The follicle epithelium in the Drosophila ovary is maintained by a small number of stem cells

The follicle epithelium of the Drosophila ovary has been a widely used and informative model for understanding epithelial tissue biology within the native, in vivo, environment (Sahai-Hernandez et al., 2012). First described over 60 years ago as a single layered epithelium that encapsulates developing germ cell cysts (Demerec, 1950; King et al., 1956), studies of this tissue have revealed insights into many aspects of epithelial biology, including diverse mechanisms that regulate the specification of cell fate in an epithelial stem cell lineage (Assa-Kunik et al., 2007; Chang et al., 2013; González-Reyes and St Johnston, 1998; Johnston et al., 2016; Pocha and Montell, 2014; Song and Xie, 2003), the establishment and maintenance of cell polarity (Bilder et al., 2000; Castanieto et al., 2014; Kronen et al., 2014; Mirouse et al., 2007; St Johnston and Ahringer, 2010), and the discovery of a novel mechanism for establishing planar polarity (Cetera et al., 2014; Chen et al., 2016).

A distinct advantage of the Drosophila ovary as an experimental model is that it has a highly consistent and well-described organization that facilitates the study of tissue biology with precise spatial and temporal resolution. Each Drosophila ovary is composed of long chains of developing follicles, called ovarioles (Miller, 1950), and oogenesis begins at the anterior tip of each ovariole in a structure called the germarium (Koch and King, 1966). The germarium has a stereotypical organization with four morphologically distinct regions, numbered from anterior to posterior as Regions 1, 2a, 2b, and 3 (Figure 1—figure supplement 1A). Germline stem cells (GSCs) reside at the anterior end of the germarium (Carpenter, 1975; Koch and King, 1966), in Region 1, and divide during adulthood to self-renew and produce daughter cells called cystoblasts. Cystoblasts undergo four rounds of mitosis with incomplete cytokinesis, as they move through Region 1 into Region 2a, which is defined by the presence of two 16 cell cysts that span the width of the germarium. Throughout Regions 1 and 2a, the germ cell cysts are covered by a population of somatic cells, referred to as inner germarial sheath (IGS) cells or escort cells. These cells provide a differentiation niche for the germ cells during these early stages of oogenesis (Kirilly et al., 2011), and may also help to propel the germ cells toward the posterior (Morris and Spradling, 2011). At the Region 2a/2b border, the cysts shed their IGS cell layer and move one at a time into Region 2b, where they become encapsulated by the follicle cell layer and take on a characteristic lens shape. Next, the cysts become more spherical in Region 3 (which is also referred to as Stage 1) and then bud off the germarium as a Stage 2 follicle. After budding, follicles rapidly grow and develop into a fully mature Stage 14 follicle that is ready for ovulation. This process, which takes approximately 8–9 days total under normal laboratory conditions (King, 1970), proceeds continuously during the first half of adult life, producing an organized tissue in which cells across the entire continuum of oogenesis are present simultaneously and arranged in order from the anterior to the posterior.

This thorough characterization of the ovarian structure and germ cell biology has facilitated the use of the ovary to study somatic cell biology as well. In a landmark study (Margolis and Spradling, 1995), the follicle epithelium was found to be maintained by a population of follicle stem cells (FSCs) in the germarium. This study was among the first to identify adult stem cells in vivo using site-specific DNA recombination to label clones of cells with a genetically heritable marker. In Drosophila studies, this is typically achieved by expressing flippase (Flp) to induce mitotic recombination between FRT sites on homologous chromosomes, whereas studies of mammalian tissues typically use cre-recombinase to induce intrachromosomal recombination at lox sites. This form of lineage tracing has become a gold standard for identifying adult stem cells (Fox et al., 2008; Simons and Clevers, 2011), and the methods used in the identification of FSCs helped to establish principles that have been applied to the study of many other Drosophila and mammalian adult stem cell lineages. Specifically, it is important to use a reliable clonal marking system in which the clonal marker is unambiguous, the method of clone induction does not significantly perturb tissue function or development, and the rate of background or ‘leaky’ clone induction is low. In addition, it is important to induce clones only sparsely, so that the majority of stem cells in the organ are not labeled with the clonal marker, to maximize the chance that each patch of labeled cells is one clone that originated from a single stem cell. With these quality control measures in place, the analysis of clone size and frequency at multiple time points can be used to infer the number of stem cells in the tissue as well as properties of the lineage such as the rate and number of transit amplifying divisions, the location of the stem cells, and the rate of stem cell turn over.

Using these principles, Margolis and Spradling induced sparse clones during adulthood and observed that single FSC clones contribute to an average of 50% of the follicle cell population on Stage 10 follicles, leading to the conclusion that each ovariole contains two actively dividing FSCs. In addition, they noticed that follicle cell clones remain coherent as a single patch of contiguous cells, even as the follicle grows and develops, and proposed that the FSCs reside at the Region 2a/2b border because persistent clones of contiguous cells at every time point always included one or more cells in this position. In contrast, a recent study concluded that there are 15–16 actively dividing FSCs per germarium that are organized into three rings, with one ring at the Region 2a/2b border, and two additional rings of cells in the adjacent positions immediately to the anterior, within Region 2a (Reilein et al., 2017). In addition, they proposed that, though all 16 FSCs are mitotically active, each FSC does not necessarily contribute to every follicle, so individual FSC clones can be discontinuous. The majority of the evidence in this study came from the use of a novel clonal marking system in which three clonal markers and two pairs of FRT sites are combined onto a single chromosome (Figure 1—figure supplement 1B). One homolog of Chromosome II has a constitutive GFP distal to FRT40A and a constitutive RFP distal to FRT42B, and the other homolog also has an FRT40A and an FRT42B with a constitutive Lac-Z that is distal to the FRT 40A. In this system, all cells initially express all three clonal markers, and heat shock-induced expression of Flp promotes recombination at one or both FRT sites, which has the potential to produce cells with 6 combinations of clonal markers (LacZ⁺, GFP^–, RFP⁺; LacZ⁺, GFP^–, RFP^–; LacZ^–, GFP⁺, RFP⁺; LacZ^–, GFP⁺, RFP^–; LacZ⁺, GFP⁺, RFP^–; LacZ⁺, GFP⁺, RFP⁺) (Figure 1—figure supplement 1B).

Here, we evaluate the conclusions of the Reilein et al. study by testing the clonal system that was reported in this study and repeating key experiments from the Margolis and Spradling study using quantitative microscopy and two standard single color marking systems: a negatively marked clonal system and a MARCM system. We find that the multicolor system used in Reilein, et al. is highly unreliable, producing an unpredictable pattern of clones across multiple time points and a high rate of ‘leaky’ clone induction in the absence of heat shock. In addition, we identify several logical flaws in the Reilein et al. study and performed new experiments to test the assumptions that underlie these flaws. Overall, our findings contradict the findings of Reilein et al. and instead support the original finding that the follicle epithelium is typically produced by two active FSCs residing at the FasIII border (Margolis and Spradling, 1995; Nystul and Spradling, 2010; Nystul and Spradling, 2007; Spradling et al., 1997). However, our quantitative approach revealed that the contribution from these two lineages is more variable than previously thought, leaving open the possibility that the number of active FSCs is not fixed but instead may fluctuate within a narrow range of one to four.

To test the LacZ, GFP, RFP (LGR) clonal marking system developed in Reilein, et al., we heat shocked the flies of the appropriate genotype for 1 hr at 37°C four times over 3 days, as described (Reilein et al., 2017), and assayed clone frequency at 5, 7, 9, 12, 14, 20, and 25 days post heat shock (dphs). Consistent with Reilein et al., we identified all six of the expected label combinations, and observed a progressive decrease in clonal diversity over the time course (Figure 1A–D and Figure 1—figure supplement 2). However, we observed less clonal diversity overall than reported in Reilein et al. For example, Reilein et al. reported that the majority of ovarioles had three uniquely labeled lineages at nine dphs, while the majority of ovarioles in our study had just two uniquely labeled lineages at this time point (Figure 1E). Notably, whereas the Reilein et al. study found a small number of ovarioles with five or six uniquely labeled lineages (8 and 4, respectively, out of 50 total at nine dphs), we never observed more than four uniquely marked clones within the same ovariole among the more than 300 ovarioles we imaged for this time course.

Figure 1 with 2 supplements see all

Download asset Open asset

Figure 1—source data 1

Frequency of clones over a time course, LRG system, 4x heat shocks.

https://cdn.elifesciences.org/articles/49050/elife-49050-fig1-data1-v2.csv

Download elife-49050-fig1-data1-v2.csv

Figure 1—source data 2

Comparison of this study with Reilein et al. (2017).

https://cdn.elifesciences.org/articles/49050/elife-49050-fig1-data2-v2.csv

Download elife-49050-fig1-data2-v2.csv

Figure 1—source data 3

Frequency of clones over a time course, LGR system, 1x heat shock.

https://cdn.elifesciences.org/articles/49050/elife-49050-fig1-data3-v2.csv

Download elife-49050-fig1-data3-v2.csv

Importantly, we also noticed that, in the large majority (88.7%, n = 62) of ovarioles with more than two distinctly labeled lineages, only 1 or 2 of the lineages spanned the entire ovariole, while the other lineage(s) covered only part of one follicle (Figure 1F) or consisted of a small patch of cells in the germarium. This suggests that just one or two FSC lineages predominated in these cases. According to the criteria used in the Reilein et al. study (Reilein et al., 2017), the small clones in these ovarioles would also be considered FSC clones. However, there are several problems with this interpretation. First, it is generally inappropriate to use multiple heat shocks with a clonal marking system that has more than one pair of FRT sites because the later heat shocks may produce subclones within clones that were induced by the earlier heat shock treatments. Thus, at least some of the cases in which small and large clones coexist in the same ovariole may have been due to the production of subclones induced by the later heat shocks. Consistent with this, clone induction with a single 1 hr heat shock substantially reduced the number of distinctly labeled clones per ovariole at 9 and 14 dphs (Figure 1G). However, this may also be due to an overall reduction in the frequency of clone induction, and we still observed small and large clones coexisting in the same ovariole with this heat shock regimen in some cases.

Second, these small clones may have arisen independent of heat shock. To test this possibility, we compared the rate of clone induction with the LGR clonal marking system to a commonly used GFP negative clonal marking system with FRT19A (Haelterman et al., 2014; Yamamoto et al., 2014) in the absence of heat shock at 7 days after eclosion (Figure 2A). As expected, we found that the rate of clone induction was very low in the GFP negative system (<2%, n = 188). However, with the LGR clonal marking system, we found that, on average, 25% of ovarioles had FSC clones (with or without transient clones), and an additional 22% had transient clones without an FSC clone (n = 112) (Figure 2A–B). This contrasts with a frequency of 2% clone induction in the absence of heat shock in the Reilein et al. study. It is unclear why we observed different distributions of clone frequencies and rates of clone induction in the absence of heat shock, though differences in experimental conditions may have been a contributing factor. Nonetheless, our inability to reproduce their results suggests that a low rate of background clone induction, which is important for any clonal marking system, is not a robust feature of the LGR system.

Figure 2

Download asset Open asset

Figure 2—source data 1

No heat shock control, LGR system vs GFPneg system.

https://cdn.elifesciences.org/articles/49050/elife-49050-fig2-data1-v2.csv

Download elife-49050-fig2-data1-v2.csv

Surprisingly, in flies with the LGR system, we identified clones that lacked expression of all three markers (Figure 2C–D) and others that expressed RFP but not LacZ or GFP (Figure 2E). This is unexpected because LacZ and GFP are both on the left arm of Chromosome II (2L), so all cells should express at least one of these two markers. We observed these types of clones in both heat-shocked and non-heat shocked cohorts. In all cases, other cells in the same ovariole clearly expressed all three markers, serving as an internal control for our ability to detect marker expression. In addition, we ruled out the possibility that the absence of LacZ and GFP signal was due to cell death because cells in the clone had normal nuclear morphology, and the clones often contained a dozen or more cells (Figure 2D), indicating that cells with these marker patterns were healthy enough to divide several times. We also ruled out the possibility that the absence of a LacZ and GFP signal was because the cells were in mitosis, when nuclear markers are more difficult to detect, because the DNA did not appear condensed. In addition, we sometimes observed this clonal pattern in post-mitotic follicle cells (Figure 2D) and multiple adjacent cells, which are unlikely to all be in mitosis at the same time, shared the same marker pattern. Taken together, these findings reveal significant flaws with the LGR clonal system that severely reduce its reliability and reproducibility. This may account for a large part of the discrepancies between the Reilein et al. study and prior studies, and strongly suggests that the LGR system is not an appropriate tool for determining the number of FSCs per ovariole.

An assumption that was integral to the calculations of FSC number in the Reilein et al. study was that the clone induction protocol induced recombination at both FRT sites in every mitotic follicle cell. Specifically, they explain that recombination at both FRT sites followed by segregation of sister chromatids at mitosis would produce nine possible genotypes with equal frequency, and that these nine genotypes would produce six detectable phenotypes (marker combinations) at a 1:1:1:2:2:2 frequency. They then stated “We therefore assumed in our statistical modeling that the different colors of FSC clone were present in those same proportions (B:G:BG:BR:GR:LGR = 1:1:1:2:2:2)’ (Reilein et al., 2017). However, we are not aware of any other example where recombination between FRT sites is reported to occur in every mitotic cell after heat shock. Cells must be in the S or G2 phase of the cell cycle for the recombination event to produce a detectable chromosomal rearrangement, and it is very unlikely that all cells will enter into this part of the cell cycle during the time that flippase is present after heat shock. Indeed, even in cases where the Flp is expressed constitutively, such as with eyelss-Flp, FRT recombination does not occur in all cells (Menut et al., 2007; Moberg et al., 2005). Using hs-Flp to generate negatively marked clones in the follicle epithelium, we typically find that only 25–50% of ovarioles have FSC clones (Castanieto et al., 2014; Johnston et al., 2016; Kim-Yip and Nystul, 2018; Kronen et al., 2014). To investigate whether FRT recombination occurs in all cells in the LGR system, we quantified the frequency of each marker combination at multiple time points after either four 1 hr heat shocks or one 1 hr heat shock. We found that the most common marker combination at every time point was the unrecombined phenotype, LacZ⁺, GFP⁺, RFP⁺, followed by LacZ⁺, GFP^–, RFP⁺, while the remaining four marker combinations were present at much lower frequencies (Figure 3A). Unexpectedly, we found that clones marked by the absence of LacZ were less frequent than clones marked by the absence of GFP, indicating that cells lacking these two markers are not equally fit. These findings challenge the underlying assumptions of the analysis in the Reilein et al. study, and suggest that, even if the LGR system were reliable, the interpretation of the data leading to their conclusions about the number of FSCs per germarium are also flawed.

Figure 3

Download asset Open asset

Figure 3—source data 1

Frequency of each type of clone over a time course, LGR system, 4x heat shocks.

https://cdn.elifesciences.org/articles/49050/elife-49050-fig3-data1-v2.csv

Download elife-49050-fig3-data1-v2.csv

Figure 3—source data 2

Frequency of each type of clone over a time course, LGR system, 1x heat shock.

https://cdn.elifesciences.org/articles/49050/elife-49050-fig3-data2-v2.csv

Download elife-49050-fig3-data2-v2.csv

Figure 3—source data 3

Frequency of each type of clone pattern, GFPneg system.

https://cdn.elifesciences.org/articles/49050/elife-49050-fig3-data3-v2.csv

Download elife-49050-fig3-data3-v2.csv

Therefore, to reexamine the original claims of the Margolis and Spradling study, we switched to the use of the FRT19A GFP negative clonal marking system that we found had a very low rate of background clone induction in the absence of heat shock (Figure 2A). Importantly, this system uses only one pair of FRT sites so there is no risk of generating subclones with multiple heat shocks. In addition, the generation of GFP negative clones with FRT19A is one of the most commonly used methods to label clones in studies of the FSC lineage (Cook et al., 2017; Dai et al., 2017; Nystul and Spradling, 2010; O'Reilly et al., 2008). With this system, all cells initially express GFP, and FRT recombination produces clones that are clearly marked by the lack of GFP expression. One of the original observations in the Margolis and Spradling study is that an FSC clone forms a coherent, contiguous patch of cells that extends from the Region 2a/2b border out to the posterior edge of the clone boundary, contributing to each new follicle that is formed after clone induction. In contrast, Reilein et al. posited that each FSC does not normally contribute to every follicle, so discontinuous clone patterns with labeled cells at the Region 2a/2b border followed by intermittent, non-contiguous patches of labeled cells downstream should be expected. However, with the GFP negative clonal marking system, we found that the clone patterns in 94% of ovarioles with clones did not fit this expectation. Specifically, 61.5% of the ovarioles with clones had large FSC clones that extended from the Region 2a/2b border through the germarium and over one or more follicles (Figure 3B–C), 20.6% had transient clones that covered only a half of a follicle or less (Figure 3B,D), and 3.6% had large clones that spanned multiple follicles but did not extend back to the Region 2a/2b border (Figure 3B,E), which is the predicted pattern for a recent FSC replacement event (Song and Xie, 2003). The remaining 14.3% of ovarioles had one of two unexpected clone patterns: 8.3% had small clones within the germarium that typically covered only a small portion of one cyst (Figure 3B,F), and 6.0% had a discontinuous pattern in which a clone near the Region 2a/2b border coexisted with a larger, non-contiguous clone of cells further downstream in the ovariole (Figure 3B,G). These two clone categories occurred frequently enough that they seem unlikely to be due solely to background recombination events, though this remains a formal possibility. Alternatively, they may be evidence that FSC lineage dynamics are not as predictable as previously thought, and that rare events cause some cells to become quiescent or delay differentiation for a period of time. Therefore, although up to 6% of the ovarioles in this population may have had discontinuous clones, our results confirm that the large majority of FSC clones typically form coherent, contiguous units and contribute to every follicle within the boundaries of the clone.

Next, we sought to estimate stem cell number by measuring clone sizes at multiple time points. In the Margolis and Spradling study, this was achieved by measuring the size of clones on Stage 10 follicles, which provides a snapshot of the events that contributed to a single follicle in each ovariole. In that case, the authors reported that an average of 50% of the follicle cell population was labeled by each FSC clone. To assay FSC behavior during the production of multiple successive follicles, we used quantitative image analysis to count the number of GFP positive and GFP negative cells in mosaic ovarioles at two time points after a standard heat shock regimen of four 1 hr heat shock treatments over 2 days (Figure 4A–B). This heat shock regimen produced sparse clonal labeling, with only 33.9% and 35.0% ovarioles containing FSC clones at 7 and 14 dphs, respectively (Figure 4C). In mosaic ovarioles with GFP negative clones that extended from the Region 2a/2b border into at least one budded follicle, we observed an average clone size of 50.1 ± 14.1% at 7dphs and 51.3 ± 18.0% at 14dphs. We took two additional approaches to further test this result. First, we quantified FSC clone sizes at 7 days after a single 30 min heat shock. In these flies, the frequency of FSC clones was much lower (11.9%), yet the size of the FSC clones in mosaic ovarioles were not significantly different (48.2 ± 14.8%) from the average sizes of clones in mosaic ovarioles at 7 or 14 days after four 1 hr heat shock treatments (Figure 4C–D). Second, we repeated this test with another clonal marking system, MARCM (Lee and Luo, 2001), that uses a different mechanism to label clones. With this system, cells are initially GFP negative and mitotic recombination at FRT sites generates GFP positive clones. Again, we found that the average size of MARCM clones in mosaic ovarioles was not significantly different (44.7 ± 10.4%) from the average sizes of clones we observed with the GFP negative system (Figure 4C–D, and Figure 4—figure supplement 1). These measurements are remarkably similar to those reported in the Margolis and Spradling study, though we observed more variability in FSC clone size than was reported previously. Thus, our data provide strong support for the conclusion that the follicle epithelium is typically maintained by just two active FSCs. However, considering values that fall within two standard deviations of the means, and the fact that we excluded non-mosaic ovarioles from our analysis, our data are consistent with the possibility that FSC number fluctuates within a range of approximately one to four FSCs per ovariole.

Figure 4 with 1 supplement see all

Download asset Open asset

Figure 4—source data 1

Frequency of clones in each experimental condition, GFPneg system and MARCM system.

https://cdn.elifesciences.org/articles/49050/elife-49050-fig4-data1-v2.csv

Download elife-49050-fig4-data1-v2.csv

Figure 4—source data 2

Clone sizes in each experimental condition, GFPneg system and MARCM system.

https://cdn.elifesciences.org/articles/49050/elife-49050-fig4-data2-v2.csv

Download elife-49050-fig4-data2-v2.csv

Lastly, to determine where the FSCs are located, we stained ovarioles with GFP negative clones at 7dphs for GFP and FasIII. The Reilein et al. study reported that the anterior-most cells labeled with the LGR clonal marker system were located within one of three adjacent positions leading up to the FasIII border, and the authors interpreted this as evidence that FSCs are located within three rings, two FasIII negative rings in Region 2a, and one ring at the boundary of FasIII expression. With the GFP negative clonal marking system, we found that the anterior-most GFP negative cell was FasIII positive and located at the boundary of FasIII expression in the large majority of ovarioles with FSC clones (85.7%, n = 63). In these germaria, all GFP negative somatic cells formed a single, contiguous clone and there were no GFP negative somatic cells anterior to the boundary of FasIII expression (Figure 5A and D), indicating that the FSC must be FasIII positive. This observation is consistent with other studies that have demonstrated that the anterior border of an FSC clone coincides with the boundary of FasIII expression (Chang et al., 2013; Kirilly et al., 2005; Nystul and Spradling, 2010; Nystul and Spradling, 2007; Spradling et al., 1997). In 7.9% of the ovarioles, the anterior-most GFP negative cell was contiguous with the rest of the clone but was located on the anterior side of the FasIII border. Thus, FasIII staining was detectable on the posterior surface of the cell (which is likely to be signal from the adjacent cell) but not the anterior surface (Figure 5B and D). IGS cells in Region 2a are also mitotic during adulthood, and labeling systems that use mitotic recombination occasionally produce clones in Region 2a (Kirilly et al., 2011). Therefore, the anterior-most GFP negative cell in these germaria could be either an FSC with little or no FasIII expression or an IGS cell that became labeled through an independent recombination event from the one that produced the FSC clone. In the remaining 6.3% of ovarioles, the anterior-most GFP negative cell was FasIII negative and was not adjacent to either the FasIII border or the large clone of contiguous GFP negative follicle cells (Figure 5C–D). Notably, we found that cells in this position could be GFP negative even in germaria that did not have an FSC clone (Figure 5E), strongly suggesting that these cells are IGS cells, not FSCs. These observations confirm the conclusion that FSCs are usually, if not always, FasIII positive, and reside at the FasIII expression boundary.

Figure 5

Download asset Open asset

An understanding about the number and position of FSCs in each germarium is an important foundation for the diverse range of studies that use the follicle epithelium as a model. Indeed, many studies have relied on the conclusions of the Margolis and Spradling study, and both the early descriptions of the ovariole and decades of subsequent work have reinforced these conclusions. For example, the prediction that there are just two active FSCs per germarium aligns well with the published rates of oogenesis and follicle cell number. Specifically, a careful measurement of cell number revealed that there are approximately 900 follicle cells per egg chamber at Stage 6, when the follicle cells exit mitosis (Kolahi et al., 2009), and measurements of the rate of oogenesis (King, 1970; Lin and Spradling, 1993; Margolis and Spradling, 1995) suggest that the transit time from the Region 2a/2b border to Stage 6 is approximately 100 hr. Two progenitor cells dividing continuously with a 9.6 hour cell cycle (King and Vanoucek, 1960; Margolis and Spradling, 1995) would take approximately 85 hr to produce 900 cells (log₂(900/2) x 9.6 = 85). Since the part of the lineage that differentiates into polar and stalk cells ceases division well before Stage 6, the actual time to produce 900 main body follicle cells would be somewhat longer. In contrast, 16 progenitor cells would take only 56 hr (log₂(900/16) x 9.6 = 56) at this average rate of division. In addition, multiple studies have found that transient clones are capable of covering up to half of a follicle (Nystul and Spradling, 2010; Skora and Spradling, 2010; Song and Xie, 2003), consistent with the idea that the follicle cell population of each follicle is typically founded by two progenitors. Moreover, images of very large persistent clones from many different studies using a wide variety of single and multicolor labeling strategies and the analysis of the clonal data are consistent with a small number of actively dividing FSCs per germarium (Dai et al., 2017; Kronen et al., 2014; Nystul and Spradling, 2007; O'Reilly et al., 2008; Song and Xie, 2002). With respect to the position of the FSCs, the supposition that some cells in Region 2a are FSCs is inconsistent with many other studies that have demonstrated that somatic cells in this region have a markedly different shape, function, and gene expression pattern than cells in Region 2b (Decotto and Spradling, 2005; Huang et al., 2014; Kirilly et al., 2011; Rust et al., 2019; Song and Xie, 2002; Spradling et al., 1997; Wang et al., 2015; Wang and Page-McCaw, 2018; Upadhyay et al., 2018). Thus, the claim that there are 15–16 FSCs per germarium at both the Region 2a/2b border and within Region 2a is incompatible with a wide range of studies that provide independent orthogonal support for the conclusions of the Margolis and Spradling study.

Our study identifies several reasons for the discrepancies between the current paradigm and the Reilein et al study. First, their estimates of FSC number relied primarily on the use of the LGR clonal marking system, which we found to be unreliable because it had a high rate of background clone induction in our hands and produced clones with marker combinations such as LacZ^–, GFP^–, RFP⁺ and LacZ^–, GFP^–, RFP^– that are not predicted by the genotype. This raises the possibility that at least some of the clones generated by this system arise through a non-canonical event, such as recombination between an FRT40A and an FRT42D. This type of recombination would produce genome rearrangements that are not likely to be compatible with clone growth and, indeed, some clones of this type contained less than five cells. However, others, such as the clone shown in Figure 2D, were larger and the cells looked healthy. Thus, an alternative hypothesis is that these clones are not the product of an FRT recombination event but instead are due to epigenetic silencing of one or more marker transgenes. Consistent with this interpretation, high levels of fluorophore expression can reduce cellular fitness in some cases, so it may be that constitutive expression of three separate markers creates a selective pressure to silence one or more markers. Moreover, epigenetic silencing of transgenes has been described in follicle cells previously (Skora and Spradling, 2010). Although this study focused on transgenes with a UAS promoter rather than a constitutive promoter like those in the LGR system, it provides clear evidence that follicle cells actively silence transgene promoters and that the changes are heritable, producing a clone of cells with similar levels of transgene expression. Regardless of the cause, a high rate of clone induction in the absence of heat shock and the presence of clones with unexpected marker combinations are problematic for the use of clone size and frequency measurements to infer stem cell number. The Reilein et al. study also used a MARCM system as an independent test of their conclusions for this part of the study, but the number of clones analyzed was significantly lower and images of the MARCM clones were not provided so it is difficult to fully assess these data.

Second, we found that the assumptions and analysis of FSC clone number was flawed. The authors arrived at an estimate of 15–16 FSCs per germarium by quantifying the number of distinct lineages at 5, 7, 14, 21, and 30 dphs and then extrapolating back to 0 dphs using neutral drift modeling, but do not explain why the average number of uniquely labeled lineages would decline from more than four at five dphs to two at 15 dphs, and then remain at approximately two for the subsequent 15 days. Neutral competition models predict that mosaic tissues will drift toward monoclonality, not biclonality. Related to this, we found no evidence for the claim that individual FSC clones can be non-contiguous and contribute only sporadically to follicles that form after clone induction. On the contrary, we found that the large majority of FSC clones labeled with the GFP negative marking system in our study closely resembled the large contiguous clones described originally (Margolis and Spradling, 1995) and reported in many subsequent studies. Indeed, if FSC clones were typically discontinuous, it would be very difficult to distinguish between persistent FSC clones and transient clones because, even at late time points, it would appear as if ‘transient’ clones had not moved out of the tissue. Thus, much of the published work on persistent FSC clones would not have been possible (Buszczak et al., 2009; Kirilly et al., 2005; O'Reilly et al., 2008; Song and Xie, 2003; Song and Xie, 2002; Su et al., 2018). Therefore, rather than interpret the decrease in clonal diversity at early time points in the Reilein et al. study as evidence for neutral drift among a large pool of FSCs, we favor the alternative hypothesis that the clonal diversity at early time points is due to the presence of transient clones that had not yet moved out of the tissue. Because the authors allowed for the possibility that FSC clones are not contiguous, these transient clones may have been misinterpreted as evidence for persistently labeled FSC lineages.

A parallel approach used in the Reilein et al. study to determine the number of FSCs per germarium was to induce LGR clones sparsely and measure the fraction of a follicle labeled by a rare FSC clone. This is a valid approach if one assumes that all stem cells in the tissue have an equal chance of becoming labeled. However, in Reilein et al., the authors argue that there are significantly different rates of proliferation among the 15–16 FSCs in each germarium. Since mitosis is required to label clones with the LGR system, more rapidly proliferating FSCs would be more likely to be labeled, though it would be difficult to predict how much the differences in proliferation rate would affect the chances of becoming labeled. The rapidly dividing FSCs would also presumably contribute a disproportionately high number of cells to the tissue, so the overrepresentation of clones from these FSCs in the dataset would skew the estimate of stem cell number by an unknown amount. Therefore, within the framework of the Reilein et al study, this would not be an appropriate approach to determine stem cell number.

Third, we challenged the assumption in the Reilein et al. study that FRT recombination at both sites occurs in 100% of mitotic follicle cells. This assumption does not accord with either our experience with the Flp/FRT system or the frequencies of marker combinations that we observed with the LGR system. Moreover, if both FRT sites always undergo recombination in every FSC, the same should occur in every mitotic follicle cell. Each ovariole has thousands of mitotic follicle cells, so at early time points such as 5 dphs, when cells produced in the germarium would not have moved out of the ovariole, every ovariole would have all six clonal marker phenotypes. Yet this clearly was not the case in either our study or in the Reilein et al. study. By assuming that the recombined phenotypes (marker combinations) should be more common than they actually are, the neutral drift analysis in Reilein et al. may have overestimated the amount of clonal diversity that was present prior to the first point of analysis (5 dphs), thus contributing to their overestimate of the number of FSCs.

Fourth, we used a standard, reliable clonal marking system and quantitative analysis to measure FSC clone size at two time points after clone induction. Our measurements were consistent with the original finding that each FSC typically produces approximately half of the follicle cells in each ovariole, though we found a wider range of clone sizes than expected. This finding, combined with the unexpected clone patterns that we observed (Figure 3C and G–H) suggest that FSC lineages are not rigid but instead exhibit some flexible, stochastic behaviors that allow for a variable number of FSCs or cause individual FSC lineages to contribute to the tissue unequally. An interesting possibility to consider in future studies is that unequal contributions between two stem cell lineages in the same germarium may precede an FSC loss and replacement event. Nonetheless, despite this variation in clone size and pattern, our findings reinforce the idea that two FSCs produce the large majority of the follicle epithelium in most cases.

Lastly, our analysis established that FSCs reside at the FasIII expression boundary. Somatic cells in Region 2a have a shape, function and gene expression pattern that is distinct from the follicle epithelium, so it would be surprising if the population of FSCs spanned this border, with some FSCs that have IGS cell characteristics and others that have follicle cell characteristics. Live imaging in the Reilein et al. study demonstrated that labeled cells near the Region 2a/2b border can move toward the anterior, but these cells could not be simultaneously labeled with cell type specific markers, so their identity is unclear. Further analysis will be required to determine whether there are any conditions in which cells convert from the IGS cell identity into a follicle cell identity or vice versa, but our data suggest that, at least under normal laboratory conditions, persistent FSC clones are typically contained entirely within the FasIII expression domain. Taken together, our study clarifies and reaffirms the current paradigm for the fundamental features of the Drosophila FSC lineage, and provides important context for a wide variety of studies that use the follicle epithelium as a model system.

All data generated or analysed during this study are included in the manuscript and supporting files. Source data files have been provided for Figures 1-5.

1. 1. Assa-Kunik E
   2. Torres IL
   3. Schejter ED
   4. Johnston DS
   5. Shilo BZ

   (2007) Drosophila follicle cells are patterned by multiple levels of notch signaling and antagonism between the notch and JAK/STAT pathways

   Development 134:1161–1169.

   https://doi.org/10.1242/dev.02800

   • PubMed
   • Google Scholar
2. 1. Bilder D
   2. Li M
   3. Perrimon N

   (2000) Cooperative regulation of cell polarity and growth by Drosophila tumor suppressors

   Science 289:113–116.

   https://doi.org/10.1126/science.289.5476.113

   • PubMed
   • Google Scholar
3. 1. Buszczak M
   2. Paterno S
   3. Spradling AC

   (2009) Drosophila stem cells share a common requirement for the histone H2B ubiquitin protease scrawny

   Science 323:248–251.

   https://doi.org/10.1126/science.1165678

   • PubMed
   • Google Scholar
4. 1. Carpenter ATC

   (1975) Electron microscopy of meiosis in Drosophila Melanogaster females

   Chromosoma 51:157–182.

   https://doi.org/10.1007/BF00319833

   • Google Scholar
5. 1. Castanieto A
   2. Johnston MJ
   3. Nystul TG

   (2014) EGFR signaling promotes self-renewal through the establishment of cell polarity in Drosophila follicle stem cells

   eLife 3:e04437.

   https://doi.org/10.7554/eLife.04437

   • Google Scholar
6. 1. Cetera M
   2. Ramirez-San Juan GR
   3. Oakes PW
   4. Lewellyn L
   5. Fairchild MJ
   6. Tanentzapf G
   7. Gardel ML
   8. Horne-Badovinac S

   (2014) Epithelial rotation promotes the global alignment of contractile actin bundles during Drosophila egg chamber elongation

   Nature Communications 5:5511.

   https://doi.org/10.1038/ncomms6511

   • PubMed
   • Google Scholar
7. 1. Chang YC
   2. Jang AC
   3. Lin CH
   4. Montell DJ

   (2013) Castor is required for Hedgehog-dependent cell-fate specification and follicle stem cell maintenance in Drosophila oogenesis

   PNAS 110:E1734–E1742.

   https://doi.org/10.1073/pnas.1300725110

   • PubMed
   • Google Scholar
8. 1. Chen DY
   2. Lipari KR
   3. Dehghan Y
   4. Streichan SJ
   5. Bilder D

   (2016) Symmetry breaking in an edgeless epithelium by Fat2-Regulated microtubule polarity

   Cell Reports 15:1125–1133.

   https://doi.org/10.1016/j.celrep.2016.04.014

   • PubMed
   • Google Scholar
9. 1. Cook MS
   2. Cazin C
   3. Amoyel M
   4. Yamamoto S
   5. Bach E
   6. Nystul T

   (2017) Neutral competition for Drosophila follicle and cyst stem cell niches requires vesicle trafficking genes

   Genetics 206:1417–1428.

   https://doi.org/10.1534/genetics.117.201202

   • PubMed
   • Google Scholar
10. 1. Dai W
    2. Peterson A
    3. Kenney T
    4. Burrous H
    5. Montell DJ

    (2017) Quantitative microscopy of the Drosophila ovary shows multiple niche signals specify progenitor cell fate

    Nature Communications 8:1244.

    https://doi.org/10.1038/s41467-017-01322-9

    • PubMed
    • Google Scholar
11. 1. Decotto E
    2. Spradling AC

    (2005) The Drosophila ovarian and testis stem cell niches: similar somatic stem cells and signals

    Developmental Cell 9:501–510.

    https://doi.org/10.1016/j.devcel.2005.08.012

    • PubMed
    • Google Scholar
12. Book

    1. Demerec M

    (1950)

    Biology of Drosophila

    John Wiley & Sons, Inc.

    • Google Scholar
13. Book

    1. Fox DT
    2. Morris LX
    3. Nystul T
    4. Spradling AC

    (2008)

    StemBook

    Cambridge: Harvard Stem Cell Institute.

    • Google Scholar
14. 1. González-Reyes A
    2. St Johnston D

    (1998)

    Patterning of the follicle cell epithelium along the anterior-posterior Axis during Drosophila oogenesis

    Development 125:2837–2846.

    • PubMed
    • Google Scholar
15. 1. Haelterman NA
    2. Jiang L
    3. Li Y
    4. Bayat V
    5. Sandoval H
    6. Ugur B
    7. Tan KL
    8. Zhang K
    9. Bei D
    10. Xiong B
    11. Charng WL
    12. Busby T
    13. Jawaid A
    14. David G
    15. Jaiswal M
    16. Venken KJ
    17. Yamamoto S
    18. Chen R
    19. Bellen HJ

    (2014) Large-scale identification of chemically induced mutations in Drosophila Melanogaster

    Genome Research 24:1707–1718.

    https://doi.org/10.1101/gr.174615.114

    • PubMed
    • Google Scholar
16. 1. Huang P
    2. Sahai-Hernandez P
    3. Bohm RA
    4. Welch WP
    5. Zhang B
    6. Nystul T

    (2014) Enhancer-Trap flippase lines for clonal analysis in the Drosophila ovary

    G3: Genes, Genomes, Genetics 4:1693–1699.

    https://doi.org/10.1534/g3.114.010710

    • Google Scholar
17. 1. Johnston MJ
    2. Bar-Cohen S
    3. Paroush Z
    4. Nystul TG

    (2016) Phosphorylated groucho delays differentiation in the follicle stem cell lineage by providing a molecular memory of EGFR signaling in the niche

    Development 143:4631–4642.

    https://doi.org/10.1242/dev.143263

    • PubMed
    • Google Scholar
18. 1. Kambadur R
    2. Koizumi K
    3. Stivers C
    4. Nagle J
    5. Poole SJ
    6. Odenwald WF

    (1998) Regulation of POU genes by Castor and hunchback establishes layered compartments in the Drosophila CNS

    Genes & Development 12:246–260.

    https://doi.org/10.1101/gad.12.2.246

    • Google Scholar
19. 1. Kim-Yip RP
    2. Nystul TG

    (2018) Wingless promotes EGFR signaling in follicle stem cells to maintain self-renewal

    Development 145:dev168716.

    https://doi.org/10.1242/dev.168716

    • PubMed
    • Google Scholar
20. 1. King RC
    2. Rubinson AC
    3. Smith RF

    (1956)

    Oogenesis in adult Drosophila Melanogaster

    Growth 20:121–157.

    • PubMed
    • Google Scholar
21. Book

    1. King RC

    (1970)

    Ovarian Development in Drosophila Melanogaster

    New York: Academic Press.

    • Google Scholar
22. 1. King RC
    2. Vanoucek EG

    (1960)

    Oogenesis in adult Drosophila Melanogaster. X. studies on the behavior of the follicle cells

    Growth 24:333–338.

    • PubMed
    • Google Scholar
23. 1. Kirilly D
    2. Spana EP
    3. Perrimon N
    4. Padgett RW
    5. Xie T

    (2005) BMP signaling is required for controlling somatic stem cell self-renewal in the Drosophila ovary

    Developmental Cell 9:651–662.

    https://doi.org/10.1016/j.devcel.2005.09.013

    • PubMed
    • Google Scholar
24. 1. Kirilly D
    2. Wang S
    3. Xie T

    (2011) Self-maintained escort cells form a germline stem cell differentiation niche

    Development 138:5087–5097.

    https://doi.org/10.1242/dev.067850

    • PubMed
    • Google Scholar
25. 1. Koch EA
    2. King RC

    (1966) The origin and early differentiation of the egg chamber of Drosophila Melanogaster

    Journal of Morphology 119:283–303.

    https://doi.org/10.1002/jmor.1051190303

    • PubMed
    • Google Scholar
26. 1. Kolahi KS
    2. White PF
    3. Shreter DM
    4. Classen AK
    5. Bilder D
    6. Mofrad MR

    (2009) Quantitative analysis of epithelial morphogenesis in Drosophila oogenesis: new insights based on morphometric analysis and mechanical modeling

    Developmental Biology 331:129–139.

    https://doi.org/10.1016/j.ydbio.2009.04.028

    • PubMed
    • Google Scholar
27. 1. Kronen MR
    2. Schoenfelder KP
    3. Klein AM
    4. Nystul TG

    (2014) Basolateral junction proteins regulate competition for the follicle stem cell niche in the Drosophila ovary

    PLOS ONE 9:e101085.

    https://doi.org/10.1371/journal.pone.0101085

    • PubMed
    • Google Scholar
28. 1. Lee T
    2. Luo L

    (2001) Mosaic analysis with a repressible cell marker (MARCM) for Drosophila neural development

    Trends in Neurosciences 24:251–254.

    https://doi.org/10.1016/S0166-2236(00)01791-4

    • Google Scholar
29. 1. Lin H
    2. Spradling AC

    (1993) Germline stem cell division and egg chamber development in transplanted Drosophila Germaria

    Developmental Biology 159:140–152.

    https://doi.org/10.1006/dbio.1993.1228

    • PubMed
    • Google Scholar
30. 1. Margolis J
    2. Spradling A

    (1995)

    Identification and behavior of epithelial stem cells in the Drosophila ovary

    Development 121:3797–3807.

    • PubMed
    • Google Scholar
31. 1. Menut L
    2. Vaccari T
    3. Dionne H
    4. Hill J
    5. Wu G
    6. Bilder D

    (2007) A mosaic genetic screen for Drosophila neoplastic tumor suppressor genes based on defective pupation

    Genetics 177:1667–1677.

    https://doi.org/10.1534/genetics.107.078360

    • PubMed
    • Google Scholar
32. 1. Miller A

    (1950)

    The Biology of Drosophila

    421–534, The internal anatomy and histology of the imago of Drosophila Melanogaster, The Biology of Drosophila, New York, Wiley.

    • Google Scholar
33. 1. Mirouse V
    2. Swick LL
    3. Kazgan N
    4. St Johnston D
    5. Brenman JE

    (2007) LKB1 and AMPK maintain epithelial cell polarity under energetic stress

    The Journal of Cell Biology 177:387–392.

    https://doi.org/10.1083/jcb.200702053

    • PubMed
    • Google Scholar
34. 1. Moberg KH
    2. Schelble S
    3. Burdick SK
    4. Hariharan IK

    (2005) Mutations in erupted, the Drosophila ortholog of mammalian tumor susceptibility gene 101, elicit non-cell-autonomous overgrowth

    Developmental Cell 9:699–710.

    https://doi.org/10.1016/j.devcel.2005.09.018

    • PubMed
    • Google Scholar
35. 1. Morris LX
    2. Spradling AC

    (2011) Long-term live imaging provides new insight into stem cell regulation and germline-soma coordination in the Drosophila ovary

    Development 138:2207–2215.

    https://doi.org/10.1242/dev.065508

    • PubMed
    • Google Scholar
36. 1. Nystul T
    2. Spradling A

    (2007) An epithelial niche in the Drosophila ovary undergoes long-range stem cell replacement

    Cell Stem Cell 1:277–285.

    https://doi.org/10.1016/j.stem.2007.07.009

    • PubMed
    • Google Scholar
37. 1. Nystul T
    2. Spradling A

    (2010) Regulation of epithelial stem cell replacement and follicle formation in the Drosophila ovary

    Genetics 184:503–515.

    https://doi.org/10.1534/genetics.109.109538

    • PubMed
    • Google Scholar
38. 1. O'Reilly AM
    2. Lee HH
    3. Simon MA

    (2008) Integrins control the positioning and proliferation of follicle stem cells in the Drosophila ovary

    The Journal of Cell Biology 182:801–815.

    https://doi.org/10.1083/jcb.200710141

    • PubMed
    • Google Scholar
39. 1. Pocha SM
    2. Montell DJ

    (2014) Cellular and molecular mechanisms of single and collective cell migrations in Drosophila: themes and variations

    Annual Review of Genetics 48:295–318.

    https://doi.org/10.1146/annurev-genet-120213-092218

    • PubMed
    • Google Scholar
40. 1. Reilein A
    2. Melamed D
    3. Park KS
    4. Berg A
    5. Cimetta E
    6. Tandon N
    7. Vunjak-Novakovic G
    8. Finkelstein S
    9. Kalderon D

    (2017) Alternative direct stem cell derivatives defined by stem cell location and graded wnt signalling

    Nature Cell Biology 19:433–444.

    https://doi.org/10.1038/ncb3505

    • PubMed
    • Google Scholar
41. Preprint

    1. Rust K
    2. Byrnes L
    3. Ks Y
    4. Park JS
    5. Sneddon JB
    6. Tward AD
    7. Nystul TG

    (2019) A Single-Cell atlas and lineage analysis of the adult Drosophila ovary

    bioRxiv.

    https://doi.org/10.1101/798223

    • Google Scholar
42. 1. Sahai-Hernandez P
    2. Castanieto A
    3. Nystul TG

    (2012) Drosophila models of epithelial stem cells and their niches

    Wiley Interdisciplinary Reviews: Developmental Biology 1:447–457.

    https://doi.org/10.1002/wdev.36

    • PubMed
    • Google Scholar
43. 1. Simons BD
    2. Clevers H

    (2011) Strategies for homeostatic stem cell self-renewal in adult tissues

    Cell 145:851–862.

    https://doi.org/10.1016/j.cell.2011.05.033

    • PubMed
    • Google Scholar
44. 1. Skora AD
    2. Spradling AC

    (2010) Epigenetic stability increases extensively during Drosophila follicle stem cell differentiation

    PNAS 107:7389–7394.

    https://doi.org/10.1073/pnas.1003180107

    • PubMed
    • Google Scholar
45. 1. Song X
    2. Xie T

    (2002) DE-cadherin-mediated cell adhesion is essential for maintaining somatic stem cells in the Drosophila ovary

    PNAS 99:14813–14818.

    https://doi.org/10.1073/pnas.232389399

    • PubMed
    • Google Scholar
46. 1. Song X
    2. Xie T

    (2003) Wingless signaling regulates the maintenance of ovarian somatic stem cells in Drosophila

    Development 130:3259–3268.

    https://doi.org/10.1242/dev.00524

    • PubMed
    • Google Scholar
47. 1. Spradling AC
    2. de Cuevas M
    3. Drummond-Barbosa D
    4. Keyes L
    5. Lilly M
    6. Pepling M
    7. Xie T

    (1997)

    The Drosophila germarium: stem cells, germ line cysts, and oocytes

    Cold Spring Harbor Symposia on Quantitative Biology 62:25–34.

    • PubMed
    • Google Scholar
48. 1. St Johnston D
    2. Ahringer J

    (2010) Cell polarity in eggs and epithelia: parallels and diversity

    Cell 141:757–774.

    https://doi.org/10.1016/j.cell.2010.05.011

    • PubMed
    • Google Scholar
49. 1. Su TY
    2. Nakato E
    3. Choi PY
    4. Nakato H

    (2018) Drosophila glypicans regulate follicle stem cell maintenance and niche competition

    Genetics 209:537–549.

    https://doi.org/10.1534/genetics.118.300839

    • PubMed
    • Google Scholar
50. 1. Upadhyay M
    2. Kuna M
    3. Tudor S
    4. Martino Cortez Y
    5. Rangan P

    (2018) A switch in the mode of wnt signaling orchestrates the formation of germline stem cell differentiation niche in Drosophila

    PLOS Genetics 14:e1007154.

    https://doi.org/10.1371/journal.pgen.1007154

    • PubMed
    • Google Scholar
51. 1. Wang S
    2. Gao Y
    3. Song X
    4. Ma X
    5. Zhu X
    6. Mao Y
    7. Yang Z
    8. Ni J
    9. Li H
    10. Malanowski KE
    11. Anoja P
    12. Park J
    13. Haug J
    14. Xie T

    (2015) Wnt signaling-mediated redox regulation maintains the germ line stem cell differentiation niche

    eLife 4:e08174.

    https://doi.org/10.7554/eLife.08174

    • PubMed
    • Google Scholar
52. 1. Wang X
    2. Page-McCaw A

    (2018) Wnt6 maintains anterior escort cells as an integral component of the germline stem cell niche

    Development 145:dev158527.

    https://doi.org/10.1242/dev.158527

    • PubMed
    • Google Scholar
53. 1. Yamamoto S
    2. Jaiswal M
    3. Charng WL
    4. Gambin T
    5. Karaca E
    6. Mirzaa G
    7. Wiszniewski W
    8. Sandoval H
    9. Haelterman NA
    10. Xiong B
    11. Zhang K
    12. Bayat V
    13. David G
    14. Li T
    15. Chen K
    16. Gala U
    17. Harel T
    18. Pehlivan D
    19. Penney S
    20. Vissers L
    21. de Ligt J
    22. Jhangiani SN
    23. Xie Y
    24. Tsang SH
    25. Parman Y
    26. Sivaci M
    27. Battaloglu E
    28. Muzny D
    29. Wan YW
    30. Liu Z
    31. Lin-Moore AT
    32. Clark RD
    33. Curry CJ
    34. Link N
    35. Schulze KL
    36. Boerwinkle E
    37. Dobyns WB
    38. Allikmets R
    39. Gibbs RA
    40. Chen R
    41. Lupski JR
    42. Wangler MF
    43. Bellen HJ

    (2014) A Drosophila genetic resource of mutants to study mechanisms underlying human genetic diseases

    Cell 159:200–214.

    https://doi.org/10.1016/j.cell.2014.09.002

    • PubMed
    • Google Scholar

Author details

1. Jocelyne Fadiga

   1. Department of Anatomy, University of California, San Francisco, San Francisco, United States
   2. Department of OB/GYN-RS, Center for Reproductive Sciences, University of California, San Francisco, San Francisco, United States

   Contribution

   Data curation, Formal analysis, Investigation, Visualization

   Competing interests

   No competing interests declared

2. Todd G Nystul

   1. Department of Anatomy, University of California, San Francisco, San Francisco, United States
   2. Department of OB/GYN-RS, Center for Reproductive Sciences, University of California, San Francisco, San Francisco, United States

   Contribution

   Conceptualization, Data curation, Software, Formal analysis, Supervision, Funding acquisition, Investigation, Visualization, Methodology, Project administration

   For correspondence

   todd.nystul@ucsf.edu

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0002-6250-2394

• 3,330

  views

• 336

  downloads

• 35

  citations

Views, downloads and citations are aggregated across all versions of this paper published by eLife.