an image of the plate for identification of plaques featuring luciferase activity.
To analyze luminescence in liquid culture medium over time, 100 μL of a fresh Kp culture (1 × 10 7 to 1 CFUs/mL) mixed with 100 μL of rTUN1::nLuc stock (1 × 10 3 PFUs/mL) was transferred into a cavity of a 96-well plate and supplemented with 2 μL of the substrate. The plate was then incubated at 37°C under continuous shaking (420 rpm) using a Varioskan Lux plate reader (Thermo Fisher Scientific, Waltham, USA). Bacterial growth (OD 600nm ) and luminescence (exposure time: 100 ms) were measured every 30 min for up to 16 h.
Clinical Matrices. For rTUN1::nLuc-based Kp detection in clinically relevant matrices, the reporter phage was added to Kpspiked urine and blood samples. Fresh urine was sterile-filtered (0.22 μm luer lock syringe filter, Merck Millipore, Cork, Ireland) and spiked with the TUN1 host Kp 7984 (final conc. 1 × 10 7 CFUs/mL). Subsequently, bacterial growth and luminescence were measured in a plate reader every 30 min for 7 h at 37°C as described above.
To test the clinical matrix blood, 5 mL of defibrinated sheep blood, spiked with Kp 7984 (final conc. 1 × 10 7 CFUs/mL), was used. Prior to measurements, the serum was collected as described elsewhere. 29 Briefly, the 5 mL spiked blood was injected into a 50 S-Monovette (Sarstedt, Nuembrecht, Germany) and centrifuged at 2000g for 15 min. The serum containing the bacterial cells was then transferred into a fresh tube and centrifuged again. After discarding the supernatant, the bacteria were resuspended to the original volume using LB medium. Subsequently, bacterial growth and luminescence in the presence and absence of rTUN1::nLuc were measured as described above.
Data Analysis. Sequence data were analyzed using Geneious Prime (Version 2021.1.1). All Varioskan data were analyzed using the Phylogenetic Classification TerL. In order to predict TUN1's DNA packaging, phylogenetic analysis of terminase large (TerL) subunit was performed using ClustalOmega for sequence alignment and Neighbor-Joining analysis was performed for tree reconstruction with a bootstrap of 1000 replicates using Geneious Prime.
Whole Genome Comparison. The whole genome comparison of Kp phage TUN1 and E. coli phage T7 was conducted using the Geneious Prime plugin MAUVE alignment.
■ RESULTS Successful Rebooting of TUN1 from Synthetic Phage Genomes Requires Circularized DNA. Generation of TUN1 (an autographivirus featuring a 41 kb sized genome) reporter phages was aimed to be achieved by synthetic biologydriven genetic engineering. As a proof of concept, the technical feasibility of in vitro assembly and phage rebooting had to be demonstrated prior to construction of reporter phages. For this, the TUN1 wild-type (WT) DNA was PCR-amplified as five or four overlapping fragments and assembled in vitro generating linear or circularized synthetic phage genomes, respectively. Synthetic WT phage TUN1 was successfully rebooted by transforming the non-replicative host E. coli NEB stable with circularized synthetic constructs, while transformation of linear constructs did not result in any rebooted phages ( Figure S2). In a next step, we aimed at integrating the nanoluciferase gene (nLuc) fused to a ribosomal binding site (RBS) into the synthetic circular construct. As the insertion site for the RBS-nLuc construct, the intergenic region between the major (gp38) and minor (gp39) capsid genes was selected ( Figure 1) as it qualifies for efficient transcription of the reporter due to the strong cps promoter regulating this operon. 30 After in vitro DNA assembly and phage rebooting, however, no plaques were observed on the Kp K64 strain 7984, indicating that no functional recombinant reporter phage particles had been generated. Thus, while synthetic WT phage can be assembled from DNA fragments and rebooted, a recombinant phage with the additional reporter gene nLuc cannot.
Genome Reduction of TUN1 Enables Insertion of nLuc-Reporter. One possible explanation for unsuccessful Figure 1. Schematic overview of TUN1 reporter phage construction. For construction of a TUN1 reporter phage, all synthetic DNA fragments (F1-5_1) were amplified with overlapping ends between consecutive segments (depicted in matching colors). The light-orange dot depicts the fusion site of F1 and F5 to F5_1, enabling circularization of the synthetic phage genome. After in vitro DNA assembly, the circular DNA was used to transform E. coli cells in which the recombinant phages were rebooted. Cells were lysed to release virions from the phage insusceptible host and the lysate used to infect Kp. Next, the recombinant phage was in vivo amplified in its host Kp forming plaques on a lawn in the bacteria. Functional reporter phages were screened by measuring plaque luminescence after addition of the luminogenic substrate.

ACS Sensors pubs.acs.org/acssensors
Article reporter integration was that the capacity of DNA encapsulation in the nascent virions is already reached with the size of the WT genome of TUN1. To address this challenge, we next aimed to reduce the TUN1 WT genome to provide space for insertion of the nLuc construct. As TUN1 comprises multiple ORFs encoding proteins of unknown functions, 24 these hypothetical genes (hpg) without any known function represent ideal candidates for targeted gene deletion. We decided to eliminate the operons gp2-4, gp7-8, and gp10-11 ( Figure 2A) as these clusters were predicted to be regulated by single promoters. For this, we used the same method as described for TUN1 WT in vitro assembly above but excluded the selected hpg clusters (hpgc) during fragment amplification ( Figure S1), resulting in TUN1 Δhpgc1, Δhpgc2, and Δhpgc3 featuring genome reductions by 518, 529, and 738 bp, respectively ( Figure 2B). As all hpgc deletion variants resulted in functional phages, a triple deletion strain (TUN1 Δhpgc123) was generated by using the single-cluster deletion mutants as PCR templates for subsequent synthetic phage genome rebooting. This resulted in a reduced genome size by 1785 bp ( Figure 2B). Analysis of the recombinant phages for their lytic activities against their host Kp 7984 in growth experiments yielded no difference between TUN1 WT and the deletion variants ( Figure 2C). While the plaque sizes of TUN1 Δhpgc1 and Δhpgc2 on Kp 7984 were similar to that of TUN1 WT, those of the deletion variants TUN1 Δhpgc3 and Δhpgc123 were slightly decreased. As the size of genome reduction in TUN1 Δhpgc2 (529 bp) matched the size of the nLuc construct (526 bp), this variant was used for reporter gene insertion in the next step.
Using the genome-reduced TUN1 Δhpgc2 as a template, we were able to insert the nLuc construct and thus to generate phage rTUN1::nLuc. Just like TUN1 WT and TUN1 Δhpgc2, rTUN1::nLuc also formed distinct clear plaques surrounded by translucent halos when tested in plaque assay on Kp 7984. To ascertain the functionality of the luciferase reporter, rTUN1::nLuc plaques were treated with furimazine, the substrate of nanoluciferase, and subsequently analyzed for luminescence. Indeed, all plaques emitted light upon addition of furimazine (Figure 3), and therefore, rTUN1::nLuc can be used for luminescence-based detection of Kp K64. rTUN1::nLuc Enables Highly Sensitive Kp K64 Detection. Plaque assays depend on bacterial growth on agar plates and are thus too time-consuming for Kp detection in clinical diagnostics. Therefore, the next step was to test the applicability of the reporter phage rTUN1::nLuc for rapid luminescence-based Kp detection in liquid culture. For this, a liquid culture of Kp 7984 (1 × 10 6 CFUs per well) was infected with 1 × 10 2 PFUs (per well) of the reporter phage (retrieved from lysate) and bacterial growth and luminescence were measured over time. We observed an immediate strong increase of luminescence reaching a maximum of 10 8 relative light units (RLUs) after only 2 h of incubation. This increase in luminescence correlated well with the start of impaired bacterial growth compared to the control containing only Kp without rTUN1::nLuc ( Figure 4A). Most probably caused by autodegradation of the substrate, some luminescent background noise was also detected for Kp 7984 without phage. Therefore, the threshold for positive luminescence results was set to RLUs ≥ 10 3 . When different rTUN1::nLuc titers were tested, we observed a high baseline luminescence (≥10 3 RLUs) for phage titers greater than 1 × 10 3 PFUs/well, even in phage-only control samples. This is most probably due to carryover of nanoluciferase during preparation of rTUN1::nLuc phage stocks. An initial titer of 1 × 10 2 PFUs/well, on the other hand, qualified as the ideal phage concentration as the baseline was comparable to the one produced by the no-phage control ( Figure S3).

ACS Sensors pubs.acs.org/acssensors Article
To investigate the sensitivity and thus the detection limit of the rTUN1::nLuc-based reporter phage assay, a dilution series of Kp 7984 (1 × 10 6 to 0.1 CFUs per well) were tested using a starting concentration of 1 × 10 2 PFUs/well of the reporter phage. Both the increase in detectable bacterial growth and the luminescence signal peak were delayed by 2−4 h with every dilution step ( Figure 4B). The results revealed that as few as one initial cell per well was sufficient to yield a luminescent signal with its peak at 10 7 RLUs after approx. 9 h. However, some of these wells with 10°Kp cells turned out negative (in luminescence and optical density), most probably because random distribution of single cells results in empty wells in some cases.
Kp K64 Cells Can Be Detected Directly in Clinical Matrices Using rTUN1::nLuc. Kp can cause severe infections with fatal clinical outcomes, including UTI or sepsis. Therefore, urine and blood represent relevant clinical matrices for Kp diagnostics. Using our reporter phage rTUN1::nLuc on

ACS Sensors pubs.acs.org/acssensors
Article both Kp-spiked (1 × 10 6 CFUs/well) urine and blood samples, respectively, distinct luminescence peaks could be measured. Conversely, no matrix-associated increase in background luminescence was detected ( Figure 4C). While the results for spiked urine samples were almost identical to those measured in spiked growth medium, the luminescence peaks retrieved from spiked blood samples appeared to be timeshifted by 2 h now, correlating with the results of an initial Kp concentration of 1 × 10 4 CFUs/well. This can most easily be explained by a loss of 10 2 bacteria per mL during serum collection from spiked blood.  Figure 4D and Figure S4). rTUN1::nLuc Enables Rapid, Real-Time Antibiotic Susceptibility Testing of Kp. The main challenge for treatment of Kp infections is the increasing number of MDR Kp strains. For instance, Kp 7984, the strain used in this work, has already been shown to be resistant to the first-line AB ertapenem. 28 Testing of antibiotic susceptibilities of Kp strains derived from patient samples is therefore crucial to ensure successful therapy. However, these tests require a pure culture of the infection-causing strain and often contain long incubation steps. To accelerate this process of selecting the adequate ABs for an efficient therapy, we tested whether rTUN1::nLuc is also suitable for the real-time, rapid AB susceptibility testing. For this, we used the experimental settings of the luminescent reporter phage assay (10 6 CFUs/ well Kp 7984 + 10 2 PFUs/well rTUN1::nLuc, shown in Figure  4A) and supplemented the growth medium with various ABs of different classes using their minimal inhibitory concentration (MIC) at breakpoints determined for Enterobacteriaceae (Table S1). We observed no increase in reporter phageassociated luminescence in the presence of gentamicin (Gent), chloramphenicol (Cmp), imipenem (Imp), meropenem (Mer), and amoxicillin/clavulanic acid (Amc) ( Figure 5A). In contrast, we detected a strong luminescence increase resulting in a distinct peak after 2 h, comparable to that of the Kp 7984 + rTUN1::nLuc-control without ABs, when levofloxacin (Lev), ceftazidim (Caz), streptomycin (Str), tigecyclin (Tgc+), ciprofloxacin (Cip), trimethoprim/sulfamethoxazole (T/S), and ertapenem (Ert) had been added ( Figure  5B). Growth experiments of Kp 7984 (without phage) in the presence of ABs further revealed that Gent, Cmp, Imp, Mer, and Amc inhibited bacterial growth, whereas Kp 7984 growth was not affected by the presence of Caz, Str, Tgc+, Cip, and T/ S ( Figure S5). Thus, Kp 7984 can be considered sensitive to Gent, Cmp, Imp, Mer, and Amc and resistant to Caz, Str, Tgc +, Cip, T/S, and Ert. In contrast to time-consuming classical growth experiments and MIC testing, our luminescent reporter phage assay using rTUN1::nLuc not only enables rapid and sensitive detection of Kp K64 strains directly from clinical samples but also allows for simultaneous, real-time antibiotic susceptibility testing within just a few hours.

■ DISCUSSION
The enormous increase of antibiotic resistance in bacteria has been estimated to lead to an excess mortality of 10 million persons per year by 2050, mostly caused by nosocomial infections. 31,32 The multiple antibiotic resistance crisis constitutes a tremendous threat to global health and has been declared a global health emergency 33 − a view supported by the World Health Organization. Rapid pathogen detection and antibiotic resistance screening could assist with nonempirical therapy and potentially save lives. Established methods for Kp detection are time-consuming and necessitate empirical antibiotic therapy in urgent cases 34 with more unfavorable outcomes. Here, we demonstrate that reporter phage-based diagnostics could be a promising alternative as integration of reporter genes into highly specific phages enables real-time detection of phage replication and thus living host cells. Besides fluorescent proteins 35 and hydrolyzing enzymes such as β-galactosidase, 36,37 luciferases can be used as reporters. 30,38−40 Since the discovery of luciferin−luciferase systems almost a century ago, 41 several luciferase enzymes have become available to researchers. Among them, the nanoluciferase enzyme (nLuc) features a comparably small size (19.1 kDa), fast substrate turnover rates and exhibits high stability under various buffer conditions and was therefore selected as a reporter candidate for this study. 42,43 Inserting reporter genes or any additional nucleotide sequence into phage genomes can be challenging. As DNA length and capsid size and thus internal pressure in the capsid correlate, 44 the DNA encapsulation capacity of most phages is limited. Hence, integration of additional genes can negatively affect capsid stability and therefore lead to severe impairments in phage assembly or replication. 45,46 DNA packaging of tailed phages relies on the recognition of specific sequences (cos-or pac-sites) by the terminase protein, followed by DNA cleavage. Depending on the phage type, this cleavage can be either sequence-specific (e.g., T7 or λ) or nonspecific (e.g., P22 or SPP1). For the latter, the cleavage does not occur at the end of the phage genome (so-called termination cleavage) but follows the mechanism of headful packaging where nonspecific DNA cleavage by TerL (terminase large subunit) is induced by increased capsid pressure. As a result, phages using the headful system often tolerate a genome size of up to 110%, while the others will not. 47 Previously, genome sequencing of TUN1 already revealed a linear genome with short direct terminal repeats. 24 In addition to that, phylogenetic analysis of the TerL sequence can also be used to predict a phage's DNA packaging system as the protein initiates DNA packaging and cleavage. For TUN1, the results of this analysis suggested that the phage belongs to the "T7-like group with directed terminal repeats" and not to the "P22-like headful group" ( Figure S6). Therefore, the fact that our nLuc reporter construct could not be inserted into the full-length TUN1 WT genome can most likely be explained by its sequence-specific DNA packaging mechanism in combination with a genome size that has already reached the maximum capsid capacity. The fact that Pulkkinen et al. successfully integrated nLuc into the T7 WT genome 48 does not necessarily contradict this hypothesis. TUN1 and T7 WT share 45% pairwise identity; however, the TUN1 WT genome comprises about 1200 bp more than T7 ( Figure S7). Therefore, the capsid capacity of TUN1 WT, compared to T7, might already be reached, limiting the integration of further nucleotides. It has already been described that insertion of additional genetic information into the genomes of other Podoviridiae or Autographiviridae requires prior genome reduction in order to overcome the problem of limited capsid capacity. 49,50 In this work, we were able to minimize the TUN1 genome by up to 4.3% (TUN1 Δhpgc123) and, at the same time, demonstrated that the hypothetical genes gp2-4, gp7-8, and gp10-11 are not essential for a successful lytic cycle of TUN1. Using the genome-reduced TUN1 Δhpgc2 as a template, insertion of the RBS-nLuc reporter construct was successful. The resulting, luminescent reporter phage, rTUN1::nLuc, enabled specific and sensitive detection of Kp K64 cells, exhibiting a limit of detection (LOD) of only one CFU per well.
Other Kp-specific nanoluciferase-based reporter phages have already been described, such as Mcoc and 8M7. These phages specifically detect Kp K21 cells with a comparably low LOD of only 10 CFUs/well and 100 bacteria per 100 mg feces, respectively. 51 Other nLuc-based reporter phage assays exhibited similar sensitivity. For instance, Pulkkinen et al. constructed a T7-nLuc phage, which was able to detect 47 E. coli cells per well. 48 Further, ΦV10nluc phage-based E. coli detection exhibited an even lower LOD of only 5 CFUs/well. These assays, however, were not tested on clinically relevant matrices. In contrast, our novel rTUN1::nLuc-based reporter phage assay enables rapid detection of Kp cells from blood and urine samples, exhibiting no matrix effects such as luciferase quenching or an unspecific increase in background luminescence. While the assay detected Kp K64 in urine with the same sensitivity as pure bacterial culture, luminescence peaks retrieved from spiked blood samples indicated a loss of bacterial cells during sample preparation. This was most probably caused by the washing steps during serum collection rather than by blood associated matrix effects as comparable studies have shown that similar optimization steps (serum separation and washing) of spiked blood samples resulted in a 20−50% reduction in bacterial concentration. 52,53 The current cutoff for diagnosis of UTIs, one of the most common diseases caused by Kp, is 1 × 10 5 CFUs/mL urine. 54 Using our novel reporter phage assay, this minimal pathological Kp load can be detected within 4 h directly from the patient's urine. Compared to gold standard methods like PCR, rTUN1::nLuc only detects viable Kp cells, which can be crucial, for example, for monitoring therapy success where differentiation between living and dead cells is essential.
In addition to that, our approach allows for rapid, real-time assessment of antibiotic resistances (real-time antibiogram) and thus facilitates treatment of Kp K64 infections. Here, susceptibility against therapeutically relevant ABs can be tested simultaneous to initial Kp detection from clinical samples, providing vital information for assessing treatment options in a few hours (depending on the Kp load in the sample). However, as our Kp collection only comprises highly resistant strains of MRGN (multi-drug-resistant Gram-negative) groups 3 and 4, 55 we were not able to test Kp K64 strains susceptible to bacteriostatic ABs such as Str, Caz, Tgc+, or T/S.
In the case of an infection with an MDR Kp K64 strain with no promising treatment options left, rTUN1::nLuc could also be used for companion diagnostics for phage therapy as positive luminescence readout from reporter phage assays may indicate infectivity of the respective wild-type phage which could then be used for phage therapy. 18

■ CONCLUSIONS
Our results clearly demonstrate the enormous diagnostic capabilities of reporter phages. However, for broad applicability directly from clinical samples, such as urine from patients suffering from UTIs, reporter phages need to be constructed not only for all locally occurring Kp K-types but also for all clinically relevant pathogens. This would then allow not only the rapid identification of the infection-causing bacterial strain from unknown samples within a few hours but also the simultaneous generation of a real-time antibiogram of the infection-causing strain as well as the selection of potential candidates for phage therapy. ■ ASSOCIATED CONTENT
Additional experimental details and materials including graphs of additional experiments like host-specificity, antibiotic testing, phylogenetic classification of TUN1 TerL, and whole genome alignment of phages TUN1 and T7 (PDF)