Adult neurogenesis occurs in two specialized regions of the mammalian brain, the subventricular
zone (SVZ) and the subgranular zone (SGZ) of the dentate gyrus (DG).
1
Adult hippocampal neural stem cells (NSCs), referred to as Type 1 cells represented
by radial glia-like cells (RGLs), generate Type 2 cells that are divided into Type
2a and Type 2 b subpopulations, the latter of which give rise to Type 3 cells (neuroblasts).
Type 3 cells exit the cell cycle and give rise to immature neurons that subsequently
mature into functional granule cells (Fig. 1A). The neural regenerative capacity depends
on maintaining a reservoir of NSCs and the integrity of the neurogenic niche, which
are regulated by neurotransmitters and numerous signals including Notch, Wnt, and
BMPs.
1
Here we showed that BMP9 (a.k.a., GDF2) deficiency altered the dynamics of adult hippocampal
neurogenesis as proliferative activity was decreased in Type 2a cells in Bmp 9 knockout
(Bmp9KO) mice (Fig. S1). Surprisingly, the number of cells committed to the neuronal
lineage was comparable to wildtype while the label retention was increased in immature
neurons, suggesting mitigation of reduced proliferation in Bmp9KO mice by increased
cell survival. We also found Bmp9 was not expressed in the adult brain, precluding
cell-autonomous regulation of neurogenic populations like other BMPs. Our findings
thus demonstrate BMP9 regulates adult homeostatic neurogenesis.
Figure 1
BMP9 regulates neurogenesis in mouse adult hippocampus. (
A
) Lineage progression and timeline of neural development in the adult hippocampus.
Type 1 cells are multipotent neural stem cells (NSCs) that develop into new neurons
and astrocytes in the SGZ. Through asymmetric division, Type 1 cells both self-renew
and produce an intermediate progenitor cell. Within the neurogenic lineage, there
are three types of intermediate progenitor cells (IPCs). These populations undergo
progressive rounds of symmetric cell division and differentiation to amplify the neural
precursor population. Most of these cells undergo apoptosis. Progenitor cells that
survive complete differentiation within one week. Over the following three weeks new-born
cells acquire morphological features of mature neurons. Neurogenic output is dependent
on combinatorial regulation of progenitor proliferation, differentiation, survival,
and maturation. Continual generation of new neurons throughout life is associated
with maintenance of the NSC pool. The production of new astrocytes similarly depends
on Type 1 NSCs. Following cell-cycle exit, post-mitotic Type 1 cells differentiate
into astrocyte progenitors. In contrast to IPCs of the neurogenic lineage, astrocyte
progeni-tors have restricted proliferation. Similar to IPCs, astrocytes progenitors
are successfully differentiated into astrocytes within one week and reach maturation
by 4 weeks. (
B
)
Bmp9KO mice do not have an altered distribution of early precursors and astrocytes.
(
a
) Coronal sections were immunostained with SOX2 and GFAP to identify astrocytes (SOX2+/GFAP+),
Type 1 cells (SOX2+/GFAP+), and Type 2a cells (SOX+/GFAP-). Astrocytes and Type 1 cells
were differentiated by morphological characteristics. Note the apical processes crossing
the granule cell layer, which is defined by the nuclear counterstain TOPRO3 in the
merged panel of the representative image. (
b-e
) Quantitative analysis using design-based stereology showed the number the total
number of (b) Type 1 cells, (c) Type 2a cells, and (d) astrocytes in wild-type and
Bmp9KO mice does not differ. (e) The relative proportions of Type 1 cells, Type 2a
cells, and astrocytes within the total SOX2+ population is not significantly different
between wild-type and Bmp9KO mice. Scale bar equals 50 μm. Whiskers on boxplots represent
maximum and minimum values and points represent values for individual mice. n = 6
for wild-type mice; n = 5 for Bmp9KO mice. Error bars on bar charts represent ± SEM
for each group. (
C
) Proliferation of early progenitors is reduced in Bmp9KO mice. (
a
) Three month old male wild-type and Bmp9KO mice were injected with CldU to label
cells in S-phase according to the experimental paradigm in Figure S1. Coronal sections
were immunostained with BrdU, SOX2, and GFAP to identify proliferating Type 1 cells
(BrdU+/SOX2+/GFAP+) and proliferating Type 2a cells (BrdU+/SOX+/GFAP-). Dashed lines
delineate the SGZ, arrowheads show clusters of BrdU + cells in each panel. Note BrdU + cells
primarily colocalize with SOX2 in both groups and that Bmp9KO mice have fewer BrdU + cells.
(
b
) Quantitative analysis using design-based stereology showed the number of BrdU + Type
1 cells is not different between wild-type and Bmp9KO mice, but (
c
) significantly fewer BrdU + Type 2a cells in Bmp9KO mice. (
d
) The proportion of BrdU + Type 1 cells and (
e
) BrdU + Type 2a cells does not differ significantly between wild-type andBmp9KO mice.
Scale bar equals 50 μm. Whiskers on boxplots represent maximum and minimum values
and points represent values for individual mice. n = 6 for wild-type mice; n = 5 for
Bmp9KO mice. Error bars on bar charts represent ±SEM for each group. ∗P < 0.05. (
D
) Deregulation of neural precursors does not decrease the success of neuronal lineage
commitment in Bmp9KO mice. (
a
) Three month old male wild-type and Bmp9KO mice were injected with CldU to label
cells in S-phase according to the experimental paradigm in Figure A.2. Coronal sections
were immunostained with BrdU and DCX to identify proliferating Type 2b/3 cells (BrdU+/DCX+).
Note colocalization of BrdU and DCX is limited in both animals and that Bmp9KO mice
have fewer BrdU + cells in the representative image. (
b-d
) Quantitative analysis using design based stereology showed the number of (b) BrdU + Type
2b/3 and (c) total number of DCX + cells is not different between wild-type and Bmp9KO
mice. (d) The relative proportion of BrdU + Type 2b/3 to all cells committed to the
neurogenic lineage is not different between wild-type and Bmp9KO groups. Scale bar
equals 50 μm. Whiskers on boxplots represent maximum and minimum values and points
represent values for individual mice. n = 6 for wild-type mice; n = 5 for Bmp9KO mice.
Error bars on bar charts represent ± SEM for each group. (
E
) Survival of new-born neurons is increased in Bmp9KO mice. (
a-f
) Three month old male wild-type and Bmp9KO mice were injected with CldU to label
cells in S-phase according to the experimental paradigm in Figure A.2 B. Coronal sections
were immunostained for BrdU, DCX, and NeuN to evaluate survival after 23 days on the
basis of label retention. Quantitative analysis using design based stereology showed
(
a
) a marked increase in the number of label-retaining cells (BrdU+) in Bmp9KO mice.
The total number of label-retaining cells is increased in Bmp9KO mice. (
b
) The number of label-retaining immature neurons in Bmp9KO mice is significantly greater
than wild-type mice, but (
c
) the number of label-retaining mature neurons does not differ between the two groups.
(
d
) The total number of immature neurons and (
e
) mature neurons is not significantly different between wild-type and Bmp9KO mice.
(
f
) The percentage of label-retaining immature neurons is significantly greater in Bmp9KO
mice. Whiskers on boxplots represent maximum and minimum values and points represent
values for individual mice. n = 5 for wild-type mice; n = 4 for Bmp9KO mice. Error
bars on bar charts represent ± SEM for each group. ∗P < 0.05, ∗∗P < 0.01. (
F
) Endogenous BMP9 is not necessary for cholinergic support in the medial septum and
diagonal band of Broca. (
a
) Coronal sections from the medial septum and vertical diagonal Band of Broca were
immunostained for ChAT and NeuN to identify the cholinergic neurons and neurons irrespective
of phenotype. (
b-d
) Quantitative analysis using design-based stereology showed (b) the number of cholinergic
(ChAT+) neurons in basal forebrain structures do not differ between wild-type and
Bmp9KO mice. (c) The total number of neurons, irrespective of phenotype does not differ
between wild-type and Bmp9KO mice. (d) There is no difference between the percentage
of ChAT positive neurons with respect to the entire neuronal population between wild-type
and Bmp9KO mice. Scale bar equals 200 μm Whiskers on boxplots represent maximum and
minimum values and points represent values for individual mice. n = 5 for wild-type
mice; n = 5 for Bmp9KO mice. Error bars on bar charts represent ± SEM for each group.
Whiskers on boxplots represent maximum and minimum values. Error bars on charts represent ± SEM
for each group.
Fig. 1
We previously identified BMP9 as one of the most potent osteogenic BMPs.
2
Emerging evidence indicates BMP9 exerts diverse functions in glucose metabolism, angiogenesis,
stem cell differentiation and maintenance of basal forebrain cholinergic neurons (BFCN)
2
;.
3
However, whether endogenous BMP9 is necessary for adult neurogenesis has not been
fully studied. We generated global Bmp 9 knockout (Bmp9KO) mice (Fig. S1A, B).
4
Using this model, we first tested whether BMP9 deficiency affected proliferation in
the SGZ. To detect subtype specific changes, we pulsed the mitotic cells with CldU
in both wildtype and Bmp9KO mice, and combined immunostaining for CldU with lineage
specific markers such as GFAP, SOX2, DCX and NeuN (Fig. S2A, B). The design-based
stereology changes were used to achieve the resolution of detecting such differences
and quantitative estimates with high precision. Based on pilot studies, we used 50%
sampling coverage and obtained CE < 1.2 for all BrdU + cell counts.
Since Type 1 cells are heterogeneous with different capacities for self-renewal and
generating neurogenic and gliogenic progeny, we quantified astrocytes based on morphology
and expression of GFAP and SOX2 to evaluate the BMP9 effect on Type 1 cell fate potential
(Fig. 1B, a & d). The median number of astrocytes for wildtype mice and Bmp9KO mice
groups did not differ (Fig. 1B, d), indicating the reduction in Type 2a cell proliferation
in Bmp9KO mice was not at the cost of astrocyte differentiation. Within the total
SOX2+ population, the proportions of Type 1 cells, Type 2 cells, and astrocytes were
not different between wildtype and Bmp9KO mice (Fig. 1B, e).
Using GFAP and SOX2 as markers to distinguish Type 1 (GFAP+/SOX2+) from Type 2a (GFAP−/SOX2+)
cells, we first evaluated the effect of BMP9 deficiency on the proliferative activity
of early precursors (Fig. 1C, a-e). Whereas Type 1 cells in wildtype and Bmp9KO groups
did not differ significantly (Fig. 1C, panel b), there were significantly fewer proliferating
Type 2a cells (BrdU+/GFAP−/SOX2+) for Bmp9KO than for wildtype mice (Fig. 1C, c),
suggesting that BMP9 deficiency may reduce proliferation of Type 2a progenitors.
To determine whether Bmp9KO mice had fewer proliferating Type 2a cells due to a smaller
Type 2a population and/or reduced proliferative activity within the population, we
analyzed the percentage of labeled Type 2a cells (BrdU+/GFAP−/SOX2+) within the total
Type 2a cell population (GFAP−/SOX2+) for each group. Bmp9KO mice demonstrated a notable,
but not significant, reduction in proliferative activity relative to wild-type mice
(Fig. 1C, e), while the total Type 2a cell population for wildtype and Bmp9KO groups
did not differ significantly (Fig. 1B, c). There was no significant difference in
the percentage of S-labeled in the Type 1 population either (Fig. 1C, d), while the
total Type 1 cell population for wildtype and Bmp9KO mice did not differ (Fig. 1B,
b). These data indicate the decreased Type 2a population in Bmp9KO mice was not caused
by aberrant Type 1 activation and consequent depletion of the Type 1 population.
As angiogenesis and neurogenesis are closely associated, we quantified the endothelial
cell progenitor-containing proportion of unidentified BrdU+ cells (BrdU+/SOX2-/GFAP−).
We found the proportion of BrdU+/SOX2-/GFAP− cells was not different between wildtype
and Bmp9KO (Fig. S3A), indicating endothelial cell hypersprouting was not present
in Bmp9KO mice. CD31 immunostaining did not reveal vasculature structural alterations
in Bmp9KO mice (Fig. S3B). These data suggest dysregulated Type 2a cells in Bmp9KO
mice may be not related to aberrant angiogenesis.
Since the Type 2a to Type 2 b transition is marked by SOX2 suppression and DCX upregulation,
we evaluated BMP9-null effect on the total number of proliferating Type 2 b progenitors
(BrdU+/DCX+) (Fig. 1D, b), the total number of DCX+ cells (Fig. 1D, c), and the percentage
of proliferating Type 2b/3 in the total DCX+ population (Fig. 1D, d). Small, but statistically
non-significant reductions were observed in Bmp9KO mice for all three measurements.
BMP9 deficiency therefore may not significantly alter Type 2b/3 proliferation and
the number of Type 2a cells committed to neurogenic lineage. Thus, BMP9 deficiency
does not alter neuronal lineage commitment.
To investigate whether BMP9 deficiency affected survival of postmitotic cells, we
used CldU pulse-labeling to birthdate cells prior to cell cycle exit (Fig. S2B), and
found Bmp9KO mice had more label-retaining-cells (LRCs; BrdU+/DCX+/NeuN+) after 23
days (Fig. 1E, a). Using DCX and NeuN staining, we found Bmp9KO mice had a significantly
greater number of LRCs (Fig. 1E, b), without change in the total number of LRCs (Fig. 1E,
c). We quantified the total number of cells in each of these population, and found
no difference in either the total population of immature (DCX+/NeuN+) (Fig. 1E, d)
or mature neurons (DCX−/NeuN+) (Fig. 1E, e) between wildtype and Bmp9KO mice. We further
found Bmp9KO mice had a significantly greater percentage of LRC immature neurons relative
to wildtype (Fig. 1E, f). Thus, BMP9 deficiency may enhance immature neuron survival
without expanding the population of either immature or mature neurons.
Lastly, we assessed Bmp 9 expression in anterior region, posterior region, and hippocampus
of wildtype mice. While highly expressed in liver as reported,
5
Bmp9 expression within the brain was low and comparable to that of hippocampus from
Bmp9KO mice (Fig. S4). We also quantified neurons expressing choline acetyltransferase
(ChaT) and the total number of neurons (NeuN+) in medial septum and diagonal band
of Broca (Fig. 1F, a-d). Surprisingly, the number of cholinergic neurons did not differ
for wildtype and Bmp9KO mice (Fig. 1F, b), neither did the total number of neurons
between wildtype and Bmp9KO mice (Fig. 1F, c). Similarly, we did not detect any difference
in proportion of cholinergic neurons (ChAT+) between the two groups (Fig. 1F, d).
Collectively, our work demonstrates that BMP9 modulates mouse adult hippocampal neurogenesis
under homeotic condition. Further studies are warranted to determine if BMP9 plays
any significant roles in neuronal ageing, neuronal degeneration, and/or acute neuronal
damage caused by stroke, ischemic and/or traumatic injuries. Furthermore, it is of
significance to investigate if BMP9 play any role in early neurogenesis, as well as
potential factors that may functionally compensate BMP9 functions during development.
Conflict of interests
The authors declare no conflict of interest.
Funding
The reported work was supported in part by research grants from the National Institutes
of Health (USA)(CA226303 to TCH and DE030480 to RRR). This project was also supported
in part by The University of Chicago Cancer Center Support Grant (P30CA014599) and
the National Center for Advancing Translational Sciences of the National Institutes
of Health (USA) (No. 5UL1TR002389). TCH was also supported by the Mabel Green Myers
Research Endowment Fund and The University of Chicago Orthopaedics Alumni Fund (USA).
Funding sources were not involved in the study design; in the collection, analysis
and interpretation of data; in the writing of the report; and in the decision to submit
the paper for publication.