This study defined the timeline of acute DXR accumulation, DNA damage, and subsequent cell death in the mouse ovary in both a cell type- and follicle type-dependent manner. These temporal data provide a necessary framework for the future design of chemoprotective agents specific for the ovary. Previous studies have elegantly identified protective agents that prevent radiation, doxorubicin, or cisplatin-induced apoptosis in the ovary 
. Apoptosis, however, is a late sequel to the initial ovarian insult. Understanding acute chemotherapy insult allows potential development of intervention approaches to prevent
initial ovarian injury and maintain oocyte fidelity. We previously identified dexrazoxane as a putative ovarian protective agent for DXR insult in vitro
. Determining the capacity of dexrazoxane or other agents to protect the ovary in vivo
requires generating a model of acute chemotherapy insult to provide metrics for assessing protection. We have identified the first 2 hours post-DXR treatment as critical to prevent induced DNA damage in ovarian stroma cell-enriched populations, which maintain the overall structural integrity of the mouse ovary, and are the first targets of DXR insult. Granulosa cells may have up to a 4-hour protection window, as DXR-induced DNA damage in granulosa cell-enriched population is not measurable until 4 hours post-injection. These quantifiable metrics, along with the detailed timeline for drug accumulation and the apoptotic response, provide a framework in which we can assess the efficacy of putative ovarian chemoprotective agents. Understanding and preventing the insult, rather than subsequent cellular demise (apoptosis), may lead to the development of ovarian protective therapeutics that will successfully translate to the clinic.
DXR time-dependent localization appeared coupled to the time line for induced DNA damage. DXR was detected via microscopy in the mouse ovary by 2 hours post-injection based on its autofluorescence, and appeared most prevalently in the core stroma cells at early (2- and 4-hour) time points (,,). Consistent with that observation, DNA damage in the stroma-enriched cells was also detected at the 2 hour time point using the comet assay (). DNA damage did not reach detectable levels in the granulosa-enriched cells until 4 hours post-injection, however (), and while the drug was visible in follicles by 2 hours, there was a time-dependent exponential increase in drug within the follicles from 2 to 4 hours (). These data suggest that DXR must penetrate follicles and reach sufficient nuclear concentrations before DNA damage occurs in granulosa cells. This model indicates a slower timeline for granulosa cell damage compared to stroma cells and is consistent with the intimate interaction between stroma cells and the blood supply.
On a daily basis, the ovary must successfully process and detoxify numerous environmental toxins to ensure viable offspring. Our study suggests one basic function of the core ovarian stromal tissue may be to shield follicles from such insults by actively sequestering toxins like DXR. The largest vessels in the ovary are located centrally in the core of the organ and radiate out, but the capillary network infiltrates the entire organ, central medulla to cortex, to ensure rapid gas and nutrient exchange throughout the ovary. Despite this typical organ capillary network, the central core ovarian stromal tissue concentrates DXR at early time points (2–4 hours post-DXR injection). The observed gradient is different from the homogenous DXR distribution observed in the canine prostate following i.p. injection, suggesting ovarian-specific DXR uptake 
. This indicates that the stromal cells may sequester DXR, which is one way to shield follicles in the face of toxic insult. Though most chemoprotection work to date has focused on follicles and oocytes, previous studies have indeed shown ovarian stroma cell disarray or reorganization following chemotherapy 
, consistent with severe insult to this tissue. The stroma cells may therefore be a key protection site in strategies to prevent chemotherapy insult in the ovary. We did not see wide-spread apoptosis in stoma cells despite dramatic degeneration of the core stromal tissue, suggesting the tissue is destroyed via other mechanisms like necrosis. These data are consistent with a study demonstrating that while ovarian ovulation returns at one month post-DXR treatment in mice, the ovary size does not recover and remains at only 50% normal weight and volume 
. The decreased ovarian size persists despite measurements showing counts for each follicle class, except secondary, were not different from control at 1 month post treatment 
. Taken with our data, the measurements by Ben-Aharon et al suggest a dramatic loss of stroma tissue that is not recovered post-treatment. The stroma cells are critical to ovarian structure and follicular recruitment as they regulate the collagen structure surrounding follicles; alterations in the collagen structure in turn affects primordial follicle recruitment 
. It is therefore possible that one mechanism contributing to long-term chemotherapy-induced ovarian insufficiency is activated reprogramming of primordial follicle recruitment as a consequence of stroma remodeling.
In contrast to stroma cells, granulosa cells appeared to package DXR into discrete puncta, or foci, which then aggregated over time. While the present study did not identify these subcellular compartments, DXR and other chemotherapy agents are found concentrated in perinuclear puncta identified as acidic organelles 
including lysosomes 
, endosomes 
, and the Golgi 
, depending on the cell type. It is unclear whether these structures play a role in the apparent sensitivity of granulosa cells to DXR-induced DNA damage or are in fact a defense mechanism. One frequently observed feature of multidrug resistant cancer cells is compartmentalization of DXR and other chemotherapy agents into acidic intracellular compartments 
; this compartmentalization is directly linked to limiting nuclear accumulation of DXR and therefore decreasing chemotherapy toxicity. Exceptions exist, however, such as the uterine drug-sensitive MES-SA cell line which accumulates DXR in lysosomes where the drug-resistant MES-SA/Dx5 cell line does not 
. The sensitivity to DXR-induced DNA damage exhibited by granulosa cells may reflect the observed subcellular compartmentalization. Alternatively, the granulosa cells in growing follicles may be highly sensitive because they are more rapidly dividing than stroma cells. Rapid cell division and DNA replication are factors that increase DXR-induced DNA damage. Future studies may test the hypothesis DXR accumulates in lysosomes or other similar compartments in granulosa cells, and further determine whether the drug compartmentalization alters granulosa cell sensitivity to chemotherapy.
Oocytes exhibited a much slower rise in DNA damage than either the stroma or granulosa cells, reaching a 50% increase in DNA damage only at 10–12 hours post-DXR injection. This suggests either the surrounding tissue functions to actively protect the oocyte or reflects the relative transcriptional quiescence of the oocytes compared to granulosa cells. A study by Bar-Joseph et al detected DXR in oocytes from secondary follicles at 3 hours post-DXR injection 
, which would suggest oocyte quiescence or resistance may account for the delayed damage response. By 10–12 hours post-injection, there is significant DXR-induced apoptosis in the granulosa cells of growing follicles (, 
) past the threshold previously defined to signify an apoptotic follicle 
. Taken together, these data suggest the oocytes may not exhibit a DNA damage response to DXR until the follicle commences apoptosis and hence oocyte attrition is secondary to follicular failure.
The time-dependent spectral changes we observed in this study suggest DXR fluorescence is environmentally sensitive in the ovary and may reflect changes in the dielectric constant and/or pH surrounding DXR 
. Ongoing studies will be required to define the specific subcellular distribution of DXR in the ovary and how that changes with time. DXR’s fluorescence spectrum has a parabolic dependence upon the dielectric constant of its solvent such the ratio of the 555 nm/590 nm peak intensities from DXR in solution is high in water (~0.8) 
. This peak ratio drops in solvents with decreasing dielectric, manifest as a “blue shift,” or a curve shift to shorter wavelengths, with decreasing dielectric constant. In heptane, however, which has the lowest dielectric tested, the peak ratio increases to over 1, manifest as a “red shift” in the peak emissions 
. DXR’s peak ratio is also dependent upon pH, such that the ratio increases (blue shift) with increasing pH. Our data show that from 2 to 4 hours post-DXR, the spectral profile exhibits a dramatic blue shift that begins to correct back toward longer wavelengths from 6 hours post-injection through a dramatic red shift at 48 hours. These data suggest that between 2 and 4 hours post-injection, DXR may experience a change in environment and become compartmentalized in cellular structures with a decreased dielectric constant or increased pH. The continual ‘red shift’, or increase in peak emissions ratio from 4 hours post-injection through 48 hours could be due in part to the observed DXR accumulation in puncta within the granulosa cells, which may correspond to acidic organelles like lysosomes, with a pH of roughly 5. Future studies may define time-dependent subcellular localization of DXR to organelles with an acidic pH, including endosomes, lysosomes, and the Golgi. The dramatic ‘red shift’ at 48 hours may also be due to the ‘red’ nuclear DXR signal which becomes consistent at this time point. Intercalation into DNA can quench DXR fluorescence up to 80%, and indeed while we see significant DNA damage in both stroma and granulosa cells at early time points (2 hours post-injection and on), we do not consistently observe ‘red’ nuclei until at least 12 hours post-DXR injection. These data suggest a model in which DXR rapidly intercalates DNA, quenching its fluorescence, until all DNA binding sites are occupied. DXR may then accumulate in the nucleus as free drug, accounting for the observed red nuclear fluorescence at later time points.
It was surprising to find that both the late antral follicles and primordial follicles accumulated significantly more DXR per area than the remaining follicle classes. It is likely there are two different mechanisms accounting for what appears to be a similar phenomenon in the two disparate follicle classes. Primordial follicles consist of a single layer of flattened granulosa cells and lack a theca ring. It is possible this single layer (small size) makes it relatively easy for DXR to penetrate throughout the follicle. Late antral follicles are, in contrast, very large. They are also comparatively metabolically dynamic, which may account for the relatively high levels of DXR uptake. Our data suggest DXR uptake was partly dependent upon follicle size, and therefore follicular cell numbers. Additional differences in drug accumulation between follicle classes remained, however, which may reflect unique metabolic activities of the different follicle classes.
Our data demonstrate that within the acute phase of DXR insult, granulosa cells of the secondary and antral follicles are the predominant sites of induced apoptosis. These data are consistent with the study by Ben-Aharon et al showing TUNEL positive granulosa cells predominantly in the antral follicles at 12 and 24 hours post-DXR injection in the mouse ovary 
. The authors also stated there was weak TUNEL signal in primary follicles 
, which may be consistent with our observation that DXR induced apoptosis in 10% of primary follicles over the time course assayed. Our data include earlier time points, however, which indicate the DXR-induced follicular apoptosis approached plateau in the acute state as early as 8 hours post-injection. This reflects the number of follicles with sufficient apoptotic granulosa cells to predict follicular demise; the number of TUNEL-positive granulosa cells in each follicle, however, continues to rise dramatically through the 48 hour time point, consistent with follicular failure. Our data highlight the necessity of understanding the initial insult and targeting intervention therapies as early as possible to prevent follicular demise. Primordial follicles did not exhibit TUNEL signal nor H2AFX phosphorylation until 48 hours post-DXR injection, despite the relatively high levels of DXR accumulation. This may reflect the relatively quiescent state of the primordial follicles as DXR-induced DNA damage requires DNA replication. Future studies will determine whether the PFs will exhibit delayed DXR-induced apoptosis as they are recruited to grow and the granulosa cells must replicate DNA. Our data propose a model in which primordial follicles sustain DXR injury by 48 hours post-injection, but may not demise until they are triggered to grow.
Delayed apoptosis in the primordial follicles also suggests a “burn out” model may be one mechanism by which DXR depletes the primordial reserve. The burn out model has been proposed for other chemotherapy agents that appear to eliminate the growing follicle population, resulting in increased primordial follicle recruitment to replace the larger follicles 
. This indirect mechanism depletes the ovarian reserve as primordial follicles are recruited to replace the ‘burnt out’ growing follicles 
. Taken together, the data in this study demonstrate DXR insult in the ovary is a complex process involving differential effects that are cell- and follicle-type dependent. The data highlight the need to understand acute insult mechanisms and suggest ovarian depletion may be the summation of multiple events including granulosa cell packaging of DXR, stroma demise and remodeling, and direct toxicity in primordial follicles as they are recruited to grow. Future studies may also determine how primordial follicle recruitment after burn out of growing follicles by DXR contributes to depleting the ovarian reserve.