Exclusive multipotency and preferential asymmetric divisions in post-embryonic neural stem cells of the fish retina

Abstract

The potency of post-embryonic stem cells can only be addressed in the living organism, by labeling single cells after embryonic development and following their descendants. Recently, transplantation experiments involving permanently labeled cells revealed multipotent neural stem cells (NSCs) of embryonic origin in the medaka retina. To analyze whether NSC potency is affected by developmental progression, as reported for the mammalian brain, we developed an inducible toolkit for clonal labeling and non-invasive fate tracking. We used this toolkit to address post-embryonic stem cells in different tissues and to functionally differentiate transient progenitor cells from permanent, bona fide stem cells in the retina. Using temporally controlled clonal induction, we showed that post-embryonic retinal NSCs are exclusively multipotent and give rise to the complete spectrum of cell types in the neural retina. Intriguingly, and in contrast to any other vertebrate stem cell system described so far, long-term analysis of clones indicates a preferential mode of asymmetric cell division. Moreover, following the behavior of clones before and after external stimuli, such as injuries, shows that NSCs in the retina maintained the preference for asymmetric cell division during regenerative responses. We present a comprehensive analysis of individual post-embryonic NSCs in their physiological environment and establish the teleost retina as an ideal model for studying adult stem cell biology at single cell resolution.

Keywords: Asymmetric division; Medaka; Multipotency; Neural progenitor cells; Neural stem cells; Retina.

Fig. 1.

A toolkit for post-embryonic clonal labeling in medaka. (A,B) The toolkit is composed of two Cre-recombinase driver lines (A) and three LoxP reporter lines (B). (A) Cre transcription can be activated via heat shock in Gaudí^HspCRE.A (top, Cre represented in gray), which contains the integration reporter cmcl2:EGFP. Tamoxifen treatment will favor Cre nuclear translocation in Gaudí^Ubiq.iCre (bottom,Cre represented in gray), which contains the integration reporter cmcl2:ECFP. (B) Gaudí reporter lines express a default FP that is lost (DSRed in Gaudí^RSG, top) or exchanged (Cerulean in Gaudí^BBW2.1, middle; DS-Red in Gaudí^LxBBW, bottom) upon Cre-mediated recombination. Scale bar: 1 mm.

Fig. 2.

Recombination can be determined in living and fixed samples. (A) Tamoxifen induction leads to expression of H₂B-EGFP in Gaudí^RSG, Gaudí^Ubiq.iCre embryos. (B) A heat-shock treatment induces expression of Cerulean, YFP or H₂B-EGFP in Gaudí^LxBBW, Gaudí^HspCre.A embryos. Scale bar: 1 mm. (C) Live imaging of a recombined Gaudí^LxBBW, Gaudí^HspCre.A fish allows identification of individual cells using native fluorescent proteins. Scale bar: 50 µm. (D) Immunofluorescence using a single anti-EGFP antibody allows detection of membrane-tagged Cerulean, cytoplasmic eYFP and nuclear eGFP in fixed samples of an adult cornea. Scale bar: 50 µm.

Fig. 3.

Gaudí driver lines induce recombination in different tissues and have a large induction range. (A) The Gaudí toolkit allows recombination in the CMZ and differentiated cells of the neural retina. (B-H) Recombination is also observed in different tissues such as cornea (B), brain (C), somites (D), intestine (E), neuromast (F), epithelia (G) and gills (H). (I-N) The number of recombined cells can be modulated from a few (I,L) to lot of cells (J,M) or almost the entire organ/tissue (K,N), modifying the intensity of the induction. Scale bars: 50 μm in A-H,L-N; 1 mm in I-K.

Fig. 4.

Post-embryonic RSCs and RPCs. (A) The medaka retina grows stereotypically by addition of cells in temporal concentric rings. (B) Transplantation of labeled blastula cells results in adult fish whose retinae contain clones of cells (ArCoSs) spanning from the embryonic to the adult retina. (C-F) Induction of Cre recombination at juvenile stage demonstrates post-embryonic retinal stem cells (RSCs) that generate induced ArCoS (iArCoS) (C,E). IdU incorporated at the time of heat-shock induction helps visualizing the induction circle (D,E), which demarcates a time-ring at which recombination was induced months before (D’). A close look at the induction circle allows functional discrimination of RSCs versus retinal progenitor cells (RPCs). Although RSCs generate iArCoS (C, filled arrowheads), transient RPCs generate smaller clones (C, empty arrowheads) that map closer to the induction circle than the origin of iArCoSs (scheme in F), reflecting a more-central location in the ciliary marginal zone (CMZ). (E) Neurons produced during embryogenesis. P₁, P₂, P_N: groups of neurons generated at different post-embryonic stages. Scale bar in C: 50 μm.

Fig. 5.

Functional differences between post-embryonic RSCs and RPCs. (A) The stereotyped addition of cells to the NR allows analyzing temporal aspects of post-embryonic neurogenesis. (B) Moderate induction of Gaudí fish results in sparse labeling of stem and progenitor cells forming isolated retinal clones. (C,D) 3D reconstruction of a clone generated by a RPC. The clone has detached from the CMZ and all labeled cells are already differentiated. (E,F) 3D reconstruction of a clone arising from a RSC. The clone is continuous with the CMZ, and a fraction of the older, i.e. more central, cells is already incorporated in the layered retina. (G) Distribution of number of cells per clone in RSCs and RPCs, 7 dpi. Most RSCs form big clones containing more than 200 cells, and most RPCs form clones of fewer than 100 differentiated cells.

Fig. 6.

Exclusive multipotency among post-embryonic RSCs. (A) The stereotypic distribution of cell types in the differentiated retina facilitates the analysis of major retinal cell types. (B) Juvenile Gaudí fish are induced by Cre-mediated recombination and grown until adulthood. (C-E) Individual NSCs in the retina are multipotent. Every clone spans through the three retinal nuclear layers (C) and contains all mayor retinal cell types (D). Gaudí^LXBBW allows unambiguous assignment of iArCoSs and demonstrates multipotency in adjacent RSCs (E, and detail in E′).

Fig. 7.

RSCs undergo asymmetrical cell divisions during homeostatic growth and regeneration. (A-C) Prediction of ArCoSs shape during growth of the retina (A), assuming symmetric divisions and neutral drift (B) or preferential asymmetric divisions (C) among RSCs. (D-G) All ArCoSs observed fit the asymmetric preference, even in old fish of 18 months of age. Inner (D) and outer (F) views of iArCoS-containing retinae, and diagrams (E,G) indicating induction circle at T₁ and final age at T_N. (H-N) RSCs do not change their behavior during regeneration responses. (H) Scheme of the experimental timeline. (I) If RSCs change to symmetric divisions upon injury, iArCoSs should either expand or reduce their width from the induction circle onwards. (J) A fixed choice for asymmetric divisions would result in constant width before and after the induction ring. Most of the iArCoSs analyzed do not change their shape during regeneration (K,N). In some cases, there is a transient expansion (L,N) or reduction (M,N) in the width of the ArCoSs, indicating a response mediated by RPCs.

Fig. 8.

RSCs also follow minor symmetrical cell divisions to increase stem cell number during homeostatic growth. (A) Based on the relative clone size, it is possible to infer the number of active stem cells in an organ. (B-E) RSCs increase in number as the fish grows. Inducing recombination in Gaudí fish at different stages (B) results in iArCoSs that differ in the relative occupancy of the retinal diameter (C, detail in D). The older the stage during which recombination was induced, the smaller the fraction of the retina occupied by each iArCoSs and, therefore, the higher the number of active RSCs at the induction time (E).