Adult neurogenesis in the short‐lived teleost Nothobranchius furzeri : localization of neurogenic niches, molecular characterization and effects of aging

1The continued production of neurons throughout adulthood, in at least in some circumscribed brain regions, has been observed in all vertebrate species so far studied. In mammals, adult neurogenesis is restricted to two principal stem cell niches in the telencephalon: the subependymal zone (or subventricular zone, SVZ) which generates GABAergic and dopaminergic interneurons of the olfactory bulb and the subgranular zone (SGZ) of the dentate gyrus. These areas contain adult neuronal stem cells (aNSCs) of glial phenotype, which are able to self-renew. aNSCs generate an intermediate precursor, which gives rise to a transient amplifying neuroblast and finally to neurons. These different stages can be identified by a combination of morphological criteria and the expression of molecular markers. For example, aNSCs in the hippocampus are characterized by the expression of SOX2, glial fibrillary acid protein (GFAP), and activated Notch signaling, whereas the amplifying neuroblast expresses a combination of PSA-NCAM, doublecortin (DCX), and the mitotic marker Ki-67. Doublecortin expression (but not expression of mitotic markers) persists for some time in young terminally differentiated neurons. Neurons generated during adulthood are functional, integrate into existing circuits, and are proposed to be involved into some specific aspects of behavioral plasticity.

2Adult neurogenesis is a highly plastic process, which can be enhanced by environmental stimulation and physical exercise and it is modulated by epigenetic mechanisms, microRNAs, and growth factors.

3Neurogenesis in the SGZ decreases exponentially with age in rodents, dogs, and humans. This decay is initiated during early postnatal life, and, in adults, neuronal production in the SGZ is < 10% of what observed during puberty. Age-dependent decrease in neurogenesis is observed in SVZ as well, but it is less dramatic than in the SGZ.

4Decreased adult neurogenesis is the consequence of quiescence of aNSCs, which are still present, but not active, in the old brain and can be re-activated by physiological and pathological stimuli.

5In contrast to mammals, adult neurogenesis is extensive in teleost fish and is observed basically in all brain regions. Systematic anatomical studies in zebrafish and medaka have identified at least 16 different neurogenic niches. In fish, aNSCs are mostly associated with the ventricular system. In the fish telencephalon, aNSCs have the typical morphology of radial glia and share several molecular markers of mammalian aNSCs. A pallial and subpallial neurogenic niche can be identified in fish and these are homolog to the SGZ and SVZ of mammals. Analysis of viral tracing has directly demonstrated that adult radial glial is self-renewing and pluripotent, proving its stem cell identity. aNSCs in the optic tectum (OT) and cerebellum are not of glial nature and express neuroepithelial markers.

6Studies on zebrafish used young adults of age 3–6 months and maximum reported lifespan for this species is 66 months and there are no quantitative reports on age-dependent modulation of adult neurogenesis. As fish show substantial growth during adult life, it is unclear to what extent adult neurogenesis is responsible for neuronal turnover or rather reflects addition of new cells in the context of continuous brain growth.

7Gliosis, measured as up-regulation of GFAP in astroglia, is a well-described hallmark of mammalian aging. In fish, however, radial glia is the predominant glia type and retains its neurogenic function in adulthood. It is therefore of interest to explore whether also radial glia undergoes gliosis as function of age.

8To investigate the effects of aging on adult neurogenesis in teleosts, we used a novel animal model: the short-lived fish Nothobranchius furzeri. This species has a captive lifespan of 3–7 months depending on the strain and it is an emerging model organism for biological investigations into aging. In particular, we demonstrated that N. furzeri undergoes age-dependent neurodegeneration and cognitive impairments, which in mammals often correlate with reduced adult neurogenesis.

9Aims of the present paper were the following:

10To localize neurogenic niches in the adult brain of N. furzeri.

11To investigate the expression of conserved molecular markers of neurogenesis in this species.

12To quantify the effects of age on adult neurogenesis in the OT.

13To study the effects of aging on the radial glia.

14To label mitotically active cells, we injected the nucleotide analog 5-ethynil-2′deoxyuridin (EdU). Whole-mount detection of Edu+ cells showed a widespread distribution in the brain of young (5–7 weeks old) subjects 4 h after injection (Fig. 1A–E). We concentrated on those areas located in telencephalon, OT, and cerebellum, which were subject of in-depth studies in zebrafish and medaka.

15In the telencephalon, three distinct regions can be recognized: A niche of tightly packed high proliferating cells, which starts quite sharply at the rostro-ventral conjunction of the two telencephalic hemispheres (Fig. 1B–E, region I). This region has a subpallial origin, is homologous to the SVZ of mammals, and is observed in the telencephalon of other teleost species. Neurogenic cells are also scattered on the dorsal and lateral surfaces of the telencephalon in various teleost species. Because of developmental eversion of the telencephalic vesicles, this region of pallial origin is facing the third ventricle and it is proposed to be homologous to the subgranular neurogenic niche of mammals. In N. furzeri, the pallial neurogenic region can be subdivided into two different areas region II and III (Fig. 1B,D): Region II is placed dorsally and rostrally and shows a much lower level of proliferation as compared to region I; region III is placed ventro-caudally and shows a level of proliferation intermediate between regions I and II. Region II and region III are separated by a ventro-lateral patch of telencephalic surface devoid of proliferative cells (Fig. 1D, indicated with *). This area is observed also in zebrafish and is suggested to correspond to the pial rather than the ventricular surface. Proliferation in region III is concentrated on its ventro-caudal portion (Fig. 1B,D arrowhead). An area of high proliferative activity is observed in the preoptic segment of the ventricular system (region IV, Fig. 1C,E). Intense neurogenic activity was observed in the dorsomedial, posterior, and ventral margins of the OT (regions V and VI; Fig. 1A,C). A clearly visible neurogenic niche is located along the midline surface of the cerebellum (Fig. 1A, region VII). Further distinct niches visible in whole-mount preparations are located at the caudal margin of the cerebellum (Fig. 1A, region VIII) and lining the margin of the fourth ventricle, (Fig. 1B, region IX).

16Sagittal overviews of the N. furzeri brain labeled with EdU are shown in Fig. S1 (Supporting Information).

17Double-labeling with EdU and the proliferation marker proliferating cell nuclear antigen (PCNA) was performed in coronal and horizontal sections of 5- to 7-week-old subjects 4 h after injection (Fig. 1F–N, Figs S2 and S3). Neurogenic niches identified by the presence of EdU+ cells always correspond to regions labeled with PCNA. In telencephalon, regions I, II, and III can be readily appreciated in horizontal and coronal sections. Region I (Fig. 1I) extends from the niche visible in whole-mount preparations along the third ventricle in caudal direction. Analysis of sections confirms the presence of a region II (Fig. 1G) and a region III. The concentration of dividing cells in the caudal margin of the telencephalon in region III is clearly visible in horizontal sections (Fig. 1F, white arrowhead). In zebrafish, the pallial neurogenic niche contains slow-cycling precursors and the subpallial niche contains faster-dividing precursors. These can be distinguished based on the percentage of colocalization between PCNA and EdU. High colocalization is diagnostic of fast cycle. Conventional microscopy is suggestive of a much higher co-localization in the subpallial area I (Fig. 1F–I, L′–L‴). Confocal microscopy confirms a higher proportion of double-labeled cells in the subpallial region I (Fig. S3).

18Caudal to region IV in the preoptic region, neurogenic niches were observed widespread along the ventricle, in the hypothalamus and caudally to the medulla oblongata and spinal cord (e.g. Fig. 1M). We do not attempt here to subdivide this into more defined areas.

19Region V corresponds to the medial margin of the OT and is placed at the border with the torus longitudinalis. Proliferating cells are present on both sides of this border in the OT and the torus longitudinalis. Region V shows a typical U-shape in coronal sections (Fig. 1M). Region VI corresponds to the posterior and ventral margins of the OT (Fig. 1H,N). Scattered Edu+, PCNA-positive cells could be detected in the subventricular gray zone of the OT (Figs 1N″ and 5Q).

20Region VII represents the cerebellar neurogenic niche. This niche was characterized in detail in zebrafish. In N. furzeri, region VII has the shape of a dorsal and a ventral wedge of highly active cells connected by a rod of dividing cells in the valvula cerebelli (Fig. 1N,N′). The size of the dorsal wedge increases and that of the ventral wedge decreases in caudal direction in corpus cerebelli and in the most posterior portion only the dorsal wedge is visible.

21A complete rostro-caudal series of coronal sections to better appreciate the full extent of these neurogenic niches is presented in Fig. S2 (Supporting Information).

22To prove the persistence of cells generated during adult life, we analyzed animals 5 weeks after a single injection of EdU. Median lifespan for the MZM-04/10 strain is 27 weeks in our fish colony, so this interval covers a significant proportion of the adult life for this species. In all brain regions, EdU-labeled cells persisted and were found into the brain parenchyma at some distance from the neurogenic niches of origin visualized with PCNA (Fig. 2). This is best appreciated in the OT posterior margin, where the newly generated cells from a dense strip (Fig. 2F,H–M). The distance between this EdU+ strip (red arrows in Fig. 2H,I) and the PCNA+ germinal zone (white arrowhead in Fig. 2H,I) represents the marginal growth of the OT within this period. In the cerebellum, neurogenesis is extensive and cells generated during adulthood migrate in the granule cell layer (Fig. 2H–I,M–N). At visual inspection, it seems that more cells are generated and integrated into the cerebellum than in any other region of the N. furzeri brain, in line with the extensive cerebellar neurogenesis reported in other fish species.

23To study the fate of the newly generated cells, we sacrificed animals 1 week after a single injection of EdU and performed a double-labeling with the neuronal marker HuC/D. Sections were then analyzed by confocal microscopy. As expected, extensive co-localization was observed in telencephalon, OT, and cerebellum (Fig. S4).

24MicroRNAs are emerging as important regulators of neurogenesis and in particular miR-9 and miR-124. miR-124 is expressed by differentiated neurons while miR-9 was reported, in zebrafish, to be up-regulated in the regions where adult neurogenesis occurs.

25MicroRNAs are extremely conserved in their sequence. A preliminary study of small RNA sequencing (M. Baumgart, M. Grothe, M. Platzer, A. Cellerino, paper in preparation) revealed that the sequences of mature miR-124 and miR-9 in N. furzeri are identical to the sequences of Danio rerio. We therefore used locked nucleic acid (LNA) in situ probes designed for D. rerio. miR-124 expression was detected in nearly all neurons throughout the adult brain, but not in the neurogenic niches. This was particularly evident in the subpallial area I of the telencephalon (Fig. 3A,B). The expression pattern of miR-9 in the telencephalon was complementary to that of miR-124 and was concentrated in the neurogenic niches (Fig. 3C,D). Confocal microscopy (Fig. 3E,F) revealed partial colocalization of miR-9 and EdU 4 h postinjection, indicating that miR-9 is expressed in mitotically active neuronal progenitors.

26In the OT, expression of miR-124 (Fig. 3G) was lower in the neurogenic niche (arrow) and in the neighboring area that contains newly differentiated cells. Expression of miR-124 was specifically reduced in a strip of cells in the most medial portion of the OT (arrowheads). This area corresponds to the location of radial glia (Fig. 5, see below), where expression of miR-9 is particularly high (Fig. 3H). miR-9, however, was not up-regulated in the germinal layer of the OT (arrow), indicating that miR-9 is not associated with neuroepithelial progenitors. Single-channel visualizations of in situ hybridization for miR-124 and miR-9 in the telencephalic structure (Fig. 3B,D,F) are shown in Fig. S6 (Supporting Information, A/B, D/E, G/H, respectively).

27In mammals, neuroblasts and newly generated neurons express DCX. Because of excellent labeling of neuronal processes, this marker is widely used to quantify neurogenesis. A DCX ortholog is present neither in the assembled zebrafish genome nor in the collections of zebrafish expressed sequence tags. We recovered a transcript, which we call NfuDCX, in a database of N. furzeri transcripts (A. Petzold, K. Reichwald, N. Hartmann, A. Cellerino, C. Englert, M. Platzer, manuscript in preparation). NfuDCX shows 86.5% aminoacidic sequence similarity to mouse DCX (Data S2). To confirm that NfuDCX is the ortholog of DCX, we performed a phylogenetic analysis aligning all DCX and DCX-like sequences present in the ENSEMBL database. NfuDCX clusters with all other DCX sequences from other vertebrate species and not with DCX-like kinase 1, or DCX-like kinase 2 (Fig. S5). This analysis also revealed that DCX is present in the genomes of the Actynoptegyian fishes Gastrosteleus aculeatus (stickleback), Tetraodon nigroviridis, and Takifugu rubipes (pufferfish) (Table S1, Fig. S5).

28Immunolabeling with DCX in the telencephalon (Fig. 3I–M, green) revealed a strip of strongly labeled cells 3–4 cell diameters in distance from the telencephalic surface that is adjacent, but does not overlap, with the ribbon of PCNA+ progenitors (Fig. 3I–M,red). Double-labeling with PCNA and DCX in the OT (Fig. 3N–O, PCNA red and DCX green) and the retina (Fig. 3P–Q, PCNA green and DCX red) revealed that a column of processes labeled for DCX can be found juxtaposed to the germinal margins of active neurogenesis. One week after EdU injections, the column of DCX-labeled processes in the OT (Fig. 3R, red) was in register with the front of EdU+ cells close to the germinal margin (Fig. 3R, green). Two weeks postinjection, the column of DCX-labeled process is found more medial with respect to the strip of EdU+ cells. In the telencephalon, the timing of DCX expression was slower. Only few cells were double-labeled with DCX (red) and EdU (green) 1 week after injection (data not shown), but double-labeling of EdU and DCX is observed 2 weeks after injection (Fig. 3S-T).

29It should be remarked that DCX does not represent a marker of newly generated neurons throughout the brain. Doublecortin expression was not observed in association with the subpallial neurogenic niche (area I) (Fig. 3L), and strong DCX labeling was observed in the olfactory nerve, in fibers entering the olfactory bulb (data not shown) and in fibers entering the medial portion of the OT (Fig. 4H,N).

30To quantify the effects of age on neurogenesis, 25-week-old animals were analyzed 4 h after EdU injection and compared to 11-week- and 5-week-old animals. Neurogenic niches are still present in old fish, but the mitotic activity in telencephalon and OT is visibly reduced (Fig. 4A,B). We quantified the number of Edu+ cells in the germinal layer of the OT using unbiased stereological estimates (see Experimental procedures). A significant age-dependent reduction in proliferating cells was detected (Fig. 4C, anova, P < 0.0001). The absolute number of EdU+ cells monotonically decreased between 5, 11, and 25 weeks. Reduced neurogenesis resulted also in visibly reduced DCX staining in the telencephalon (Fig. 4I–L) and on the posterior margin of the OT (Fig. 4M) of 25-week-old fish when compared to 5-week-old fish (Fig. 4E–G). No differences were seen in the central region of the OT (Fig. 4H–N), where DCX expression is not associated with neurogenesis.

31Comparison of neurogenesis decay with somatic growth reveals that there is no direct relationship between these two processes (Fig. 4D). Indeed, neurogenesis is already reduced at age 11 weeks, when fish are still in the growing phase, and it is present – even if reduced – at 25 weeks of age, when the fish have practically ceased to grow.

32Age-dependent up-regulation of GFAP is a well-characterized hallmark of aging in mammals. We therefore studied age-dependent regulation of glial markers in N. furzeri.

33The expression of glial markers in telencephalic aNSCs was studied in great detail in zebrafish. In zebrafish, pallial aNSCs, but not the aNSCs in the subpallial region, expresses the typical glial markers S100B and GFAP. Cells on the germinal zone of the OT do not express glial markers and have a neuroepithelial phenotype.

34Glial fibrillary acid protein is not detected in the telencephalon of young N. furzeri. Only faint labeling of GFAP+ radial fibers is observed in the OT of young N. furzeri (Fig. 5A). S100B is expressed in the neurogenic telencephalic areas II and III in cells with the typical morphology of radial glia, but not in the subpallial area I (Fig. 5I–L, Fig. S8). In the OT, S100B marks the radial glia, which constitutes the most medial margin of the OT and is excluded from the germinal layer (Fig. 5M, Fig. S8). Double-labeling with EdU 4 h postinjection and confocal analysis revealed a partial colocalization, demonstrating that at least some S100B+ cells in the pallium are mitotically active (Fig. 5L white arrows, Fig. S8).

35Mushasi-1 (Msi-1) is an RNA-binding protein necessary for neurogenesis and enriched in aNSC. The expression domain of S100B largely overlaps with that of Msi-1. Msi-1, however, also labels the germinal layer of the subpallial neurogenic niche and OT (Fig. 5N–Q), which are not labeled for S100B. At a cellular level, confocal analysis revealed complete co-localization between Msi-1 and Edu 4 h postinjection, revealing that Msi-1 labels both the radial glia and the amplifying neuroblasts (Fig. 5N–Q, Fig. S8). This analysis also revealed that scattered EdU+ cells in the OT are Msi-1+ (Fig. 5Q).

36A dramatic up-regulation of GFAP immunoreactivity is observed in the OT of 25-week-old fish (Fig. 5D) when compared to 5-week-old fish (Fig. 5A). A dense network of fibers running tangentially to the medial margin of the OT is observed, and also, radial processes become strongly positive. These changes are not accompanied by up-regulation of S100B (Fig. 5B,E) or Msi-1 (Fig. S5), although GFAP immunoreactivtiy co-localizes with S100B (Fig. 5C,F) and Msi-1 (Fig. S7). Quantification of GFAP/S100B ratio revealed a ∼ 20-fold increase between 5-weeks and 25-weeks of age (Fig. 5G). Up-regulation of GFAP mRNA in the entire brain could also be detected by real-time PCR (Fig. 5H).

37In the present paper, we analyzed the niches of adult neurogenesis in N. furzeri. We found that the neurogenic niches have a similar distribution to that described in detail in zebrafish. In particular, in the telencephalon of both species, there is a clear distinction between a less active pallial and a more active subpallial niche. The pallial zone contains slow-cycling mitotically active radial glia positive for S100B, while the subpallial zone contains fast-cycling aNSCs lacking S100B labeling.

38Organization of stem cell niches in the OT of N. furzeri is also conserved and almost identical to what described in medaka: a clearly defined, continuous, and very active germinal layer is observed in the lateral, posterior, and inferior border of the OT. Progenitors in the OT are not labeled with S100B antibodies. In addition, a sparse neurogenic activity can be observed throughout the medial margin of the OT. This position corresponds to the location of the radial glia in the OT and double-labeling of EdU, and Msi-1 indicates that also the radial glia in the OT is mitotically active, although at a much lower level as compared to the telencephalon. This is a very similar situation to what described in the retina, where there is a neuroepithelial ciliary marginal zone which can produce all cells with exception of rods and neurogenic Müller glia which produces only rods. It remains to be investigated whether radial glia in the OT produces a specific neuronal type or not.

39We further characterized the neurogenic niches in telencephalon and OT to define expression of conserved molecular markers. The microRNAs, miR-9 and miR-124, are known to control neurogenesis and neuronal differentiation. Here, we report that miR-9 is highly expressed in the radial glia in telencephalon and OT but not in the neuroepithelial precursors of the OT. miR-124, on the other hand, is specifically excluded in neuronal precursors and newly generated neurons. This result is in line with the expression domains reported in adult zebrafish brain and suggests that miR-9 controls the activity of the neurogenic radial glia, while miR-124 is necessary for differentiation of postmitotic neurons.

40The RNA-binding protein Musashi-1 (Msi-1) was found to be expressed in all neurogenic niches in telencephalon and OT and its expression was not restricted to radial glia. All cells labeled with EdU after short intervals were found also to be positive for Msi-1; therefore, this essential neurogenic gene is expressed both in the aNSCs and in the amplifying progenitors in N. furzeri.

41An interesting finding was the identification of DCX in N. furzeri. This gene is necessary for adult neurogenesis in mammals and is the marker of choice to label newly generated neurons, but it is absent from the zebrafish genome. In N. furzeri, DCX labels newly generated neurons in pallium and OT and this strongly suggests a conservation of DCX function between mammals and Actinopterygian fishes. Doublecortin expression in N. furzeri is dramatically down-regulated with age, as in all mammalian species studied thus far.

42The most important finding of the present paper is the demonstration of a strong age-dependent regulation of neurogenesis in N. furzeri. This age-dependent regulation does not directly correlate with somatic growth and was observed already during early adult life, as observed in mammals.

43Gliosis is a widespread phenomenon in the aging mammalian brain, which is easily detected as up-regulation of GFAP. Aging-dependent gliosis was not investigated in any fish species so far. We could detect massive gliosis in the radial glia of the OT as a consequence of aging in N. furzeri. Gliosis is the typical response of the brain to injury and is observed in many neurodegenerative diseases. Age-dependent gliosis is therefore a further indication of age-dependent neurodegeneration in N. furzeri. This increase in GFAP is not paralleled by a similar dramatic increase in S100B or Msi-1, indicating that the number of radial glia does not change dramatically but these cells up-regulate GFAP expression.

44In summary, we have shown conserved aspects of adult neurogenesis in N. furzeri both in the location of neurogenic niches and in the expression of molecular markers. More importantly, we report that N. furzeri shows age-dependent decrease in neurogenesis and age-dependent gliosis. Intervention studies in aging mammals are associated with high costs and long experimental times. Our data qualify N. furzeri as an alternative model organism to study mechanisms underlying age-dependent regulation of neurogenesis and gliosis.

45Detailed experimental procedures are provided as Data S1.

46All experiments were performed on group-house N. furzeri of the MZM-04/10 strain. Details can be found in. The protocols of fish maintenance were approved by the local authority in the State of Thuringia (Veterinär- und Lebensmittelüberwachungsamt).

47Immunohistochemistry was performed using commercially available antibodies (Table S2). In situ hybridization was performed using LNA probes labeled with DIG at 5′ and 3′ at an hybridization temperature of 42°C. EdU (50 μL of a 10 μm solution) was injected intraperitoneally and detected according to manufacturer’s instruction (Invitrogen, Grand Island, NY, USA).

48Glial fibrillary acid protein (Data S3) was cloned using Rapid Amplification of cDNA Ends (RACE; Clonthech, Mountain View, CA, USA), and expression levels were quantified using the Qiagen protocols on a Rotorgene (Qiagen, Hilden, Germany).

49To quantify EdU+ cells, brains were serially sectioned and every third section was analyzed. To count cells, two confocal planes at a distance of 4 μm were superimposed and cells present in both planes were excluded. Unbiased cell estimates were so derived:

50Let the ordered series of sections for a specimen be, and let the sampling density be 1/X, where Q = M × X, then a random number is extracted and every section is analyzed. Then, the unbiased estimate of cell numbers can be calculated by the formula

51= number of profiles present only in the reference (upper) plane of section i, = number of profiles present only in the bottom plane of section i, FS = fraction of total section volume contained between reference and bottom planes.