Centromere proteins are asymmetrically distributed between newly divided germline stem and daughter cells and maintain a balanced niche in Drosophila males

Abstract

Stem cells can undergo asymmetric cell division (ACD) giving rise to one new stem cell and one differentiating daughter cell. In Drosophila germline stem cells (GSCs), the centromeric histone CENP-A (CID in flies) is asymmetrically distributed between sister chromatids such that chromosomes that end up in the GSC harbor more CID at centromeres. A model of “mitotic drive” has been proposed in GSCs such that stronger and earlier centromere and kinetochore interactions with microtubules bias sister chromatid segregation. Here we show that in Drosophila males, centromere proteins CID, CAL1, and CENP-C are asymmetrically distributed in newly divided GSCs and daughter cells in S phase. We find that overexpression of CID (either with or without CAL1) or CENP-C depletion disrupts CID asymmetry, with an increased pool of GSCs relative to daughter cells detectable in the niche. This result suggests a shift toward GSC self-renewal rather than differentiation, important for maintaining tissue homeostasis. Overexpression of CAL1 does not disrupt asymmetry, but instead drives germ cell proliferation in the niche. Our results in male GSCs are comparable to female GSCs, indicating that despite differences in signaling, organization, and niche composition, the effects of centromere proteins on GSC maintenance are conserved between the sexes.

INTRODUCTION

Asymmetric cell division (ACD) in stem cells generates two daughter cells with unequal fates: one self-renewing daughter cell to replenish the stem cell pool and another daughter cell that differentiates (Sunchu and Cabernard, 2020). ACD is important for maintaining tissue homeostasis and its misregulation can lead to several diseases, including cancer and infertility (Morrison and Kimble, 2006; Knoblich, 2010). The silent sister hypothesis (SSH) was put forward as a mechanism to explain how ACD can give rise to genetically identical daughter cells with two different fates (Lansdorp, 2007). It proposed that distinct epigenetic marks on sister chromatids can result in differential gene expression in the two daughter cells accounting for different cell fates. Evidence from germline stem cells (GSCs) in Drosophila males have provided an initial support for this idea (Tran et al., 2012; Xie et al., 2015). These studies demonstrated a bias in the incorporation and inheritance of preexisting “old” histones and newly synthesized “new” histones (and associated posttranslational modifications) in GSCs undergoing ACD. More recently, differential condensation of sister chromatids via the prereplication complex component Cdc6 was also reported in GSCs (Ranjan et al., 2022). Critically, disruption of histone inheritance patterns leads to stem cell maintenance and cellular differentiation phenotypes in Drosophila testes (Tran et al., 2012; Xie et al., 2015; Ranjan et al., 2022).

An additional element of the SSH proposed the existence of epigenetically distinct sister centromeres (Lansdorp, 2007). Evidence from Drosophila intestinal stem cells provided a first support for this notion (García Del Arco et al., 2018). In this study, labeling and tracking of “old” and “new” versions of the centromeric histone CENP-A (called Centromere Identifier CID in flies) showed that “old” CENP-A was preferentially retained in the stem cell. More recent experiments performed in Drosophila GSCs have now shown that in mitosis, centromeres of sister chromatids that end up in the stem cell harbor a higher level of CID compared with those that end up in the differentiating daughter cell (Ranjan et al., 2019; Dattoli et al., 2020). In male GSCs, an approximate 1.5-fold asymmetric distribution of CID between sister centromeres was measured at prophase/prometaphase, anaphase, and telophase (Ranjan et al., 2019). In female GSCs, 1.2-fold CID asymmetry was measured at prometaphase/metaphase (Dattoli et al., 2020). In addition to CID, Drosophila centromeres are comprised of two major components, the centromere assembly factor chromosome alignment defect 1 (CAL1; Chen et al., 2014) and centromere protein CENP-C that serves as the platform for kinetochore assembly and microtubule attachment (Przewloka et al., 2011). In male GSCs, an approximate 1.7-fold asymmetry for CAL1 was reported at prometaphase (Ranjan et al., 2019), while in females 1.4-fold asymmetry for CENP-C was measured at anaphase and telophase (Carty et al., 2021). Also in males, the kinetochore component NDC80 displayed an overall 1.49-fold enrichment in the sister chromatids segregated to the stem cell compared with those segregated to the daughter cell (Ranjan et al., 2019). Spindle microtubules are also asymmetrically distributed in GSCs. In males, there is a temporal asymmetry in microtubule dynamics such that centromeres on the “stem cell” side, with more CID, are captured first (Ranjan et al., 2019); in females, more spindle microtubules were found to emanate from the pole positioned in the direction of the future stem cell (Dattoli et al., 2020). Based on these observations, a mitotic drive model has been proposed as a mechanism that contributes to ACD via centromere strength (Kochendoerfer et al., 2021; Ranjan and Chen, 2022). According to this model, during ACD, a cascade of spatial and temporal events occurs in an asymmetric manner (centrosome and microtubule activity, nuclear membrane breakdown, centromere and kinetochore assembly). Ultimately, these events bias which sister chromatid ends up in the stem or daughter cell that in turn sets up cell fate differences through the inheritance of a distinct set of epigenetic marks.

To determine the importance of centromere asymmetry for ACD in GSCs, and ultimately in cell fate, studies so far have focused on centromere depletion and overexpression strategies. Overexpression of CID together with CAL1 in female GSCs disrupted CID asymmetry leading to more stem cells relative to daughter cells in the ovary; whereas CID or CAL1 knockdown in GSCs blocked cell division (Dattoli et al., 2020). While overexpression of CENP-C was not sufficient to disrupt centromere asymmetry in female GSCs, knockdown of CENP-C also resulted in more GSCs relative to daughter cells in the ovary (Carty et al., 2021). CAL1 knockdown experiments in male GSCs have shown loss of CID asymmetry and an eventual reduction in GSC number (Ranjan et al., 2019). However, the impact of CAL1 depletion on the balance of stem and daughter cells in the male niche was not assessed. Moreover, CENP-C asymmetry and function in male GSCs is so far unexplored. In this study, we investigate the effects of CID, CAL1, CENP-C depletion, and overexpression on centromere asymmetry and function in male GSCs and impacts on the balance of stem and daughter cells in the niche.

RESULTS

CID, CAL1, and CENP-C are asymmetrically distributed between GSCs and GBs at S phase

In Drosophila males, the GSC niche, the specific microenvironment in which GSCs are present in an undifferentiated and self-renewable state, is located at the apical end of each testis. Depending on the fly line, between 6 and 15 GSCs are positioned surrounding a cluster of nondividing stromal cells called the hub (Kochendoerfer et al., 2021). GSCs frequently undergo ACD to give rise to a new GSC that remains associated with the hub, as well as a daughter cell—the gonialblast (GB)—that moves away from the hub and initiates differentiation (Figure 1A). GBs undergo four rounds of mitosis to generate 2-, 4-, 8- and 16-cell cysts (CC) that undergo meiosis to eventually generate 64 haploid spermatozoa. GSCs or GBs can be identified based on the spectrosome, which is a spectrin-rich cytoplasmic organelle. The spectrosome changes morphology through the cell cycle and it can be used to indicate GSC cell cycle phase, first documented in females (Ables and Drummond-Barbosa, 2013). In males, the spectrosome structure between GSC and GB can be used as a cellular marker to recognize the two daughter cells derived from a newly divided GSC (Tran et al., 2012).

FIGURE 1:

(A) Schematic of the male germline stem cell (GSC) niche and cell divisions occurring in the Drosophila adult testis. GSCs are positioned in close contact with a patch of condensed nuclei that comprise the hub (gray circles). GSCs and daughter gonialblasts (GBs) can be identified by a round spectrin-rich organelle, the spectrosome. The spectrosome forms a bridged shape in S phase that connects newly divided GSC and GB pairs. GBs undergo four rounds of mitosis generating 2-, 4-, 8- and 16-cell cysts (CC). 16CC undergo meiosis to give bundles of 64 haploid spermatozoa. (B) Immunofluorescence image of the GSC niche of wild-type Drosophila testes stained for CID (white), the spectrosome together with the hub (red), DAPI (cyan), and pulse labeled with EdU (green). Scale bar = 5 μm. Red dashed line outlines the hub. White dashed line highlights a GSC and GB pair in S phase; zoom shows CID foci in highlighted GSC and GB nuclei. (C) Immunofluorescence image of the GSC niche of wild-type Drosophila testes stained for CAL1 (white), the spectrosome together with the hub (red), CID to mark centromeres (magenta), and pulse labeled with EdU (green). Scale bar = 5 μm. Red dashed line outlines the hub. White dashed line highlights a GSC and GB pair in S phase; zoom shows CAL1 foci in highlighted GSC and GB nuclei. Scale bar = 5 μm. (D) Immunofluorescence image of the GSC niche of wild-type Drosophila testes stained for CENP-C (white), the spectrosome together with the hub (red), DAPI (cyan), and pulse labeled with EdU (green). Scale bar = 5 μm. Red dashed line outlines the hub. White dashed line highlights a GSC and GB pair in S phase; zoom shows CENP-C foci in highlighted GSC and GB nuclei. (E) Quantitation of the ratio of total CID, CAL1, or CENP-C fluorescent intensity (integrated density) between GSC and GB S-phase pairs in the wild type. Each point represents the ratio of total intensity between GSC vs. its corresponding GB. Values 1.46 (for CID), 1.51 (for CAL1), and 1.35 (for CENP-C) indicate mean fold differences in intensity. At least 20 GSC-GB pairs were analyzed per quantitation.

Previous studies showed that in Drosophila male GSCs, CID and CAL1 are asymmetrically distributed between the centromeres of sister chromatids in mitosis (Ranjan et al., 2019). To assess whether this asymmetry is maintained in the next cell cycle, we pulse labeled testes with EdU for 30 min to identify cells undergoing DNA replication in S phase and used immunolabeling of the bridge-shaped spectrosome to identify newly divided GSC-GB pairs. Immunofluorescence staining for CID, CAL1, and CENP-C indicated that more centromere proteins are present in the GSC compared with the GB (Figure 1, B–D). Quantitation showed an approximate 1.46-fold enrichment for CID in GSCs compared with GBs (Figure 1E). Previously, 1.5-fold CID asymmetry was reported in male GSCs in mitosis (Ranjan et al., 2019). Therefore, the level of asymmetry appears to be comparable between mitosis and S phase. An earlier replication timing was recently reported to occur in GBs, meaning that newly divided GSCs and GBs can enter asynchronously into S phase (Ranjan et al., 2022). To determine any impact on CID asymmetry measurements, we performed short 10-min EdU pulse labeling that effectively captured GBs showing a more intense EdU signal than GSCs (Supplemental Figure 1A). Quantitation of CID asymmetry showed no significant difference between short and long EdU pulse labeling or between GSC-GB pairs in which the GB had a stronger EdU signal and all other GSC-GB pairs (Supplemental Figure 1B). For CAL1, we measured a 1.51-fold enrichment in GSCs compared with GBs in S phase. Previously, 1.7-fold asymmetry was reported in mitosis (prophase; Ranjan et al., 2019). Here, differences in the level of asymmetry detected might reflect different CAL1 dynamics between M and S phases or between GSCs and GBs. Indeed, we noted that measured intensities for CAL1 were variable in S phase. Finally, quantitation of CENP-C showed a 1.35-fold enrichment in GSCs, which is comparable to previous measurements made in female GSCs in mitosis (1.4-fold; Carty et al., 2021) and in S phase (1.3-fold; Dattoli et al., 2020). Our experiments now show that the asymmetric distribution of centromere proteins CID, CAL1, and CENP-C between sister chromatids first observed in mitosis, is carried through into S phase of the next cell cycle.

RNAi depletion of CID, CAL1, or CENP-C leads to a reduced number of GSCs per testis

We next aimed to determine whether the asymmetric distribution of CID, CENP-C, and CAL1 observed at S phase is important for GSCs maintenance. Previous knockdown experiments in male GSCs showed that CAL1 depletion disrupted the asymmetric CID distribution between sister chromatids at anaphase/early telophase (Ranjan et al., 2019). A reduced number of GSCs per testis was also reported upon CAL1 RNAi. Using a similar strategy (spatiotemporally controlled knockdown using nanos-GAL4 activated by the tubGAL80 temperature-sensitive driver, nanos-GAL4; tubGAL80ts), we performed knockdowns of CID and CENP-C, in addition to CAL1 knockdown with an independent RNAi line. We first confirmed respective knockdowns by immunofluorescent staining of testes for CID, CAL1, or CENP-C 10 d after RNAi induction (Figure 2A). We quantified a 40% reduction of CID in GSCs, a 40% reduction of CAL1 in GSCs, and a 37% reduction of CENP-C in GSCs in respective RNAi experiments (Figure 2B). We also noted that CID levels decreased in CAL1 RNAi and CENP-C RNAi (Supplemental Figure 1C), which is expected given that CAL1 and CENP-C function in CID assembly (Erhardt et al., 2008). In parallel, we confirmed germline defects by immunofluorescent staining of testes for the germ cell marker VASA, the spectrosome, and the hub (Figure 2A).

FIGURE 2:

(A) Immunofluorescence image of testis from nanos-GAL4; tubGAL80ts control, CID RNAi, CAL1 RNAi, and CENP-C RNAi 10 d after RNAi induction. Testes from respective RNAi experiments and controls are stained for either CID, CAL1, or CENP-C (yellow), the spectrosome together with the hub (red), VASA (gray), and DAPI (cyan). Scale bar = 10 μm. (B) Quantitation of CID, CAL1, and CENP-C knockdown in GSCs (n = 25–30) 10 d after respective RNAi induction. ****, p < 0.0001; ***, p < 0.001; *, p < 0.05; error bars = SEM. (C) Immunofluorescence image of testis from nanos-GAL4; tubGAL80ts control, CID RNAi, CAL1 RNAi, and CENP-C RNAi 10 d after RNAi induction. Testes are stained for pMAD (gray) and the spectrosome together with the hub (red). Yellow dashed circles indicate pMAD-positive GSCs with a round spectrosome located in contact with the hub selected for quantitation in 2D. Scale bar = 10 μm. (D) Quantitation of the number of GSCs per testes (n = 20–30) in the nanos-GAL4; tubGAL80ts control, nontarget mCherry RNAi control, CID RNAi, CAL1 RNAi, and CENP-C RNAi 10 d after RNAi induction. GSCs were identified as pMAD-positive cells with a round spectrosome that were attached to the hub. ****, p < 0.0001; ns, nonsignificant; error bars = SEM.

We next utilized staining for phosphorylated mothers against decapentaplegic (pMAD), a bone morphogenetic protein signaling indicator present in GSCs (Song et al., 2004), together with the presence of the round spectrosome in cells in direct contact with the DAPI-dense cluster of hub cells to identify and quantify GSC number per testis (Figure 2C). Using these criteria, we counted between four and five GSCs per testis in nanos-GAL4; tubGAL80ts and in nontarget mCherry RNAi controls 10 d after RNAi induction (Figure 2D). In the CID, CAL1, and CENP-C RNAi, the number of GSCs per testis was significantly reduced to approximately two for each line. This result indicates that depletion of CID, CAL1, or CENP-C in GSCs leads to a reduced number of GSCs per testis. Despite this reduction in GSCs observed 10 d after RNAi induction, all lines continued to produce mature sperm indicating that the final stage of spermiogenesis was not yet affected at this RNAi timepoint (Supplemental Figure 1D). Finally, we performed fertility tests and measured no significant difference in percentage fertility between RNAi lines and the control, measured up to 23 d after RNAi induction (Supplemental Figure 1E).

Overexpression of CID or CID together with CAL1, but not CAL1 alone, disrupts the asymmetric distribution of CID between GSCs and GBs

The observed GSC loss in testis depleted for CID, CAL1, and CENP-C is indicative of stem cell self-renewal failure. However, this phenotype could also arise due to cell division defects related to these essential genes. As a complementary means to disrupt centromere asymmetry in GSCs, we utilized the GAL4-UAS system to induce the overexpression of centromere components in GSCs. Importantly, we previously employed this strategy to successfully disrupt centromere asymmetry in female GSCs (Dattoli et al., 2020). For this purpose, we crossed flies carrying CID-mCherry (CID_OE), CAL1-YFP (CAL1_OE), or both CAL1-YFP and CID-mCherry (CAL1-CID_OE) transgenes to a nanos-GAL4 driver line. We first validated the induced expression of the transgenes in male GSCs from 3-d-old F₁ progeny and confirmed expected localization patterns for CID-mCherry, CAL1-YFP, and CID-mCherry_CAL1-YFP (Supplemental Figure 2A). Next, we stained the overexpression lines with anti-CID antibody to quantify total CID intensity and determine the level of CID overexpression in GSCs (Figure 3A). For this we isolated GSCs with a round spectrosome, which are mostly at G2 phase, the time of CID assembly in these cells (Ranjan et al., 2019). Quantitation revealed an approximate twofold increase in total CID intensity per GSCs in all three overexpression lines (Figure 3B). We noted that CID intensity was variable, which might reflect gradual CID assembly during a long G2 phase in GSCs. Additionally, we quantified the total number of CID foci per GSC nucleus to determine whether CID and/or CAL1 overexpression resulted in extra CID foci per nucleus (Figure 3C). In control GSCs, less than four foci were detected indicative of centromere clustering in these cells. In CID_OE and CAL1-CID_OE lines the number of foci is significantly increased to approximately five, possibly representing ectopic centromeres or reduced centromere clustering, whereas no significant increase was detected in the CAL1_OE. We next assayed for CID asymmetry in GSC-GB pairs in S phase (Figure 3D). We found that CID or CAL1-CID overexpression disrupts CID asymmetry, giving a symmetric distribution of CID between GSCs and GBs (GSC/GB CID intensity ratio of approximately 1.0) compared with the asymmetric distribution measured in the nanos-GAL4 control (GSC/GB CID intensity ratio = 1.43), which was consistent with the wild-type value (Figure 1E). In contrast, CAL1 overexpression did not significantly change CID distribution and the GSC/GB CID intensity ratio remained at 1.48 (Figure 3E). F₁ males from all three overexpression lines produced mature sperm stored in the seminal vesicle (Supplemental Figure 2B). Fertility tests showed no significant difference in percentage fertility between OE lines and the control (Supplemental Figure 2C).

FIGURE 3:

(A) Immunofluorescence image of control nanos-GAL4 testis, or testis overexpressing CID-mCherry (CID_OE), CAL1-YFP (CAL1_OE), and CAL1-YFP together with CID-mCherry (CAL1-CID_OE) stained for CID (cyan), the spectrosome (red), and VASA marking germ cells (gray). White arrows and dashed circles indicate typical GSCs isolated for quantitation that were VASA positive, with a round spectrosome and located adjacent to the hub (identified using DAPI; unpublished data). Scale bar = 10 μm. (B) Quantitation of total CID fluorescent intensity (integrated density) in GSCs (n = 25–30) in the control nanos-GAL4, and in CID_OE, CAL1_OE, or CAL1-CID_OE lines. ****, p < 0.0001; **, p < 0.01; error bars = SEM. n = 30 GSCs per quantitation. (C) Quantitation of total number of CID foci in GSCs (n = 28) in the control nanos-GAL4, and in CID_OE, CAL1_OE, or CAL1-CID_OE lines. ****, p < 0.0001; ***, p < 0.001; ns, nonsignificant; error bars = SEM. n = 30 GSCs per quantitation. (D) Immunofluorescence image of control nanos-GAL4 testis, or testis overexpressing CID-mCherry (CID_OE), CAL1-YFP (CAL1_OE), or CAL1-YFP together with CID-mCherry (CAL1-CID_OE) stained for CID (white), the spectrosome together with the hub (red), DAPI (cyan), and pulse labeled with EdU (green). Red dashed line outlines the hub. White dashed line highlights a GSC and GB pair in S phase. The cluster of DAPI-dense nuclei was used to identify the hub in CAL1-CID_OE. Scale bar = 10 μm. (E) Quantitation of the ratio of total CID fluorescent intensity (integrated density) between GSC and GB S-phase pairs in nanos-GAL4, CID_OE, or CAL1-CID_OE lines. Values 1.43 (for nanos-GAL4), 1.0 (for CID_OE), 1.0 (for CAL1-CID_OE), and 1.48 (for CAL1_OE) indicate mean fold differences in intensity. Between 25 and 30 GSC-GB pairs were analyzed per quantitation, with the exception of CAL1-CID (n = 21).

Overexpression of CID or CID together with CAL1 disrupts the stem and daughter cell balance in the niche; CAL1 overexpression leads to more germ cell proliferation

To determine the effects of CID and/or CAL1 overexpression on the balance of stem and daughter cells in the GSC niche, we devised two functional assays (Figures 4 and 5). The first assay was based on spectrosome morphology and VASA staining to mark germ cells (Figure 4A). VASA-positive cells with a round spectrosome located in close association with the hub were annotated as GSCs. In addition, all VASA-positive cells with a round spectrosome up to and including the two-cell cyst stage (2CC) were counted. Compared to the nanos-GAL4 control, CID_OE, CAL1_OE, and CAL1-CID_OE lines showed a disrupted and less organized GSC niche (Figure 4B). An expansion of the hub diameter and number of hub nuclei was also measured in the case of CID_OE and CAL1_OE (Supplemental Figure 3, A–C), which are likely secondary effects of germline disruption (Gönczy and DiNardo, 1996; Dinardo et al., 2011).

FIGURE 4:

(A) Schematic of the assay used to quantify the balance of GSCs (in yellow), GBs and 2CC (in white) in Drosophila adult testis. (B) Immunofluorescence image of control nanos-GAL4 testis, or testis overexpressing CID-mCherry (CID_OE) or CAL1-YFP (CAL1_OE) or CAL1-YFP together with CID-mCherry (CAL1-CID_OE) stained for the spectrosome (red), VASA marking germ cells (gray), and DAPI (cyan). Yellow dashed circles indicate GSCs with a round spectrosome located in contact with the hub. White dashed circles indicate GBs or 2CCs with a round spectrosome not in contact with the hub. Scale bar = 10 μm. (C) Quantitation of total number of GSCs or germ cells up to 2CC in the control nanos-GAL4, and in CID_OE, CAL1_OE, or CAL1-CID_OE lines. ****, p < 0.0001; ***, p < 0.001; *, p < 0.05; error bars = SEM. n = 30 testes were analyzed per quantitation. (D) Ratio of all germ cells up to 2CC divided by the number of GSCs in the control nanos-GAL4, and in CID_OE, CAL1_OE, or CAL1-CID_OE lines. **, p < 0.01; ****, p < 0.0001; ns, nonsignificant; error bars = SEM.

FIGURE 5:

(A) Schematic of the assay used to quantify the balance of GSCs (in magenta), GBs and 2CC (in white) in Drosophila adult testis. GSCs were counted based on a pMAD-positive signal with a round spectrosome adjacent to the hub. All germ cells up to the 2CC were VASA positive and BAM negative. BAM expression in 4CC is shown in red. (B) Immunofluorescence image of control nanos-GAL4 testis, or testis overexpressing CID-mCherry (CID_OE) or CAL1-YFP (CAL1_OE) or CAL1-YFP together with CID-mCherry (CAL1-CID_OE) stained for pMAD (magenta), BAM (red), VASA marking germ cells (gray), and DAPI (cyan). Magenta dashed circles indicate pMAD-positive GSCs in contact with the hub. White dashed circles indicate VASA-positive cells counted up to 2CC. Scale bar = 10 μm. (C) Quantitation of total number of GSCs or germ cells up to 2CC in the control nanos-GAL4, and in CID_OE, CAL1_OE, or CAL1-CID_OE lines. ****, p < 0.0001; ***, p < 0.001; **, p < 0.01; ns, nonsignificant; error bars = SEM. n = 30 testes were analyzed per quantitation. (D) Ratio of all germ cells up to 2CC divided by the number of GSCs in the control nanos-GAL4, and in CID_OE, CAL1_OE, or CAL1-CID_OE lines. ****, p < 0.0001; *, p < 0.05; ns, nonsignificant; error bars = SEM.

Using our criteria to identify GSCs, we quantified the number of GSCs per testis (Figure 4C). Between four and five GSCs were detected in the control, whereas this number increased in CID_OE (6.3), CAL1_OE (6), and CAL1-CID_OE (8.4) lines. Quantitation of all VASA-positive cells with a round spectrosome up to 2CC also revealed an increase in CID_OE (17.3), CAL1_OE (18.4), and CAL1-CID_OE (17.6) lines compared with the control (15.8). When expressed as a ratio, for every one GSC we observed approximately 3.4 “daughter cells” in the control (Figure 4D). In CID_OE and CAL1-CID_OE, for every one GSC there are 2.8 and 2.2 daughter cells, respectively, meaning that there are less daughter cells for every stem cell. For CAL1 overexpression, the ratio (3.1) is not significantly different from control; however, we observed an overall increase in the total number of germ cells in the niche. Longer term CAL1 overexpression studies (after 10 d) showed a further increase in the total number of round-spectrosome and VASA-positive germ cells per testis (Supplemental Figure 3D). In this case, for every one GSC, 4.5 daughter cells were detected (Supplemental Figure 3E). Taken together, the results for CAL1 overexpression after 3 and 10 d indicate that CAL1 expression might drive germ cell proliferation, including GSCs, in the niche.

To validate these findings, we developed a second functional assay to measure the balance of stem and daughter cells in the male GSC niche based on an independent set of cellular markers (Figure 5A). We noted that an ectopic pMAD signal, localized away from the hub region, was sometimes observed in control and OE lines. For our experiments, GSCs were specifically identified as cells with a pMAD signal directly in contact with the DAPI-dense nuclei that comprise the hub region. In addition, all VASA-positive cells that were negative for the differentiation marker bag of marbles (BAM) that is turned on at the four-cell cyst (4CC) stage. Again, we noted that compared with the nanos-GAL4 control, CID_OE, CAL1_OE, and CAL1-CID_OE lines showed a disrupted and less organized GSC niche (Figure 5B). Using this second set of criteria to identify GSCs, we quantified the number of GSCs per testis (Figure 5C). Approximately five (4.6) GSCs were detected in the control, whereas this number increased in CID_OE (7.5), CAL1_OE (7.6), and CAL1-CID_OE (9.7) lines. Quantitation of all VASA-positive cells up to the 4CC when BAM is expressed also revealed an increase in CID_OE (13.3), CAL1_OE (13), and CAL1-CID_OE (12.2) lines compared with the control (9.7). When expressed as a ratio, we observed approximately 2.5 daughter cells for every GSC in the control (Figure 4D). In CID_OE and CAL1-CID_OE, for every one GSC there are approximately 1.5 daughter cells, meaning that there are less daughter cells for every stem cell. For CAL1 overexpression, the ratio is not significantly different from the control; however, we observe an overall increase in the total number of germ cells in the niche and more variability in these samples (see SEM error bars).

Taken together, both assays of stem and daughter cell balance indicate that upon CID_OE or CAL1-CID_OE the number of daughter cells relative to GSCs is reduced, meaning that the pool of GSCs is increased. In the case of CAL1_OE, both assays indicate no change in the balance of stem to daughter cells. Instead, the overall total number of germ cells (GSCs up to 2CCs) is increased upon CAL1 overexpression.

Extra GSCs observed upon CID/CAL1 overexpression are not due to dedifferentiation

Given that we detected an increased pool of GSCs upon CID and CAL1-CID overexpression, we next wanted to determine the origin of such cells. One possibility is that extra GSC-like cells have arisen due to dedifferentiation of GBs, as previously documented (Cheng et al., 2008). Centrosome positioning can be used to assay for GSC-like cells derived from dedifferentiated GBs (Cheng et al., 2008). We stained CID_OE, CAL1_OE, and CAL1-CID_OE lines with the centrosome marker anti-gamma tubulin and calculated the percentage of GSCs with misorientated centrosomes per testis. GSCs in which one centrosome is normally positioned adjacent to the hub interface are defined as oriented; whereas GSCs with both centrosomes positioned away from the hub are defined as misoriented (Supplemental Figure 3F). Quantitation in CID_OE (8%), CAL1_OE (3%), and CAL1-CID_OE (10%) lines did not reveal any significant increase in the percentage of GSCs with misoriented centrosomes compared with the nanos-GAL4 control (6%), suggesting that extra GSCs are not derived from GB dedifferentiation (Supplemental Figure 3G).

Overexpression of CENP-C does not affect niche composition

Previous experiments in Drosophila female GSCs showed that in contrast to CID and CAL1 overexpression, overexpression of CENP-C did not impact on centromere asymmetry or cell fate (Carty et al., 2021). To investigate whether this observation holds true in the differentially regulated GSC niche in males, we overexpressed HA-tagged CENP-C (HA-CENP-C_OE) in GSCs in testes using the nanos-GAL4 driver (Figures 6 and 7). We first confirmed the induction of HA-CENP-C in pMAD-positive GSCs using an anti-HA antibody (Supplemental Figure 4A). We also confirmed that HA-CENP-C localized at centromeres by costaining with anti-CID antibody (Supplemental Figure 4B). For GSCs captured at S phase, no significant change in CID intensity was detected upon HA-CENP-C expression compared with the nanos-GAL4 control (Figure 6, A and B), indicating that excess CENP-C is not sufficient to drive centromere assembly. Next, we assessed the effects of HA-CENP-C overexpression on CID asymmetry (Figure 6C). Again, we measured no significant change (ratio GSC/GB = 1.43) compared with the nanos-GAL4 control (ratio GSC/GB = 1.43) (Figure 6C). Finally, we quantified GSCs (pMAD positive with a round spectrosome and in contact with the hub) and germ cells up to 2CC (VASA positive with a round spectrosome; Figure 7). For HA-CENP-C_OE, we did not find a significant change in stem and daughter cell balance compared with the nanos-GAL4 control (Figure 7, B and C). This result is very comparable to that reported in female GSCs, in which the overexpression of CENP-C did not affect CID assembly, asymmetry, or stem and daughter cell balance in the ovary (Carty et al., 2021).

FIGURE 6:

(A) Immunofluorescence image of control nanos-GAL4, HA-CENP-C_OE, CENP-C RNAi and HA-CENP-C; CENP-C RNAi (HA-CENP-C rescue) testis stained for CID (white), the spectrosome together with the hub (red), DAPI (cyan), and pulse labeled with EdU (green). White dashed line highlights a GSC and GB pair in S phase. Scale bar = 10 μm. (B) Quantitation of total CID fluorescent intensity (integrated density) in GSCs in the control nanos-GAL4, HA-CENP-C_OE, CENP-C RNAi and HA-CENP-C; CENP-C RNAi (HA-CENP-C rescue) lines. **, p < 0.01; ns, nonsignificant; error bars = SEM. n = 25–30 GSCs per quantitation. (C) Quantitation of the ratio of total CID fluorescent intensity (integrated density) between GSC and GB S-phase pairs in nanos-GAL4, HA-CENP-C_OE, CENP-C RNAi and HA-CENP-C; CENP-C RNAi (HA-CENP-C rescue) lines. Values 1.43 (for nanos-GAL4), 1.43 (for CENP-C_OE), 1.96 (for CENP-C RNAi), and 1.51 (for HA-CENP-C rescue) indicate mean fold differences in intensity. **, p < 0.01; ns, nonsignificant; error bars = SEM. n = 25–30 GSCs per quantitation.

FIGURE 7:

(A) Immunofluorescence image of control nanos-GAL4, HA-CENP-C_OE, CENP-C RNAi and HA-CENP-C; CENP-C RNAi (HA-CENP-C rescue) testis stained for pMAD (gray), the spectrosome together with the hub (red), and DAPI (cyan). Yellow dashed circles indicate GSCs with a round spectrosome located in contact with the hub. White dashed line indicates area in which cells with a round spectrosome up to 2CC were counted. Scale bar = 10 μm. (B) Quantitation of total number of GSCs or germ cells up to 2CC in the control nanos-GAL4, HA-CENP-C_OE, CENP-C RNAi and HA-CENP-C; CENP-C RNAi (HA-CENP-C rescue) lines. **, p < 0.01; ns, nonsignificant; error bars = SEM. n = 30 testes were analyzed per quantitation. (C) Ratio of all germ cells up to 2CC divided by the number of GSCs in the control nanos-GAL4, HA-CENP-C_OE, CENP-C RNAi and HA-CENP-C; CENP-C RNAi (HA-CENP-C rescue) lines. **, p < 0.01; ns, nonsignificant; error bars = SEM.

CENP-C depletion disrupts the balance of stem and daughter cells in the niche

While RNAi knockdown of CENP-C using the nanos-GAL4; tubGAL80ts driver impacted on the total number of GSCs per testis (Figure 2), we were unable to detect a major reduction in CID intensity (Supplemental Figure 1C). We reasoned that in this RNAi experiment CENP-C kinetochore function might be perturbed, affecting cell division without impacting on CENP-C’s function in CID assembly or maintenance. Instead, CENP-C knockdown driven by the nanos-GAL4 driver, which is active earlier in the development of the F₁ progeny, led to a more robust CENP-C knockdown and we quantified a significant reduction in CID intensity at GSC centromeres (Figure 6, A and B, and Supplemental Figure 4B). In this CENP-C RNAi experiment, rather than a loss of CID asymmetry, we observed an unexpected increase in CID asymmetry in S-phase GSC-GB pairs (ratio GSC/GB = 1.96; Figure 6C). Furthermore, the number of GSCs increased (pMAD positive with round spectrosome and in contact with the hub) with no change in the number of germ cells up to 2CC (VASA positive with round spectrosome; Figure 7, A–C). This resulted in a significant change in the stem and daughter cell balance, with approximately 2 daughter cells for every GSC, compared with 2.5 daughter cells for every GSC in the nanos-GAL4 control (Figure 7, B and C). Using the nanos-GAL4 driver, neither CID asymmetry nor stem and daughter cell balance was affected in the nontarget mCherry RNAi control (Supplemental Figure 4, C and D). Rescue experiments in which we overexpressed HA-CENP-C that is resistant to the shRNA in the CENP-C RNAi background showed no significant changes in CID intensity at centromeres, CID asymmetry (1.51), nor in the stem to daughter cell balance (2.5:1; Figures 6 and 7). Finally, CENP-C overexpression or depletion did not impact on fertility, with mature sperm detected in testes from these lines (Supplemental Figure 4E) and no significant effect on the percentage fertility (Supplemental Figure 2C). These findings for CENP-C in male GSCs are strikingly similar to those reported in female GSCs, in which CENP-C RNAi increases CID asymmetry between GSCs and cystoblasts, the GB equivalent in females. Moreover, CENP-C RNAi in female GSCs shifted the balance of stem and daughter cells such that more GSCs were detectable in the ovary (Carty et al., 2021).

DISCUSSION

This study investigates CID, CAL1, and CENP-C inheritance in newly divided stem (GSC) and daughter (GB) cells that have entered S phase of the next cell cycle. We find an approximate 1.4-fold asymmetric distribution for CID, CAL1, and CENP-C, with more centromere proteins detectable in GSCs compared with GBs. This result is consistent with previous measurements of CID and CAL1 asymmetry at mitosis in male GSCs (Ranjan et al., 2019), as well as asymmetries reported for CID and CENP-C at S phase in female GSCs (Dattoli et al., 2020; Carty et al., 2021). Recently, asynchrony in DNA replication timing between the GSC and GB was documented, in which the GB can enter earlier into S phase (Ranjan et al., 2022). Given that CID assembly occurs in G2 phase in GSCs, no new CID assembly is expected in S phase and our results indicate that CID asymmetry measurements do not change significantly as GSCs and GBs progress through S phase. It is now important to understand when and how such asymmetries are established and maintained during the cell cycle. One possibility is that asymmetry is established during the time of centromere assembly, occurring at G2 phase up to prophase in GSCs. Differential pools of old and new CAL1 at sister centromeres in GSCs supports asymmetric CID assembly and maintenance at this cell cycle time (Ranjan et al., 2019). Asymmetric CID assembly in G2 phase might also explain the unique cell cycle timing of centromere assembly in GSCs (Dattoli et al., 2020). In these stem cells undergoing ACD, assembly occurs before chromosome segregation, as opposed to after in most other (symmetrically diving) cell types (Carty and Dunleavy, 2020). A second, and not mutually exclusive, possibility is that the redistribution of parental CID nucleosomes during S phase sets up this asymmetry between sister centromeres. Indeed, unidirectional DNA replication forks appear to be enriched in GSCs, which might provide a mechanism for this (Wooten et al., 2019). Further single-cell investigations in S-phase GSCs in which sister centromeres might be resolved are required to determine whether this is the case.

To disrupt centromere function and asymmetry in GSCs, we employed overexpression and depletion of CID, CAL1, and CENP-C. Spatiotemporally controlled knockdown of CID, CAL1, and CENP-C resulted in reduced numbers of GSCs per testis 10 d after RNAi induction, which is indicative of stem cell self-renewal failure or defective GSC maintenance. Our result for CAL1 depletion is consistent with GSC loss and decreased expression of the GSC marker Stat92E reported to occur 10 d after RNAi (Ranjan et al., 2019). In CENP-C RNAi experiments using the nanos-GAL4; tubGAL80ts driver, although we recorded a reduction in GSCs per testis, the CID level in GSCs appeared to be unaffected. We reasoned that in this RNAi experiment CENP-C kinetochore function might be perturbed, affecting cell division without impacting on CENP-C’s function in CID assembly or maintenance. Therefore, we utilized the nanos-GAL4 driver to induce a stronger depletion of CENP-C, which resulted in an approximate 22% drop in CID at GSC centromeres. In this case, we noted an expansion of the GSCs pool. We suppose that CENP-C depletion must reach a threshold before it impacts on CID level at centromeres, which in turn is required to impact on GSC maintenance. Our result in males is in line with previous CENP-C RNAi experiments performed in female GSCs. GSCs in the ovary with a 60% reduction in CENP-C displayed a 35% drop in CID at centromeres with an initial expansion of the GSC pool and a subsequent depletion of the GSC pool upon long-term CENP-C RNAi (Carty et al., 2021).

Given that CID, CAL1, and CENP-C are essential genes, we adopted an overexpression strategy to overcome analysis of complex cell division phenotypes related to centromere function. Overexpression of CID, CAL1, and CID together with CAL1 resulted in a twofold increase in CID intensity in GSCs. For overexpression of CID or CID together with CAL1, but not for CAL1 overexpression alone, this perturbed CID asymmetry and gave a symmetric distribution between stem and daughter cells. To determine the effects on stem cell fate, we developed two functional assays using independent sets of cellular markers to measure the effect on stem and daughter cell balance in the male niche. Both assays showed that overexpression of CID or CID together with CAL1 resulted in more stem cells relative to daughter cells. We interpret this result as more stem cell self-renewal occurring in the niche, which we can correlate with a symmetric CID distribution between GSCs and GBs. It is now important to understand the origin of these extra GSC-like cells in the niche. One possibility is that they arise due to extra symmetric cell divisions of GSCs. According to our fixed cell analyses of centrosome orientation, these GSCs do not appear to have come from dedifferentiation of GBs. However, live imaging of the testis stem cell niche might be required to accurately detect symmetric cell divisions (Greenspan and Matunis, 2017). Overexpression of CAL1 alone did not change the asymmetric distribution of CID between GSCs and GBs nor did it change the GSC and daughter cell balance. Instead, both assays showed that CAL1 overexpression correlated with a general increase in germ cell number in the niche. It is possible that CAL1 might drive cell proliferation in this context and this function could be independent of its role at centromeres. Indeed, a centromere-independent function for CAL1 in terminally differentiated enteroblasts in the Drosophila midgut has already been proposed (García Del Arco et al., 2018). In summary, these results in male GSCs are very comparable to our previous findings in female GSCs in which CID or CAL1-CID overexpression resulted in more stem cell self-renewal, while CAL1 overexpression increased germ cell proliferation (Dattoli et al., 2020).

Different to CID and CAL1, overexpression of CENP-C did not impact on CID intensity at centromeres and had no effect on CID asymmetry or stem cell maintenance. Instead, knockdown of CENP-C (using the nanos-GAL4 driver) increased CID asymmetry between GSCs and GBs. Again, this result in males is strikingly similar to our findings in female GSCs in which CENP-C depletion increased the level of CID asymmetry (Carty et al., 2021). Results for CENP-C RNAi indicate that more CID asymmetry can also increase the GSC pool in testes and ovaries. Taken together, it appears that either a loss (OE experiments, GSC/GB ratio 1:1) or an increase (CENP-C RNAi, GSC/GB ratio ∼2:1) in the level of CID asymmetry correlates with an increase in the GSC pool. This suggests that an optimal level of centromere asymmetry (approximately 1.5-fold more CID on GSC chromosomes) is potentially recognized by spindle microtubules, which can drive selective sister chromatid segregation. This bias in segregation can then lead to changes in gene expression that ultimately affect cell fate, contributing to a balanced niche. Although centromere depletion and overexpression approaches perturbed the number and balance of germ cells in the niche, adult males continued to produce mature sperm and were fertile. This observation questions the functional impact of having one less stem cell in the niche. Perhaps it is only under conditions of stress or advanced aging that this deficiency might come into play.

Our results show that manipulation of centromere function impacts on the stem and daughter cell balance within the niche. This effect could be mediated through gene expression due to changes in niche composition and extrinsic cell signaling or alternatively might be due to the inheritance of sister chromatids carrying specific gene expression patterns. Our results also identify conserved features between Drosophila male and female GSCs. The male GSC niche is more complex than in females and is distinct in terms of JAK-STAT/Upd signaling, spectrosome positioning, Bam expression pattern, and germline-soma interactions (Vidaurre and Chen, 2021). Moreover, although in males inheritance of the mother centrosome is critical for GSC maintenance, in females, GSCs inherit the daughter centrosome (Chen and Yamashita, 2021). Yet, despite these well-characterized differences between the sexes, centromere asymmetry and effects on cell fate are conserved between male and female GSCs. Although centromere strength alone is unlikely to be the major determinant of stem cell identity, our results place the centromere as an important player in stem cell ACD.

MATERIALS AND METHODS

Fly stocks and husbandry

Stocks were cultured on standard cornmeal medium (NUTRI-fly) preserved with 0.5% propionic acid and 0.1% tegosept at 20°C under a 12-h light-dark cycle. All fly stocks used were obtained from Bloomington Drosophila Stock Center or Vienna Drosophila Stock Center (#) unless otherwise stated. The following fly stocks were used: wild-type Oregon-R (BDSC #25211), UAS-dcr2; nanos-GAL4 (BDSC #25751), nanos-GAL4; tub-GAL80ts (gift from Yukiko Yamashita, Massachusetts Institute of Technology), UAS-CID RNAi (Vienna Drosophila Resource Center [VDRC] #102090), UAS-CAL1 RNAi (VDRC #45248), UAS-CENP-C RNAi (BDSC #38917), UASp-HA-CENP-C; SM6 Cy (gift from Kim S. McKim, Rutgers University), HA-CENP-C; UAS-CENP-C-RNAi (Carty et al., 2021), mCherry RNAi (gift from Gary Karpen, University of California Berkeley). Fly lines used for overexpression of CID, CAL1, and CAL1-CID under UAS control were generated in Dattoli et al. (2020). CENP-C knockdown (and rescue) using the nanos-GAL4 driver was performed at 22°C. For knockdown using the nanos-GAL4; tub-GAL80ts crosses were set at 18°C and progeny were shifted to 29°C upon eclosion. HA-CENP-C was induced using nanos-GAL4 at either 25°C or at 22°C for rescue experiments. F₁ progeny were dissected 3 d after eclosion unless stated otherwise. Results obtained from each experiment rely on three biological replicates, unless otherwise specified.

Immunofluorescence

Testes were squashed and snap-frozen in liquid nitrogen. After fixation in 4% paraformaldehyde for 10 min (5 min for anti-SMAD3/5), samples were immediately washed in 1X PBS-0.4% Triton-X100 (0.4% PBST). Samples were then blocked in 0.4% PBST with 1% bovine serum albumin for 2–4 h at room temperature, incubated with primary antibodies (in blocking buffer) overnight at 4°C. Samples were then washed in 0.4% PBST for 3 × 30 min. Secondary antibodies are added (1:500 in blocking buffer) for 2 h at room temperature in the dark. Samples are again washed 3 × 30 min in 0.4% PBST followed by the addition of DAPI (1:1000) for 15 min in 1X PBS.

EdU incorporation

Testes were dissected and incubated for 10 or 30 min with EdU (0.01 mM; Berry & Associates) in 1X PBS and then fixed as described. After washing in 0.4% PBST, testes were incubated for 30 min in the dark with 2 mM CuSO₄, 300 µM fluorescent azide, and 10 mM ascorbic acid. Samples were then washed with 0.4% PBST for 10 min and then blocked and stained as described above.

Antibodies

For immunostaining, the following primary antibodies were used: rabbit anti-CENP-A (CID) antibody (active motif 39719; 1:500), sheep anti-CENP-C (Dattoli et al., 2020); 1:2000), rabbit anti-VASA (Santa Cruz sc-30210; 1:500), rat anti-VASA (Developmental Studies Hybridoma Bank [DSHB]; 1:500), sheep anti-CAL1 (this study; 1:1000), rabbit anti-SMAD3/5 (pMad; Abcam ab52903; 1:500), DAPI (1:1000), mouse anti-Armadillo (DSHB N2 7A1; 1:700), mouse anti-1B1 (Hts; DSHB; 1:100), mouse anti-BAM (DSHB; 1:50), mouse anti-FIBRILLARIN (Abcam; ab5821; 1:500), mouse anti-HA (ThermoFisher; 26183; 1:500), mouse anti-gamma-tubulin (Abcam; ab44928; 1:500), mouse anti-FASCILLIN III (DSHB 7G10; 1:200), rat anti-mCherry (Chromotek 5F8; 1:200). Sheep anti-CAL1 antibody was raised against CAL1 amino acids 748–979 with a C-terminal His tag produced according to Bade et al. (2014). Species-specific Alexa Fluor 405, 488, 546, and 647 secondary antibodies were used at a 1:500 dilution.

Widefield microscopy

Images of immunostained testes mounted in SlowFade Gold antifade reagent (Invitrogen; S36936) were acquired using a DeltaVision Elite microscope system (Applied Precision) equipped with a 100×, 60×, or 40× oil immersion UPlanS-Apo objective (NA 1.4). Images were acquired as z-stacks with a step size of 0.5 μm. Fluorescence passed through a 435/48 nm; 525/48 nm; 597/45 nm; 632/34 nm band-pass filter for detection of, respectively, DAPI, Alexa Fluor 488, mCherry, and Alexa Fluor 647 in sequential mode.

Quantification

For CID, CAL1, or CENP-C intensity: For each quantification one cell/testis was considered. Images from a single cell (nucleus) were projected (max intensity) to capture all the centromeres present in the cell at a specific cell cycle phase. ImageJ software was used to measure fluorescent intensity of CID in the following way: The background was subtracted from the projected image. Threshold was adjusted and the image size was adjusted in order to eliminate unwanted objects. Following, the command “analyze particles” was used to select centromeres. Finally, integrated density (MGV*area) from each centromere foci were extracted and used as fluorescent intensity to measure the total amount of fluorescence per nucleus. GSC and daughter cells were counted in 3D from all layers of z-stack images.

Fertility tests

For each test, two virgin wild-type females were crossed with two (OE or RNAi) males and the total number of adult progeny was scored after 23 d. For RNAi, males were incubated at 29°C for 7 d before crossing. For OE, fertility crosses were carried out at 25°C.

Statistical analyses

Data distribution was assumed to be normal, but this was not formally tested. The p value in each graph shown was calculated with an unpaired t test with Welch’s correction. All statistical analysis was performed using Prism 9 software.

Abbreviations used:

ACD

asymmetric cell division

BAM

bag of marbles

CAL1

chromosome alignment defect 1

CC

cell cyst

Cdc6

cell division control protein 6

CID

centromere Identifier

DAPI

4’,6-diamidino-2-phenylindole

EdU

5-ethinyl-2’-deoxyuridine

GAL

galactose induced protein

GB

gonialblast

GSC

germline stem cell

HA

haemagglutinin

JAK-STAT

janus kinase-signal transducer and activator of transcription

NDC80

nuclear division cycle 80

pMAD

phosphorylated mothers against decapentaplegic

SEM

standard error of the mean

SMAD

suppressor of mothers against decapentaplegic

SSH

silent sister hypothesis

Stat92E

signal transducer and activator transcription protein at 92E

UAS

upstream activating sequence

Upd

unpaired

YFP

yellow fluorescent protein