|Home | About | Journals | Submit | Contact Us | Français|
The generation of new neurons in the mammalian hippocampus continues throughout life, and lineage progression is regulated by transcription factors, local cues and environmental influences. The ability to direct stem/progenitor cell fate in situ may have therapeutic potential. Using an in vivo retroviral delivery and lineage tracing approach, we compare the lineage instruction factors Pax6 and Olig2, and demonstrate that both participate in regulation of adult hippocampal neurogenesis in adult rats. We show that overexpression of the pro-neuronal factor Pax6 pushes neuronal precursor cells to early maturation and increases the frequency of neuronal phenotypes. However, Pax6 overexpression results in no net increase in neurogenesis at three weeks. Blocking of Olig2 function reduces and slows neuronal commitment and differentiation and decreases net neurogenesis. Altering expression of both factors also changes gliogenesis. Our results establish that Pax6 decreases the number of NG2 (Neuron-Glia 2) progenitor cells and prevents oligodendrocytic lineage commitment, while repression of Olig2 results in an expanded astrocytic lineage. We conclude that selectively modifying transcriptional cues within hippocampal progenitor cells is sufficient to induce a cell fate switch, thus altering the neurogenesis-gliogenesis ratio. In addition, our data show the competence of multiple progenitor lineages to respond divergently to the same signal. Therefore, directing instructive cues to select phenotype and developmental stage could be critical to achieve precise outcomes in cell genesis. Further understanding the regulation of lineage progression in all progenitor populations within the target region will be important for developing therapeutic strategies to direct cell fate for brain repair.
Adult hippocampal neurogenesis is a dynamic process within the dentate gyrus (DG), where neural stem cells (NSCs) retain fate plasticity and can adapt to external stimuli [1, 2], injury, or pathology [3, 4]. The hippocampal subgranular zone (SGZ) contains radial glia-like stem cells (type-1 cells) that give rise to amplifying progenitor cells (type-2 and type-3 cells), which mostly become neurons [5, 6]. In addition, NSCs are multipotent and can generate glia [7–9]. NG2 progenitor cells also reside in the DG as a potentially multipotent subpopulation  that makes no apparent contribution to adult neurogenesis .
As in development, the progression from type-1 cells to mature neurons is regulated by soluble intrinsic factors that regulate stem cell maintenance (e.g. Pax6, Tbr2) and direct cell fate (e.g. Olig2, Sox2, Pax6, Tbr2) [5, 12, 13]. Transcription factors controlling cell differentiation have been elucidated for the developing CNS  (reviewed in [15, 16]), and to some extend for the postnatal brain [5, 12, 17]. Although adult neurogenesis can be modulated by many factors, the regulation of cell fate decisions in the adult DG is less well understood. Here, we investigate the role of the lineage-related genes Pax6 and Olig2 in directing cell fate of stem/progenitor cells in the adult SGZ in vivo, and asked if manipulating their activity in situ could direct the outcome of neurogenesis.
The transcription factor Pax6 essentially regulates the balance between stem cell maintenance and neuronal lineage progression in development [15, 18, 19], and postnatal neurogenesis [17, 19, 20]. Its expression is confined to type-1 stem cells and early progenitor cells of the SGZ, where Pax6 positively influences neuronal differentiation [5, 17, 21]. So far, it has not been investigated how Pax6 regulates lineage progression in the adult DG. In contrast, the transcription factor Olig2 is required in glial fate commitment, and has been studied in various brain regions and injury models [20, 22, 23]. Olig2 directs neuronal-glial cell fate decision in development [24, 25], and postnatal subventricular zone [20, 26].
In our study, we retrovirally overexpressed the neurogenic factor Pax6 and blocked the pan-gliogenic factor Olig2, to better define how multiple signals dynamically regulate the neurogenic niche. By examining both, early events in cell lineage decision and the outcome of cell fate following differentiation, we were able to ascertain the effect of altering these cell lineage instructions signals on the population of differentiated cells in the adult hippocampal niche. This is the first study to characterize the role of a gliogenic signal in the DG. Understanding the process of lineage commitment in the context of balancing the production of all cell populations derived from type-1 stem cells provides fundamental insight into both the normal process of adult neurogenesis and the way in which the population balance is altered in response to environmental cues and injury. Our data revealed that overexpression of Pax6 acts on the neuronal and NG2 progenitor population and pushed maturation of newly born neurons. The absence of Olig2 signaling led to delayed neuronal maturation of hippocampal precursors and induced infected cells to become astrocytes.
Neural stem cell therapies for the diseased brain will require methods to selectively and efficiently drive cell fate of endogenous stem/progenitor cells. It has been shown that altered expression of a single gene can direct cell fate of adult hippocampal progenitors toward oligodendrocytes . Our results show how both neuronal and glial lineage instruction factors regulate the generation of new neurons, and that their modulation disrupts neurogenic homeostasis and changes the composition of the neurogenic niche.
Seven to eight week-old young adult female Fisher 344 rats were purchased from Harlan Sprague Dawley (Indiana, USA). Rats were housed two to three per cage under standard laboratory conditions with a light/dark cycle of 12 hours each and free access to food and water. All experiments were performed according to and approved by national and institutional guidelines (NIH, IACUC). At day 0, rats (n = 32, distributed across three groups for 2 time points) were deeply anesthetized with a mixture of ketamine (75 mg/kg), xylazine (4 mg/kg), and acepromazine (5.6 mg/kg), and 1μl of a retroviral suspension containing 1) green fluorescent protein (GFP), 2) Olig2-VP16-IRES-GFP (Olig2-VP16) or 3) Pax6-IRES-GFP (Pax6) was infused into the left hemisphere of the DG via a stereotaxic system. Dentate gyrus coordinates from Bregma in mm are: A/P −3.6, M/L +2.5, D/V −4.0. The retroviral vectors used have been described elsewhere , and were produced by transient transfection of the construct plasmids (kind gifts from M. Goetz) with gag/pol and vsv-g packaging plasmids as previously described . Viral titer typically were ~5×107 TU/ml.
Retrovirally-mediated transgene expression is a useful tool to determine the effect of specific genes in single cells. Because insertion of the transgene is random, retroviral GFP expression may be variable, and may be reduced over time. We have designed our experimental approaches to minimize variability in delivery and detection, including using antibodies to GFP to compensate for reduced GFP protein and for any quenching of native GFP fluorescence following pretreatment for thymidine analog detection.
Retrovirally-labeled cells and their progeny were investigated five days and 21 days following gene delivery. In addition, two thymidine analog injection paradigms were used to determine the effect of the transgenes on a broader population of progenitor cells. For analyzing proliferation (short-term, group A), animals received three i.p. injections of bromodeoxyuridine (BrdU, 50 mg/kg body weight; SIGMA, St. Louis, MO) 2 hours apart on day 5 following gene delivery and were killed 2 hours after the last injection. Cell survival and long-term cell division activity of progenitor cells three weeks following gene delivery were determined using the combination of two other halogenated thymidine analogs, 5-chloro-2′deoxyuridine (CldU) and 5-iodo-2′deoxyuridine (IdU) . First, on day 1, animals received three i.p. injections of CldU (42.5 mg/kg body weight; SIGMA, St. Louis, MO) 2 hours apart and were left three weeks to assess survival of the labeled cohort (group C). Prior to tissue collection at three weeks, IdU (57.5 mg/kg body weight; MP Biomedicals) was administered 3x a 2 hours apart to label cell proliferation, and animals were killed on the next day (day 22, group B).
Rats were deeply anesthetized and perfused transcardially with 0.9% sodium chloride followed by 4% paraformaldehyde (PFA) in 0.1 M phosphate buffer. Brains were removed from the skulls, postfixed in 4% PFA at 4°C over night, and transferred into 30% sucrose. Brains were cut on a freezing sliding microtome (Leica SM2000R, Deerfield, IL, USA) in the horizontal plane in 40 μm thick sections and cryoprotected. Sections were stained free-floating with all antibodies diluted in Tris-buffered saline containing 3% donkey serum and 0.1% Triton X-100. For BrdU, CldU and IdU staining, DNA was denatured in 2N HCl for 30 minutes at 37°C.
Primary antibodies were applied in the following concentrations: anti-BrdU (rat, 1:500; Accurate), anti-CldU (rat anti-BrdU, 1:500; Accurate), anti-IdU (mouse anti BrdU, 1:500; Becton Dickinson), anti-GFP (chicken, 1:5000, Aves, USA), anti-NG2 (rabbit, 1:500; Millipore, Temecula, CA), anti-NG2 (mouse, 1:2000; Millipore, Temecula, CA), anti-RIP (mouse, 1:250; Millipore, Temecula, CA), anti-S100β (rabbit, 1:2500; Swant, Belinzona, Switzerland), anti-S100β (mouse, 1:5000; Abcam), anti-GFAP (rabbit, 1:2500; Dako), anti-DCX (goat, 1:200; Santa Cruz Biotechnologies, Santa Cruz, CA), anti-NeuN (mouse, 1:1000; Millipore; Temecula, CA).
For immunofluorescence, Alexa488, Cy3- or Cy5-conjugated secondary antibodies were all used at a concentration of 1:500 (Jackson ImmunoResearch, Laboratories, West Grove, PA) with the exception of DyLight488 (anti-chicken, 1:1000). Fluorescent sections were coverslipped in polyvinyl alcohol with diazabicyclooctane as anti-fading agent (PVA-DABCO; prepared with glycerol, 0.2M Tris-HCL, and double-distilled water) . Immunohistochemistry followed the peroxidase method with biotinylated secondary antibodies (all: 1:500; Jackson ImmunoResearch Laboratories, West Grove, PA), ABC Elite reagent (Vector Laboratories, Burlingame, CA) and diaminobenzidine (DAB; SIGMA) as chromogen.
For thymidine labeling, one-in-six series of sections of each brain were DAB stained, and immunoreactive cells were counted throughout the rostrocaudal extent of the dentate gyrus. BrdU, CldU or IdU-positive cells were quantified by the optical fractionator approach using the Stereo-Investigator software (MBF Bioscience Williston, VT) with an unbiased counting frame and the optical dissector principle (grid size: 300×150, frame size: 180×100, mean contour area: approximately 900 μm2, estimated mean Schmitz-Hoff CE values were between 0.09 to 0.16 with a group mean of 0.11). For retrovirally-labeled cells, one-in-twelve series of sections were labeled for multiple-immunofluorescence as described above for phenotypic analysis, and evaluated for three-dimensional co-localization by examining orthogonal views from a series of confocal microscope focal planes (Olympus FV500). To quantify cell number, the following steps were used: 1) for each staining combination, the phenotype of GFP+ cells was recorded as a percentage of all GFP+ cells detected within that animal. From these data, the mean frequency of each phenotype was determined for each experimental condition. 2) To determine differences in the number of GFP+ cells between conditions, the total number of GFP+ cells was estimated from the fractionated sampling for each condition. 3) The frequency distribution determined above (step 1) was then normalized against this total GFP cell number (step 2) to estimate actual population differences between phenotypes. For qualitative imaging, appropriate gain and black level settings were determined on control tissues stained with secondary antibodies alone to discriminate true signal from background. All images were taken in sequential scanning mode and minimally processed and composed in Adobe Photoshop. Only general adjustments to signal distribution were made and figures were not otherwise manipulated.
Statistical tests to detect differences between group means were performed by an ANOVA analysis, followed by Tukey’s post hoc tests, in cases where a significant F statistic was obtained (PRISM software). Student’s t-test was used for individual pair wise comparisons. All values are expressed as mean ± standard error of the mean (SEM). P - values of ≤ 0.05 were considered statistically significant.
We used retroviral vectors to target dividing cells, and to overexpress or repress, respectively, the genes Pax6 and Olig2 to determine changes in cell fate of hippocampal stem/progenitor cells. To discriminate the effect of transgene expression, we first assessed the phenotypic identity of GFP-only transduced cells. Newly committed neurons were identified by expression of the transient immature neuronal marker doublecortin (DCX) and the mature neuronal marker NeuN. Approximately half of newly generated, GFP-positive (GFP+) cells showed commitment to a neuronal phenotype whether assessed shortly after retroviral transduction into the dentate gyrus (five days, Fig. 1A), or when the progeny of cells were analyzed at 21 days (Table 1). Confocal microscopic analysis of cell phenotype revealed the overall distribution of infected cells in the DG. Five days following gene delivery with the GFP-only construct, the majority of retroviral-labeled cells with neuronal lineage commitment were horizontal type-2 cells with medium processes (Fig. 1B). At three weeks following GFP delivery, lineage-traced cells displayed a mature neuronal morphology as seen by long apical dendritic branching into the molecular layer (ML; Fig. 1C), and migration into the inner GCL (Fig. 1E). These results are consistent with the developmental progression and morphological changes of newly generated neurons that have been described previously [32, 33].
Besides becoming neurons, some retroviral-infected cells within the SGZ and GCL also adopted glial fates, demonstrated by morphology and marker coexpression of the proteoglycan NG2 (Fig. 2A, C), the oligodendrocytic lineage marker RIP (Fig. 2B), and S100β at both time points. Adoption of oligodendrocytic fate was confirmed by coexpression of CNPase and APC; however, RIP was chosen as the most effective marker for quantifying co-localization in the adult rat DG. S100β+ cells represent postmitotic astrocytes with many short and highly branched, bushy processes (Fig. 2D). Five days following gene delivery, a few cells also expressed the glial fibrillary acidic protein GFAP (Fig. 2E), often colabeled with the SRY-related HMG-box gene 2 (Sox2). As neural stem cells enter cell cycle with low-frequency, there is a low probability of their inclusion within the GFP-expressing cell population. GFAP and Sox2 mark both stem and non-stem astrocytes distinguished by radial and stellate morphology  and the characteristic radial morphology of type 1 neural stem cells was not observed in cells coexpressing GFP, GFAP, and Sox2. A portion of the GFP-expressing cells did not label with any of the phenotypic markers used for stem or progenitor cells and may represent a population that becomes arrested in regard to lineage commitment following exit from cell cycle.
Three weeks following GFP-only gene delivery, newly generated neurons in the SGZ adopted a characteristic morphology and extended an apical dendrite that branched within the molecular layer (Fig. 3A). Pax6-transduced newly generated neurons exhibited a morphology, typical of GFP-only cells, but showed somewhat abnormal arborization, having smaller dendrites in any direction, and no branching within the ML detectable by GFP-expression (Fig. 3B). Furthermore, a transitional phenotype of cells coexpressing both DCX and NeuN primarily in the perikaryal cytoplasm was observed; this phenotype is not seen in non-infected tissue. These novel DCX/NeuN double-labeled cells were also found in the hilus (Fig. 3C).
Three weeks following gene delivery of the dominant negative Olig2-VP16, a disruption of normal neuronal maturation was observed. The majority of newly generated neurons still appear immature, with stunted, less elaborate dendritic arbor. Those cells revealed only one strong dendrite reaching into the ML without branching (Fig. 3D). Blocking Olig2 resulted in fewer transduced neurons expressing NeuN, with those detected possessing a single apical dendrite with little additional dendritic arbor, a morphology consistent with an immature neuroblast phenotype (Fig. 3E). Blocking Olig2 also resulted in GFP-positive cells that exhibited small processes and appeared to have a neuronal morphology, but were DCX- and NeuN-negative (Fig. 3F).
Quantitative assessment of lineage commitment soon after transgene delivery revealed that GFP-only cells had a neuronal distribution of 43% (DCX 39.3 ± 1.3%, NeuN 4.3 ± 2.5%; Fig. 4A) and a smaller glial distribution of 30% (NG2 9%, RIP 8%, S100β 6%, GFAP 6.6%) with the remaining GFP-positive population expressing no detectable phenotype (Table 1). Population estimates reflect the frequency distribution with 109 out of 248 GFP+ cells expressing either DCX or NeuN, 73 cells that followed a glial lineage, and 66 cells remaining with no detectable phenotype (Table 1). When the progeny of GFP-infected cells was analyzed at three weeks, an increase in neuronal maturation was observed (to 60% or 239 out of 402 GFP+ cells detected; Fig. 4B, Table 1) accompanied by a shift from primarily early to primarily mature neuronal marker expression (DCX 8.1 ± 5.1%, NeuN 51.6 ± 3.7%, Fig. 4B). The population of GFP/NG2+ cells doubled, accounting for 19% of observed cells. Of the remaining glial phenotypes, 6% expressed RIP, and 5% were S100β+. No GFAP labeling was found at 21 days. The number of GFP-positive cells with no detectable phenotype had fallen to <10% by 21 days post-transduction (Table 1).
Next, we assessed overexpression of the pro-neuronal transcription factor Pax6 on lineage commitment of infected cells and their progeny in the DG. Retrovirus-mediated overexpression of Pax6 led to a significant increase in the ratio of cells committed to a neuronal lineage as determined by expression of DCX and NeuN at both time points (Fig. 4, Table 1). Five days following infection, already more than a third of Pax6-transduced cells colabeled with NeuN, a significant higher amount compared to the other groups (NeuN 37.9 ± 9.4%, p vs. GFP-only = 0.0054, p vs. Olig2-VP16 = 0.0025; Fig. 4A), but only 8% expressed DCX (8.1 ± 4.9%, p vs. GFP-only = 0.0207, p vs. Olig2-VP16 = 0.024). Interestingly, we observed the emergence of a novel population of cells coexpressing DCX/NeuN that accounted for 17% of the GFP-population. Thus, when combined with the NeuN+ population, 55% of cells already expressed the mature neuronal marker five days following Pax6 overexpression. When the progeny of Pax6-infected cells was assessed at 21 days, the ratio of cells committed to a neuronal lineage has increased to approximately 80% with 21% expressing DCX/NeuN, 58% expressing NeuN-only (total NeuN p vs. GFP-only = 0.0202), and 2.8% expressing DCX-only (Fig. 4B).
At five days, overexpression of Pax6 increased both the proportion of newly generated neurons as a function of the GFP-labeled population and their total number (Fig. 4A), which resulted in a significant net increase of neurogenesis (Pax6: 157 cells vs. GFP-only: 11 cells, p = 0.002; Table 1). At 21 days following Pax6 overexpression, it may appear that neurogenesis was also increased as the percentage of NeuN-positive cells was enhanced (Fig. 4B). However, this percentage increase was due to a reduced number of GFP-positive cells detected at 21 days following Pax6 (Table 1). Comparison of the actual number of NeuN-positive cells reveals no difference between Pax6 and control-GFP conditions at this time (Pax6: 215 cells vs. GFP-only: 207 cells, p = 0.7740; Table 1). Thus, there was no net increase in neurogenesis following Pax6 gene delivery at 3 weeks.
To evaluate the effect of Pax6 overexpression on the fate of glial progenitors, we examined relevant phenotypic markers. The distribution pattern of S100β+ astrocytes did not differ from that seen with GFP-transduction alone at both early and later times of differentiation. Pax6-infected cells expressing GFAP were detected at five days (4.3 ± 2.5%; Fig. 4A) following gene delivery, but this phenotype was not detected in the progeny at 21 days (Fig. 4B, Table 1). Overexpression of Pax6 also led to a decrease in the number of NG2+ cells over time (from 28 ± 16 cells at 5 days to 7 ± 7 cells at 3 weeks) and in comparison to control at 21 days (GFP-only 77 ± 26 cells, p = 0.05; Table 1). Furthermore, no cells committed to an oligodendrocytic lineage (RIP-expression), normally detected in the GFP-only group, at both time points (Fig. 4, Table 1).
Our study is the first to investigate the role of Olig2 in adult hippocampal neurogenesis by retroviral gene delivery of the dominant negative Olig2-VP16. By fusion to Olig2, the transcriptional activator VP16 replaces the Olig2 repressor domain and thus interferes with the normal function of Olig2. The viral constructs used has been described elsewhere [25, 28]. Olig2-VP16 gene delivery produced changes in neurogenesis, slowing maturation of cells and reducing overall neurogenesis. At five days, 38% were DCX+ cells indicating early neuronal commitment was similar to that seen in the GFP-only group (Fig. 4A). Unlike the GFP-only group, expression of Olig2-VP16 resulted in the appearance of a transition population coexpressing DCX/NeuN (9%), and in the absence of NeuN-only positive cells. By three weeks, the number of lineage-traced cells expressing NeuN slightly increased to 18.5% but was still significantly lower compare to GFP-only conditions with 51.6% (p = 0.0217; Fig 4B). Corresponding to the increase in NeuN-only expression, the frequency of DCX/NeuN (4%) and DCX+ cells (16%) decreased, but still was more prominent compared to control virus (Fig. 4B). Taking all DCX and NeuN expression together as an indicator of neuronal lineage commitment at three weeks, only 39% of retrovirally-infected cells adopted a neuronal fate when Olig2-VP16 was expressed, compared to 60% in the GFP-only condition and approximately 80% in the Pax6 condition.
In assessing the effect of blocking Olig2 on net neurogenesis at three weeks, the significant decrease in the frequency of newly generated neurons (Fig. 4B) corresponded to a significant decrease in the actual number (total NeuN in Olig2-VP16: 102 cells vs. GFP-only: 207 cells, p = 0.038; Table 1). Blocking Olig2 also resulted in an increase in the population of GFP+ cells for which no phenotype could be determined using the panel of markers in this analysis (145 out of 450 cells, n = 4, p vs. GFP-only = 0.043, Table 1).
Similar to the alteration of neurogenic maturation, blocking of Olig2 effected gliogenesis and resulted in a higher frequency of astrocytes (Fig. 4). The percentage of retrovirus-infected cells coexpressing GFAP doubled at five days when compared to GFP-only (Olig2-VP16 13.2 ± 5.2% vs. GFP 6.6 ± 3.9%; Fig. 4A), but was lower at 21 days (6.9% that account for 31 cells out of 450 cells detected; Fig. 4B, Table 1). This is in contrast to the GFP-only and Pax6 conditions, where no GFAP+ cells were detected at 21 days. Although some co-staining with S100β was found (GFAP-negative cells; Fig. 4, Table 1), it did not significantly differ between conditions or time points. Hence, Olig2-VP16 increases the ratio of GFAP+ but not S100β+ astrocytes. Expression of the dominant negative Olig2-VP16 did not alter the frequency and number of NG2 relative to control (Fig. 4, Table 1), suggesting no disruption of glial precursor cells. However, coexpression of RIP was only detected at five days, and was not observed in the progeny at three weeks. Therefore the dominant negative Olig2-VP16 appears to inhibit the relatively infrequent oligodendrocytic maturation of proliferating cells in the SGZ, while increasing frequency and number of retrovirally-labeled cells that produced no detectable phenotype (Table 1).
Changing the distribution of cell lineage specification in the hippocampal neurogenic niche could produce a response to reestablish homeostasis in cell production. To determine if overexpression or repression, respectively, of the lineage-instruction factors Pax6 and Olig2 influenced proliferative activity in the hippocampal dentate gyrus through a secondary effect of the environmental niche, subjects were probed with halogenated thymidine markers at five, 21 and 22 days following gene delivery (Fig. 5A). The distribution of thymidine analog-labeled cells (that includes both virally transduced and non-infected progenitor cells) is illustrated in Figure 5B.
The number of BrdU+ cells (group A at 5 days) following retroviral-GFP delivery into the dentate gyrus of the left hemisphere did not differ from the non-injected hemisphere (Control-GFP 3,463 ± 493 vs. non-injected site 2,552 ± 1061; Fig. 5C), revealing no effect due to gene delivery. The expression of the Pax6 or Olig2-VP16 transgenes produced a significant elevation in proliferation relative to GFP alone (Control-GFP vs. Olig2-VP16 6,401 ± 505, p = 0.0045 and Control-GFP vs. Pax6 5,282 ± 416; p = 0.0224; n = 5; Fig. 5C). Here too, the number of BrdU-positive cells in the contralateral, non-injected DG of those animals was not effected (data not shown), suggesting elevated proliferation was due solely to local transgene expression. However, no difference was found in the ultimate survival of these newly generated cells detected by CldU labeling (group C) irrespective of which transgene was expressed (Fig. 5C). The stimulation of proliferation observed with expression of Pax6 and Olig2-VP16 was largely transitory as the number of newly generated cells at three weeks following gene delivery (detected by IdU labeling; group B) had returned to normal levels with the exception of Olig2-VP16, which still produced a significant elevation by 3,736 ± 480 cells (p = 0.0247; n = 7, Fig. 5C).
Our data demonstrate that both neuronal and glial lineage instruction factors, known to shape the developing CNS, are active in adult hippocampal neurogenesis. We show a differential effect of Pax6 and Olig2 on distinct progenitor cell populations and their lineage progression. The results establish that it is possible to manipulate the stem cell niche to direct cell fate outcome. We found that retroviral-labeled progenitor cells in the adult dentate gyrus predominantly differentiate into neurons. Yet, another population of lineage-traced NG2 precursor cells differentiates into oligodendrocytes . We present new estimates of gliogenesis in the adult rat dentate gyrus, and show that under normal conditions approximately 30% of newly generated cells are glia-committed; a frequency higher than previously reported in mice [27, 36].
Lineage instruction factors act on different cell types and multiple developmental steps [27, 33, 37]. Pax6 is normally expressed by early neuronal progenitors [5, 12, 21], where its expression influences neuronal differentiation [17, 19, 38]. GFAP+ cells also express Pax6, but due to the low probability of targeting quiescent type-1 stem cells, our data primarily address the competence of neuronal progenitors to respond to Pax6. In this regard, overexpression of Pax6 accelerated differentiation more quickly through the DCX-expressing stage, generating both more NeuN+ cells and giving rise to a novel DCX/NeuN transitional phenotype. Pax6-transduced cells mostly displayed neuronal morphology and maturation at both time points examined, however, inducing early differentiation resulted in abnormal granule cell dendritic arborization. Potentially, this atypical granule cell morphology may represent differentiation into interneurons, as has been suggested for NG2-derived cells [10, 39]. Despite Pax6-induced rapid maturation there was no increase in net neurogenesis at three weeks relative to control, but rather a decrease in the number of virally transduced cells. It may be that Pax6 overexpression hastened maturation (increased NeuN expression) and reduced the time spent as a transient-amplifying cell (DCX expression), thus causing loss of transduced proliferating immature progenitor cells contributing to the lower number of newly generated cells at three weeks. It is also possible, that early maturation forced cells to terminally exit cell cycle prematurely so that fewer GFP+ cells were detected. In this case, a therapeutic increase of neurogenesis may require an expansion of progenitor populations prior to delivery of a lineage instruction signal.
The decrease at three weeks was despite a transient elevation in overall proliferation induced by Pax6 gene delivery when assessed by BrdU-only (thymidine-analog labeled cells include both virally infected and non-infected progenitor cells). Increased proliferation may be an indirect response of the niche due to forced cell cycle exit or death of progenitor cells as a result of Pax6 pressure. In turn, disruption of proliferation of Pax6-GFP+ cells may be on a background of the niche’s transient increased expression of mitogens.
This is the first study to estimate the influence of a gliogenic factor on adult hippocampal neurogenesis. Our data demonstrate that Olig2 function is required for the proper maturation of newly generated neurons in the hippocampal niche. Although blocking Olig2 signaling increased the transient DCX progenitor population, relatively few cells expressed NeuN at either time. In contrast to Pax6, Olig2-VP16 acts negatively on the neuronal population by slowing maturation of neuronal progenitor cells resulting in a decrease in net neurogenesis. At three weeks, only 39% of transduced cells followed a neuronal lineage. These newly generated neurons together with a large number of marker-negative (N.D.) GFP-expressing cells exhibited characteristic morphology of type-2 and type-3 progenitor cells such as bipolarity and few processes [32, 33]. Olig2-VP16-tranduced cells thus may be arrested at a transient amplifying progenitor stage. Interestingly, the modulation of cell fate decision in the relatively small number of infected cells stimulated a proliferative response among the general progenitor cell pool. Results from BrdU and IdU pulsing showed a significant increase in the number of proliferating cells after Olig2-VP16 transgene expression. However, no difference was found in the ultimate survival of these newly generated cells. The transient elevation of proliferation may reflect increased expression of mitogens as a homeostatic response of the niche following retroviral gene delivery.
Under normal conditions, Pax6 is not expressed in NG2+ cells in the adult dentate gyrus. However, we are able to retrovirally target dividing NG2 cells to express Pax6. NG2 glia precursor cells are multipotent ex vivo  and following transplantation . While Pax6 expression did not change the number of NG2+ cells at five days, fewer cells were detected at three weeks, and no mature oligodendrocytes were observed relative to control at either time. Here, elevated Pax6 exerted a selective pressure on NG2+ cells, reducing cell survival and thus preventing their differentiation into oligodendrocytes. As there is no induction of alternate glial fate, it is likely that Pax6 expression leads to the loss of NG2 cells and their progeny, or indirectly impairs NG2 cell survival by directing a neuronal fate. The selective drive to neuronal versus glial fate decision (a cell fate switch) appears to limit the total progenitor pool, and in turn decreases the number of lineage-traced GFP+ cells found at three weeks.
Olig2 is normally expressed by oligodendrocytes throughout the CNS, and precursor cells during development [24, 25], and is present in radial glia astrocytes and transient-amplifying progenitors in the postnatal and adult SVZ, where it functions as a regulator for neuronal-glial fate decisions [20, 26]. Olig2 mRNA is also expressed in the adult mouse hippocampus  but we detected few positive cells by immunohistochemistry in the rat hippocampus. One result of delayed neuronal maturation caused by Olig2-VP16 in our study is the generation of a significant number of astrocytes. This is in line with previous studies of neurospheres cultures and injury models, where Olig2-VP16 transduction decreased the number of oligodendrocytes and neurons, suggesting a role for Olig2 in neuronal-glial fate decision by acting as a repressor of the astrocytic lineage .
We show that blocking Olig2 function can direct adult hippocampal precursor cells, largely committed to a neuronal lineage, towards a glial fate, consistent with a previous report where early lineage instruction with mash1 directed cell fate in the DG . We suggest that blocking Olig2 in the progeny of type-1 cells results in their direction to astrocytes at very early stages of fate decision, e. g. type-2a cell stage that marks a transition between glial and neuronal differentiation , while blocking Olig2 in cells at later stages of commitment (type-2 and type-3) has a negative effect on neuronal lineage commitment as seen by distinct changes in their morphology, maturation and survival. Interestingly, our results showed an increase in GFAP+ cells, but not S100β+ astrocytes. One possibility is that Olig2 inhibition affects GFAP+ astrocytes that potentially could be type-1 stem cells. However, based solely on morphology, no lineage traced GFP+ type-1 stem cells were found. This may be a technical limitation due to the low probability of targeting type-1 cells by retroviral infection. Furthermore, there was no alteration of the NG2 cell population following Olig2-VP16 expression, but there was a disturbance in their commitment to an oligodendrocytic lineage over time.
Consistent with earlier studies, GFP-only transduced cells largely matured into granule neurons over time [27, 33]. However, our results establish that at least a quarter of newly generated cells showed commitment to a gliogenic lineage, which in turn was susceptible to engineered expression of the transgenes used in this study. Our findings demonstrate that inductive signals can positively regulate cell fate outcome, but also produced a selective pressure on the niche, e.g. a shift in the ratio of neurogenesis vs. gliogenesis that in turn produced a homeostatic response as shown by an elevation in overall proliferation. In addition, we showed the competence of multiple progenitor cell populations to respond to the same lineage instruction signal. Thus, targeting the expression of one lineage instruction factor to the adult neurogenic niche may have multiple consequences. Furthermore, in the absence of other influences on the neurogenic niche, there appears to be an upper limit on neuronal production. While homeostasis is enforced in this direction, there does not seem to be an immediate compensatory response to restore phenotype ratios following redirection of cell fate or loss within progenitor cell pools. Therefore, increasing output of a particular mature cell phenotype from the neurogenic niche will likely require a combination of lineage instruction factors with other signals favorable for cell survival and differentiation. The development of such strategies will be required to manipulate neurogenesis for therapeutic use.
This work was supported by NIH awards AG20047 and AG22555 and DoE award DE-SC0001810 to DP. We thank Magdalena Goetz for the pMXIG Pax6 and pMXIG Olig2-VP16 constructs, Sarah Schuck and Carol Galioto for technical assistance, and Rupert W. Overall and Isabelle Aubert for critical reading of the manuscript.
Disclosure of Potential Conflicts of Interest
The authors declare no conflicts of interest.
Author contribution summary:Friederike Klempin: Concept and design, data collection and analysis, interpretation and manuscript writing
Robert Marr: Producing retroviral vectors, manuscript writing
Daniel A. Peterson: Financial support, concept and design, writing and final approval of manuscript