Stem-cell niche self-restricts the signaling range via receptor-ligand degradation

Stem-cell niche signaling is short-range in nature, such that only stem cells but not their differentiating progeny experience self-renewing signals. At the apical tip of the Drosophila testes, 8 to 10 germline stem cells (GSCs) surround the hub, a cluster of somatic cells that function as the major component of the stem cell niche. We have shown that GSCs form microtubule-based nanotubes (MT-nanotubes), which project into the hub cells, serving as the platform for niche signal reception: the receptor Tkv expressed by GSCs localizes to the surface of MT-nanotubes, where it receives the hub-derived ligand Decapentaplegic (Dpp), ensuring the reception of the ligand specifically by stem cells but not by differentiating cells. Here we show that receptor (Tkv)-ligand (Dpp) interaction at the surface of MT-nanotubes serves a second purpose of dampening the niche signaling: we found that the receptor Tkv and the ligand Dpp are internalized into hub cells and are degraded in the hub cell lysosomes. Perturbation of hub lysosomal function or MT-nanotube formation leads to excess receptor retention within germ cells as well as excess Dpp that diffuses out of the hub, leading to ectopic activation of niche signal in differentiating germ cells. Our results demonstrate that MT-nanotubes plays dual roles in ensuring the short-range nature of the niche signaling by 1) providing exclusive interphase of the niche ligandreceptor interaction and 2) limiting the amount of available ligand-receptor via their degradation.


Introduction
Many stem cells reside in a special microenvironment, called niche, to maintain their identity 1 . Niche signaling is believed to be short-range such that only stem cells but not differentiating cells activate self-renewal signaling above the threshold. In the Drosophila testis, germline stem cells (GSCs) reside in the niche formed by post-mitotic somatic cells called hub cells.
GSCs typically divide asymmetrically, giving rise to one daughter cell that retains the attachment to the hub to self-renew and the other daughter cell, gonialblast (GB), that is displaced away from the hub and differentiate into spermatogonia (SG). Hub cells secrete ligands, Dpp and Unpaired (Upd).
Dpp activates Bone Morphogenetic Protein (BMP) pathway, whereas Upd activates JAK/STAT pathway in GSCs, both of which are required for maintenance of GSCs 2 3 4 5 . These niche-derived ligands must act over a short range so that signaling is only active in GSCs, but not in GBs. Defining the boundary of niche signaling between abutting GSCs and GBs is of critical importance in maintaining stem cell population while ensuring differentiation of their progeny 2 . However, how this short-range nature of niche signaling is achieved is poorly understood.
We have previously demonstrated that the cellular projections, microtubule-based (MT)nanotubes present on GSCs project into hub cells ( Figure 1A). Tkv is produced by GSCs and trafficked to the surface of MT-nanotubes where it interacts with hub-derived Dpp, leading us to propose that MT-nanotubes serves as a signaling platform for productive Dpp-Tkv interaction ( Figure 1A). Because MT-nanotubes are formed only by GSCs, this likely contributes to short-range nature of the niche signaling by excluding GBs from receiving Dpp. Here we show that MTnanotubes serve a second function: in addition to serving as a platform for Dpp-Tkv interaction, it also promotes internalization of Tkv-Dpp by hub cells, where they are degraded in lysosomes.
Perturbation of hub lysosomes or MT-nanotube formation resulted in signal activation in non-stem cell populations, suggesting that Dpp-Tkv degradation in the hub plays a critical role in dampening the signal outside the niche. We propose here that MT-nanotubes not only promote the specificity of niche signal reception by stem cells, but also limits the range of niche signal by degrading the receptor-ligand complex.

Tkv receptors expressed in GSC are transferred to hub lysosomes
We have previously shown that Tkv, the receptor for one of the hub-derived self-renewal ligands, Dpp, is produced in GSCs and trafficked to the surface of the MT-nanotubes. At the surface of MT-nanotubes, Tkv engages in signaling by interacting with Dpp derived from hub cells. We showed that Intraflagellar transport-B (IFT-B; oseg2, osm6, and che-13) genes are required for MTnanotube formation 6 . In the absence/reduction of MT-nanotubes, Tkv is retained in the cell body of GSCs, leading to compromised Dpp-Tkv interaction and thus reduced the activation of downstream BMP signal 6 .
In addition to Tkv localization on MT-nanotube surface, we noticed that Tkv-mCherry expressed in germline were observed within the cell body of hub cells, not necessarily colocalizing with the MT-nanotubes ( Figure 1B, white arrowheads). We found that this Tkv population found in the hub cells colocalizes with lysosomes ( Figure 1C, D). Tkv's lysosome localization was further examined by using chloroquine (CQ) treatment, a drug that raises lysosomal pH and inhibits lysosomal enzymes [7][8][9] . CQ treatment is known to enlarge the lysosome size 10  To further confirm the translocation of Tkv from one cell to the other (GSC to hub cells), we followed the temporal order of Tkv localization after its production by inducing GSC clones expressing Tkv-mCherry. When GSC clones expressing Tkv-mCherry was induced by heat shock (hs) (see methods), Tkv-mCherry was first observed on the GSC plasma membrane and in the GSC cytoplasm as puncta (day 1 after hs). After 2 days post hs, its signal in the hub became evident. Finally, after 3 days, the Tkv signal along the GSC plasma membrane reduced and the Tkv signal in the hub increased further ( Figure 1N-Q). These results indicate that Tkv is transferred from GSC to the hub cells.
In addition to Tkv, the type II receptor Punt, a co-receptor of Tkv, was observed in the hub lysosomes ( Figure 1R). Moreover, Dpp ligand (visualized via a knock-in line in which endogenous Dpp is fused to mCherry 11 and completely colocalized with Tkv-GFP trap in the hub, Figure S1B), was also seen in lysosomes in the hub ( Figure 1S). A reporter of ligand-bound Tkv, TIPF 12 was also the same ( Figure 1T). Taken together, these results suggest that GSC-derived Tkv is trafficked to the MTnanotubes and then to lysosomes in the hub, together with co-receptor Punt and the hub-derived ligand Dpp. Based on these results, we hypothesized that Tkv, Punt and Dpp may be degraded within the hub cell lysosomes.

Hub lysosomes degrade Tkv to downregulate Dpp signal in GSCs
To test the possibility that Tkv and Dpp are degraded within the hub cell lysosomes, we examined the effect of impairing lysosome function in the hub cells. To this end, we knocked down genes required for lysosomal function using a hub cell-specific driver. Spinster (Spin) is a putative late-endosomal/lysosomal efflux permease and a known regulator of lysosomal biogenesis 13 . It has been also reported that Spin regulates Dpp signaling through degrading Tkv in Drosophila eye discs 14 . Lysosomal-associated membrane protein-1 (Lamp1) is an abundant protein in the lysosomal membrane that is required for lysosomes to fuse with endosomes 15 . Hub-specific knock down of these genes led to increased sizes of Tkv punctae in the hub (Figure 2A, B, and C), likely reflecting defective degradation of Tkv in the hub lysosome. This led to a significant increase in pMad levels in GSCs and their immediate progeny ( Figure 2D, E, and F), indicating that hub lysosomes are responsible for niche signal attenuation. In contrast, germ cell-specific knockdown of these two lysosomal genes did not alter pMad level ( Figure 2G, H, and F). It should be noted that dpp mRNA levels showed no detectible alteration in lysosomal-defective hub cells ( Figure S2), indicating that niche signal attenuation by hub cells is not caused by a change in dpp transcription level. These results suggest that hub lysosomes are responsible for the degradation of GSC-derived Tkv receptor, which may serve to attenuate the niche signaling to the appropriate level.

Ubiquitination mediates Tkv degradation by promoting translocation of Tkv from GSC to the hub lysosome
Ubiquitination of membrane proteins is known to be required for recognition by the endosomal sorting complexes required for transport including endocytosis, lysosomal fusion, and degradation 16 . SMAD ubiquitination regulatory factor (Smurf) is a HECT (Homologous to the E6-AP Carboxyl Terminus) domain-containing protein with E3 ubiquitin ligase activity, and disruption of Smurf function has been shown to enhance Dpp-Tkv signal in GSCs 17,18 . It has been reported that phosphorylation of Tkv Ser238 residue is required for Smurf-dependent ubiquitination 17 .
We found that Tkv-S238A-GFP, in which the Serine residue required for Smurf-dependent ubiquitination was mutated, exhibits striking difference from wild type Tkv-GFP on two accounts. Taken together, these results demonstrate that ubiquitination of Tkv, induced by phosphorylation of S238, is required for its translocation from GSC to the hub. Our data further suggest that this translocation of Tkv to the lysosome is critical to downregulate excess amount of Tkv.

Degradation of Dpp-Tkv in the hub lysosome is required for proper differentiation
The above results suggest that degradation of Dpp-Tkv in the hub is required to downregulate Dpp-Tkv signaling. Although, shutting down Dpp-Tkv signal is required for proper differentiation, Bag of marbles (Bam), a master differentiation factor, whose expression is suppressed by Dpp signal within GSCs, typically peaks around 4-8 SG stage 19 . In the IFT-KD/Tkv-OE testes, Bam peak was never observed ( Figure 3G, H), consistent with the idea that the germline tumor was caused by ectopic or prolonged Dpp-Tkv signal.
In sum, these results show that combination of compromised Dpp-Tkv degradation and overexpression of Tkv leads to defective attenuation of Dpp-Tkv signaling, revealing the importance of Dpp-Tkv degradation in the hub in promoting differentiation.

Dpp ligand diffuses out from the hub upon inhibition of lysosomes
If the Tkv digestion in adjacent hub cells is happening only in GSCs, how can IFT knockdown cause a tumor located outside of the niche? We speculate that Dpp ligand may diffuse away from the hub when the lysosomal degradation is impaired. To test this idea, we examined the localization of Dpp-mCherry (mCherry knock-in strain) 11  To determine if the Dpp-mCherry detected outside the hub is due to its diffusion from the hub, we used fluorescence recovery after photobleaching (FRAP) analysis. After photobleaching a 10 m diameter circle located approximately 10 m away from the hub center, the photobleached region was quickly equalized by the Dpp-mCherry signal from neighboring regions. The recovery was rapid with an average time of 7.4 ± 2.1 seconds (n=7) to reach 50% of the original intensity ( Figure 4I, K, movie1).
In contrast, after photobleaching of the entire hub region, signal did not recover ( Figure 4J, L, movie2), indicating that the Dpp signal outside the hub is likely the Dpp protein diffused from the hub, and that the hub is likely the sole source of Dpp. It should be noted that the photobleached signal did not always fully recover up to 100% (averaging 70 ± 15%, n=7) possibly because rest of Dpp protein fraction might be trapped in the photobleached field. We speculate that Dpp protein may bind to the extracellular matrix components as reported in Drosophila embryo 20 . Alternatively, diffused Dpp ligand might be endocytosed into germ cells after binding to Tkv. Indeed, we observed lysosomal localization of Dpp ligand together with Tkv in germ cells located away from the hub ( Figure 4B arrowheads, Figure S3 arrowheads).
In summary, these observations suggest that in the absence of proteolysis of Dpp, the ligand can diffuse from the hub. MT-nanotube-mediated degradation of receptor together with ligand is essential for preventing signal overload outside of the niche.

Discussion
We have shown previously that niche cells and stem cells interact in a contact-dependent manner, with GSCs and hub cells engaging in productive signaling via MT-nanotubes, enabling highly specific cell-cell interactions and excluding non-stem cells from receiving stem cell signals. Here, we demonstrate MT-nanotubes also contribute to proteolysis of receptor and ligand via transferring the stem cell receptor together with ligand to the niche cell lysosomes ( Figure 4M). This ensures the removal of receptor and ligand after the signal engagement and prevents ligand-receptor interactions outside of the niche.
Cytonemes, another type of actin-dependent signaling protrusion 21 22 , also transfer ligand and receptor, allowing the interaction between cells at a distance. Ligand-producing and receptorproducing cells both form cytonemes and both cells have been observed to exchange signaling proteins: receptor into the ligand-producing cells and ligand into the receptor-producing cells 22 . Similarly, exchange of receptor tyrosine kinases, bride of sevenless and sevenless (sev), during Drosophila retinal development has also been documented 23 . Exchange of plasma membrane proteins also occurs in mammalian systems; "Trogocytosis" is the phenomena reported in lymphocytes and antigenpresenting cells through the immunological synapse 24 25 . These studies, together with our study, illustrate the universality of such transfer in general contact-dependent signaling.
What could be the potential benefit(s) of this mechanism in which stem cells transfer and degrade their receptor in niche cells instead of digesting by themselves? Since Tkv transport occurs from MT-nanotube membrane, only ligand-bound receptors but not ligand-unbound receptors may be subjected to degradation. In this way, stem cell can secure at least required amount of Tkv. Another possibility is that this might be the strategy of GSCs to avoid receptor endocytosis. Since Tkv is transferred into hub cells, activated receptor never come back to the cell body of GSCs. Such signaling endosomes can act as intracellular platforms for signaling in many other systems 26 27 . BMP signal is also known to utilize signaling endosomes. The activated receptor complex phosphorylates Mad (or receptor-regulated Smad 2 and 3 in mammalian systems) at the C-terminal of BMP receptor presented on the surface of SARA endosomes 28 . Although signaling endosomes can enhance the activation of downstream signal, endosomes can be inherited into differentiating daughter cells after division, which may compromise the specificity of the niche-stem cell signaling. Therefore, transferring the receptor to other cells may be the stem cell-specific strategy to achieve high specificity of the niche signal, such that signal activation only occurs in stem cells, but not in differentiating daughter cells.
It remains to be investigated whether lysosomal proteolysis of ligand and receptor in niche cells, as demonstrated by our study, might also regulate other stem cell systems.
Bonin for manuscript editing; this work was supported by an NIH grant 1R35GM128678-01 and start-up funds from UConn Health (to M.I.).

Author Contributions
M.I, Conception and design, acquisition of data, analysis and interpretation of data, drafting and revising the article; S.L, T.S, M.A and N.G., Acquisition of data, analysis and interpretation of data, drafting or revising the manuscript.

Declaration of Interests
The authors declare no competing interests.

Fly husbandry and strains
All fly stocks were raised on standard Bloomington medium at 25°C (unless temperature control was For the overexpression of Dpp-mCherry, updGal4 ts driver was used, and a combination of updGal4 and tubGal80 ts 31 was used to avoid lethality. Temperature shift crosses were performed by culturing flies at 18°C to avoid lethality during development and shifted to 29°C upon eclosion for 4 days before analysis. Control crosses for RNAi screening were designed with matching gal4 and UAS copy number using TRiP control stock (BDSC35785) at 25 ℃. RNAi screening of candidate genes for Tkv trafficking and degradation was performed by driving UAS-RNAi constructs under the control of nosGal4 or updGal4 (see below for validation method).

Quantitative reverse transcription PCR
Males carrying nos-gal4 driver were crossed with males of indicated RNAi lines or UAS-GFP-Tubulin Transgenic flies were generated using strain attP2 by PhiC31 integrase-mediated transgenesis (BestGene Inc.).

Immunofluorescent Staining
Immunofluorescent staining was performed as described previously 34 . Briefly, testes were dissected in phosphate-buffered saline (PBS) and fixed in 4% formaldehyde in PBS for 30-60 minutes. Next, testes were washed in PBST (PBS + 0.3% TritonX-100) for at least 30 minutes, followed by incubation with primary antibody in 3% bovine serum albumin (BSA) in PBST at 4 °C overnight. Samples were washed for 60 minutes (three times for 20 minutes each) in PBST, incubated with secondary antibody in 3% BSA in PBST at 4 °C overnight, and then washed for 60 minutes (three times for 20 minutes each) in PBST.
Samples were then mounted using VECTASHIELD with 4′,6-diamidino-2-phenylindole (DAPI) Images were taken using a Zeiss LSM800 confocal microscope with a 63× oil immersion objective (NA=1.4) and processed using Image J and Adobe Photoshop software. Three-dimensional rendering was performed by Imaris software.

In situ hybridization
In situ hybridization on adult testes was performed as described previously 35 .
Briefly, testes were dissected in 1X PBS and then fixed in 4% formaldehyde/PBS for 45 min. After fixing, they were rinsed 2 times with 1X PBS, then resuspended in 70% EtOH, and left overnight at 4°C.

Chloroquine or Lysotracker/LysoSensor treatment
Testes from newly eclosed flies were dissected into Schneider's Drosophila medium containing 10% fetal bovine serum and glutamine-penicillin-streptomycin. Then testes were incubated at room temperature with or without 100 M chloroquine (Sigma) in 1 mL media for 4 hours prior to imaging. For the lysosome staining, testes were incubated with 50 nM of LysoTracker Deep Red (ThermoFisher L12492) or 100 nM of LysoSensor Green DND-189 (ThermoFisher L7535) probes in 1 mL media for 10 min at room temperature then briefly rinsed with 1 mL of media for 3 times prior to imaging.
These testes were placed onto Gold Seal™ Rite-On™ Micro Slides two etched rings with media, then covered with coverslips. An inverted Zeiss LSM800 confocal microscope with a 63× oil immersion objective (NA=1.4) was used for imaging.

Quantification of pMad intensities
Mean intensity values in the GSC nuclear region were measured for anti-pMad staining. To normalize the staining condition, data were further normalized by the average of measurements of pMad from randomly picked three cyst cells in the same testes, and the ratios of relative intensities were calculated for each GSC (see Figure 3C).

FRAP analysis
Fluorescence recovery after photo-bleaching (FRAP) of Dpp-mCherry signal was undertaken using a Zeiss LSM800 confocal laser scanning microscope with 63X/1.4 NA oil objective. Zen software was used for programming of each experiment. Encircled (approximately 10 m diameter) areas of interest were photobleached using laser powers to achieve an approximately 70-90% bleach using the 561 nm laser.
Fluorescence recovery was monitored every second.

Statistical analysis and graphing
No statistical methods were used to predetermine sample size. The experiments were not randomized. The