In the course of adult hippocampal neurogenesis, the newly generated neuronal cells pass through a stage that is characterized by transient DCX expression. In the present study we have added a number of important insights to the knowledge about this crucial phase, which links the precursor cell stage with a postmitotic immature neuronal stage. We show that the phase of increasing dendritic maturation can have variable length but that this pattern, despite its variability, is not influenced by either positive or negative neurogenic stimuli. We propose that these results indicate that although newborn neurons follow a rather fixed pattern of dendritic development and maturation (as captured by the categories A – F), initiation of this development is not only dependent on proliferation per se and exit from the cell cycle, and that regulation of neurogenesis does not overtly include aspects of dendritic maturation.
Whereas about 22% of the DCX positive cells were in cell cycle, 68% were showing a co-localization with the strictly postmitotic marker calretinin. This implies that at the given time-point, the remaining 10% of DCX-positive cells are in the progression from a progenitor cell to an early postmitotic stage. The most mature cells (based on dendritic morphology), were always CR-positive. But at one day after BrdU, some CR-positive cells without processes were found in the SGZ. About one third of the DCX-positive cells represent progenitor cells (type-2b and type-3 cells), whereas the remainder are immature postmitotic neurons. This ratio might not be fixed and may be subject to regulation or age-dependent shifts. Within the group of postmitotic cells the time point of CR expression after exit from the cell cycle also differed.
We used DCX as a marker here. DCX protein binds to elements of the cytoskeleton, including intermediate filament nestin, which is expressed in precursor cells. In general, DCX expression has been primarily brought into connection with neuronal migration, although other aspects of pathfinding including a role in the growth cones of neurites and in synapse formation have been suggested . Because both processes involve cytoskeletal dynamics and DCX binds to the microtubular apparatus it is tempting to speculate that this association is causal. Our finding that although dendrite development follows a predictable pattern it is still variable in onset and duration, suggests that additional levels of regulation exist that become effective after DCX has started to be expressed. Neurogenic stimuli that act on precursor cells, such as those studied here, do not seem to belong to these regulators. We propose that regulation of dendritic maturation is influenced by those stimuli that rather affect the selective survival of newborn cells.
A future study will address this hypothesis. However, the experiment is not as straightforward as it might seem because activity-dependent survival is dependent on a critical postmitotic period. So far, no studies in mice have been able to unambiguously demonstrate selective survival, e.g. in response to a learning stimulus [39, 44]. Also in rats, where positive results exist, the data are not yet fully conclusive [50–52].
The largest population of proliferative progenitor cells in the SGZ is DCX-negative. Pro-proliferative stimuli affect DCX-positive and DCX-negative cells . Activity-dependent regulation of cell survival, in contrast, primarily affects cells that have left the cell cycle but still might be DCX-positive. From our previous  and the present data and consistent with other reports [53, 54] we hypothesize that the phase of DCX and CR co-expression is the phase during which the newborn cells are primarily selected for long-term survival. Postmitotic cells within the neuronal lineage (e.g. expressing TUC4, associated with axon elongation, or the polysialylated form of the neural cell adhesion molecule, PSA-NCAM) have been shown to have a reduced threshold of LTP induction and thus are presumably particularly plastic [53–55].
We found that the newborn cells leave the close contact to astrocytes, which presumably represents the precursor cell niche, at varying times in the course of development, but separation from the niche is not predictive of the degree of (dendritic) maturation. This might indicate that once the stem cells have produced their differentiating progeny, other factors but the engulfing astrocytic processes become necessary. We and others hypothesize that these factors depend on synaptic input.
First synaptic input reaches the newly generated cells of the dentate gyrus on a nestin-positive progenitor cell level and this first input is GABAergic . Ambrogini and colleagues have characterized the stages of development electrophysiologically and found GABAergic input as well . They did not test for DCX expression in their study. Because nestin-GFP and DCX show an overlap in expression (type-2b cells) both studies should have identified the identical cell type. GABA, which acts excitatory at these early stages of development promotes further maturation of new granule cells [7, 57].
Many current hypotheses on the activity-dependent regulation of adult hippocampal neurogenesis include a Hebbian mechanism, in which newly generated cells are recruited into the network based on the successful establishment of connectivity early after becoming postmitotic [51, 58–60]. This hypothesis is supported by the finding that the number of surviving cells declines rapidly after the cells have turned postmitotic. We add to this idea of a selective induction of survival early after exit from the cell cycle by demonstrating that a number of the apoptotic cells in the SGZ were DCX- or CR-positive. Due to the very low number of TUNEL-positive cells in our sample, no quantitative statement is possible. We hypothesize from our data that dendritic development might in fact depend on this initial immature synaptic integration.
Our data indicate that the selection and elimination process might take place, when the newly generated cells already have acquired comparatively complex apical dendrites. Because this phase is also associated with TUC4 expression (not shown here) and TUC4 is involved in axon elongation, the selected cells also should have extended their axon to CA3. Nevertheless, it is not yet known, whether the synaptic input that mediates the survival-promoting effect is direct (e.g. from the perforant path), or indirect (e.g. via recurrent axon collaterals and GABAergic interneurons). In total, synaptic and functional maturation of new neurons takes several weeks and extends beyond the period of DCX and CR expression [25, 61, 62].
Synaptic activity is likely to promote cell survival in a similar manner than described for synaptic plasticity. For example, NMDA receptors are likely to be involved in mediating this effect, because blockade of NMDA receptors increased adult neurogenesis [63–66]. Massive synaptic (and extra-synaptic) stimulation of kainic acid and other excitatory receptors causes a strong and lasting increase in cell proliferation [40, 41]. We have shown that at the peak of proliferative activity this increase is largely due to a division of DCX-expressing cells . In the present study we found a corresponding increase in the total number of DCX-positive cells as well as an increase in the number of DCX-positive cells in the proliferative stages (A – D). Surprisingly, however, we also found similar increases in the remaining categories (E and F) and an overall distribution of dendrite morphologies that was not different from control conditions (Fig. 4).
This might suggest that the overall dynamics of apical dendrite development were not altered by the seizures. This is not to say that other parameters related to dendritic development not investigated here would have to remain unaffected as well. The morphology, extent and persistence of basal dendrites, for example, is strongly affected by experimental seizures [18, 67].
DCX is thought to be primarily involved in neuronal migration , either by stabilizing and destabilizing protofilaments or by modulating dynein action . Migration of new neurons in the SGZ is limited, and we have found previously that the immature cells find their final position early in development . The extent and duration of DCX expression during adult hippocampal neurogenesis thus seems surprising. However, certain cytoskeletal rearrangements in the course of neurite extension might also rely on DCX function, although this remains to be shown. Alternatively, the variable duration of DCX expression as seen in our study might also reflect a varying migrational state, even if the actual migration is minimal. The migrational condition might in turn be associated with other developmentally relevant events. We found that in the initial phase of dendrite elongation, the immature neurite often paralleled the SGZ (or was oblique to it), whereas in the mature cells the stem of the dendritic tree is perpendicular to the SGZ. This might indicate that the cell bodies could follow the lead of the outgrowing dendrites and relocate to an appropriate position perpendicular to the stem. Migration in the SGZ might thus actually consist of a relocation of the cell body along the pioneering dendrite.
As a final note, our data indicate that assessing the number of DCX-positive cells in the dentate gyrus alone as a measure of "adult neurogenesis" might give a misleading impression, because the phase of DCX expression is not only associated with possible cell death (thus reducing quantitative precision) but also encompasses a wide range of qualitative changes.