In the present study we have used lentiviral vectors expressing EGFP from three different promoters in the mouse hippocampus and have identified sub-field specific differences in transgene expression in various cell types of the GCL of the DG. Furthermore, we have characterized the cell types transduced by these lentiviral vectors, concluding that they target primarily NPC and immature neurons present in the SGZ and more immature layers of the GCL. Our observations suggest the existence of intrinsic differences in the permissiveness to lentivirus transduction among populations of granule cells of the GCL. In particular, we show for the first time that mature neurons of the outer granule cell layer do not express lentivirus-delivered transgenes, despite successful expression in other hippocampal cell types. Therefore, only adult-generated neurons may be target for lentivirus-mediated transgene delivery within the GCL.
The DG of the mammalian hippocampus is progressively constructed through a complex developmental program. Embryology studies have demonstrated that the GCL can be divided into an outer shell and an inner core, originated from separate embryonic progenitor pools. These progenitors generate first the outer shell followed by the development of the inner core by later-born granule cells . Therefore, the outer shell of the GCL is partially assembled during embryogenesis and the majority of dentate granule cells, located in the inner shell are generated after birth [36–38]. These and other observations have generated the hypothesis that, in contrast to the neocortex, the DG is built up following a life-long outside-in arrangement, where new cells are incorporated in the GCL following a downward gradient of positional cues .
In rodents, proliferative cells become largely confined to the SGZ at the base of the GCL after postnatal day 30 . Therefore, during the juvenile and adult periods the SGZ is the source of newly produced granule cells .
Several groups have shown heterogeneous functional properties of granule cells in the adult hippocampus. In particular, new neurons generated by adult neurogenesis display increased synaptic plasticity and increased excitability suggesting that maturation of the neuronal phenotype includes changes in membrane excitability and morphology, as well as the establishment of appropriate connectivity [24, 39, 40]. Interestingly, it has been proposed that functional and morphological differences among granule cells are a function of their location within the GCL rather than of their relative age [23, 41].
Herein we report that the three different lentivirus systems tested in this study, transduced mainly cells located in the SGZ and inner layers of the GCL. Cells expressing the reporter transgene EGFP one week after viral injection were mainly immature neurons expressing DCX. These observations resemble the EGFP expression profile achieved using MMLV-derived vectors that transduce only proliferating cells . Therefore, the initial cell population hit by the lentivirus was most probably a subpopulation of NPC that evolved into the neuronal lineage as judged by the predominance of DCX+ cells one week after transduction, similar to reports using MMLV-vectors [22, 24]. Moreover, retro- and lenti-viral vectors have been shown to target similar, although not completely overlapping, populations in the hippocampus . Therefore, the use of adeno-associated virus-derived vectors may be more adequate to target mature neurons of embryonic origin in the adult dentate gyrus .
An indubitable characterization of the cell type originally transduced by the lentiviral vector may request the use of cell type specific promoters restricted to NPCs . However, in the adult dentate gyrus, DCX is only expressed in cells contributing to adult neurogenesis and therefore can be used as a bona fide marker of newborn adult-generated neurons [44, 45].
Our observations are in agreement with the described ability of lentiviral vectors to transduce adult NPC in vivo . The presence of subpopulations of EGFP+ cells expressing the NPC marker nestin and Ki67, a cell proliferation marker expressed during the active phases of the cell cycle  emphasize our conclusions.
Moreover, the reduced numbers of EGFP+/NeuN+ cells found, their morphology and their location in the inner layers of the GCL, indicate that these EGFP+/NeuN+ cells have most probably originated from a population of immature cells originally hit by the virus.
Crucial to sustain these conclusions are our experiments in which we delivered the lentiviral vector to the SR, situated between the CA1 and the outer shell of the GCL. If the pattern of EGFP expression restricted to the inner layers of the GCL would have been a mere mechanical effect of the steric hindrance generated by the tightly packed structure of the GCL , the lentiviral vector should have been able to transduce cells in the outer layers of the GCL, when delivered to the SR. Conversely, we observed strong EGFP expression in cells within the ML and CA1, demonstrating adequate diffusion of the lentivirus across different cellular structures. Moreover, EGFP+ cells were homogenously distributed within the CA1 layer, with profuse EGFP expression in the soma, axons and dendrites of cells phenotypicaly resembling mature pyramidal neurons. Our experiments using a peptide-cy5 conjugate, depicted in Fig. 6, showed that this construct delivered into the SR, could effectively transduce the neurons located in the outer layers of the suprapyramidal blade of the GCL and beyond into the hilus and the infrapyramidal blade. These experiments demonstrated that stereotaxic injection to the SR permits effective delivery to the GCL.
Our data from the CA1 cells demonstrated as well that the CMV promoter is indeed able to promote transgene expression in mature postmitotic neurons, as previously described . These observations made us to conclude that, although the use of different (cell-type specific) promoters is useful to promote different patterns of transgene expression in the GCL, cells present in the outer shell of the GCL only scarcely express transgenes delivered by lentiviral vectors. Interestingly, the Synapsin I promoter rendered an EGFP expression profile more similar to that of the CMV promoter than to that of the CaMKII promoter, in accordance to its expression in earlier neuronal developmental stages . Therefore, although further experiments to investigate transgene expression mediated by different promoters at later times post-injection seems important to address the relevance of differential promoter use, it escapes the objective of the present study.
One potential drawback of the use of the CMV promoter may be its potential activation in astrocytes short time after injury, described in the cerebral cortex and caudate-putamen . Nevertheless, this activation could be dependent on virus titers and other factors such as the particular CMV sequence used and the time after the injection . For the interpretation of the data presented herein it is worth to take into account that sections surrounding the injection site were routinely discarded.
Specific transgene silencing after lentiviral vector-mediated delivery has been described before . Although we can not exclude from this set of experiments the possibility that transgene expression driven by the three promoters used in this study were selectively silenced in mature neurons present in the outer layers of the GCL, the fact that the CMV promoter was able to promote expression in cells of the CA1 makes this possibility unlikely.
Overall, our observations are in agreement with previous reports showing that lentiviral vectors can successfully transduce mitotic and postmitotic cells [26, 46, 52]. However, the exact nature of the cell types and hippocampal sub-fields targeted by lentiviral vectors remains controversial. Previous reports did not find sub-field specific differences in GFP expression. This could be due to technical differences such as the use of different GFP variants and constructs, analysis of the samples at different time points after stereotaxic injection or differences in the CMV promoter sequence used to control transgene expression [6, 12]. Nevertheless, the disparity in EGFP expression reported herein between cells located in the inner or outer layers of the GCL seems to be a function of intrinsic differences between cells generated by embryonic or adult neurogenesis. In this context, disparities in transgene expression in granule cells, depending on their relative location within the GCL and their progression into the neuronal differentiation program, emphasize the heterogeneity between newly adult-generated neurons and pre-existing ones, probably originated during embryonic and/or early postnatal development.
Although further experiments will be required to clarify the exact nature of this heterogeneity among granule cells of the DG, regarding their permissiveness to lentivirus-delivered transgene expression, one possible explanation could be the differential expression of receptor proteins that recognize pseudotyping proteins by subpopulations of granule cells. However, VSV-G pseudotyped viruses have been shown to effectively transduce cells within the GCL of the DG [6, 12]. This suggests that, although pseudotyping proteins can influence transduction efficiency and tropism to hippocampal cell types [11, 25], the receptors for VSV-G glycoprotein are present in granule cells of the DG. Moreover, transgene expression from VSV-G pseudotyped lentivirus is pantropic in the rat brain, labelling a variety of glial and neuronal cell types depending on the promoter used to control transgene expression .
Interestingly, even though cell mitosis is not a requisite for integration, transduction efficiency of lentiviral vectors is dependent on cell-cycle progression of target cells, with cells actively growing or arrested in phases other than G0 being more efficiently transduced in vivo [26, 53–55]. As demonstrated here, lentivirus transduced EGFP+ cells are in their vast majority positive for progenitor (nestin), astrocyte (GFAP), proliferation (Ki67) and immature neuron (DCX) cell markers. Furthermore, Schmetsdorf et al  have demonstrated that cells from distinct hippocampal fields, including CA1, CA3 and DG, express completely different repertoires of cell cycle-related proteins. Therefore, although a more thorough elucidation of the factors regulating lentivirus transduction of postmitotic granule cells is beyond the scope of this article, our observations demonstrating lentivirus-mediated transgene expression in NPC and immature neurons suggest that cell-cycle progression is an important determinant in lentivirus transduction efficiency of hippocampal granule cells in vivo.