Conceived and designed the experiments: RK IS RPM GK. Performed the experiments: RK IS MD GP PC RPM VH. Analyzed the data: RK IS GP PC RPM BV GK. Contributed reagents/materials/analysis tools: RK MD GP RPM VH BV. Wrote the paper: RK RPM GK.
The authors have declared that no competing interests exist.
Essentially all knowledge about adult hippocampal neurogenesis in humans still comes from one seminal study by Eriksson et al. in 1998, although several others have provided suggestive findings. But only little information has been available in how far the situation in animal models would reflect the conditions in the adult and aging human brain. We therefore here mapped numerous features associated with adult neurogenesis in rodents in samples from human hippocampus across the entire lifespan. Such data would not offer proof of adult neurogenesis in humans, because it is based on the assumption that humans and rodents share marker expression patterns in adult neurogenesis. Nevertheless, together the data provide valuable information at least about the presence of markers, for which a link to adult neurogenesis might more reasonably be assumed than for others, in the adult human brain and their change with increasing age.
In rodents, doublecortin (DCX) is transiently expressed during adult neurogenesis and within the neurogenic niche of the dentate gyrus can serve as a valuable marker. We validated DCX as marker of granule cell development in fetal human tissue and used DCX expression as seed to examine the dentate gyrus for additional neurogenesis-associated features across the lifespan. We studied 54 individuals and detected DCX expression between birth and 100 years of age. Caveats for post-mortem analyses of human tissues apply but all samples were free of signs of ischemia and activated caspase-3. Fourteen markers related to adult hippocampal neurogenesis in rodents were assessed in DCX-positive cells. Total numbers of DCX expressing cells declined exponentially with increasing age, and co-expression of DCX with the other markers decreased. This argued against a non-specific re-appearance of immature markers in specimen from old brains. Early postnatally all 14 markers were co-expressed in DCX-positive cells. Until 30 to 40 years of age, for example, an overlap of DCX with Ki67, Mcm2, Sox2, Nestin, Prox1, PSA-NCAM, Calretinin, NeuN, and others was detected, and some key markers (Nestin, Sox2, Prox1) remained co-expressed into oldest age.
Our data suggest that in the adult human hippocampus neurogenesis-associated features that have been identified in rodents show patterns, as well as qualitative and quantitative age-related changes, that are similar to the course of adult hippocampal neurogenesis in rodents. Consequently, although further validation as well as the application of independent methodology (e.g. electron microscopy and cell culture work) is desirable, our data will help to devise the framework for specific research on cellular plasticity in the aging human hippocampus.
Adult hippocampal neurogenesis, i.e. the production of new granule cell neurons in the adult hippocampus, has captured the imagination of a wide audience and is beginning to influence hypotheses for clinical medicine. Adult neurogenesis is conserved in all mammalian species studied so far including non-human primates
Adult hippocampal neurogenesis in mice has been described in considerable detail, and distinct developmental stages have been identified
DCX is a brain-specific microtubule-associated protein whose exact function is not yet known. It appears to act as microtubule stabilizer in a way that is particularly pertinent to migration
DCX can serve as a marker for new neurons in the adult rodent hippocampus as long as the transient nature of its expression is taken into account. In addition, DCX is expressed elsewhere in the rodent (and human) brain, where no link to adult neurogenesis can be made
We also made a first attempt to gather semi-quantitative information on how expression of such markers might change across the lifespan as previously described for rodents
We performed western blot analyses of normal hippocampal tissue taken from a deceased child and resected hippocampus from a subject suffering from temporal lobe epilepsy with ammonshorn sclerosis (
Lane 1: Fetus of GW 20, elective abortion, cortical tissue; lane 2: child of age 3 years, healthy hippocampus; lane 3: adult of age 35 years, suffering from temporal lobe epilepsy (TLE), resected hippocampal tissue. The doublecortin antibody (sc-8066) reacted with proteins in the 45 kDa range, which corresponds to the MGrel. of DCX. The signal is strong in the fetal brain and much weaker in the juvenile and adult samples. The GAPDH signal at the 36 kDa position serves as MGrel and loading control.
We first examined DCX expression in the developing human DG. At GW11 we found indication of dense DCX expression on the level of both
A, B, The overlapping staining indicates the co-expression of DCX mRNA and protein. In cortical tissue (A), as well as in the ganglionic eminence (B), the mRNA is not restricted perinuclearly, but is also found in the transverse oriented processes of the round-shaped cells. C, Negative control of the
A, B, From birth to adulthood (A: 1 day, B: 28 years of age), hippocampal DCX labeling reveals a distinct staining pattern in the granule cell layer and hippocampus proper. C, In most granule cells (neonatal DG), the mRNA signal occurs outside the nucleus in a small cytoplasmic rim. D, Higher magnification of the DG area marked in (B). E, Application of a labelled sense cRNA probe to a serial section of (B). The absence of sense-probe labelling confirmed the specificity of the detection system for DCX-mRNA/cRNA antisense probe hybrids. Differential Interference Contrast. Scale bars A, B, 1 mm; C-E, 25 µm.
We focused on DCX expression in the DG. DCX-positive (DCX+) cells in the DG could be found in samples across the entire lifespan between 1 day and 100 years of age (exemplary in
Postnatal, numerous DCX+ neurite-bearing cells are scattered throughout the GCL (A). In the adulthood, DCX+ cells are gradually less differentiated in relation to growing age. Initially, some cells show a neuron-like morphology (B,D), whereas the majority of them appears increasingly more undifferentiated and strongly DCX-stained (B-F). In the juvenile mouse hippocampus a multitude of DCX+ cells delineate the GCL formation (G,I). Very old mice, however, exhibit only a few DCX expressing granule cells (H,K). DAB immunohistochemistry of DCX in paraffin embedded brain samples. Length of scale bars as indicated.
Increasing cellular differentiation (dendrites and axons) can be observed starting from the hilar polymorphic layer/subgranular layer (PML/SGL) throughout the GCL up to the inner molecular layer. However, undifferentiated round-shaped cells were always found throughout the entire GCL. This picture is a montage of DCX labeled granule cell and sorted increasingly maturated from left to right. Dendrites are directed towards the granule cell layer (top) and the axons towards the hilus (bottom). Scale bar, 10 µm.
We next established whether DCX expression in the human DG could be linked to other markers that in the rodent are associated with adult hippocampal neurogenesis (Complete list in
C–E, The SGZ of juvenile and adult subjects exhibits immature DCX+ cells co-expressing
N, Immature neurons express DCX together with postmitotic neuronal marker NeuN; here at age of 75 years. O, Expression of PCNA in GFAP+ cells might be indicative of ongoing stem cell activity in the SGZ but might also reflect classical astrocytic proliferation or DNA repair (see
We first examined markers for cell proliferation. PCNA expression in DCX+ cells could be detected at all ages (
Based on our previous studies in rodents and reports from the literature we next focused on nestin
Sox2 is a precursor cell marker that in rodents can be detected in a proportion of hippocampal progenitor cells
Prox1 expression, a transcription factor closely related to granule cell development in rodents
Although being glutamatergic neurons, granule cells can co-express GABA and GABA synthesizing enzymes. We detected GAD67/DCX co-localization up to 89 years of age (
Finally, we attempted a semi-quantitative assessment of neurogenesis in the human DG (including the hilar polymorphic layer) across the lifespan. We found that the density of DCX+ cells showed a log-log linear course of the regression curve over the period of life (
Comprising data from 45 subjects, between 1 day and 94 years, the number of labeled cells in the DG were plotted against the age of the individual. The regression curve shows a log-log-linear decrease (Pearson’s Correlation Coefficient (PPC): -0.95102;
Research on adult hippocampal neurogenesis is often justified by the possible implications that this process might have for human cognition in health and disease. Given this claim the paucity of direct information about adult neurogenesis in the adult and aging human brain is disturbing. We here offer a large data set that provides qualitative and semi-quantitative information about neurogenesis-associated features in the human hippocampus. These data do not offer per se proof of adult neurogenesis in each particular sample but together provide valuable information about the presence of neurogenesis-associated markers in the adult human brain and their change with increasing age.
We show that during the first years of life the human DG shows an expression pattern of DCX in numerous combinations with markers of proliferation, precursor cells, maturation, and differentiation that is completely consistent with the information about hippocampal neurogenesis in rodents (
The frame indicates the stages of adult neurogenesis covered in this study: all additional markers were tested together with DCX. Marker combinations that we could detect in our samples are marked with a green check. Not all markers were found at all ages, see
This pattern changes considerable with increasing age but does so in a way that is again more or less consistent with what is known about the age-related changes in adult neurogenesis in rodents (see
Adult hippocampal neurogenesis in mice and rats shows a sequence of marker expressions that has been described in increasing detail
Presence of stem-like cells was suggested by the detection of Sox2
Because precursor cells are by definition proliferating cells markers of cell division help to substantiate the nature of the DCX-positive cells. Reif et al., who based their conclusions on adult human neurogenesis on immunohistochemistry for proliferation marker Ki67 and studied 60 samples between 25 and 68 years, did not detect positive cells in all cases and found Ki67 up to 62 years (
In rodents, the phase of DCX expression is also the phase during which cells are eliminated by cell death
DCX is associated with neurogenesis in the dentate gyrus of adult rodents but expression of DCX is not limited to the context of adult neurogenesis. The overlap with other markers corroborates the idea that the link might exist in the human dentate gyrus as well but in humans as in rodents, DCX is not a neurogenesis marker with high specificity. In the rodent hippocampus its sensitivity is high.
In the murine dentate gyrus DCX shows a nearly complete overlap with the expression of the polysialilated form of the Neural Cell Adhesion Molecule (PSA-NCAM)
Of particular interest are the round-shaped DCX-positive cells, which from their morphology and co-marker expression might relate to the DCX-positive progenitor cells, type-2b and -3 that have been described for the rodent brain. This extrapolation is speculative because further information about the exact course of adult neurogenesis would be needed. Our data might provide at first indication. More detailed analysis is hindered by the fact, however, that these cells are only a subset of all DCX-positive cells and in any given section only few markers can be tested. To describe the course of adult neurogenesis and its dynamics a study based on a cohort of cells that can be followed through the stages of development would be needed. This, however, requires the BrdU-method (or a virus-based approach), which is not readily applicable to humans.
If we take the BrdU-method as the Gold standard (which appears to be generally accepted in the field), there is currently no evidence of adult hippocampal neurogenesis in humans older than 72 years, the oldest sample in the Eriksson study. Because a new BrdU-based study in even older subjects is unlikely to be presented anytime soon, surrogate markers and cumulative supportive evidence gains weight for this age bracket. Single markers will generally not suffice to prove or disprove adult neurogenesis in a particular human sample but contextual information as offered here might help to judge these cases. The use of markers is generally affected by the many caveats associated to investigating human post mortem samples. Combination of markers will decrease the likelihood of false-positive results, but the most serious concern is that with increasing age, markers that in young age are indeed associated with neurogenesis might be increasingly indicative of degeneration and cell death. DNA-synthesis
In any case, however, and contrary to the apprehension that with increasing age the amount of non-specific labeling and false-positive marker expression might generally increase, we found a reduction in DCX expression and a reduction in marker overlap. In no case, overlaps not present at a younger age were found in old age. To our knowledge ours it the first study to undertake the analysis of so many histological markers in human samples and this marker loss by itself constitutes an interesting result. Most published studies have been concerned with the age-dependent appearance of histological features, most notably neurofibrillary tangles, senile plaques, deposition of age-pigment, or the increased proliferation of macro- and microglial cells.
Still, the subjects in our series were not healthy but died for extracerebral reasons listed in
PCNA is co-factor of DNA polymerases and is expressed during G1 and S phase of the cell cycle. PCNA is critically involved in DNA replication
We also assessed the temporal change in the number of DCX+ cells with age and found an exponential decrease, again similar to the decrease in neurogenesis in rodents
In fact we found in the present study that the panel of overlapping markers was exhaustive up to 30 to 40 years (
Stereological studies of the human DG, covering samples up to 101 years of age, determined that the number of granule cells in humans is very stable across the lifespan but varies greatly
The reduced presence of neurogenesis-associated features in the aged human hippocampus suggests that, as in rodents, neurogenesis is likely to occur on a very low scale, at least if compared to younger age. One of the intriguing questions is thus, how so few new neurons might be functionally beneficial at all. The alternative position to this approach is the theory that because in the course of evolution the ability to produce neurons in adulthood became increasingly restricted, humans are different from animals in that their hippocampus does not rely on this option to alter its neuronal network structure any longer
We have proposed models that help to explain why already very few neurons might make relevant functional contributions in the particular network situations of the hippocampus and we believe that such models also explain why postnatal and young-adult neurogenesis is necessary but increasingly dispensable with increasing age
Overall, our results suggest a pattern of adult neurogenesis in the aging human dentate gyrus that shows large similarities to rodents. Our data alone cannot prove or disprove the true presence or absence of neurogenesis at any age and the consideration of isolated samples might be misleading. Our key results lie in the fact that we see an expression pattern for key markers for adult neurogenesis also in the human brain and that this expression pattern changes with time both qualitatively and quantitatively.
Full description of the experimental procedures can be found in the supplemental material (
Autopsy cases were selected from the archive at the Department of Neuropathology, University Hospital, Freiburg, Germany, according to age, sex, postmortem interval and lack of clinical or postmortem evidence of neuropathology. Paraffin sections from 3 fetal brain tissue samples (GW11, 20 and 40), hippocampal pieces from a patient suffering from temporal lobe epilepsy and from 51 deceased persons without central nervous damage aged from 1 day to 100 years were analyzed (
Western blotting and
Standard protocols were followed. The detailed description is found in the Supplementary Material. Antibodies are listed
Supplemental material and methods.
(0.06 MB DOC)
Test of juvenile and adult human GCL specimens for potential perimortal hypoxia. A, No co-expression of DCX with the Heat Shock Protein 27 (HSP27) was detected and no signs of microglia activation (CD68) were found in the samples. B, Similarly, microglial marker Glut1 and Hypoxia-Induced Factor 1α (HIF-1α) were not increased. C, Matrix MetalloProteinases (MMP) are important for migration of young cells within established neural networks. Here, MMP9 shows strong co-labeling with DCX, whereas the same cell is not stimulated to express Vascular Endothelial Growth Factor A (VEGF-A), which would be indicative of a hypoxic milieu. D, Proliferation markers like PCNA might show expression in cells undergoing apoptosis. Here, a DCX+ cell in a 2 month-old GCL presents signs of programmed cell death by activated-Caspase-3 (Casp-3). E, Whereas apoptosis is very physiologic in neonatal age and crucial for neuronal network consolidation, the DG of a 58 years-old subject revealed no signs of degradation. The lack of PCNA- and activated caspase-3 labeling in the presented cell argues against the interpretation that DCX expression in higher age is induced by stress factors. Scale bars, 5 µm and 10 µm as indicated.
(0.33 MB JPG)
A, B, Proof of reactivities of anti activated-caspase-3 (A) and anti heat-shock protein HSP27 antibodies (B) in a glioblastoma multiforme specimen. Positive cells are devoid of DCX and PCNA immunoreactions, resp. C, Technical control incubation: Omitting primary antibodies, the incubation of serial sections with secondary antibodies alone failed to generate any specific fluorescence signal. D, E, Search for DCX expressing neuronal cells outside the hippocampus. Bipolar DCX+ cells could be detected in the parietal neocortical parenchyma (LII) (D) as well as in the temporal migratory stream (TMS) of the piriform Cortex/Amygdala (E). There are neither co-labelings with the microglial marker CD68 and the heat-shock-protein HSP27 nor with the cell type specific markers NeuN and GFAP. Scale bar length as indicated.
(56.30 MB TIF)
PSA-NCAM expresssion. PSA-NCAM is expressed in a small hippocampal SGZ cell of a 68 years-old male. DAB staining.
(0.60 MB JPG)
Estimation of PCNA+ cell densities in the DG across the entire human lifespan. Comprising data from 25 patients, aged between 1 day and 100 years (see
(0.10 MB TIF)
Doublecortin mRNA. Generation, purification and cloning of human DCX-specific cDNA inserts aimed to generate DIG-labeled cRNA probes for northern blotting and in situ hybridization. A, After bulk separation in a preparative 1.5% agarose gel, the fetal human brain DCX-PCR product (510 bp, lane 2) was recovered and in B, ligated into a pSPT19 vector (3.604 bp, lane 1), cloned and approved for correct insertion by digestion with EcoRI and HindIII resulting in a linearized vector (3.104 bp) and the DCX insert (lane 2) designed for the cRNA-DIG probe generation by in vitro transcription. C, Separation of total RNA from fetal brain in an ethidium bromide stained acrylamide gel (lane 1) showing strong fluorescence of the 18S and 28S ribosomal rRNAs. D, After blotting on nylon membrane the immobilized RNA was simultaneously hybridized with the DIG-labeled antisense probes against DCX and β-actin mRNAs and processed for visualization of the mRNA/cRNA-DIG binding sites by the enhanced chemiluminescence method. Besides the very strong signal of β-actin mRNA serving as housekeeping transcript and loading control a faint band of DCX mRNA was visible at the correct position of 9.54 kb.
(0.26 MB TIF)
Analysis of perimortal hypoxic changes.
(0.04 MB DOC)
Data of subjects included in the study.
(0.13 MB DOC)
Quantification of PCNA and GFAP co-expressing undifferentiated cells in the dentate gyrus.
(0.05 MB DOC)
List of primary antibodies.
(0.10 MB DOC)
Fluorochrome conjugated secondary antibodies and optical/technical parameters for their detection in the Leica TCS NT confocal laser scanning microscope.
(0.03 MB DOC)
We dedicate this article to the memory of Peter Eriksson. We thank Marco Prinz for generous support, Jürgen Schulte-Mönting for statistical calculations, Gesina Grothe for cell counting, Monika Hock, Christina El-Gaz, Barbara Herbstritt, Karin Strasser, and Viviana Sverdlick for excellent technical support, and Marcia Machein, Andrea Ullrich and Klaus Müller for valuable discussions.