We describe here the development of a new system of mutagenesis of cellular genes aimed at identifying genetic variants of interest for applications in basic and applied research. A crucial feature of the system is the possibility of performing, easily and in a controlled manner, a straightforward screening of the library in the human cell. This pragmatic approach allows to overcome the generation of false positives variants, a frequent problem encountered upon in vitro screening of libraries. The application of the Retrovolution system led to the identification of a dCK mutant that fulfils the long-sought feature of increasing sensitivity of tumour cells to Gemcitabine treatment, based on the presence of mutations unpredictable on a rational basis. The screening step was performed following a protocol that ensures that the isolated mutant sensitizes the cells independently from the integration site of the provirus, therefore constituting a good candidate as “suicide gene” in gene therapy.
The mutagenic potential of Retrovolution relies on the error-prone nature of the replication machinery of HIV-1, and on the frequent occurrence of recombination, that reshuffles the pre-existing mutations contributing to increase the diversity of the library. We show here that these two sources of genetic diversification prompt enough diversity as to lead to the identification of mutants of interest. With a relatively reduced number of cycles the system allowed the creation of a library in which each variant of the target gene was characterized by 1.5 mutated positions on average. A higher proportion of mutations would have increased the chances of producing a high percentage of non-functional proteins. An alternative approach for generating a library of mutants of cellular genes contained in lentiviral vectors would have been constituted first by a mutagenesis step of the cellular gene through conventional methods as error-prone PCR, followed by the insertion of the library in the lentiviral gRNA plasmid. With respect to this approach, the method we describe, while slower in the generation of a complex library, provides the advantage of circumventing the unavoidable loss of complexity inherent to the step of cloning the library in the lentiviral gRNA plasmid. Another way to accelerate the acquisition of mutations would have been the use of error-prone reverse transcriptases or mild mutagens. These procedures were not privileged, though, due to the inherent drawback of reduced yield of vector production 
Depending on the nature of the target gene, the experimental settings of Retrovolution can be adapted to introduce selection ongoing during the mutagenesis steps. In the present work this was not possible since we targeted a gene for which a negative selection, as the induction of cell death, needed to be applied. The targeting of the dCK gene was aimed, despite the intrinsic difficulty of the screening procedure, at the isolation of variants of a gene with a relevant biological interest for biomedical applications. The properties of the mutant identified, indeed, constitute the second major point emerging from this work.
Improving the efficiency of the currently clinically employed anticancer drugs and overcoming resistances arising during treatments is a major goal of cancer research. To improve the efficiency of the treatments with nucleoside analogues like Gemcitabine, gene therapy approaches aimed at sensitizing cancer cells through the introduction of a transgene that improves the efficacy of drug activation have been proposed. The hdCK is the best candidate transgene for this purpose as the kinase is responsible for the limiting step in intracellular activation of deoxycytidine analogues currently used in clinical treatments. The competition between the natural substrate of dCK and the prodrug inside the cell constitutes an obstacle to the efficient activation of the prodrug itself. The rational design of mutants in the catalytic site of the kinase attained the goal of improving phosphorylation of the prodrug in the past, but the mutants isolated displayed a concomitant, and stronger, increase in the efficiency of phosphorylation of the natural substrate 
. As a result, phosphorylation of the prodrug is expected to remain disfavoured with respect to the natural substrate and these mutants fail, as we confirmed (), to confer a drug sensitization phenotype to the cell. The G12 mutant, on the contrary, has an increased specificity for the phosphorylation of Gemcitabine due to the fact that it completely lost the ability to phosphorylate its natural substrate. This results in a reduction of the competition between the two substrates in vivo
and in a sensitization of the cells to the prodrug. The biochemical mechanism through which an increased sensitization to the drug is obtained is therefore opposed to what generally tried to obtain by engineering the dCK on a rational basis.
The screening for the isolation of the G12 mutant was performed on Messa 10K cells, where the most striking sensitization was observed. These cells constitute a model of tumoral cells acquiring resistance to deoxycytidine analogues due to a loss of dCK activity, a fundamental problem that has to be bypassed for an improvement of cancer treatment. Nevertheless, the G12 mutant has an effect also on tumour cells presenting an endogenous dCK activity, like HT29 and HEK-293T. Therefore, besides a potential application in cancer treatment, G12 also constitutes a promising suicide gene to use as negative selection marker in cells used in transplantation medicine. Suicide genes are inserted in cells transplanted for therapeutical purposes to specifically ablate them in the event they would undergo uncontrolled proliferation in vivo
. To this end exogenous genes are generally used, the most frequently employed being the Herpes Simplex Virus thymidine kinase gene in association with gancyclovir treatment. A problem of increasing relevance in clinical gene therapy, though, is constituted by the immune response raised by the patient against the protein encoded by the suicide gene 
. A suicide gene of human origin, as the mutant described here, is expected to have a faint possibility of being highly immunogenic and therefore represents an ideal candidate for this application.
Besides conferring an increased sensitivity to Gemcitabine through an unexpected mechanism, the G12 mutant is characterized by mutations localized in regions unpredictable on a rational basis. The human dCK is a globular, dimeric protein in which each monomer is formed by a core of 5 beta-sheets surrounded by 10 alpha-helices 
(). G12 dCK carries the mutations E171K, E247K, and L249M, the first two of which induce a charge change in conserved residues (). Amino acids 247 and 249 are located in the “base-sensing loop” (alpha-helix 10), which influences the folding of the protein upon binding of ATP or UTP as phosphate donor 
and, consequently, also affects substrate binding. Indeed, a variant containing only these mutations, which was identified during the screening of the library (mutant E8, ), slightly increased Messa10K cells sensitivity to Gemcitabine treatment, although with a cell death ratio that did not significantly differ from that of the wt-dCK (p>0.05). The marked sensitization observed with G12 thus requires the presence of the additional mutation, E171K. Residue 171 is located at the base of alpha-helix 7 that, with alpha-helix 4, is involved in the generation of the interface of the dCK dimer 
, and its mutation in a residue of opposite charge could potentially influence the efficiency of formation of the active dimer itself. Gel filtration analyses, though, revealed that the extent of dimerization of G12 does not differ from that of the wt-dCK (Figure S2
). It is therefore likely that, as observed for other proteins 
, the mutation rather triggers long-distance changes in the quaternary arrangement of the protein, possibly involving the region of the active site.
In conclusion, Retrovolution allowed the identification of a promising suicide gene to use in cancer treatment but also as negative selection marker in transplantation medicine by allowing isolating a variant of the dCK that would not have been predictable on a rational basis. The property of producing the library and allowing its screening directly in human cells, ensuring that each mutant emerging from the procedure will be relevant for modifying the phenotype of the cells, is central to these findings. Retrovolution thus opens new avenues for the modification of genes conferring complex phenotypes of interest for a broad field of applications in basic and applied research. With the exploitation of its combination of genetic flexibility and ability to deliver transgenes to human cells, the lifestyle of one of the most important pathogens of our recent history appears far from being fully exploited.