Reduced Levels of Lagging Strand Polymerases Shape Stem Cell Chromatin

Stem cells display asymmetric histone inheritance while non-stem progenitor cells exhibit symmetric patterns in the Drosophila male germline lineage. Here, we report that components involved in lagging strand synthesis, such as DNA polymerase α and δ (Polα and Polδ), have significantly reduced levels in stem cells compared to progenitor cells. Compromising Polα genetically induces the replication-coupled histone incorporation pattern in progenitor cells to be indistinguishable from that in stem cells, which can be recapitulated using a Polα inhibitor in a concentration-dependent manner. Furthermore, stem cell-derived chromatin fibers display a higher degree of old histone recycling by the leading strand compared to progenitor cell-derived chromatin fibers. However, upon reducing Polα levels in progenitor cells, the chromatin fibers now display asymmetric old histone recycling just like GSC-derived fibers. The old versus new histone asymmetry is comparable between stem cells and progenitor cells at both S-phase and M-phase. Together, these results indicate that developmentally programmed expression of key DNA replication components is important to shape stem cell chromatin. Furthermore, manipulating one crucial DNA replication component can induce replication-coupled histone dynamics in non-stem cells in a manner similar to that in stem cells.

The polα50 P-element insertion (BL-27205) was verified by sequencing to be a null allele using the following primers: 5'-AGCTCCAATCGTGTATCTCTCT-3' (specific to the 5' UTR of the polα50 gene locus) and 5'-CAATCATATCGCTGTCTCACTC-3' (specific to the Pelement sequences of the EP insertion) were used to amplify the genomic sequences corresponding to the 5' end of the polα50 gene locus, where the P-element insertion was located based on the Flybase (https://flybase.org/).Sequencing with this pair of primers confirmed that the P-element is inserted at a position nine base pairs downstream of the start codon, resulting in the new coding sequence 5'-ATGCCCGAAcatgatgaaataacataa (lowercase sequences indicate the P-element insertion).This leads to eight codons followed by a stop codon (underlined).Hence, this allele results in an early stop codon that very likely represents a null loss-of-function allele of the polα50 gene.This allele is not homozygous viable and is maintained as a heterozygous stock over a balancer chromosome.All experiments using the polα50 +/-background were outcrossing the polα50/Balancer stock to a wild-type stock to have the polα50 P-element insertion allele over a wild-type chromosome.

Heat shock scheme. Flies with UASp-FRT-H3-EGFP-FRT-H3-mCherry along with any relevant
genotypes were crossed with hs-flp; nanos-Gal4 and raised at 25°C.Within two days of eclosure, adult male flies were transferred to a vial and the vial was submerged underwater at 37°C for 90 minutes.Flies were then recovered at 29°C for 18 hours prior to dissection for experiments, with the exception of experiments using the PolA1 inhibitor, as described below.
Confocal images were taken on the Zeiss LSM800 (with Airyscan mode) with a 63x oil objective lenses or on the Leica SPE with 63x oil immersion lenses.
Quantification of protein levels in the early germline.Images were analyzed using the ImageJ software FIJI.Germline cyst stages were identified using Arm signal to label the two cyst cells encapsulating each cyst.Average intensity values were recorded for the center Z-slice of each cell/nucleus of interest.For germ cells within one cyst, only one germline nucleus from the entire cyst was measured as one data point.For the comparison of protein levels of endogenously tagged proteins, immunostaining signals in GSCs, 4-cell and 8-cell SGs were measured, and a background was subtracted using the post-mitotic hub cells, which are devoid of signals from any of these replication components.Signal intensity from 4-cell and 8-cell SGs were then normalized to the average intensity of GSCs from the same batch of testes.For the batch-based normalization, within one experimental batch, each data point is normalized to the average of WT GSCs in this corresponding batch.To compare data among different batches, the resulting values were then used to calculate the relative amount of GSC protein level to SG protein level (set to 1 to facilitate comparison), and plot on a log2 scale (Fig. 1b).The dataset shown in Figure 1b are from germ cells at each corresponding differentiation stages (Table S1).We also labeled S-phase germ cells using a EdU pulse and quantified them separately.The results using S-phase germ cells were similar to those using germ cells without distinguishing S-phase from G2-phase (data not shown).
For the comparison of the stage-specificity of each driver or driver combination, nanos-Gal4 by itself, nos-Gal4ΔVP16; bam-Gal80 combination, or bam-Gal4 by itself was crossed to the UASp-FRT-H3-EGFP-FRT-H3-mCherry transgene without hs-flp.The EGFP signals reflecting the relative strength of each driver or driver combination were quantified in the corresponding germline cyst stages, identified using Arm to label the two encapsulating cyst cells.The central slice of a representative nucleus was taken for each cyst measured as one data point.The cytoplasmic space was used as a background for subtraction.The EGFP signals were normalized to the stage with the highest relative signal intensity: For nanos-Gal4 by itself, all quantifications were normalized to the signals in GSCs; for the nos-Gal4ΔVP16; bam-Gal80 combination, all quantifications were also normalized to the signals in GSCs; for bam-Gal4 by itself, all quantifications were normalized to the signals in the 8-cell SGs (Fig. S1e).

S-Phase colocalization imaging and analysis.
To visualize potentially differential histone incorporation during S-phase, we applied a clearance buffer which effectively removes nucleoplasmic protein as previously described (7,8).Briefly, the clearance buffer is prepared by mixing 989µls of the clearance buffer stock solution (8.4 mM HEPES, 100 mM NaCl, 3 mM MgCl, 1 mM EGTA, 300 mM Sucrose, 2% Triton X-1000, and 2% BSA in ddH2O) with 1 µl DTT and 10 µl protease inhibitor (100x Leupeptin).After dissection, tissue samples were incubated in 10 μM EdU (Invitrogen Click-iT EdU Imaging Kit, catalog # C10340) for 15 minutes in Schneider's media at room temperature.At the end of the 15 minutes, the Schneider's media were drained and the clearance buffer was added for two minutes at 4°C in darkness.
Samples were then fixed in 4% PFA, washed with 1xPBST, and then blocked in 3% BSA for 30 minutes.For robust signals, both the old H3-EGFP and new H3-mCherry were immunostained with antibodies (e.g., anti-EGFP and anti-mCherry) using standard procedures.The CLICK reaction was performed according to manufacturer's instructions to label EdU.The DNA dye Hoechst was also added at this step.
Images were acquired on the Zeiss LSM800 using Airyscan mode on a 63x oil immersion objective.All samples were imaged using the identical settings.GSCs were identified by their proximity to the hub region.When 4-cell stage SGs were used, only one SG per cyst was analyzed to represent one data point.All images were analyzed using FIJI software.The Pearson score was recorded using the Coloc2 plugin for each nucleus, which was cropped to include just the nucleus as much as possible as delineated by the Hoechst signals.For each batch of images, the average measurement of the control GSCs was set to 1 and the other treatments are normalized to control GSCs, in order to avoid batch variability.The resulting values are then used to calculate mean and standard error of the mean (Mean± SEM).
Inhibitor treatment and analysis.For S-phase colocalization experiments using the inhibitor, flies were heat shocked as described above and left at 29°C to recover for 14 hours.Testes were then dissected and placed in incubation media for four hours, resulting in 18 total hours of post-heat shock recovery.After incubation with the inhibitor at the designated concentrations, these tissues were processed for S-phase colocalization analysis as described above.
Polα180 inhibitor (MedChemExpress Cat# HY-147812), a derivative of the classical inhibitor adarotene, was prepared in DMSO as stock and stored at -20°C (for short term) and -80°C (for long term) according to manufacturer's instructions.Drug incubation was performed on testes in "live cell media" containing Schneider's insect medium with 200 μg/ml insulin, 15% FBS by volume, and 0.6x pen/strep (9).Prior to experiments, incubation media was prepared by diluting inhibitor solution (or DMSO vehicle) to the proper concentration in live cell media.
Testes were dissected and placed in 100μl of incubation media as quickly as possible following dissection.Incubated testes were left in open tubes in darkness at room temperature (RT) for four hours.Because four hours are longer than the standard S-phase of the early male germline (10- 14), all S-phase cells at the end of the incubation should have been exposed to the inhibitor for the entirety of their current S-phase.
For EdU incorporation, 20μM EdU was added to the incubation media for the last 15 minutes of the drug incubation before tissue fixation.Only cells in early-to mid-S-phase were used for quantifications, as denoted by EdU staining covering all or most of the nucleus.Cells with focal EdU signal, indicative of late S-phase, were excluded to avoid skewing of the data.When a cyst was considered, only one nucleus from each cyst was measured as one data point.S2b were based on all early-stage germ cells, as no significant difference of EdU incorporation was detected among GSCs, GBs, and SGs from the same sample (data not shown).

Data shown in Figure
Generation of chromatin fibers from the Drosophila male germline.Chromatin fibers were prepared as previously described (4,15).Briefly, after adding EdU to the testis samples and incubating for 15 minutes, lysis buffer was added (100 mM NaCl, 25 mM Tris-base, 0.2% Joy detergent, pH=10).The testis tip was then micro-dissected on the slide and the rest of the testis was removed.Cells were allowed to fully lyse for approximately 5 minutes and then a Sucrose/Formalin (1M sucrose; 10% formaldehyde) solution was added and left for 2-minute to incubate, before a cover slip was gently placed on the top.The slide was then transferred to liquid nitrogen for two minutes before the cover slip was removed.The slide was then transferred to 95% EtOH for 10 min at -20°C in a freezer.Afterwards, the slide was fixed in 1% PFA for 1 minute.Samples were washed 3× in a Coplin jar with 1×PBST followed by blocking the sample with 3% BSA in 1×PBST for 30 minutes.Primary antibodies were then added for overnight incubation in a humidity chamber at 4°C.To assess histone asymmetry, anti-PCNA, anti-H3K27me3, and anti-GFP primary antibodies were added to chromatin fibers from the testes from the males with the following genotypes: each of the drivers (nos-Gal4 itself, nos-Gal4ΔVP16; bam-Gal80 combination, or bam-Gal4 itself) crossed with UASp-FRT-H3-EGFP-FRT-H3-mcherry without hs-flp.For cdc45-mCherry; DNA Polymerase-HA fibers, mCherry and HA primary antibodies were used.After the incubation with the primary antibodies, the slides are washed in a coplin jar with 1×PBS.Then the secondary antibodies were added and incubated for two hours at room temperature in a humidity chamber.The click chemistry was performed to label EdU following the manufacturer's instruction.When DNA needs to be labeled, Hoechst is included at 1:1,000 to stain the samples.Additionally, for samples that need DNA labeling, ProLong™ Gold Antifade Mountant with DNA Stain DAPI (Thermo Fisher catalog # P36931) was used.For samples that do not need DNA labeling, ProLong Diamond mounting media without DAPI (Thermo Fisher catalog# P36961) was used.

Sequential labeling using EdU and BrdU analogs on DNA fibers. After sample dissection, 10
μM EdU was added for a 10-minute incorporation, followed by washing out EdU. BrdU was subsequently added for another 10 minutes.After this sequential labeling, DNA fibers were prepared using the same procedure as described above for chromatin fibers, with the exception of using a different lysis buffer to strip proteins from the DNA (200 mM Tris-HCl, pH 7.5, 50 mM EDTA, 0.5% SDS).The fibers were then treated with 1M HCl for 30 minutes at room temperature to expose the incorporated BrdU.After washing with 1×PBST, BrdU antibodies were added for incubation overnight at 4°C in a humidity chamber.Secondaries antibodies against the BrdU primary antibodies were then added for two hours at room temperature in a humidity chamber.The click reaction to recognize EdU was performed subsequently along with Hoechst incubation at 1:1000.Samples were then mounted in ProLong Diamond mounting media with DAPI.The EdU-positive DNA fibers representing regions that undergo DNA replication during EdU pulse (and thus have EdU on at least one side) were used for subsequent analyses as shown in Figure 5b.Quantification of DNA fibers and chromatin fibers.All images were analyzed using FIJI software.To quantify the asymmetry between sister chromatids, line plots were drawn on both strands, using the PCNA-enriched side to denote the lagging strand.Most fibers have relatively short separable regions (≤ 2m), for which the entire fiber was used for quantification.For fibers with longer separable regions (> 2 μm), they were divided into 2μm-long non-overlapping segments along the length of the chromatin fiber and each of them was used for analyses.The region with no overlap with any of the chromatin fibers was used as background signal for subtraction from the measured signals from both strands.The ratio of signals = log2 (leading strand signal ̶ background signal) / (lagging strand signal ̶ background signal).
For the sequential EdU-and BrdU-labeled DNA fibers, there is no strandedness indicator such as PCNA.As such, the strand with higher BrdU signals was used as the reference strand, allowing EdU signal to be independently measured, which could be on the same or the opposite strand.All quantifications were performed similar to the chromatin fibers, with the ratio of signals = log2 (BrdU-enriched strand signal ̶ background signal) / (BrdU-depleted strand signal ̶ background signal).
For the Cdc45-and DNA Polymerase-labeled fibers, the distance between Cdc45 signal and the HA signal (labeling either Pol or Pol) was quantified from the center of the Cdc45 focus to the nearest HA signal.
A quantitative assay for chromosomal condensation state.We used an area-based method to monitor the chromosomal condensation state as previously described (8), using a dual-color histone transgene UASp-FRT-histone-EGFP -FRT-histone-mCherry.A maximum intensity projection was generated for old H3-(EGFP) and new H3-(mCherry) enriched areas.The intensity of each pixel was determined and scaled individually, setting the minimum intensity to 0 and the maximum to 65,535 (a 16-bit range).We monitored the pixels across the image with a threshold of 35% of the maximum intensity.Condensation kinetic profiles were generated to compare old H3-versus new H3-enriched regions by calculating the percentage of pixels above the threshold (the condensation parameter).Relative compaction index was measured and plotted by taking a ratio of the percentage of pixels of the new H3-enriched to the old H3-enriched regions as described previously (8).

Statistics and reproducibility.
For all comparisons between two groups, Mann-Whitney tests were used unless otherwise noted.For one-group datasets, one sample t-test was used with a null hypothesis that the data is symmetrically distributed (e.g., ratio= 1 for datasets without logarithmic transformation, log2= 0 for logarithmically transformed data).

Details for Figure 1b
Endogenously expressed Polα-HA levels are significantly depleted in GSCs relative to SGs according to the Mann-Whitney test with a P-value < 10 -4 (****).
Endogenously expressed Polδ-HA levels are significantly depleted in GSCs relative to SGs according to the Mann-Whitney test with a P-value < 10 -4 (****).
Endogenously expressed Polε-HA levels are not significantly different between GSCs and SGs according to the Mann-Whitney test with a P-value > 0.05 (= 0.0806, ns).
Endogenously expressed Cdc45-HA levels are not significantly different between GSCs and SGs according to the Mann-Whitney test with a P-value > 0.05 (= 0.8469, ns).
Endogenously expressed Ctf4-EGFP levels are not significantly different between GSCs and SGs according to the Mann-Whitney test with a P-value > 0.05 (= 0.8595, ns).

Details for Fig. S4b
For DNA fibers with both EdU and BrdU signals, EdU and BrdU exhibit significantly different distribution according to the Mann-Whitney test with a P-value < 10 -4 (****).
For DNA fibers with both EdU and BrdU signals, BrdU is significantly asymmetrically localized by a one sample t-test with a null hypothesis of log2=0 and a P-value < 10 -4 (****).
For DNA fibers with both EdU and BrdU signals, EdU is significantly asymmetrically localized by a one sample t-test with a null hypothesis of log2=0 and a P-value < 10 -4 (****).

Details for Figure 5g
The chromatin fibers co-labeled with Cdc45-mCherry and Polα-HA exhibit significantly longer distances between mCherry focus and HA signal relative to the chromatin fibers co-labeled with Cdc45-mCherry and Polε-HA by the Mann-Whitney test with a P-value < 10 -4 (****).
In nos-Gal4>H3-EGFP chromatin fibers, EdU was not significantly asymmetrically localized by a one sample t-test with a null hypothesis of log2= 0 and a P-value > 0.05 (= 0.1028, ns).

Figure S4: Visualization and quantification of delayed lagging strand synthesis. (a)
Assessment of observed patterns on DNA fibers, wherein symmetric fibers refer to fibers with EdU on both strands, weak asymmetric fibers have both EdU and BrdU but less than a 2-fold asymmetry on both strands, while strong asymmetric fibers have a greater than 2-fold difference for at least one of the signals (i.e., either EdU or BrdU).The distribution of the three categories are as follows: 60% symmetric, 23% weak asymmetric, 17% strong asymmetric.See Table S9 for details.(b) Log2-scale 1D quantification of EdU and BrdU from the DNA fibers with both signals, where the positive side is the strand with higher BrdU and the negative side is the strand with higher EdU: log2BrdU= 1.23± 0.23 (n=21), log2EdU= -0.93± 0.17 (n= 21).****: P< 10 -4 , Mann-Whitney test for the comparison between two groups, one tailed t-test with a null hypothesis of log2= 0 (symmetric pattern).See Table S10 for details.(c) Assessment of EdU asymmetries wherein ≥ 2-fold are considered strong asymmetry, < 2-fold are considered weak asymmetry.The distribution of the three categories are as follows: 11% strong asymmetry toward the leading strand, 33% weak asymmetry toward the leading strand, 15% weak asymmetry toward the lagging strand, 41% strong asymmetry toward the lagging strand.See Table S13 for details.(d) Quantification of H3K27me3, PCNA, and EdU asymmetry from nos-Gal4>H3-EGFP labeled chromatin fibers using log2 scale: log2H3K27me3= 0.95± 0.18 (n=27), log2PCNA= -1.52± 0.24 (n=27), log2EdU= -0.52± 0.31 (n=27).****: P< 10 -4 , one tailed t-test with a null hypothesis of log2= 0 (symmetric pattern).See Table S14 for details.(e) A model of replication patterns that could explain the observed EdU patterns.(f) Quantification of EdU distribution on chromatin fibers labeled with H3-EGFP driven by the following drivers with regard to the strandedness using log2 scale: nanos-Gal4= -0.52± 0.31 (n=27, P= 0.103, ns), nanos-Gal4; polα50 +/-= -1.06± 0.32 (n=26, P< 0.01), bam-Gal4= -0.03± 0.     5d and log2 values in Fig. S4f) Germ cells were determined by endogenously tagged Vasa-mApple signals.Following imaging, EdU incorporation was quantified by measuring the mean EdU signal intensity in EdU-positive germline nuclei and subtracting the background measured from the nearby EdU-negative cells.
Identifying and imaging replicative DNA fibers and chromatin fibers.All DNA fibers and chromatin fibers in this study were imaged with the Airyscan mode on a Zeiss LSM800 using a 63× oil immersion lens.Germline-derived chromatin fibers were identified using the H3-EGFP signal expressed with different germ cell-specific drivers or driver combination.Replicative regions were identified by both PCNA and EdU signals, or the presence of Cdc45, DNA Polymerase, and EdU.Fibers regions with detectable separation between sister chromatids were imaged and analyzed.Quality controls to select appropriate chromatin fiber regions for further analyses included fiber length, shape, and the molecular specificity of signals.For example, for quantifying old histone-enriched H3K27me3 with strandedness information, the EdU labeled fibers positive with PCNA, H3-EGFP and H3K27me3 signals were used.For analyzing the Cdc45 signals with DNA polymerases, fibers with EdU-labeling regions, clear Cdc45 and anti-HA signals were used.For sequential EdU and BrdU labeled DNA fibers, two patterns were imaged and quantified at DNA regions that replicate during the EdU pulse (thus incorporating EdU on at least one side of the duplicated sister chromatids): First, regions with clear sister chromatid separation with Hoechst and EdU signals but no discernable BrdU signal.Second, regions with clear sister chromatid separation with clear Hoechst, EdU, and BrdU signals.For detailed description of the analyses of sister chromatids using chromatin fibers, refer to (4, 15). :

Figure S1 :
Figure S1: Expression pattern of a replication machinery component CTF4 and distinct

Table S11 : Log2 Raw Data Related to Figure 5d and Figure S4f*:
Data are provided in absolute values in Fig.