Varied effects of doxorubicin (DOX) on the corpus luteum of C57BL/6 mice during early pregnancy^†

Abstract

Certain chemotherapeutic drugs are toxic to ovarian follicles. The corpus luteum (CL) is normally developed from an ovulated follicle for producing progesterone (P4) to support early pregnancy. To fill in the knowledge gap about effects of chemotherapy on the CL, we tested the hypothesis that chemotherapy may target endothelial cells and/or luteal cells in the CL to impair CL function in P4 steroidogenesis using doxorubicin (DOX) as a representative chemotherapeutic drug in mice. In both mixed background mice and C57BL/6 mice, a single intraperitoneal injection of DOX (10 mg/kg) on 0.5-day postcoitum (D0.5, postovulation) led to ~58% D3.5 mice with serum P4 levels lower than the serum P4 range in the phosphate buffer saline-treated control mice. Further studies in the C57BL/6 ovaries revealed that CLs from DOX-treated mice with low P4 levels had less defined luteal cords and disrupted collagen IV expression pattern, indicating disrupted capillary, accompanied with less differentiated luteal cells that had smaller cytoplasm and reduced StAR expression. DOX-treated ovaries had increased granulosa cell death in the growing follicles, reduced proliferating cell nuclear antigen-positive endothelial cells in the CLs, enlarged lipid droplets, and disrupted F-actin in the luteal cells. These novel data suggest that the proliferating endothelial cells in the developing CL may be the primary target of DOX to impair the vascular support for luteal cell differentiation and subsequently P4 steroidogenesis. This study fills in the knowledge gap about the toxic effects of chemotherapy on the CL and provides critical information for risk assessment of chemotherapy in premenopausal patients.

Our finding that doxorubicin may target proliferating endothelial cells in the developing corpus luteum to impair the vascular support for luteal cell differentiation and subsequent progesterone synthesis fills in the knowledge gap about toxic effects of chemotherapy on the corpus luteum.

Introduction

Millions of prepubescent girls and reproductive aged women suffer from cancers. A unique side effect from anticancer treatments, including chemotherapy and radiotherapy, in premenopausal cancer patients is fertility impairment [1, 2]. Oncofertility thus becomes an emerging discipline with the expressed goal to protect the future reproductive health of cancer patients [3–5]. Female fertility depends on a functional female reproductive system, in which the ovaries, the Fallopian tubes (oviducts in mice), and the uterus are the physical sites for supporting pregnancy events, from oocyte production in the ovary, to fertilization in the Fallopian tubes as well as early embryo development and transport in both Fallopian tubes/oviducts and uterus, to embryo implantation and postimplantation embryo/fetus development in the uterus. Because of the prominent gonadotoxicities of many oncologic treatments, the research in oncofertility has been mainly focused on the ovarian follicles [6–11]. Besides the follicles that produce oocytes to initiate a pregnancy, the ovary also has the corpus luteum (CL, pl: corpora lutea) for producing progesterone (P4) to support early pregnancy events, from preimplantation embryo development and transport to embryo implantation [12–17]. In addition to its critical function in supporting pregnancy, the CL is also essential for the luteal phase of the ovarian cycle, which includes follicular phase, ovulation, and luteal phase. A functional ovarian cycle is important for the wellbeing (e.g., menstrual cycle) of females, especially premenopausal females. Any toxic effects of chemotherapy on the CL remain largely unknown.

The CL is a temporary endocrine gland normally developed from an ovulated follicle. It has three main stages during its life span: development, maintenance, and regression. The two main cell types in the CL are luteal cells and endothelial cells [18]. The vasculature in the CL supports luteal cell functions, particularly P4 steroidogenesis [19, 20]. The precursor cholesterol for P4 steroidogenesis in luteal cells is mainly imported from the circulation via endocytosis or selective uptake and it has a minor source via de novo cholesterol synthesis [20]. Cholesterol can be stored in the cytoplasmic lipid droplets as cholesteryl esters, which will undergo hydrolysis via the intracellular neutral hormone-sensitive lipase (also named cholesteryl ester hydrolase) and/or lysosomal acid lipase to free cholesterol for P4 steroidogenesis [21–23]. Transport of cholesterol from the outer to the inner mitochondrial membrane for P4 steroidogenesis is the rate-limiting step carried out by steroidogenic acute regulatory protein (StAR) [20, 24]. P450 side chain cleavage (P450SCC/CYP11A1) converts cholesterol to pregnenolone on the inner mitochondrial membrane, and 3β-hydroxysteroid dehydrogenase (3β-HSD) then converts pregnenolone to P4 in the smooth endoplasmic reticulum (SER) [20]. P4 steroidogenesis is accompanied by dramatic changes in morphology and numbers of mitochondria and SER complexes [20]. Serum P4 levels increase with CL development and reach a plateau by 3.5-day postcoitum (D3.5, embryo implantation initiates ~D4.0 in mice) in mice to support early pregnancy events, such as preimplantation embryo transport from the oviduct to the uterus and establishment of uterine receptivity for embryo implantation [15]. If pregnancy does not occur, such as during pseudopregnancy, the CL will undergo luteal regression, which is marked by structural and functional degradation of the CL, such as lipid droplet accumulation in luteal cells [25, 26], luteal cell death and a sharp drop of serum P4 level [27].

Doxorubicin (DOX, Adriamycin, a cytotoxic anthracycline antibiotic) is a commonly used chemotherapeutic agent in premenopausal cancer patients [28, 29]. Its chemotherapeutic effects on cancer cells may involve different mechanisms, such as intercalation into DNA and disruption of topoisomerase-II-mediated DNA repair, as well as generation of free radicals to damage cellular membranes, DNA, and proteins [30]. DOX chemotherapy increases the risk for cardiovascular diseases, which could be partially contributed by oxidative stress-induced endothelial dysfunction in the conduit arteries [31]. DOX has toxicities on ovarian follicles in rodents [6, 10, 11, 32], such as follicular atresia and overactivation [10, 32], and impaired secretion of 17beta-estradiol (E2) [6], and toxicities of DOX are exacerbated by deficiency of multidrug resistance protein 1 (MDR1) [11], which is also called P-glycoprotein/ABCB1/CD243 that belongs to the ATP-binding cassette family of transporters (ABC transporters). The CL is a highly vascularized transient organ in the ovary that could represent a prime target for DOX-induced toxic action. In addition, DOX can target steroidogenic tissues, such as testis [33, 34]. The main objective of this study was to determine any toxic effects of DOX on the CL. We hypothesized that DOX may target endothelial cells and/or luteal cells to impair CL functions, especially in P4 production. Since CLs are normally derived from the ovulated follicles and DOX can induce follicular toxicity, we tested this hypothesis in newly mated and ovulated mice to avoid secondary effects in the CL from DOX-induced follicular toxicity.

Materials and methods

Animals

In the initial study, mixed background (C57BL/6 and 129/SvJ [35]) wild-type mice (10–22 weeks old at dissection) that were available in our mouse colony were used. They were mated with stud males and checked for the presence of a vaginal plug (an indication of mating) the following morning. The day of plug detection was designated as 0.5-day postcoitum (D0.5). Thirteen D0.5 females were randomly assigned in phosphate buffer saline (PBS) group (N = 6) and DOX-treated group (N = 7). In the second study, pure background C57BL/6 female mice (6 weeks old) were purchased from the Jackson Laboratory (Ellsworth, Maine). They were acclimated to the Coverdell animal facility at the University of Georgia for 2 weeks with body weight monitored twice a week to ensure well acclimation to the new environment. At 8 weeks old, the acclimated females were mated with stud males and checked for the presence of a vaginal plug the following morning. A total of 11 females were plugged within 11 days of cohabitation and they were randomly assigned into phosphate buffer saline (PBS) group (N = 6) and DOX-treated group (N = 5). The Coverdell animal facility is on a 12-h light/dark cycle (6:00 a.m. to 6:00 p.m.) at 23 ± 1°C with 30–50% relative humidity. All mice had free access to regular chow 5053 (Labdiet, St. Louis, MO) and water. All methods used in this study were approved by the University of Georgia IACUC Committee (Institutional Animal Care and Use Committee) and conform to National Institutes of Health Guidelines and Public Law.

DOX treatment and tissue collection

Dose selection

In humans, the recommended dosing of DOX follows a 21–28 day 40–60 mg/m² cycle and maximal 4 cycles based on patient information (age, body weight, disease status, etc.) and cancer type [36]. We used a single intraperitoneal (i.p.) injection of 10 mg/kg body weight of DOX in mice to determine its effects in ovarian follicles [6, 10, 11] and the uterus [37]. Based on FDA animal–human dose conversions, 10 mg/kg in adult mice is equivalent to ~30 mg/m² in adult humans [38], which is within the clinically used chemotherapy dose range.

Treatment timing

A CL is normally developed from an ovulated follicle and ovulation in cyclic mice occurs in the same night of mating [39], which can be detected by the presence of a vaginal plug the next morning. Since DOX is toxic to the ovarian follicles that could affect subsequent ovulation and CL formation and development, DOX treatment was applied on postovulation D0.5, therefore avoiding any secondary effects on the CL from the follicular toxicity of DOX.

DOX treatment

DOX (D-4000, LC Laboratories, Woburn, MA) was dissolved in dimethyl sulfoxide (DMSO) to make a stock at 100 mg/ml, aliquoted, and kept at −20°C. A thawed aliquot was kept at 4°C and used within 3 days of thawing. Upon detection of a vaginal plug, D0.5 females were randomly assigned into two groups to receive a single i.p. injection of vehicle control (100 μl of <2% DMSO in sterile 1× PBS, the % of DMSO was the same as that in DOX-treated group) or DOX (10 mg/kg, mouse body weight (kg) × 10 mg/kg (final dose) /100 mg/ml (stock concentration) = ml of stock solution diluted into sterile 1× PBS in 100 μl), respectively. Although it is unclear about the half-life of DOX in the mouse ovaries via i.p. injection, studies reported a half-life of DOX ranging from 11 to 45 h in different mouse tissues upon a single i.p. injection of 11–12 mg/kg [40, 41].

Dissection

On D3.5, mice were anesthetized via isoflurane inhalation. Blood was collected via the orbital sinus for serum collection as previously described [13, 14]. The left ovary was snap frozen in liquid nitrogen and kept at −80°C, whereas the right ovary was fixed in Bouin solution for 24 h, then kept in 70% ethanol at 4°C. In the D3.5 C57BL/6 mice, one side of oviduct and uterine horn was flushed to determine the effects of DOX on oocytes and embryos.

Serum progesterone (P4) and 17β-estradiol (E2) measurement

Serum was collected from the blood samples after clotting at room temperature for 45 min and stored at −80°C. Serum P4 and E2 were measured in the Ligand Assay and Analysis Core of the Center for Research in Reproduction at the University of Virginia (Charlottesville, Virginia). P4: %CV > 20 and reportable range = 0.150–40.00 ng/ml; E2: %CV > 20 and reportable range = 3–300 pg/ml.

Ovary histology and the number of CLs

The fixed C57BL/6 ovaries were processed for paraffin embedding as previously described [13, 14, 42, 43]. Paraffin sections (6 μm) through the widest middle portion of the ovaries were collected, processed, and stained with hematoxylin and eosin (H&E). The numbers of CLs in ovarian sections were independently examined by four individuals who were blinded to the treatments and obtained consistent CL counts. The number of CLs from one middle section per ovary per mouse was used for statistical analysis.

Immunohistochemistry and immunofluorescence

Immunohistochemistry was employed to detect proliferating cells using anti-proliferating cell nuclear antigen (PCNA, 1:500, D3H8P, Cell Signaling Technology) in fixed C57BL/6 ovarian sections (6 μm) as previously described [44]. The sections were counterstained with hematoxylin. PCNA-positive endothelial cells in the CLs were identified based on the typical elongated nucleus of endothelial cells. One representative field (as shown in Figure 4A1–C1) per CL was analyzed and the average number of PCNA-positive endothelial cells per field per CL in the section/mouse was calculated as on data point. Immunofluorescence was used to detect collagen IV (Col IV), StAR, and heat-shock protein 60 (HSP60). Briefly, paraffin/frozen sections were processed and subjected to antigen retrieval in 0.01 M sodium citrate (pH 6.0) at 95°C for 20 min. Sections were washed with 1× PBS followed by membrane permeabilization with 0.15% Triton X-100. The slides were then washed and blocked with 10% goat serum (16210064, ThermoFisher) and 1% bovine serum albumin (BSA) (B14, ThermoFisher) in 1× Tris-Buffered Saline (TBS) for fixed sections or 1× PBS for frozen sections for 1 h at room temperature; they were subsequently incubated with anti-collagen IV (1:200, Abcam, ab19808), anti StAR (1:700, Abcam, ab96637), or anti-HSP60 (1:400, Cell Signaling Technology, mAb #12165) in a humidified chamber for overnight at 4°C. The following day, the sections were washed in 1× PBS and incubated with secondary Alexa Fluor 488-conjugated goat anti-rabbit IgG antibody (1:200, Invitrogen, A11034) for 1 h. The sections were counterstained and mounted in 4′,6′-diamino-2-phenylindole (DAPI)-containing Vectashield (Vector Laboratories, Burlingame, CA). The negative control was processed together except without the primary antibody. Images were captured using a Carl Zeiss imaging system with an AxioCam MRc5 digital camera. The presented images had a resolution of 300 pixels/inch.

Figure 4.

PCNA immunohistochemistry and Col IV immunofluorescence in D3.5 C57BL/6 ovaries. (A–C1) PCNA staining in fixed ovaries. (D) Number of PCNA-positive endothelial cells per representative field as shown in (A1–C1). N = 4/group; ^*, P = 0.01021; error bar, SD. (E–G1) Col IV staining in frozen ovaries. (A and E) PBS- (B and F) DOX-treated with low progesterone (P4) level. (C and G) DOX-treated with normal P4 level. (A1–C1, E1–G1) Enlarged from the box in (A–C) and (E–F), respectively; scale bar: 200 μm (A–C, E–G), 25 μm (A1–C1, E1–G1); #, CL; ^*, follicle; green arrow in (A1–C1) PCNA-positive endothelial cells. No specific staining in the negative control (data not shown).

TUNEL staining

Frozen C57BL/6 ovarian sections (10 μm) were fixed in freshly prepared 4% paraformaldehyde in 0.02 M PBS (pH 7.4) at 23°C for 10 min. Slides were washed 2 × 5 min in 1× PBS and permeabilized in 1× PBS-T (0.1% Triton X-100) with 0.1% sodium citrate. Terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL) reaction mixture was made by adding 5 μl TUNEL-enzyme solution to 45 μl TUNEL-label solution. Slides were incubated with the mixture for 1 h at 37°C in a humidified chamber. Slides were rinsed three times in 1× PBS and counterstained and mounted in DAPI-containing Vectashield. The negative control was processed together except without the 5 μl TUNEL-enzyme solution.

Lipid droplet staining

Frozen C57BL/6 ovarian sections (10 μm) were fixed in 4% paraformaldehyde at room temperature for 20 min, washed twice in 1× PBS, then covered with 1.6 μg/ml Nile red (N3013, Sigma-Aldrich) in 1× PBS at room temperature for 15 min. Sections were then washed in 1× PBS and counterstained with DAPI. The relative sizes of lipid droplets (expressed as pixel ^{^2}) were quantified using ImageJ [13, 14]. Briefly, original TIF images were converted to single-channel, 8-bit images. Quantile-based normalization was used to standardize the image intensities. Finally, background subtraction, threshold segmentation, and quantification of lipid droplets were performed using ImageJ.

Phalloidin staining

Frozen C57BL/6 ovarian sections (10 μm) were fixed in freshly prepared 4% paraformaldehyde in 0.02 M PBS (pH 7.4) at 23°C for 10 min. Slides were washed 2 × 5 min in 1× PBS. Slides were permeabilized and blocked for nonspecific staining in 1× PBS-T (0.1% Triton X-100) in 1% BSA for 30 min, washed 2 × 5 min in 1× PBS, and incubated with 200 μl 488-conjugated phalloidin solution (1:100 dilution) for 30 min at 23°C. The negative control received 200 μl 1× PBS. Slides were washed 2 × 5 min in 1× PBS. Slides were counterstained and mounted in DAPI-containing Vectashield (H-1200-10, Vector Laboratories).

Statistical analysis

Data are presented as individual data points or mean ± standard deviation (SD) where applicable. A linear mixed model fit by REML using Sattherthwaite method was used to analyze the percent weight change from D0.5 to D3.5. The percent weight change data were analyzed in R (Version 4.0.2, package lmerTest 3.1-3). Variances between two groups were tested using the “F-test Two-Sample for Variances” in Excel. Correlation analyses between percentage body weight change and serum P4 levels were done using “Regression” in Excel. Two-tailed equal or unequal variance student t-test was used to compare two groups. Significance level is set at P < 0.05.

Results

Effects of DOX treatment on body weight and serum progesterone levels

In the initial study, we used wild-type mice with mixed background (C57BL/6 and 129/SvJ [35]) that were available in our mouse colony. Their age range was from 10.0 to 22.1 weeks at dissection (16.2 ± 4.2 weeks in PBS group, N = 6; and 14.5 ± 5.4 weeks in DOX group, N = 7). DOX treatment significantly reduced body weight gain on D1.5 (−3.0 ± 1.9% in PBS group and −5.1 ± 1.4% in DOX group, P = 0.0262) (Figure 1A). The DOX-treated mice quickly regained body weight and the body weight gains compared with D0.5 were comparable between the two groups on both D2.5 and D3.5 (Figure 1A).

Figure 1.

Body weight changes and serum progesterone levels. (A–C) Mice in mixed background at 10.0–22.1 weeks old; (D–F) mice in C57BL/6 pure background at 9.6–10.9 weeks old. Black dots, PBS-treated vehicle control group (N = 6 in both sets); red triangles, DOX-treated group (N = 7 in mixed background and N = 5 in C57BL/6 background). (A and D) Daily body weight changes of individual mice from treatment on 0.5-day postcoitum (D0.5) to dissection on D3.5. #, P = 0.0262 (D1.5) in (A) and P = 0.00774 (D1.5) in (D). (B and E) Serum progesterone levels of individual mice. Line, average of the group; #, P = 0.097 in (B) and P = 0.068 in (E), two-tailed equal variance t-test. (C and F) Lack of significant correlation between body weight changes on D1.5 and serum progesterone levels on D3.5. P = 0.153 in (C) and P = 0.390 in (F).

A main function of the CL during early pregnancy is to produce P4 for supporting early pregnancy events. The serum P4 levels reach a plateau by D3.5 in mice. Among the 13 samples in this initial experiment, four samples with the lowest P4 levels were in the DOX-treated group, which had three other samples within the full range of P4 levels in the control group (Figure 1B, P = 0.097). There was no significant correlation between the P4 levels and the changes of body weight on D1.5 (Figure 1C, P = 0.153). There was no significant difference in serum 17β-estradiol (E2) levels between the two groups (P = 0.828, data not shown). Since the mice in the initial study were in mixed background and ranging from 10.0 to 22.1 weeks old, the varied effects of DOX on P4 levels of individual mice promoted us to test the hypothesis in mice with C57BL/6 pure background and a narrower age range.

To control the potential variables of mouse strain, mouse age, and housing and mating environment, we ordered C57BL/6 females in the second set of study and only included the mice plugged within 11 days of cohabitation with stud males. They were randomly assigned into two groups on D0.5 to receive a single i.p. injection of PBS (N = 6, vehicle control) and DOX (N = 5), respectively. The plugging latency (duration from cohabitation to vaginal plug detection) and the body weight at the time of treatment on D0.5 were comparable between the two groups. DOX treatment also significantly reduced body weight gain on D1.5 (−0.3 ± 1.8% in PBS group and −3.9 ± 3.0% in DOX group, P = 0.00774). The temporal patterns of body weight change were comparable between the two sets (Figure 1A and D). In the second set, the ages at dissection on D3.5 were at a narrow range of 9.9 ± 0.4 weeks in the PBS group (N = 6) and 10.1 ± 0.7 weeks in the DOX group (N = 5). However, even with all the potential variables controlled in the second set, varied effects of DOX on P4 levels of individual mice were still present. Among the 11 samples, the three lowest ones were in the DOX-treated group, which had two other samples with comparable P4 levels as the control group (Figure 1E, P = 0.068). There was no significant correlation between the P4 levels and the changes of body weight on D1.5 (Figure 1F, P = 0.390) in the second set of mice either. There was no significant difference in E2 levels between the two groups (P = 0.215, data not shown). These data indicate that a single injection of DOX treatment on D0.5 can have varied effects on CL function in P4 synthesis. In the two sets (Figure 1B and E), the ranges of progesterone levels in the PBS control group were different, which could be contributed by batch-to-batch difference in detecting progesterone levels, mouse strain difference, and age difference, etc. The two sets (total N = 12 in both PBS- and DOX-treated groups) were not combined for t-test. However, if they were combined as the percentage of mice in the DOX-treated group with “low” progesterone levels, it would be 4/7 (57.1%) in the first set (Figure 1B, mixed background) and 3/5 (60%) in the second set (Figure 1E, C57BL/6 pure background). The average percentage of mice with low P4 levels (below the range in control mice) would be 0% (0/12) in the PBS-treated control mice and 7/12 (58.3%) in the DOX-treated mice (Figure 1B and E). Fisher exact test gave a two-tailed P-value of 0.0046. We focused on the second set of mice in C57BL/6 pure background for further analyses.

Preimplantation embryo development

Control mice

The timing of mouse preimplantation embryo development in wild type control mice follows an order of zygote (D0.5/E0.5), two-cell (D1.5/E1.5), morula (D2.5/E2.5), and blastocyst (D3.5/E3.5) stages [45, 46]. The mouse embryos are in the oviduct by D3.0 and are in the uterus by D3.5 under the normal condition [15]. When the C57BL/6 mice were dissected on D3.5, one side of oviduct and uterine horn from each mouse was flushed to determine the location of oocytes and embryos as well as embryo development stages. Among the flushing from five of the six oviducts (one with nothing) in the control group, there were four germinal vesicle (GV) oocytes (Figure 2Ai), eight degenerated possible oocytes with low-density cytoplasm (Figure 2Aii), and three possible later stage oocytes, which had dense aggregates in the cytoplasm detached from the zona pellucida but had no identifiable nucleus or polar body (Figure 2Aiii); there were 12 blastocysts (Figure 2Aiv) flushed from three out of the six uterine horns but no oocytes nor earlier stage embryos flushed from any of the six uterine horns. These observations in the control group indicate that only blastocysts were present in the uterus, whereas unfertilized oocytes were retained in the oviduct.

Figure 2.

Images of oocytes and embryos from C57BL/6 oviduct and uterine horn. The number under each image indicating the total number of oocyte/embryo with similar appearance at the same location in the same group. (A) PBS control group (N = 6 mice). Oviduct: Ai, GV oocyte; Aii, degenerated possible oocyte; Aiii, possible metaphase I (MI), or metaphase II (MII) oocyte. Uterus: Aiv, blastocyst. (B) DOX-treated group (N = 5 mice). Oviduct: Bi, degenerated possible oocyte; Bii, possible MII oocyte or zygote; Biii, zygote with two visible polar bodies (one fragmented); Biv, zygote with one visible polar body and two nuclei; Bv, two-cell embryo. Uterus: Bvi, degenerated possible oocyte; Bvii, zygote with two polar bodies.

DOX-treated mice

In the oviducts, among the flushing from four of the five oviducts in the DOX group (N = 5 mice), there were three degenerated possible oocytes with low-density cytoplasm (Figure 2Bi), five possible metaphase II (MII) oocytes or zygotes with dense aggregates in the cytoplasm detached from the zona pellucida and with a polar body but without an identifiable nucleus (Figure 2Bii), one zygote with dense aggregates in the cytoplasm and two polar bodies (one fragmented) (Figure 2Biii), one zygote with two nuclei and one visible polar body (Figure 2Biv), and one 2-cell embryo with dense aggregates in the cytoplasm (Figure 2Bv). In the uterine horns, there were two degenerated possible oocytes (Figure 2Bvi) and three zygotes with dense aggregates in the cytoplasm and two polar bodies but without identifiable nuclei (Figure 2Bvii) flushed from one uterine horn but no oocytes or embryos were flushed from the other four uterine horns in this group. These observations indicate that DOX treatment inhibited early embryo development in vivo and that DOX treatment might interfere with oocyte/embryo transport as well.

Ovary histology

The CL numbers from histology of D3.5 ovaries were comparable between the control and DOX-treated groups (Figure 3A), and there was no apparent difference in the general appearance of the CLs under low magnification between the two groups (Figure 3B–D). Under higher magnification, there were three obvious morphological changes in the CLs from DOX-treated mice with low P4 levels: (1) the corpus luteal cords, which were permeated by microvasculature, were less defined (Figure 3C1) than that seen in the control CL (Figure 3B1); (2) there was scattered cell debris that may indicate scattered endothelial cell and luteal cell degeneration (Figure 3C1); and (3) the luteal cells did not show the typical large polygonal cytoplasm and the luteal cell cytoplasm often showed small foamy areas (Figure 3C1). In the ovaries from DOX-treated mice with normal P4 levels (Figure 3D1), there were defined corpus luteal cords outlined by endothelial cells in the CLs; there was no obvious cell debris, but there was foamy cytoplasm and the nuclei appeared larger in the luteal cells (Figure 3D1). The foamy areas in the cytoplasm were most likely occupied by lipid droplets that were dissolved during histological processing. These data indicate that DOX treatment has various effects on both endothelial cells and luteal cells in the CL, even though some effects may not be correlated with impaired P4 synthesis. Regardless of the effect of DOX treatment on the serum P4 levels, the histology also confirmed that DOX treatment increased granulosa cell death in the developing follicles (Figure 3B2–D2), and this observation is consistent with previous studies in nonpregnant mice [6–11]. The increased granulosa cell death in the developing follicles was not reflected in the serum E2 levels most likely due to the normally low E2 levels during early pregnancy.

Figure 3.

Histology of D3.5 C57BL/6 ovaries. (A) Numbers of CLs in PBS- and DOX-treated groups. N = 5–6; error bar, SD. (B, B1, B2) PBS- (C, C1, C2) DOX-treated with low progesterone (P4) level. (D, D1, D2) DOX-treated with normal P4 level. (B1, C1, D1) CL enlarged from the smaller box in (B, C, D), respectively; B2, C2, D2: follicles enlarged from the bigger box in (B, C, D), respectively; scale bar: 200 μm (B–D), 25 μm (B1–D1), or 50 μm (B2–D2); #, CL; ^*, follicle; green arrow in C1 and D1, vacuolated luteal cells; black arrow in C1, C2, D2, degenerated cells. H&E staining.

PCNA staining to detect cell proliferation in the ovary

PCNA immunostaining indicated that the overall strongest staining was in the granulosa cells of developing follicles of both control and DOX-treated ovaries (Figure 4A–C) despite more intense staining in the control follicles than in the DOX-treated follicles (Figure 4A–C), which most likely reflects reduced cell density due to the increased granulosa cell death in the DOX-treated follicles as seen in the histology (Figure 3B2–D2). PCNA-positive cells were scattered in the CLs (Figure 4A1–C1). In the control CLs, most of the PCNA-positive cells were endothelial cells typically with an elongated nucleus surrounding the luteal cords (Figure 4A1); in the DOX-treated CLs from mice with low P4 levels, the luteal cords were not defined, and most of the scattered PCNA-positive cells did not appear to be endothelial cells that should have an elongated nucleus surrounding the luteal cords (Figure 4B1); in the DOX-treated CLs from the two mice with normal P4 levels (Figure 1E), PCNA-positive endothelial cells were also reduced (Figure 4C1). The reduction of PCNA-positive endothelial cells in the DOX-treated CLs was significant (Figure 4D). These data indicate that DOX treatment reduced endothelial cell proliferation in the CL.

Col IV staining to detect basal lamina of endothelial cells in the ovary

Col IV is a marker of the basal lamina of endothelial cells. Immunofluorescence of Col IV revealed that compared with the control group (Figure 4E and E1), the expression level of Col IV was reduced and the expression pattern of Col IV was disorganized in the CLs from the DOX-treated mice with low P4 levels (Figure 4F and F1), consistent with the lack of defined luteal cords and reduced endothelial cells in histology (Figure 3C1). The Col IV staining pattern in the CLs from DOX-treated mice with normal P4 levels was less disrupted (Figure 4G and G1) compared with that in the CLs from the DOX-treated mice with low P4 levels (Figure 4F and F1). CL histology (Figure 3), PCNA staining (Figure 4A–D), and Col IV staining (Figure 4E–G1) consistently show adverse effects of DOX on the CL microvasculture, more severe in the CLs from DOX-treated mice with low P4 levels.

TUNEL staining to detect cell death in the ovary

A distinctive pattern in the ovary sections from both groups was the intense and massive TUNEL staining in some follicles (Figure 5A–C) due to physiological ovarian follicle atresia in the control group and additional toxic effect of DOX in the follicles of DOX-treated group. The numbers of identifiable growing follicles in the TUNEL and DAPI double-stained sections were comparable between the control (7.5 ± 2.1) and the DOX-treated (7.2 ± 2.4, P = 0.831) groups (Figure 5D); however, the percentage of TUNEL-positive growing follicles was significantly higher in the DOX-treated group (87.8 ± 12.6%) than that in the control group (45.9 ± 14.4%, P = 0.00062) (Figure 5E). It was consistent with the histological observation (Figure 3B2–D2) and served as another indication of the DOX effect on the ovarian follicles. In addition, the TUNEL-positive follicle cell density was lower in the DOX-treated group (Figure 5A2–C2), consistent with the histology (Figure 3B2–D2). This study revealed that DOX-induced cell death in the growing follicles during early pregnancy remained significant 3 days after treatment.

Figure 5.

TUNEL staining of D3.5 C57BL/6 ovaries. (A, A1, A2) PBS- (B, B1, B2) DOX-treated with low progesterone (P4) level. (C, C1, C2) DOX-treated with normal P4 level. (A1, B1, C1) CL enlarged from the smaller box in (A, B, C), respectively; (A2, B2, C2) follicles enlarged from the bigger box in (A, B, C), respectively; scale bar: 400 μm (A–C), 25 μm (A1–C1), or 100 μm (A2–C2); #, CL; ^*, follicle. (D) Numbers of identifiable growing follicles in PBS- and DOX-treated groups. (E) Percentage of TUNEL-positive follicles in identifiable growing follicles in PBS- and DOX-treated groups. (D and E) N = 5–6; ^*, P = 0.00062; error bar, SD. No specific staining in the negative control (data not shown).

In the CLs from the control group, there was no TUNEL staining in the sections from five of the six mice in this group (Figure 5A and A1). In the section from the sixth mouse with the lowest P4 level (Figure 1E), there were scattered TUNEL-positive cells in the CLs (data not shown). In the CLs from the DOX-treated group, the CLs from mice with low P4 levels had scattered TUNEL-positive cells (Figure 5B and B1), some of which are expected to be endothelial cells, and a few of which with the largest green staining areas might be luteal cells; whereas those from the mice with normal P4 levels had no TUNEL-positive cells (Figure 5C and C1). These data indicate varied effects of DOX on cell apoptosis in the CL and increased cell apoptosis in the CL was associated with low-serum P4 levels.

StAR staining to detect the protein for the rate-limiting step of steroidogenesis in the CL

An essential step of steroidogenesis is the conversion of the substrate cholesterol to pregnenolone in the mitochondria; whereas the rate-limiting step of steroidogenesis is the transport of the substrate cholesterol from the outer to the inner mitochondrial membrane, a step carried out by the StAR [20]. StAR immunofluorescence revealed that all mice except the one with the lowest P4 level (Figure 1E) in the control group had intense StAR staining in the cytoplasm of almost all luteal cells (Figure 6A and A1); in the DOX-treated mice with low P4 levels, the luteal cells were less differentiated (Figures 3C1 and4B1), most of the luteal cells had minimal StAR staining and only a few scattered luteal cells had strong StAR staining (Figure 6B and B1); and in the DOX-treated mice with normal P4 levels, the intensity of StAR staining ranged from weak to strong in the luteal cells (Figure 6C and C1). These data demonstrate varied effects of DOX treatment on StAR staining in the luteal cells and a dramatic reduction of StAR expression in luteal cells was associated with less differentiated luteal cells and low P4 levels in the DOX-treated group.

Figure 6.

Immunofluorescence detection of StAR and HSP60 in D3.5 C57BL/6 ovaries. (A–C1) StAR; (D–F1) HSP60; (A and D) PBS- (B and E) DOX-treated with low progesterone (P4) level. (C and F) DOX-treated with normal P4 level. (A1, B1, C1) Enlarged from the box in (A, B, C), respectively; green: StAR staining. (D1, E1, F1) Enlarged from the box in (D, E, F), respectively; green: HSP60 staining. #: CL; scale bar: 200 μm (A–F) or 25 μm (A1–F1). No specific staining in the negative control (data not shown).

HSP60 staining to detect mitochondria in the CL

Since StAR is expressed in the mitochondria and DOX can accumulate in the mitochondria to cause oxidative stress [47, 48], to determine if DOX could affect the mitochondrial density, and thus indirectly affect the StAR expression levels in the luteal cells, we detected the expression of HSP60, a mitochondrial marker localized in the matrix of mitochondria [49]. HSP60 immunofluorescence indicated that all CLs in different groups had strong HSP60 staining (Figure 6D–F) and the strongest HSP60 staining in the CL was in the cytoplasm of luteal cells (Figure 6D1–F1). Therefore, the reduction of StAR expression in the DOX-treated luteal cells (Figure 6B1 and C1) was not caused by lack of mitochondria.

Nile red staining to detect lipid droplets in the CL

The key function of StAR in steroidogenesis is to transport the substrate cholesterol and lipid droplets are a main source of cholesterol in luteal cells. We detected the lipid droplets in the CLs using Nile red staining. At low magnification, the Nile red staining in the control CLs was relatively lighter than that in the surrounding interstitial compartment (Figure 7A); whereas that in the DOX-treated CLs was more comparable to the surrounding interstitial compartment (Figure 7B and C). At a higher magnification, compared with the sizes of the lipid droplets in the control luteal cells (Figure 7A1), the lipid droplets in the DOX-treated luteal cells were more variable in sizes with some large ones regardless of the serum P4 levels (Figure 7B1 and C1). Quantification data revealed enlarged lipid droplets in DOX-treated group (Figure 7D). These data demonstrate lipid droplet accumulation in DOX-treated luteal cells that was not correlated with StAR expression or serum P4 levels.

Figure 7.

Nile red staining of lipid droplets and Phalloidin staining of actin filaments (F-actin) in D3.5 C57BL/6 CLs. (A–C1) Nile red staining; (E–G1 Phalloidin staining; (A and E) PBS- (B and F) DOX-treated with low progesterone (P4) level. (C and G) DOX-treated with normal P4 level. (A1, B1, C1) Enlarged from the CL marked with a red # in (A, B, C), respectively; green, Nile red staining of lipid droplets; blue, DAPI staining of nuclei; #, CL; scale bar, 200 μm (A–C) or 12.5 μm (A1–C1). (D) Relative sizes of lipid droplets. The average size of lipid droplets (pixel ^{^2} using ImageJ) in all representative areas of all CLs from the same mouse is considered as one data point. N = 5–6; ^*P = 4.42E−05; error bar, SD. (E1, F1, G1) Enlarged from the boxed area in (E, F, G), respectively; green, Phalloidin staining of F-actin; blue, DAPI staining of nuclei; scale bar, 50 μm (E–G) or 12.5 μm (E1–G1); white arrows in (E and E1), pointing to the same luteal cell whose nucleus was outlined by the green Phalloidin staining at a lower magnification (E) and a higher magnification (E1). No specific staining in the negative control (data not shown).

Phalloidin staining to detect cytoskeleton integrity in the CL

The cytoskeleton is critical for directional lipid droplet movement in the cytoplasm. Phalloidin staining was used to detect F-actin in the CLs. In the control CLs, the individual luteal cells (w/round nuclei) were readily identifiable because the continuous sheet-like Phalloidin staining clearly outlined the positions of the nuclei (Figure 7E and E1). However, in the DOX-treated CLs (Figure 7F and G), most luteal cells were not as readily identifiable as those in the control CLs (Figure 7E), because the Phalloidin staining was often seen as clusters (Figure 7F–G1), especially in the CLs from DOX-treated mice with low P4 levels (Figure 7F and F1). These data demonstrate disrupted cytoskeleton in DOX-treated luteal cells.

Discussion

DOX is a widely used chemotherapeutic agent. A major side effect of DOX is intestinal mucositis. In our early pregnancy mouse model, the first response observed was a transient body weight loss detected 24 h after injection, which disappeared within 48 h of treatment. It was demonstrated in pigs and rodents that reduced food consumption and gastrointestinal damage are contributing factors for DOX-induced weight loss [50–52]. We did not examine food consumption nor the intestine on D1.5 in this study when the weight loss was transiently evident. Although there was an overall reduction of body weight in the DOX-treated group on D1.5, there were different levels of reduction, indicating varied individual sensitivity. In addition, no significant correlation between body weight changes and serum P4 levels could suggest different sensitivities on different parameters within the same individual toward DOX treatment.

DOX has toxic effects on cultured embryos [53, 54]. Information about effects of DOX on early embryo development in vivo remains lacking. Our limited data indicated that a single therapeutical-relevant dose of DOX (10 mg/kg) prevented mouse embryo development beyond two-cell stage, mainly arrested at one-cell zygote stage in vivo. Since DOX was delivered via i.p. injection shortly after fertilization when ovarian hormones P4 and E2 are at low levels [15, 55], the toxic effects on the embryos were most likely caused by direct exposure to DOX in the oviduct. How does DOX reach the oviductal lumen? DOX via i.v. or i.p. injection in mice reached the ovaries and uterus within hours [10, 56]. Although there is insufficient literature on DOX pharmacokinetics in the mouse oviduct, which is localized in between the ovary and the uterus, it is reasonable to speculate that DOX could be distributed to the oviduct within hours of injection, and in our experimental setting, DOX distribution in the oviduct should peak within hours of injection before an embryo reaches the two-cell stage on D1.5. Based on a study that i.v. injected tracers on D0.5 mice were detected in the multivesicular bodies and in numerous small vesicles in the apical portion of the preampulla oviductal epithelial cells [57] and another study showing the role of P-glycoprotein/MDR1 in exporting DOX from ovarian cells [11], we expect that both exocytosis and ABC transporters on the oviductal epithelial cells could be involved in transporting i.p. injected DOX into the oviductal lumen to affect the zygotes on D0.5 in this study. How could DOX cause zygote arrest in the oviductal lumen? One possibility could be the adverse effect of DOX on F-actin to impair DNA damage repair [58].

Luteal cells are normally differentiated from the remaining granulosa cells and theca cells in the ovulated follicles. This study revealed that during early pregnancy, in contrast to massive apoptosis of granulosa cells in the developing follicles, the luteal cells in the CLs were rarely TUNEL-positive upon DOX treatment and luteal cell apoptosis might be occasionally present in the CLs from DOX-treated mice with low P4 levels. One explanation for the cell-type specific sensitivity to DOX treatment is the expression of ABC transporters, such as P-glycoprotein/MDR1, that pump drugs out of the cells. DOX was accumulated in cardiac tissue of mice lacking mdr1a P-glycoprotein [56], which also had increased ovarian toxicity [11] due to the expected intracellular accumulation of DOX. P-glycoprotein has a spatiotemporal expression pattern in the ovary. It was shown to be highly upregulated in the granulosa cells of rat preovulatory follicle after equine chorionic gonadotropin stimulation for 42 h (preovulation) and it remains highly expressed in the luteal cells postovulation [59]. The upregulation of P-glycoprotein in the CLs compared with follicles was also evident in adult mouse ovaries at estrus stage [11]. Since DOX treatment was administrated on D0.5 (postovulation), P-glycoprotein expression is expected to be already upregulated to pump out DOX from the luteal cells. Therefore, one reasonable explanation for the resistance of luteal cells in the CLs but vulnerability of granulosa cells in the follicles from DOX-induced cell death would be the upregulation of ATP transporters, such as P-glycoprotein, to reduce DOX levels in the luteal cells. Since chemotherapeutic drugs, including DOX, are preferentially targeting proliferating cells (e.g., cancer cells), and luteal cells are not typically proliferating, whereas granulosa cells are highly proliferating, minimal cell proliferation in luteal cells can be another contributing factor for their resistance to DOX-induced apoptosis.

In addition to the luteal cells, the other main type of cells in the CL is the endothelial cell [18]. Although the area of an ovarian follicle occupied by granulosa cells lacks vasculature, the CL is highly vascularized [19]. Endothelial cells are highly proliferating during CL development. The vasculature in the CL supports CL development and luteal cell functions, including P4 steroidogenesis and transport of steroid hormones to the systemic circulation [19, 20]. Endothelial cells are sensitive to DOX treatment in general [60, 61] and vascular toxicity of DOX contributes to the placental toxicity in ICR mice [62]. In the CLs from DOX-treated mice with low P4 levels, the vascular toxicity of DOX could be the main contributing factor for the lack of defined luteal cord structure. However, in the CLs from DOX-treated mice with normal P4 levels, there were defined luteal cords. Although we have not investigated the contributing factors for the varied individual sensitivity to DOX treatment, we speculate that varied individual ovulation timing could be the main one. It is generally believed that ovulation in mice occurs during the dark period (night); however, there could be up to 10 h of difference in the ovulation timing among different mice [39]. Therefore, the time gap between ovulation and DOX treatment could vary for hours among different mice even though the DOX treatment timing was kept consistent. Capillary growth in the granulosa cell compartment initiates right before ovulation and continues during CL development [19]. It is expected that endothelial cell proliferation is most active during the initial hours of CL development to support luteal cell differentiation. Chemotherapeutical drugs, including DOX, preferentially target proliferating cells; therefore, it is expected that the proliferating endothelial cells are the preferred targets of DOX in the developing CL. The mice with earlier ovulation timing would have a longer period between ovulation and DOX treatment for establishing a more mature capillary network to support luteal cell differentiation, therefore, DOX treatment had less adverse effects on CL development. While those with later ovulation timing that was closer to the time of DOX treatment, the endothelial cells were targeted by DOX before sufficient capillary network was established to support CL differentiation. This rationale is supported by our observation that CLs from mice with low P4 levels had less defined luteal cord structure and disrupted Col IV expression patterns, indicating impaired vasculature in the CLs, and these changes were accompanied with less differentiated luteal cells with diminished StAR expression. These observations indicate that endothelial cells in the CLs are likely the primary target of DOX and that there is individual variation in sensitivity to DOX treatment, which might be contributed by the varied ovulation timing in mice.

The individual variation in the sensitivity to DOX treatment is also reflected in StAR expression. Diminished StAR expression in the luteal cells of DOX-treated CLs was correlated with disrupted vasculature, less differentiated luteal cells, and reduced P4 levels. DOX suppressed StAR mRNA expression in male rat testes that was correlated with decreased plasma testosterone levels [33, 34]. Since StAR is the rate-limiting protein for transporting the substrate cholesterol from the outer to the inner mitochondrial membrane for P4 steroidogenesis, the quantity of mitochondria in the cytoplasm could affect the overall StAR expression level in the luteal cells. DOX can accumulate in the mitochondria due to its specific binding to the abundant phospholipid cardiolipin located in the inner mitochondrial membrane. The accumulated DOX can disrupt the electron transport chain in the mitochondria to overproduce reactive oxygen species (ROS) [47, 48, 63]. Although the quantity of mitochondria in the luteal cells, based on HSP60 expression levels, is not significantly affected by DOX treatment, DOX-induced overproduction of ROS in mitochondria could inhibit StAR expression, which will reduce the substrate cholesterol from reaching the inner mitochondrial membrane and disrupt the mitochondrial function in supporting the enzymatical conversion of cholesterol to pregnenolone (by P450SCC/CYP11A1) on the inner mitochondrial membrane. The molecular mechanisms in DOX-induced suppression of StAR expression, which could be the cause of DOX-induced P4 deficiency in some mice, remain to be elucidated.

Lipid droplets store neutral lipids, including cholesteryl ester that is a main source of cholesterol for P4 steroidogenesis. They are surrounded by a phospholipid monolayer and coated with perilipins (PLIN1-5) [23]. The CLs undergo development and maintenance for P4 synthesis to support early pregnancy. If pregnancy does not occur or P4 production in CL is no longer needed during pregnancy, the CL undergoes luteal regression/luteolysis, which is hallmarked by lipid droplet accumulation [26], which is also present in structurally regressing CLs from a previous cycle that coexist with current cycle CLs at maintenance stage [25]. DOX treatment leads to lipid droplet accumulation in the luteal cells on D3.5 when the CLs are normally at the maintenance stage. Since DOX could disrupt actin cytoskeleton in the luteal cells and cytoskeleton plays an important role for directional lipid movement, e.g., to the mitochondria, the disrupted cytoskeleton inevitably halted directional lipid movement leading to lipid droplet accumulation. On the other hand, there was a dramatic reduction of StAR expression in the luteal cells from DOX-treated mice with low P4 levels; therefore, the utilization of lipid droplets as the substrate resource for P4 synthesis is diminished, which may also lead to lipid droplet accumulation. The disrupted cytoskeleton in DOX-treated CLs may be caused by DOX-induced oxidative stress [64].

In summary, we have identified multiple effects of DOX treatment on D0.5 in both endothelial cells and luteal cells of D3.5 preimplantation CLs. There are three types of effects from DOX treatment during early pregnancy: (1) effects are present in all DOX-treated CLs but not correlated with the key function of CLs in P4 steroidogenesis, such as lipid droplet accumulation and disrupted cytoskeleton; (2) effects varied greatly and are correlated with P4 levels, such as StAR expression in the luteal cells and impaired morphology of luteal cords, which are surrounded by endothelial cells; and (3) varied effects uncorrelated with P4 levels, such as body weight change on D1.5. The molecular mechanisms for DOX-induced effects, e.g., endothelial cell toxicity and reduced luteal cell StAR expression, in the CL remain to be investigated. This study fills in the knowledge gap about toxic effects of chemotherapy on the CL and provides critical information for risk assessment of chemotherapy in premenopausal patients.

Data availability

Data available on request.

Grant Support: This work was supported by the National Institutes of Health (NIH R03HD100652 and R03HD097384 to XY; and NIH K01ES030014 to SX).