Herpes simplex virus thymidine kinase (HSVTK) with ganciclovir (GCV) is currently the most widely used suicide gene/prodrug system in cancer gene therapy. A major limitation in this therapy is the inefficient activation of GCV by HSVTK to its active antimetabolites. We described earlier two strategies to overcome this limitation: (1) generation of HSVTK mutants with improved GCV activation potential and (2) construction of a fusion protein encoding HSVTK and mouse guanylate kinase (MGMK), the second enzyme in the GCV activation pathway. As a means to further enhance GCV activation, two MGMK/HSVTK constructs containing the HSVTK mutants, mutant 30 and SR39, were generated and evaluated for their tumor and bystander killing effects in vitro and in vivo. One fusion mutant, MGMK/30, shows significant reduction in IC(50) values of approximately 12 500-fold, 100-fold, and 125-fold compared with HSVTK, mutant 30 or MGMK/HSVTK, respectively. In vitro bystander analyses show that 5% of MGMK/30-expressing cells are sufficient to induce 75% of tumor cell killing. In an xenograft tumor model, MGMK/30 displays the greatest inhibition of tumor growth at a GCV concentration (1 mg kg(-1)) that has no effect on wild-type HSVTK-, MGMK/HSVTK-, or mutant 30-transfected cells. Another fusion construct, MGMK/SR39, sensitizes rat C6 glioma cells to GCV by 2500-fold or 25-fold compared with HSVTK or MGMK/HSVTK, respectively. In vitro analyses show similar IC(50) values between cells harboring SR39 and MGMK/SR39, although MGMK/SR39 seems to elicit stronger bystander killing effects in which 1% of MGMK/SR39-transfected cells result in 60% cell death. In a xenograft tumor model, despite observable tumor growth inhibition, no statistical significance in tumor volume was detected between mice harboring SR39- and MGMK/SR39-transfected cells when dosed with 1 mg kg(-1) GCV. However, at a lower dose of GCV (0.1 mg kg(-1)), MGMK/SR39 seems to have slightly greater tumor growth inhibition properties compared with SR39 (P< or =0.05). In vivo studies indicate that both mutant fusion proteins display substantial improvements in bystander killing in the presence of 1 mg kg(-1) GCV, even when only 5% of the tumor cells are transfected. Such fusion mutants with exceptional prodrug converting properties will allow administration of lower and non-myelosuppressive doses of GCV concomitant with improved tumor killing and as such are promising candidates for translational gene therapy studies.