Hierarchical Stem Cell Topography Splits Growth and Homeostatic Functions in the Fish Gill

While lower vertebrates contain adult stem cells (aSCs) that maintain homeostasis and drive un-exhaustive organismal growth, mammalian aSCs display mainly the homeostatic function. Understanding aSC-driven growth is of paramount importance to promote organ regeneration and prevent tumor formation in mammals. Here we present a clonal approach to address common or dedicated populations of aSCs for homeostasis and growth. Our functional assays on medaka gills demonstrate the existence of separate homeostatic and growth aSCs, which are clonal but differ in their topology. While homeostatic aSCs are fixed, embedded in the tissue, growth aSCs locate at the expanding peripheral zone. Modifications in tissue architecture can convert the homeostatic zone into a growth zone, indicating a leading role for the physical niche defining stem cell output. We hypothesize that physical niches are main players to restrict aSCs to a homeostatic function in animals with a fixed adult size.

Besides being the respiratory organ of fish, the gill has additional functions as a sensory and  The massive post-embryonic growth of teleost gills occurs by increasing the number but also the 2 length of filaments. Previous data on stationary samples suggest that filaments grow from their tip 3 (Morgan, 1974), and we followed two complementary dynamic approaches to characterise stem cells 4 during filament growth. First, we exploited the high rate of cellular turnover previously observed by 5 a pulse of IdU ( Figure 1D) which labels mitotic cells all along the filament. We reasoned that during 6 a chase period, cells that divide repeatedly -as expected for stem cells driving growth -would 7 dilute their IdU content with every cell division, as previously reported for other fish tissues

24
while we observed that the central filaments showed a longer basal signal that becomes shorter in 25 more peripheral filaments, the upper non-labelled fraction seemed rather stable along central-to-26 periphery axis of the branchial arch ( Figure 3F). This suggested that individual filaments had grown 27 at comparable rates during the chase phase, highlighting a coordination among the stem cells that

29
We revealed in the previous sections that growth stem cells at the periphery of branchial arches (br-30 archSCs) generate new filaments, and we showed that each filament contains, in turn, growth stem labelled -and therefore clonal -filaments observed at the periphery of branchial arches in adult 1 Gaudí RSG Gaudí Ubiq.iCRE fish induced for recombination during embryogenesis ( Figure 2B, 4A, B). We 2 reasoned that if a labelled br-archSC is fate restricted, the consecutive filaments formed from it 3 should display an identical recombination pattern, since filamSCs would have inherited the same 4 fate-restriction from their common br-archSC. Alternatively, if filamSCs would acquire the fate-5 restriction when each filament is formed, then a stretch of clonal filaments should display different 6 recombination patterns, based on the independent fate acquisition at the onset of filament 7 formation (schemes in Figure 5A). We have focussed on 153 branchial arch extremes that started 8 with a labelled filament (N= 83 for rec. pattern 1, N= 44 for rec. pattern 2, N= 22 for rec. pattern 3 and 9 N= 4 for rec. pattern 4), and 97.4% were followed by a filament with the same recombination pattern 10 (Supplementary Table 4). Moreover, 81.7% of stretches maintained the same recombination pattern 11 for 6 or more filaments, indicating that the labelled cell-of-origin for post-embryonic filaments was 12 already fate restricted. Altogether, our data revealed that a branchial arch contains fate restricted 13 growth br-archSCs at its peripheral extremes that produce growth filamSCs stem cells with the same    analysis of clonal progression. We used double transgenic Gaudí Ubiq.iCre Gaudí RSG adults that were 1 grown for 3 additional weeks after clonal labelling, and focussed on those containing only a few 2 recombined lamellae per branchial arch (labelling efficiency less than 0.5%). A detailed analysis on 3 lamellae located far away from the filaments' growing tip revealed clones of labelled cells spanning 4 from the proximal to the distal extreme of the lamella ( Figure 6D, E). The clones ranged from a few 5 pillar cells ( Figure 6D

23
We speculated that modifying the close environment of homeostatic stem cells by ablating the 24 growing zone of a filament could elicit a growth response from the homeostatic domain. We

8
To address the nature of the cells re-establishing the growth domain, the same injury assay was 9 performed on Gaudí Ubiq.iCRE Gaudí RSG transgenic fish that had been induced for sparse recombination 10 at late embryonic stage (8 dpf) and grown for two months. When we analysed these samples 3 weeks 11 after injury, we observed that the recombination pattern of the basal, non-injured region was In this study, we use mathematical modelling and genetic lineage analysis to reveal the rationale 2 behind the permanent post-embryonic growth in a vertebrate. We introduce the fish gill, and 3 particularly branchial arches, as a new model system that displays an exquisite temporal/spatial 4 organisation, and use it to characterise growth and homeostatic stem cells. We reveal two domains 5 harbouring growth stem cells: both extremes of each branchial arch contain br-archSCs, which in 6 turn generate filamSCs that locate to the tip of newly formed filaments. Additionally, filamSCs 7 generate homeostatic stem cells at the lamellae along the longitudinal axis of the filament. The 8 peripheral-to-central axis of branchial arches reflects a young-to-old filament order, and the 9 longitudinal axis of a filament reflects a young-to-old lamellae order. The two growth stem cells 10 and the one homeostatic stem cell types are clonal and organised in a hierarchical manner.

11
Our observations indicate that the relative position within the organ has a major impact on the 12 growth vs homeostatic activity of stem cells. We have found that when the growth domain of a 13 filament is lost, the homeostatic domain is able to generate a new, functional growth domain. This

22
In most adult mammalian organs, stem cells maintain homeostasis by generating new cells that will 23 replace those lost during physiological or pathological conditions. We have functionally identified 24 homeostatic stem cells in the fish gill, and focussed on the ones generating pillar cells. Our lineage 25 analysis demonstrates that growth and homeostatic stem cells are clonal along a filament, where 26 the former generate the latter. The most obvious difference between these two stem cell types is 27 their relative position; growth stem cells are located at the growing tip, beyond the rigid core that 28 physically sustains the structure of the filament, while homeostatic stem cells are embedded inside 29 the tissue, adjacent to the collagen-rich chondrocyte column. It is to note that both the function 30 and the relative location of the gill homeostatic stem cells match those of the mammalian 1 gill suggests the existence of a physical niche that would restrict stem cells to their homeostatic 2 role, preventing them to drive growth. We believe that during vertebrate evolution, the transition 3 from lower (ever-growing) to higher (size-fixed) vertebrates involved restraining the growth activity

7
To stain gills, adult fish were sacrificed using a 2 mg/ml Tricaine solution (Sigma-Aldrich, A5040-8 25G) and fixed in 4% PFA/PTW for at least 2 hours. Entire Gills were enucleated and fixed overnight 9 in 4% PFA/PTW at 4C, washed extensively with PTW and permeabilised using acetone (10-15 10 minutes at -20C). Staining was performed either on entire gills or on separated branchial arches.

11
After staining, samples were transferred to Glycerol 50% and mounted between cover slides using a 12 minimal spacer.