Asymmetric division of Drosophila male germline stem cell shows asymmetric histone distribution

Abstract

Stem cells can self-renew and generate differentiating daughter cells. It is not known whether these cells maintain their epigenetic information during asymmetric division. Using a dual-color method to differentially label “old” versus “new” histones in Drosophila male germline stem cells (GSC), we show that preexisting canonical H3, but not variant H3.3, histones are selectively segregated to the GSC whereas newly synthesized histones incorporated during DNA replication are enriched in the differentiating daughter cell. The asymmetric histone distribution occurs in GSCs but not in symmetrically dividing progenitor cells. Furthermore, if GSCs are genetically manipulated to divide symmetrically, this asymmetric mode is lost. This work suggests that stem cells retain preexisting canonical histones during asymmetric cell divisions, probably as a mechanism to maintain their unique molecular properties.

Although all cells in an organism contain the same genetic material, different genes are expressed in different cells, allowing them to differentiate along distinct pathways. Epigenetic mechanisms regulate gene expression and maintain a specific cell fate through many cell divisions (1–3). Stem cells have the remarkable ability to both self-renew and generate daughter cells that enter differentiation (4). Epigenetic mechanisms have been reported to regulate stem cell activity in multiple lineages (5–7). However, to date, there has been little direct in vivo evidence demonstrating whether stem cells retain their epigenetic information.

The Drosophila male GSCs are well characterized in terms of their physiological location, microenvironment (i.e. niche), and cellular structures (8, 9) (Fig. 1A, 1B). Male GSCs can be identified precisely by their distinct anatomical positions and morphological features. A GSC usually divides asymmetrically to produce a self-renewed GSC and a daughter cell gonialblast (GB) that undergoes differentiation. Therefore, GSCs can be examined at single-cell resolution for a direct comparison between the two daughter cells from one GSC division.

Figure 1. Experimental design and potential results.

(A) A diagram of the GSC niche. HUB-hub cells, CySC- cyst progenitor/somatic stem cell. (B) Immunofluorescent image of the niche: HUB (anti-Fas III, red, asterisk), GSC-GB pair expressing H3-GFP (green, dotted outline) connected by a spectrosome (anti-α-Spectrin, red, arrow). (C) The UASp-FRT-histone-GFP-PolyA-FRT-histone-mKO-PolyA transgene. UAS: upstream activating sequence; FRT: FLP (flippase) recombination target; histone: H3, H2B, or histone variant H3.3. nanos-Gal4: a germline-specific driver. hs-FLP: the yeast FLP recombinase controlled by the heat shock (hs) promoter. (D–E) Two potential results: for simplicity, only one GSC-GB pair is shown, and each entire cell is colored according to histone fluorescence.

In eukaryotes, the basic unit of chromatin called nucleosome contains histone octamer (2XH3, H4, H2A, H2B) and DNA wrapping around them. Indeed, histones are one of the major carriers of epigenetic information (10). To address how histones are distributed during the GSC asymmetric division, we developed a switchable dual-color method to differentially label “old” versus “new” histones (Fig. 1C) that employs both spatial (by Gal4; UAS system) and temporal (by heat shock induction) controls to switch labeled histones from green [Green Fluorescent Protein (GFP)] to red [monomeric Kusabira-Orange (mKO)]. Heat shock treatment induces an irreversible DNA recombination to shut down expression of GFP-labeled old histones and initiate expression of mKO-labeled new histones. If the old histones are partitioned non-selectively, the GFP will initially exhibit equal distribution in the GSC and GB, and will be gradually replaced by the mKO (Fig. 1D). However, if the old histones are preferentially retained in the GSCs to constitute potentially GSC-specific chromatin structure, the GFP will be detected specifically in the GSCs (Fig. 1E). During DNA replication-dependent canonical histone deposition, histones H3 and H4 are incorporated as a tetramer, and histones H2A and H2B are incorporated as dimers (11–15). Therefore, we generated independent transgenic strains for H3 and H2B, respectively. On the other hand, histone variants are incorporated into chromatin in a transcription-coupled, but DNA replication-independent manner (16, 17). Therefore, the histone variant H3.3 was used as a control for canonical histones.

To avoid potential complications caused by heat shock-induced DNA recombination on either one or both chromosomes in GSCs, each of the three transgenes (H3, H2B and H3.3) was integrated as a single copy and analyzed in heterozygous flies. Examination of testes with transgenes revealed nuclear GFP but little mKO signal before heat shock. After heat shock, mKO signals were detectable (Fig. S1). Different GSCs undergo mitosis asynchronously, and an average cell cycle length of GSCs is approximately 12 to 16 hours. Among all GSCs, 75–77% are in G2 phase, 21% are in S phase, less than 2% are in mitosis, and G1 phase GSCs are almost negligible (18–22). Moreover, the GSC and GB arising from an asymmetric division remain connected after mitosis by a cellular structure known as the spectrosome, when they undergo the next G1 and S phases synchronously (19, 21).

To examine the distribution of old vs. new histones in GSC and GB after a round of DNA replication-dependent histone deposition, testes were studied 16 to 20 hours after heat shock (Fig. 2A). In particular, GSC-GB pairs connected by spectrosomes were examined (Fig. 2B, 2D, arrows). Based on the cell cycle length of GSCs, these GSC-GB pairs were from GSCs that switched from histone-GFP to histone-mKO genetic code during their G2 phase, underwent the first mitosis followed by G1, S, G2 phase and the second mitosis (Fig. 2A). With this time frame, both old histones and new histones were detectable in GSCs at the second G2 phase (Fig. 2E-E”, Table S1), because new histones had been synthesized and incorporated during the first S phase. For histone H3, the GFP signal was detected primarily in the GSC, but not in the GB (Fig. 2B'). By contrast, the mKO signals were present in both the GSC and the GB, with a relatively higher level in the GB (Fig. 2B, 2B”). The asymmetric distribution of histone H3 was specific for GSC divisions, because both the GFP and the mKO signals were equally distributed in spermatogonial cells derived from a symmetric division of the GB, in the same testis samples (Fig. 2C-C”). Quantification of fluorescence intensity revealed that the old H3 (GFP-labeled) signal was ~5.7-fold more enriched in the GSC compared to the GB, whereas new H3 (mKO-labeled) signal was ~1.6-fold more enriched in the GB compared to the GSC (H3 GSC/GB data in Fig.2H, Table S1–S2). By contrast, this differential distribution of old vs. new histone was not detected for symmetrically dividing spermatogonial cells (H3 SG1/SG2 data in Fig.2H, Table S1–S2: H3-GFP ratio in SG1/SG2= 1.09; H3-mKO ratio in SG1/SG2= 1.02).

Figure 2. H3 is asymmetrically segregated during the second GSC division after heat shock.

(A) Heat shock regime. H3 is distributed asymmetrically in GSC vs. GB (B-B”) but symmetrically in two-cell spermatogonia (C-C”). (D-D”) H3.3 is distributed symmetrically in GSC vs. GB. H3 distribution pattern in GSCs: (E-E”) G2 phase, (F-F”) anaphase, (G-G”) telophase. Scale: 5μm. Asterisk: HUB (anti-FasIII); arrow: spectrosome (anti-α spectrin). (H) Quantification of GFP and mKO fluorescence intensity ratio (Table S2). H3 GSC/GB GFP ratio > 1 (* P<10⁻⁴), GSC/GB mKO ratio < 1 (* P<10⁻⁴), N=15. H3 two-cell spermatogonial (SG) SG1/SG2 GFP ratio (# P=0.103) and mKO ratio (# P=0.684) insignificantly different from 1, N=16. H3.3 GSC/GB GFP ratio (# P=0.513) and mKO ratio (# P=0.532) insignificantly different from 1, N=12. Error bars: S.E. P-value: one-sample t-test.

In contrast to the asymmetric distribution pattern for the canonical histone H3, the histone variant H3.3 did not show this asymmetry during GSC divisions, by fluorescence images (Fig. 2D-D”) and by quantification (H3.3 GSC/GB data in Fig. 2H, Table S1–S2: H3.3-GFP ratio in GSC/GB= 1.03; H3.3-mKO ratio in GSC/GB= 1.03). The symmetry of the histone variant H3.3 suggests that the asymmetric mode is specific for canonical histone H3.

Because mitotic GSCs are less than 2%, all analysis above were based on post-mitotic GSC-GB pairs. To further examine the histone segregation pattern during mitosis, we screened for mitotic GSCs. Indeed, old histones were mainly associated with the chromatids segregated to the GSC-side at meta-to-anaphase (Fig. S2A-A”), anaphase (Fig.2F-F', Fig. S2B-B”, arrowheads), and telophase (Fig.2G-G', arrowheads). By contrast, new histones were more enriched at the chromatids segregated to GB-side (Fig.2F, 2F”, 2G, 2G”, Fig. S2B-B”, arrows). These results suggest that the sister chromatids preloaded with old histones are preferentially retained in GSCs and the ones enriched with new histones are partitioned to GBs during GSC mitosis.

Next, we examined the histone distribution pattern during the first GSC division by recovering GSCs for 4 to 6 hours following heat shock (Fig. 3A). An asymmetric distribution pattern was also found in the GSC-GB pairs with the H3 transgene (Fig.3B-B”). By contrast, symmetric distribution pattern was observed for both dividing spermatogonial cells with the H3 transgene (Fig. 3C-C”) and H3.3 during GSC division (Fig. 3D-D”). Quantification of fluorescence intensity revealed that the old H3-GFP signal was ~13-fold more enriched in the GSC compared to the GB, whereas the new H3-mKO signal was ~2.4-fold more enriched in the GB compared to the GSC (H3 GSC/GB data in Fig.3F, Table S3–S4). By contrast, there was no differential distribution of the old vs. new histone for the symmetrically dividing spermatogonial cells (H3 SG1/SG2 data in Fig.3F, Table S3–S4: H3-GFP ratio in SG1/SG2=1.07; H3-mKO ratio in SG1/SG2=1.06), or H3.3 during GSC division (H3.3 GSC/GB data in Fig. 3F, Table S3–S4: H3.3-GFP ratio in GSC/GB =1.00; H3.3-mKO ratio in GSC/GB =1.02). Although asymmetric histone distribution pattern was detected in post-mitotic GSC-GB pairs, examination of the mitotic GSC at this stage did not show any asymmetry (Fig.3E-E”'). These data suggest that the asymmetric segregation mode (Fig. 2F-F”, 2G-G”) relies on replication-dependent histone incorporation prior to mitosis. However, the greater than 10-fold difference of GFP signal between GSC and GB could be contributed by faster turnover of old histones in GBs, probably as a mechanism to reset the chromatin for differentiation. By contrast, the difference of mKO in GSC and GB was less dramatic, probably due to new histone synthesis in both cells. Furthermore, the H2B transgene showed similar pattern to H3 after the first GSC division (Fig. S3).

Figure 3. H3 is asymmetrically distributed after the first GSC division after heat shock.

(A) Heat shock regime. H3 is distributed asymmetrically in GSC vs. GB (B-B”) but symmetrically in two-cell spermatogonia (C-C”). (D-D”) H3.3 is distributed symmetrically in GSC vs. GB. (E-E”) A telophase GSC. Scale: 5μm. Asterisk: HUB (anti-FasIII); arrow: spectrosome (anti-α spectrin). (F) Quantification of GFP and mKO fluorescence intensity ratio (Table S4). H3 GSC/GB GFP ratio > 1 (* P< 10⁻⁴), GSC/GB mKO ratio < 1 (* P< 10⁻⁴), N=12. H3 two-cell spermatogonial (SG) SG1/SG2 GFP ratio (# P=0.225) and mKO ratio (# P=0.365) insignificantly different from 1, N=11. H3.3 GSC/GB GFP ratio (# P=0.970) and mKO ratio (# P=0.594) insignificantly different from 1, N=13. Error bars: S.E. P-value: one-sample t-test.

The consistent asymmetric cell divisions of GSCs could be lost under certain conditions, such as ectopic activation of the key JAK-STAT signaling pathway in the niche (23–25). It has been shown that overexpression of the JAK-STAT ligand unpaired (OE-upd) induces overpopulation of GSCs (23, 24). Consistent with the loss of asymmetry in expanded GSCs, the asymmetric distribution pattern of the histone H3 was not observed in OE-upd testes 16–20 hour after heat shock (Fig.4A-A”). These results demonstrate that the asymmetric histone distribution pattern is dependent on GSC asymmetric divisions. We propose a two-step process as our favorite explanation (Fig.S4A; an alternative explanation discussed in Fig. S4B): old and newly synthesized histones are incorporated to different sister chromatids during S phase, then during mitosis, the sister chromatid preloaded with old histones is preferentially segregated to GSC.

Figure 4. Loss of asymmetric H3 distribution pattern upon overexpression of upd.

(A-A”) H3-GFP (A') and H3-mKO (A”) in nanos-Gal4; UAS-upd testis. Asterisk: HUB (anti-FasIII).

In summary, these data reveal that stem cells preserve preexisting histones through asymmetric cell divisions. The JAK-STAT signaling pathway required for the asymmetric GSC divisions contributes to the asymmetric histone distribution pattern. This work provides a critical first step toward identifying the detailed molecular mechanisms underlying old histone retention during GSC asymmetric division. These findings in the well characterized GSC model system will help understanding how epigenetic information could be maintained by stem cells, or reset in their sibling cells which undergo cellular differentiation.