1. Doxorubicin leads to decreased ovarian reserve and declined ovarian function
Several studies have indicated that chemotherapy-induced damage to the ovary includes microvascular damage, interstitial necrosis, and fibrosis 11,15. In our study, ovarian damage was induced by the intraperitoneal injection of doxorubicin (Dox, 10 mg kg-1) (Fig. 1a). Compared with the blank group, ovarian weight and ovary index were significantly decreased (81.8% vs 25.0%, P < 0.05) (Fig. S1, 1b and 1c). The proportion of irregular estrus cycles increased in the Dox-treated group (Fig. 1b). The levels of estrogen, progesterone, and AMH in the Dox group decreased, and the level of follicle-stimulating hormone (FSH) was increased (Fig. 1c-f). Furthermore, the number of mouse follicles at all levels was counted after H&E staining of paraffin sections of mouse ovaries (Fig. 1g). Compared with the blank group, the number of primordial follicles (PMF), secondary follicles (SF), and total number of healthy follicles (THF) in the Dox group decreased significantly (Fig. 1h). Moreover, the proportion of primordial follicles and secondary follicles in the Dox group was lower than that in the Blank group, with atretic follicles (ATF) increasing (Fig. 1i), indicating that Dox causes a significant loss of primordial follicles and an increase in atretic follicles. In addition, fibrosis of the ovary was evaluated by red Sirius staining (Fig. 1j), showing that the proportion of collagen area in the Dox group is higher than that in the blank group (31.58% ± 1.702%, 9.975% ± 0.6826%, P < 0.001), which was also confirmed by polarized light photography. Masson’s trichrome staining (Fig. 1k) revealed that the proportion of collagen-positive (blue) fiber area in the ovary of the Dox group is higher than that of the Dox group (3.639% ± 0.3098%, 18.59% ± 0.7567%, P < 0.001). The expression of the fibrosis marker α-SMA in the Dox group was higher than that in the blank group by IHC (Fig. S1, d and e). Together, these results indicate that Dox causes ovarian damage in mice by inducing ovarian interstitial fibrosis.
2. Tgfbr2 and Sirt1 are markedly reduced in human and mouse damaged ovarian tissues
To identify genes that might be responsible for doxorubicin-induced ovarian damage, we performed an RNA seq analysis in RNAs that were extracted from the ovarian tissues of the Dox and blank groups. Differentially expressed genes (DEGs) were obtained through data quality control, mapping, and normalization. Compared with the blank group, the DEGs (119 upregulated genes and 178 downregulated genes) were identified in the ovaries of the Dox group compared with the blank group (FDR < 0.05, fold change > 2). The DEGs were displayed in a volcano plot (Fig. 2a) and heatmap (Fig. S2a). Gene Ontology (GO) enrichment analysis of upregulated DEGs revealed that they were mainly enriched in biological processes, such as “Adherens junction,” “Focal adhesion,” and “Extracellular matrix” (Fig. 2b). KEGG pathway enrichment analysis found that it was enriched in “ECM-receptor interaction,” “Cell adhesion molecules (CAMs),” and other pathways, indicating that Dox may increase the degree of ovarian fibrosis (Fig. S2b), which is also consistent with the ovarian phenotype in the above experiment. The genes related to steroid hormone synthesis and ovarian follicle development, such as Hsd17b7, Lhcgr, Hmgcs1, and Sirt1, were significantly downregulated, and fibrosis-related genes that inhibit fibrosis, such as Tgfbr2 and Mmp15, were also downregulated (Fig. 2a). GO and KEGG pathway enrichment analysis of down-regulated genes revealed that these genes are mainly enriched in “Sterol biosynthetic process”, “Cholesterol metabolic process”, and other steroid hormone synthesis-related biology processes (Fig. 2c). KEGG pathway analysis includes “Ovarian steroidogenesis,” “Estrogen signaling pathway” and other KEGG pathways (Fig. S2c), confirming that Dox causes the decline of ovarian function. These results suggest that Dox increases the degree of ovarian fibrosis while reduces the hormone synthesis function of the ovary.
The transforming growth factor-beta 1 (Tgfb1) signaling pathway plays a key role in the progression of multiple organ fibrosis, such as renal fibrosis, age-related hypertrophic cardiomyopathy, immune cell recruitment, and extracellular matrix 16 17. Exogenous administration of Tgfbr2-targeted inhibition of the Tgfb1 pathway can alleviate myocardial fibrosis and myocardial infarction by reducing myocardial remodeling18, preventing the progression of heart failure, and improving the survival rate 19. In addition, sirtuin 1 (Sirt1) can affect follicle development by regulating apoptosis. Studies have found that activating Sirt1 can inhibit the activation of primordial follicles, reduce follicle assumption, and increase the reserve of follicles 20. Therefore, it is speculated that increasing the expression of Sirt1 can play a protective role against chemotherapy-induced ovarian damage.
Our qPCR results showed that the expression of ovarian fibrosis-related molecules Tgfb1, Acta2, IL-1b, and Tnf-α in the Dox group is upregulated, whereas the expression of Sirt1, Tgfbr2, and Timp2 is downregulated (Fig. 2d). Western blot (WB) and immunohistochemistry (IHC) analysis revealed that the expression of Sirt1 and Tgfbr2 is also decreased in the Dox group (Fig. 2e and 2f). To validate the difference in the expression of Sirt1 and Tgfbr2 in human ovarian tissue between Dox and normal controls, 8 samples of human ovarian tissue aged between 33 and 67 years old were collected for culture in vitro. After culturing in medium containing Dox (1 μg mL-1), the relative expression of SIRT1 and TGFBR2 mRNA (Fig. 2g) and protein (Fig. 2h and 2i) was significantly lower than that of the ovaries cultured in the control medium.
In summary, our results suggest that Dox might cause ovarian damage by reducing the expression of Tgfbr2, a key factor that inhibits the process of fibrosis, and Sirt1 that is related to follicle development.
3. AdV serves as a suited vector for ovary infection
To identify a better vector more suited for the gene therapy experiments in ovaries, we explored the transductional tropism of three commonly used vectors by carrying out ovarian culture in vitro with empty viral vectors (AdV-GFP, AAV-GFP, and LV-GFP) to observe the fluorescence intensity. The results showed that AdV-GFP has a better affinity for the ovaries, as there was almost no fluorescence intensity in the LV-GFP and AAV-GFP groups (Fig. 3a and 3f). Also, the media containing 108, 109 and 1010 titers of AdV-GFP were used for ovarian culture in vitro. With time and the virus titer increased, the fluorescence also increased (Fig. 3b and 3g). The ovaries cultured with 1010 titer of AdV-GFP were frozen and sectioned. These samples were analyzed by GFP fluorescence and the oocyte marker Ddx4 immunofluorescence labeling (Fig. 3c). The result showed that the oocytes have almost no green fluorescence. Furthermore, the ovarian tissues were digested and separated into single cells for culture and then cultured with 109 empty AdV-GFP. The results showed that most of the GFP-positive cells are stroma cells, and the proportion of GFP-positive stromal cells is 61.55 ± 2.651%. The oocytes were hardly transfected (Fig. 3e and 3h). The cells were stained with Cyp17a1 (red fluorescence), a marker of ovarian stromal cells (Fig. 3d). The statistical results showed that the ratio of red light to green light overlapping cells is 52.23 ± 9.805%. These results indicate that AdV-GFP mainly infects ovarian stromal cells and has almost no infectivity to oocytes. Also, these results along with literature 21 suggest that AdV is a better gene carrier for the following ovarian gene therapy experiments.
4. Safety assessment of AdV gene therapy in situ injection of ovary
To further explore the safety of ovarian gene therapy and determine the duration of overexpression in the ovary, we constructed AdV-Sirt1 and AdV-Tgfbr2 for the following in vivo experiments. A schematic diagram of the vector construction is shown in Fig. 4a. The experimental flowchart is presented in Fig. 4b. The mouse ovaries of the four groups were injected with AdV-GFP, AdV-Sirt1, AdV-Tgfbr2, and PBS in situ, and then tested at four time points (2, 4, 6, and 8 weeks). By the end of the observation (Day 61), there was no significant difference in body weight among all of the four groups. Also, the hair color and vitality of the mice displayed no changes among the four groups. The levels of alanine aminotransferase (ALT), aspartate aminotransferase (AST), and creatine kinase (CK) were not significantly altered in each time point compared with the PBS group (Fig. 4c-e). However, blood urea nitrogen (BUN) and creatinine (CREA) in the AdV-Sirt1 and AdV-Tgfbr2 groups were higher than those in the PBS group at different time points (Fig. 4f-g), though the levels of the inflammation marker TNF-α in the other groups were not significantly different at various time points compared with the PBS group (Fig. 4h). Furthermore, H&E staining analyses of the hearts, livers, spleens, kidneys, and uteruses of the mice showed that there are no apparent pathological changes in these tissues (Fig. S3a). Also, there was no significant difference in organs’ weights among all of the 8-week groups (Fig. S3b). By calculating the organ index, we found that there is a decrease in liver index in the AdV-Tgfbr2 group (Fig. S3c). The changes in the liver index and the level of CREA in the AdV-Tgfbr2 group might be due to the influence of Tgfbr2 or other yet unknown factors, which requires further investigation.
To test whether ovarian gene therapy is toxic to the ovaries, we also tested the ovaries. There was no significant difference in the ovarian weight at each time point. AdV-Sirt1 and AdV-Tgfbr2 groups showed an increasing trend at 8 weeks, which may be related to the increase in body weight over time (Fig. S3d). Further calculations revealed that the ovarian index of each group did not change significantly at different time points. However, only in the AdV-Tgfbr2 group, the 4-week group showed a significant decrease compared with the 2-week group, but it was compensated in the subsequent time (Fig. S3e). The estrous cycle of the mice in each group was monitored one week after ovarian injection in situ (Fig. 4i). The results showed that there is no significant difference in the ratio of regular and irregular estrous cycles among the groups (Fig. 4j). In conclusion, AdV-Sirt1 and AdV-Tgfbr2 appear to be safe agents because no significant damage to ovarian function in mice.
We also assessed the expression of Sirt1 and Tgfbr2 in mouse ovarian tissues by extracting RNAs from of ovarian tissues in each group and conducting qPCR analysis at each time point. As a result, the mRNA expression of Sirt1 increased and lasted 2 weeks and 4 weeks after injection of AdV-Sirt1, but there was no significant difference by 6-8 weeks (Fig. 5a). The mRNA expression of Tgfbr2 increased and lasted 6 weeks after injection of AdV-Tgfbr2, but not by week 8 (Fig. 5b). This was confirmed by WB analysis of their protein levels (Fig. 5c-f). As a control, we also detected GFP expression in mouse ovaries after injection of AdV-GFP by IHC analysis. As a result, the expression of GFP increased from 2 to 6 weeks and gradually decreased by 8 weeks (Fig. 5g-h and S4a-b). In summary, in situ injection of the AdV vector into the ovary can maintain the overexpression effect of target genes for 4-6 weeks.
5. The protective effects of adenovirus mediated Sirt1 and Tgfbr2 gene therapy on Dox-induced ovarian damage
Next, we determined whether in vivo restitution of Sirt1 and Tgfbr2 could repair or improve ovarian damage caused by Dox. As shown in Fig. 6a, mice were treated with a single dose of AdV-Sirt1, AdV-Tgfbr2, AdV-Sirt1 combined with AdV-Tgfbr2 or AdV-GFP control, and Dox ( the model of ovarian damage was induced by the intraperitoneal injection of doxorubicin (Dox, 10 mg kg-1), or N.S. control was administered intraperitoneally to the animals one week later for studying their effects on ovarian function. The animals’ body weights were monitored after the intervention (Fig. S5a). Compared with their weights before surgery (day 0), the animals’ body weights in each group on the first postoperative day (day 1) decreased significantly, but gradually recovered from day 2. By day 38, their body weights in the Dox, S+Dox, T+Dox, and ST+Dox groups showed a decreasing trend, compared with that of the N.S. group. The survival rate of T+Dox was 90%, and that of ST+Dox was 88.89%, showing no statistical difference when compared to the N.S., Dox, and S+Dox groups (Fig. 6b). The ovarian endocrine function and the estrous cycle of mice were monitored for 2 weeks. Interestingly, the regularity of the estrous cycle in the Dox group was 30% while the regularity of the Sirt1+Dox group increased to 44%, and the ratio of the ST+Dox group was increased to 50% (Fig. 6c and S5b). Also, sex hormone levels, such as the levels of estrogen, progesterone, and AMH, in the ST+Dox group increased significantly, compared with the Dox group. Moreover, the levels of estrogen and AMH in the T+Dox group tended to increase. There was no significant difference in FSH levels among the five groups, but the T+Dox and ST+Dox groups showed a downward trend (Fig. 6d-g).
Compared with the Dox group, the ovarian weight of the mice in the T+Dox and ST+Dox groups were significantly increased (Fig. 6i), and the ovary index of S+Dox, T+Dox, and ST+Dox groups also increased (Fig. 6j). H&E staining of ovarian sections was performed to detect mice ovarian reserve with representative images as shown in Fig. 6h. The number of ATFs decreased in the S+Dox, T+Dox, and ST+Dox groups by follicle counting in comparison with the Dox group, (Fig. 6k). Follicle ratio results showed that compared with the Dox group, the proportion of growing follicles in the T+Dox group appeared to incline (Fig. S5d). These results suggest that AdV-Tgfbr2 and AdV-Sirt1+AdV-Tgfbr2 could improve the recovery of ovarian endocrine and reserve function after Dox treatment.
Further, we tested whether the gene therapy could improve the reproductive function of the ovary after Dox treatment. Randomly selected mice from each group were caged with wild-type male mice for 10 days, and the pregnancy rate of the female mice was tested to evaluate reproductive function. The pregnancy rates of the S+Dox, T+Dox, and ST+Dox female mice increased, compared with the Dox group (Fig. 7a). The average litter size of post-delivery mice (Fig. 7b) and all mated mice of T+Dox and ST+Dox groups (Fig. 7c) increased compared with the Dox group. There was no significant statistical difference in the average birth weight per litter (Fig. S6a) and the ratio of male to female (Fig. S6b) in each group. As shown in the representative images of the offspring of each group (Fig. 7d), all of the pups looked healthy without any apparent birth defects. These data indicate that AdV-Tgfbr2 alone or combined with AdV-Sirt1 via in vivo restitution could rescue ovarian endocrine and reproductive damage caused by Dox.
To verify whether the offspring of each group of female mice carry the AdV gene, we extracted the genomic DNA of each group of offspring mice and designed primers for the E4 region of AdV. The results of agarose gel electrophoresis after PCR showed that Gapdh in each group had a positive band (except for the negative control), and there was no positive band in AdV-E4 in each group (except the positive control) (Fig. 7e). This result indicates that the ovarian in situ injection of the AdV vector does not pass the genomic sequence of this vector to the offspring through vertical transmission. This suggests that the approach is safe to the offspring. Also, we simultaneously detected the primordial follicles and primary follicles of the ovaries in the offspring of PND3 and PND7. As shown in the representative H&E images in Fig. 7f, the number and proportion of follicles in PND3 and PND7 mice as quantified and presented in Fig. 7g and 7h were not significantly different among all of the groups, suggesting that ovarian in situ gene therapy does not affect the ovarian reserve of the offspring.
6. Signature transcriptome in response to overexpression of Sirt1 and Tgfbr2 likely underlying the protection of ovaries from doxorubicin-induced damage
To gain molecular insights into how Sirt1 and Tgfbr2 might improve the Dox-induced ovarian damage, we carried out transcriptome sequencing in the Dox and ST+ groups, which had a better rescuing effect. The FPKM value was obtained through quality control, mapping, and quantification. Since the gene expression value of RNA-Seq is usually expressed by FPKM, we first corrected the data and then visualized the distribution of gene expression levels before (Fig. 8a) and after (Fig. 8b) correction through box plots, respectively. In addition, principal component analysis (PCA) was also used to assess the differences between groups and sample duplication within groups, and to perform dimensionality reduction and principal component analysis on genetic variables. The PC1 coordinate axis is the method with the largest variance in data. The PC2 coordinate axis selects the direction orthogonal to the PC1 coordinate axis and has the second largest variance. As shown in Fig. 8c, the samples were divided into two groups after PCA analysis. In order to show the correlation of gene expression between samples, Pearson correlation calculation was conducted on all gene expression levels between two samples, and the results were presented in the form of a heat map (Fig. 8d), indicating the sample differences between groups, which was used for downstream differential expressed gene (DEG) analysis. We used the Deseq2 package 22 to analyze the DEGs of the ST+Dox and Dox groups. Genes with FDR (P.adj) < 0.05, and log2 foldchange > 1 were extracted as DEGs. Furthermore, a heatmap plot was drawn and clustered as demonstrated in Fig. 8e. In the volcano map, some of the DEGs of interest were marked, and it was found that genes related to the fibrosis process, such as Mmp2, Mmp12, Mmp19, Col22a1, and Timp1, were significantly downregulated. Reproduction-related genes, such as Fshr, were significantly upregulated (Fig. 8f).
Table 1. GO enrichment analysis of DEGs
Category
|
Term
|
Description
|
Count
|
%
|
P value
|
BP
|
GO:0008406
|
gonad development
|
13
|
7.26%
|
1.25E-08
|
BP
|
GO:0048608
|
reproductive structure development
|
15
|
8.38%
|
2.27E-06
|
BP
|
GO:0061458
|
reproductive system development
|
15
|
8.38%
|
2.53E-06
|
BP
|
GO:0043062
|
extracellular structure organization
|
13
|
7.26%
|
1.10E-06
|
BP
|
GO:0030198
|
extracellular matrix organization
|
11
|
6.14%
|
8.81E-06
|
BP
|
GO:0046626
|
regulation of insulin receptor signaling pathway
|
6
|
3.35%
|
2.51E-05
|
BP
|
GO:1900076
|
regulation of cellular response to insulin stimulus
|
6
|
3.35%
|
6.62E-05
|
CC
|
GO:0031012
|
extracellular matrix
|
18
|
9.90%
|
3.58E-08
|
CC
|
GO:0062023
|
collagen-containing extracellular matrix
|
12
|
6.59%
|
2.57E-05
|
CC
|
GO:0030017
|
sarcomere
|
8
|
4.40%
|
0.000119391
|
CC
|
GO:0019898
|
extrinsic component of membrane
|
10
|
5.49%
|
0.000181577
|
CC
|
GO:0044449
|
contractile fiber part
|
8
|
4.40%
|
0.00019496
|
MF
|
GO:0008237
|
metallopeptidase activity
|
9
|
5.02%
|
1.08E-05
|
MF
|
GO:0004175
|
endopeptidase activity
|
13
|
7.26%
|
5.28E-05
|
MF
|
GO:0004222
|
metalloendopeptidase activity
|
6
|
3.35%
|
0.000142372
|
MF
|
GO:0046935
|
1-phosphatidylinositol-3-kinase regulator activity
|
3
|
1.68%
|
0.000164419
|
MF
|
GO:0005201
|
extracellular matrix structural constituent
|
6
|
3.35%
|
0.00086498
|
GO functional enrichment analysis was performed using the ClusterProfiler package 23. The results of gene enrichment showed that in the biological process (BP) category, DEGs were significantly enriched in biological processes, such as “reproductive structure development”, “reproductive system development”, and “gonad development”. The DEGs in the cellular component (CC) category were significantly enriched in “extracellular matrix”, “collagen-containing extracellular matrix”, collagen-containing extracellular matrix, and other cytological components. The DEGs in the molecular function (MF) category were enriched in “metallopeptidase activity” and “metalloendopeptidase activity” (Fig. 8g and Table 1). In addition, gene set enrichment analysis (GSEA) results indicated that the BP terms, such as “ovarian follicle development” and “response to gonadotropin,” in ST+Dox group were up-regulated than Dox group, and the corresponding genes of BP terms, such as Inha, Inhba, Foxl2 and Amh, were ranked higher in the gene set (Fig. 8h). At the same time, the GSEA of KEGG pathway revealed that the pathways, such as “Oocyte meosis,” “Progesterone-medieated oocyte maturation,” and “Insulin signaling pathway,” in the ST+Dox group are also higher in the gene rank list. The corresponding genes Akt1 and Igf1r were ranked higher in the pathway (Fig. 8i).
In addition, we used the differentially expressed pathways as the analysis object to conduct a differential pathway variation analysis (Gene set variation analysis, GSVA) 24 based on the hallmark pathway gene set provided by the molecular signature database (version 7.2) 25. GSVA analysis showed that the term “ESTRONGEN_RESPONSE_LATE” has a higher score and that other terms, such as inflammation-related responses such as “INTERFERON_GAMMA_RESPONSE,” and “INFLAMMATORY_RESPONSE,” also have higher GSVA scores (Fig. S7a). Additionally, using the String website (https://string-db.org/) and Cytoscape (Windows 3.8.2 version), a protein-protein interaction (PPI) network of DEGs was constructed and the degree value was filtered through the Centiscape 2.2 plug-in. Genes with a degree value > 5.0 were selected as hub genes, and Timp2 and Mmp2 were identified as hub genes (Fig. S7b).