Considerations for Assessing the Extent of Hippocampal Neurogenesis in the Adult and Aging Human Brain

Adult hippocampal neurogenesis (AHN) is implicated in brain adaptations and disease pathogenesis. A seminal study showed adult-born neurons in the subgranular zone (SGZ) of the dentate gyrus (DG) in cancer patients 58–72 years of age, detecting bromodeoxyuridine co-localization with neuronal markers (Eriksson et al., 1998). Subsequently, human AHN was reported using immunohistochemistry targeting markers expressed by neuronal cells at different maturational stages, in situ hybridization (ISH), and ¹⁴C decay-defined neuronal age (Spalding et al., 2013). Detection of AHN markers is dependent on methodological approaches such as brain tissue processing and fixation, postmortem interval, and other factors affecting tissue antigenicity and preservation of proteins and mRNAs (Kempermann et al., 2018). This discussion intends to clarify the divergent findings between our group (Boldrini et al., 2018) and the Alvarez-Buylla group (Sorrells et al., 2018; Paredes et al., 2018) and elucidate key factors surrounding conflicting conclusions.

In subjects age 14–79 years without neuropsychiatric disease or treatment, we found declining quiescent neural progenitors (QNPs), neuroplasticity, and angiogenesis in the anterior-mid DG of older individuals, but not fewer intermediate neural progenitors (INPs, co-expressing nestin and the transcription factor Sox2), immature neurons (co-expressing doublecortin [DCX] and polysialylated neural cell adhesion molecule [PSA-NCAM]), or mature granule neurons (GNs, expressing neuronal nuclear marker [NeuN]). An important element to take into account when evaluating why some report lack of AHN past adolescence (Sorrells et al., 2018) is the potential impact of missing information on subjects’ neuropsychiatric diagnosis, medications, and recent drug usage, which affect AHN levels selectively in anterior DG (Boldrini et al., 2012).

It is pertinent to consider caveats associated with stem cell characterization based on cell surface and proliferation markers since interpretations of cellular morphology may be subjective. Ki-67, expressed in all cell-cycle phases except G0, exhibits phase-specific staining patterns. Recognizing phase-specific expression patterns is essential. We showed that Ki-67 expression is strictly nuclear during DNA replication, surrounded by eosinophilic cytoplasm immunoreactive for nestin (Figures 1I, 1K, and 1L in Boldrini et al., 2018). In metaphase of mitosis, however, Ki-67 is visualized in the cytoplasm (Braun et al., 1988). Ki-67+ cells encompass all cell types including non-neuronal proliferating cells, such as proliferating endothelial cells in capillaries (Figure 2 in Boldrini et al., 2012; Figure 4C in Boldrini et al., 2018), explaining their higher prevalence than nestin+ NPCs.

Despite showing evidence of DCX+ processes in human DG (Figure 3C in Boldrini et al., 2018), the differential staining pattern of DCX remains scrutinized. While DCX+ dendrites are visible in mice and embryonic/perinatal human development, they are harder to visualize in adult human DG (Kempermann et al., 2018). In rats, dendritic growth slows with age (Hattiangady and Shetty, 2008), possibly explaining why DCX+ dendrites can be easily observed in the younger samples of Sorrels et al.

We hypothesized that commercially available anti-DCX antibodies may work best in animal tissue, and less effectively detect adult human DCX+ dendrites. We performed double immunohistochemistry and double immunofluorescence for DCX and neurofilament (NF) to identify dendrites of DCX+ cells that did not immunoreact to anti-DCX antibodies. We found DCX+ cells with NF+ dendrites extending through the GCL and elongated DCX/NF+ cells, showing a migrating morphology in the adult human SGZ (Figures S1A–S1I). The known neuronal expression of NF supports that the DCX/NF+ cells we detected are not glia. The NF antibody we used (Sigma-Aldrich:N0142) is specific for the 200 kDa, high-molecular-weight neurofilament not expressed by mature GNs, as reported in developing rat hippocampus (Lopez-Picon et al., 2003).

Our data showing stable INPs with age, combined with an age-related decline of the QNP pool, support mouse data reporting a sharper decrease in QNPs than in amplifying progenitors with age (Encinas et al., 2011). QNPs expressing Sox2 are necessary for maintaining multipotency. The “disposable stem cell” model hypothesizes that QNPs exit quiescence after undergoing asymmetric division and eventually differentiate into mature astrocytes after generating amplifying progenitors (Encinas et al., 2011). As such, we expect that some Sox2+ QNPs differentiate into glia. However, Sox2/Nestin+ expression is followed by a period of neuronal differentiation in which cells continue to express Nestin and not Sox2. These INPs are early indicators of AHN and were steady across ages.

We further showed stable numbers of DCX/PSA-NCAM+ immature neurons in the DG across our samples, viewing larger PSA-NCAM+ pyramidal cells as indicators of plasticity (Figure 2 in Boldrini et al., 2018), not evidence of AHN. Age-related declines in PSA-NCAM+ cells with stable DCX/PSA-NCAM+ cells may reflect age-associated declines in neuroplasticity or migration not diminishing neuronal maturation or survival.

We previously showed that AHN occurs within an angiogenic niche in the adult human SGZ with clusters of nestin+ NPCs associated with the vasculature; a higher density of capillaries and NPCs are associated with antidepressant treatment (Figures 1A–1C, 4A, 4B, and S2 and Table S1 in Boldrini et al., 2012). We observed similar clustering of Sox2/nestin+ INPs and DCX/PSA-NCAM+ immature neurons in the SGZ.

Limitations to marker studies were addressed using RNAscope technology (https://acdbio.com/) on flash-frozen hippocampal tissue. We employed unbiased stereology to quantify DCX mRNA⁺ cell density using ISH in three sections of the anterior-mid DG from subjects with no neuropsychiatric disease or treatment, antidepressant-treated, and untreated subjects with major depressive disorder (MDD) (n = 5/group, age 19–67 years, postmortem interval = 6–27 hr) and found a trend toward lower DCX mRNA⁺ cell density in SGZ/GCL in MDD in this preliminary sample (Figures S1J–S1N). Comparison between cells positive for DCX mRNA and protein (Boldrini et al., 2018) cannot be made since the experimental groups differ with regard to psychiatric illness. In controls, we find no decline with age of DCX mRNA⁺ cell density (Figure S1N), comparable to our results related to DCX-immunoreactive cells. We suspect our detection of DCX mRNA⁺ cells and the negative findings of Sorrels et al. could be attributed to tissue quality or processing. We followed a protocol for fresh frozen tissue that requires shorter fixation times and does not necessitate harsh antigen retrieval that may denature mRNA.

A range of methodologies is necessary to characterize the complex neurogenic niche. Studying high-quality brain tissue from extensively characterized subjects is essential to reproducibility and unbiased interpretation of results. We thank our colleagues for their interest in our work and welcome further discussion and collaboration.