Human adult neurogenesis seems unlikely: Found signatures of adult neurogenesis in mouse, pig, and macaque dentate gyrus, but not in humans, adding to a growing body of evidence that this process is likely lost in humans

Transcriptomic taxonomy and neurogenic trajectories of adult human, macaque, and pig hippocampal and entorhinal cell. Daniel Franjic et al. Neuron, Volume 110, Issue 3, February 2 2022, Pages 452-469.e14. https://doi.org/10.1016/j.neuron.2021.10.036

Highlights

• Single-nucleus RNA-seq of adult hippocampal-entorhinal cells in human, monkey, and pig

• Transcriptomic signatures of adult neurogenesis in mouse, pig, and monkey but not human

• Excitatory neuron diversification delineates transitions from 3- to 6-layered cortex

• METTL7B defines subregion-specific excitatory neurons and astrocytes in primates

Summary: The hippocampal-entorhinal system supports cognitive functions, has lifelong neurogenic capabilities in many species, and is selectively vulnerable to Alzheimer’s disease. To investigate neurogenic potential and cellular diversity, we profiled single-nucleus transcriptomes in five hippocampal-entorhinal subregions in humans, macaques, and pigs. Integrated cross-species analysis revealed robust transcriptomic and histologic signatures of neurogenesis in the adult mouse, pig, and macaque but not humans. Doublecortin (DCX), a widely accepted marker of newly generated granule cells, was detected in diverse human neurons, but it did not define immature neuron populations. To explore species differences in cellular diversity and implications for disease, we characterized subregion-specific, transcriptomically defined cell types and transitional changes from the three-layered archicortex to the six-layered neocortex. Notably, METTL7B defined subregion-specific excitatory neurons and astrocytes in primates, associated with endoplasmic reticulum and lipid droplet proteins, including Alzheimer’s disease-related proteins. This resource reveals cell-type- and species-specific properties shaping hippocampal-entorhinal neurogenesis and function.

Discussion

We report an extensive single-cell transcriptomics analysis of several anatomically defined cell populations in the adult human, macaque, and pig hippocampal-entorhinal system. Our findings reveal fundamental species differences in adult hippocampal neurogenesis and delineate the molecular diversity of the cytoarchitectural transition from the allo- to the neocortex. These results also outline genes that are selectively enriched in certain species and cell types that may have a role in the specific biology and/or pathology of the hippocampal-entorhinal system

Unlike recent studies that mostly rely on one or two key markers (e.g., progenitors [nestin]; neuroblasts and immature granule cells [DCX]) (Boldrini et al., 2018; Moreno-Jiménez et al., 2019; Tobin et al., 2019), single-cell RNA-seq studies are more comprehensive because they leverage combinatorial gene expression profiles to identify cell populations more robustly (Hochgerner et al., 2018). This approach also allows cross-species analysis, amplifying rare signals within a single species that may be masked when analyzed separately. Our cross-species analysis allowed identification of the neurogenic lineage in the mouse, pig, and macaque, which was virtually absent in the human. We only detected one cell with the transcriptomics profile characteristic of nIPCs and one with putative neuroblast profile among 32,067 granule cells (0.003%) in our adult human DG samples, a proportion considerably lower than the expected 0.09%–3.8% neuroblasts according to previous DCX immunostaining or 14C incorporation studies of the adult human HIP (see Table S2 for data and relevant studies).

The same analytic strategy detected much higher proportions of neuroblasts in the other analyzed species (mouse, 6.6%; pig, 55.6%; macaque, 2.0%; Figure 2B; Table S3). These proportions were higher than those estimated previously based on progenitor proliferation and identification of neuroblasts markers such as DCX (Table S2), suggesting that more studies are needed to fine-tune detection of these neurogenic populations. However, this apparently lax detection protocol confirms that our parameters are unlikely to have missed any appreciable neuroblast populations among the large pool of surveyed human DG granule cells, even if they might exhibit an ambiguous profile.

Alternative confounding of our cross-species integrative analysis from possible human-specific transcriptomics changes was ruled out when human UMAP layouts did not include any clustering of neurogenic cells adjacent to the mature granule cell cluster. Likewise, the possibility that human neuroblasts exist in our samples but that their transcriptomics profile differs from other species and blends with related cell populations is lessened by findings that all neurogenic lineages preceding mature granule cells were absent in human DG samples (Tables S2 and S3).

We also extended our findings to existing snRNA-seq data of the adult human HIP. We reappraised the identity of a recently reported neural progenitor cluster (Ayhan et al., 2021) marked by LPAR1, a gene reported to mark mouse DG neural progenitors (Walker et al., 2016; Hochgerner et al., 2018). Our analyses indicated that this cluster actually represented doublets formed by oligodendrocytes and granule cells (Figure S3S). In addition, reanalysis of the pioneer HIP data (Habib et al., 2017) by Sorrells et al. (2021) showed that the cell cluster labeled as neural stem cells was actually characteristic of ependymal cells.

Analysis of DCX transcripts in all the analyzed species showed expression in mature neurons, mostly InNs, and in glial cells, indicating that DCX expression is not exclusive to DG neuroblasts (Figures 3A and 3B). This pattern is in agreement with the reanalysis of the data from Habib et al. (2017) (Sorrells et al., 2021). All transcriptomics analyses performed so far suggest a lack of neurogenic cell populations in the adult human DG.

At the protein level, DCX was, with a few exceptions (Figure S3L), present exclusively in DG cells resembling neuroblasts and immature granule cells in all analyzed non-human species. Also, cells with immature morphology could be detected in other areas, such as the EC of the macaque or the pyriform cortex of the mouse, as described previously (Gómez-Climent et al., 2008; Zhang et al., 2009). In humans, there is intense controversy regarding DCX immunostaining in the human DG, with some reports showing negative results (Dennis et al., 2016; Cipriani et al., 2018; Sorrells et al., 2018, 2021) and others describing DCX-IL cells (Knoth et al., 2010; Epp et al., 2013; Boldrini et al., 2018; Le Maître et al., 2018; Moreno-Jiménez et al., 2019, 2021; Tobin et al., 2019). We detected clear DCX-IL cells in the amygdala and occasionally in the EC, but we could not find DCX-IL cells resembling neuroblasts in the DG in the same tissue sections. These inconsistencies in detecting DCX-IL cells in the adult human DG cannot be fully attributed to postmortem denaturation and degradation of DCX protein because DCX-IL cells were clearly detected in samples with prolonged PMIs (Figures S3D, S3E, and S3I–S3L). Moreno-Jiménez et al. (2019) reported an intensive protocol for antigen retrieval as a necessary step to label DCX cells in the human DG. However, they reported no positive cells in the EC, a relatively common finding in our study (Figure S3D) and another (Sorrells et al., 2021) using conventional antigen retrieval. Because our analysis did not reveal neuroblasts at the RNA or protein level (using diverse antigen retrieval methods), the question remains what those previously reported cells could be. Apart from underappreciated non-specific and off-target effects (Sorrells et al., 2021), those studies could label mature granule cells and InNs that might contain low levels of DCX protein that was detected specially after multi-step antigen retrieval. In support of this hypothesis is the fact that the faintly immunolabeled cells we detected mostly in the vicinity of the granule cell layer, exhibited the morphology of mature InNs, and some co-labeled with antibodies against GAD1, a marker of InNs (Figures 3E and S3M–S3Q). This faint staining is far from the strong staining and well-defined morphology of somata and dendrites revealed in the EC and in the amygdala (Figures S3D and S3E) and is similar to the light DCX immunostaining reported previously (Seki et al., 2019). Thus, our conclusion is that DCX protein might be expressed at very low levels in InNs or in some mature granule cells that can be lightly immunolabeled under normal antigen retrieval but can show more intense and widespread staining under more elaborate tissue treatment and stringent conditions of antigen retrieval. In fact, Figure 2I from Moreno-Jiménez et al. (2019) shows that around 75% of DCX-IL cells were colocalized with NeuN (RBFOX3, 75%), a marker of mature granule cells, and 91% of DCX-IL cells were also positive for Prospero homeobox1 (PROX1), a transcription factor expressed by granule cells that is also expressed by InNs generated in the caudal ganglionic eminence (Ma et al., 2013; Laclef and Métin, 2018), supporting the possibility that most DCX-IL cells might actually be mature granule cells or InNs.

Regarding RNA analysis, although the PMI for humans was longer than for other analyzed species, human brains were kept at 4°C for most of the PMI period, whereas the pigs used as controls for PMI were kept at room temperature. This warm PMI will likely exacerbate the postmortem effects, but those conditions were not an obstacle to detect the neurogenic pathway in this species. It could be argued that the neurogenic pathway in the human DG is not detected because our snRNA-seq strategy might preferentially exclude neurogenic cells in humans. However, it seems extremely unlikely that it will affect all cell types in the neurogenic lineage, from progenitors to neuroblasts, and only in human. Overall, the most parsimonious interpretation of the combined results from our RNA transcript analysis and the DCX protein study is that, contrary to the other analyzed mammals, ongoing baseline neurogenesis does not occur or is extremely rare in the adult human DG.

Similar species-related and cell-specific transcriptomics profiling that characterizes neurogenic potential also outlines the transition from allocortical to neocortical domains in the hippocampal-entorhinal system and shows that ExNs are the main drivers of the differences between subfields (Figure 4), which evidences a richer complement of ExNs than traditional descriptions based on cytoarchitecture. Our analysis provides a primer to further study these populations and characterize the possible implications for hippocampal-entorhinal physiology. These data refine our understanding of the evolution of the allo-, meso-, and neocortex. The transcriptomics signatures we developed strongly suggest homology between the mammalian allocortex and specifically deep layers of the EC and neocortex.

Among the genes contributing to the layer transition, we identified METTL7B to be important in hippocampus physiology and function. We found that METTL7B, equipped with methyltransferase activity, interacts with important AD-related proteins (e.g., APP, LRP1, RTN3, and RTN4). Importantly, our results suggest that these functional interactions in a subset of ExNs and astrocytes seem to be phylogenetically specific to Old World monkeys and apes (parvorder Catarrhini), species that show more marked signs of pathology related to aging, such as AD, than other species (Perez et al., 2013; Finch and Austad, 2015; Edler et al., 2017; Paspalas et al., 2018). Overall, our analyses provide multiple vignettes of how this resource can be used to identify cell types and genes that might be functionally relevant for the biology of the hippocampus, allowing inter-species comparisons.