The increasing spread of MRGN is a global problem . Especially carbapenem-resistant K. pneumoniae (CRKP) pose a high risk in hospitals and other medical facilities. Liangfei Xu et al., reported 2-fold increased mortality among CRKP infected patients compared to those infected with carbapenem-susceptible K. pneumoniae (CSKP) .
Thus, alternative therapeutic approaches to antibiotic treatment are required. For CRKP infections, capsule-specific lytic bacteriophages are promising candidates [14, 17–19]. Knowledge of the K-type of the respective K. pneumoniae strains is therefore imperative.
This study reports the use of a rapid method to predict the K-type of Kp first described by Brisse et al. . In total, 163 CRKP strains from a previously described Tunisian clinical specimen collection  were capsule-typed. K-type K64 was not only identified as the most predominant K-type, but also as the K-type associated with the highest case/fatality rates of CRKP infected patients (> 70 % in patients treated at the MHT ICU). This fits recent observations, both regarding the high case/fatality [6, 7] as well as the increased virulence of capsule type K64 .
However, the following limitations to this analysis have to be considered: i. the clinical specimen collection available lacks power in differentiating death caused by CRKP from any other factors, and it is difficult to draw definitive conclusions from current evidence because of the residual confounding factors and small sample sizes in many studies; ii. the study is retrospective in nature and thus susceptible to selection bias.
Kp can be found ubiquitously not only in human tissues but also in non-clinical sources, such as soil, drinking or surface waters and sewage . This indicates that Kp phages should also be present in such environments and therefore easy to be isolated from nature.
In search of a potent tool to fight CRKP infections we looked for Klebsiella phages present at the MHT. Thereby we were able to obtain the lytic phage TUN1 from wastewater samples at the ICU using CRKP 7984 as a host strain. Based on BlastN analysis of the complete genome as well as morphologic characteristics phage TUN1 could be classified as bacteriophage of the genus Przondovirus, family Autographiviridae order Caudovirales, according to the newest species demarcation criteria set by the International Committee on Taxonomy of Viruses (ICTV) .
Phage TUN1 exhibited lysis specificity for Tunisian K64-type Kp strains. The replication cycle of 20 min and the burst size of 76 phage particles/cell are within the range described elsewhere [15–19]. The growth curve indicates that MOI = 1 reached approximately 60 % of Kp cells, leading to a three-phased growth curve. While not classical, the three phases are distinguishable and similar enough to each other, to support the interpretation as a growth curve with 3 steps of 20 minutes, each, and a burst size of 76 particles per cell. As K64 was the most predominant K-type at the MHT, especially in the ICU, and all experiments were performed at 37°C representing human body temperature, phage TUN1 appears to be a good candidate for phage therapy. Furthermore, unlike e.g. the mycobacterium abscessus phage used for therapy by Dedrick et al.  TUN1 is a purely lytic phage and therefore does not need removal of lysogenic elements form the genome.
However, with ~ 10 minutes, the latent time of phage TUN1 was found to be relatively short and the burst size with 76 phage particles/cell was on the lower end of productivity. Furthermore, the putative depolymerase seems to be very specific for K64 Kp strains.
Additional restriction factors other than capsule type are most likely responsible for phage TUN1 being able to lyse more than two thirds but not all Kp strains of capsule type 64.
For future phage therapy applications, all desirable features such as strong lytic activity, absence of virulence factors and high phage stability could be combined via genetic engineering. The use of genetic elements with depolymerase activity, targeting additional clinically relevant capsule types, may lead to phages with broader hostranges, and therefore more promising candidates for phage therapy based on phage TUN1.
Conducting initial proof-of-principle experiments, we were able to show that phage TUN1 can be rescued in a non-replicative host (E. coli). In vitro assembly of therapeutic bacteriophages in the absence of live bacterial hosts would simplify the regulatory approval as the most common factors of concern like the presence of endotoxins or the possibility of host-DNA contamination, would be absent by default. A collaboration with the Walter Reed Army Institute of Research (WRAIR) is aiming for the development of technical approaches to field deployable therapeutic phage cocktails [31–32].
The future goals of our current project at the Bundeswehr Institute of Microbiology are to i. identify and characterize bacteriophages against a broad range of Kp K-types found in German/Tunisian military and civilian hospitals; ii. develop technical approaches that allow genetic modification for the removal of unwanted genetic/lysogenic phage genome elements, possibly the addition of capsule specific factors and iii. work out a pipeline for therapeutic phage production, that implements the highest safety standards, facilitating regulatory approval.
Such a pipeline could then be applied in both civilian and military hospitals, and eventually extended to products able to withstand the harsher conditions in military field settings.