The oestrous cycle stage affects mammary tumour sensitivity to chemotherapy

Patient samples

All retrospective medical data/biospecimen studies at the Netherlands Cancer Institute have been executed pursuant to Dutch legislation and international standards. Before 25 May 2018, national legislation on data protection applied, as well as the International Guideline on Good Clinical Practice. From 25 May 2018 we also adhered to the General Data Protection Regulation of the European Union. Within this framework, patients are informed and have always had the opportunity to object or actively consent to the (continued) use of their personal data and biospecimens in research. Hence, the procedures comply both with (inter)national legislative and ethical standards. All retrospective medical data/biospecimens provided by LUMC were part of the DIRECT study (NCT02126449 (ref. ⁴⁵)), were conducted in accordance with the Declaration of Helsinki (October 2013) and were approved by the Ethics Committee of LUMC in agreement with Dutch law for medical research involving human subjects.

Assessment of HER2 status in this study followed established guidelines, with scoring based on both immunohistochemistry (IHC) and in situ hybridization techniques. Immunohistochemistry results were categorized as 0/1+ (HER2-negative), 2+ (equivocal; considered positive if amplification was detected with in situ hybridization) or 3+ (HER2-positive). Assessment of oestrogen receptor was performed according to Dutch guidelines, with scores of 10% or above considered positive, and those below 10% negative. Pathological complete response (pCR) scoring adhered to local standard guidelines. For one patient, pCR was determined based on radiological images because surgery was not performed due to pCR having been achieved. For measurement of tumour reduction from baseline, we compared pretreatment radiological images with post-treatment residual disease as assessed by pathological evaluation.

Experimental model and subject data

All mice were adult females, housed under a 12/12 h light/dark cycle and under specific-pathogen-free laboratory conditions and received food and water ad libitum. All experiments were approved and performed according to the guidelines of the Animal Welfare Committees of the Royal Dutch Academy for Sciences (Hubrecht Institute), the Netherlands Cancer Institute or KU Leuven. Sample size was not determined a priori. Mice were randomly assigned to experimental groups. All experiments were performed in a blinded manner, except for tumour volume measurement, which was performed by the same investigator who administered chemotherapy treatment, making it impossible to work in a blinded manner.

MMTV-PyMT²⁶ and MMTV-Cre⁶¹ mice were purchased from Jackson Laboratory. E-cad-mCFP mice⁶² were a gift from H. Clevers, R26-loxP-stop-loxP-YFP (R26R-YFP) mice a gift from J. Deschamps and MMTV-Wnt1 (ref. ³⁰) mice a gift from J. Hilkens. MMTV-PyMT;R26R-Confetti;R26-CreERT² mice^62,63, of a mixed BL6/FVB genetic background, were used to label and trace single cells by IVM. Following tumour development, mice were intraperitoneally injected with tamoxifen (Sigma-Aldrich; 1.5 mg 25 g^−1, diluted in sunflower oil) to activate Cre recombinase and induce colour randomization of the confetti cassette. MMTV–Wnt1;R26R-Confetti:R26-CreERT² mice^62,63, of a mixed BL6/FVB genetic background, were used to isolate tumour pieces, which were transplanted into NOD.Cg-Prkdcscid Il2rgtm1Wjl/SzJ (NSG) female mice (8–12 weeks of age) Following tumour development, mice were intraperitoneally injected with tamoxifen (Sigma-Aldrich; 1.5 mg 25 g^−1, diluted in sunflower oil) to activate Cre recombinase and induce colour randomization of the confetti cassette. Staining and treatment experiments in transgenic mice were performed on MMTV-PyMT;MMTV-Cre;R26-LSL-YFP;E-cad-mCFP and MMTV-Wnt1;MMTV-Cre; R26-LSL-YFP;E-cad-mCFP mice of a pure FVB genetic background.

For staining and treatment experiments in transplanted models, tumours from three independent, treatment-naive MMTV-PyMT;MMTV-Cre;R26-LSL-YFP;E-cad-mCFP donors were harvested. Mammary tumours were minced and enzymatically digested by gentle shaking for 30 min at 37 C in digestion mix (0.2% trypsin from bovine pancreas; Sigma) and 0.2% collagenase A (Roche)). The digested tumours were spun down and cell fragments embedded in basal membrane extract (BME) (RGF BME type 2 PathClear). Mammary tumour organoid medium contained DMEM/F12 GlutaMAX (GIBCO), 2% B27 (Invitrogen) and 10 ng ml^−1 fibroblast growth factor. Brca1^{−/−}Trp53^{−/−} organoids were a gift from J. Jonkers. Both MMTV-PyMT and Brca1^{−/−}Trp53^{−/−} organoids were orthotopically transplanted into both fourth mammary glands of 8–12-week-old FVB/NRj female mice (Janvier Labs).

For transplantation of MMTV-PyMT organoids, 250,000 single cells were plated 3 days before transplantation. On the day of transplantation, BME was dissolved by mechanical disruption and cells dissolved in 100 μl of BME type 2 (RGF BME type 2 PathClear)/PBS. For transplantation of Brca1^{−/−}Trp53^{−/−} organoids, 10.000 single cells were transplanted in both fourth mammary glands of FVB/NRj mice.

Staging of mice

For determination of the oestrous stage of mice, a vaginal smear was collected and analysed as described in refs. ^24,46,64. In short, the vagina was flushed with 50 µl of PBS which was then transferred to a glass slide. Following air drying, the slide was stained with 0.01% crystal violet and oestrus stage determined by examination of cytological characteristics using a light microscope. Mice were categorized as regularly cycling when they showed cycles of length 4–6 days³⁷.

Ovariectomy

In a subgroup of mice, ovariectomy was performed when mammary tumours of volume 50 mm³ could be palpated, after which mice were allowed to recover for at least 7 days before IVM or treatment.

Chemotherapeutic treatment of mice

When tumours had reached a cumulative volume of 500–750 mm³ following short-term treatment, or either 250–450 mm³ (transgenic MMTV–PyMT model) or 50–200 mm³ (transplanted models) for long-term treatment, mice were treated with either vehicle (PBS) or chemotherapeutic—either doxorubicin hydrochloride (5 mg kg^−1; catalogue no. D1515, Sigma-Aldrich) or cyclophosphamide monohydrate (100 mg kg^−1; catalogue no. C7397, Sigma-Aldrich)—once per week by intravenous injection. Treatment cycles were continued for a maximum of 5 weeks or until the humane endpoint was reached (cumulative tumour load over 750 mm³ or single tumour over 350 mm³). Tumour diameter was measured three times per week by calipers. Following the last treatment cycle, primary tumours and lungs were collected from these mice for further analyses.

Depletion experiments

When mice approached a cumulative tumour volume of 500–750 mm³, they began receiving depletion antibodies at least 1 week before chemotherapeutic treatment and this continued until the end of the experiment. Individual mice were treated every week with either 60 mg kg^−1 of either InVivoMAb anti-mouse CSF1R (CD115) monoclonal antibody (BioXCell, clone AFS98, catalogue no. BE0213) or InVivoMAb rat IgG2a isotype control (BioXCell, clone 2A3, catalogue no. BE0089), or received InVivoMAb anti-mouse CD4 (BioXCell, clone GK1.5, catalogue no. BE003-1) and InVivoMAb anti-mouse CD8 (BioXCell, clone YTS 169.4, catalogue no. BP0117), three times per week, every other day, with a first dose of 400 µg followed by 200 µg per antibody.

Mammary imaging window implantation and IVM

To follow in vivo tumour clone dynamics, a mammary imaging window was inserted into tumour-bearing mice 2–3 days following induction of lineage tracing. Mice were anaesthetized using a mixture of 2% isoflurane and compressed air, with surgery performed under aseptic conditions. Before surgery, the skin overlying the tumour was shaved and the skin disinfected using 70% ethanol. An incision was made through the skin overlying the tumour and an imaging window inserted^65,66. The imaging window was secured using a non-absorbable, non-woven, purse‐string suture (4‐0 prolene suture) in the skin. For imaging, mice were sedated using isoflurane inhalation anaesthesia (roughly 2.0% isoflurane/compressed air mixture) and received 200 µl of sterile PBS by subcutaneous injection to prevent dehydration. Mice were placed in a custom-designed imaging box on the microscope stage and maintained under constant anaesthesia under the microscope within a temperature-adjusted climate chamber (34.5 °C). Following each imaging session, mice were allowed to recover on a heat mat. Imaging was performed on either an inverted Leica TCS SP5 AOBS multiphoton microscope with a chameleon Ti:Sapphire pumped Optical Parametric Oscillator (Coherent Inc.; www.coherent.com) or a Leica SP8-DIVE confocal microscope equipped with an Insight X3 dual-beam pulsed layer (Spectra Physics; www.spectra-physics.com), equipped with non-descanned detectors. CFP was excited at a wavelength of 840 nm, and yellow fluorescent protein (YFP), green flourescent protein and red flourescent protein at a wavelength of 960 nm. All images were in 12-bit and acquired with a ×25 (HCX IRAPO, numerical aperture 0.95, working distance 2.5 mm) water objective with a free working distance of 2.40 mm. Three-dimensional tile scans of large tumour areas were taken at the indicated time points (at least 5.0 × 5.0 × 0.5 mm³ (x, y, z)) with z-steps of 5–10 µm. Tile scans were collected at regular intervals over time periods of up to 5 weeks following induction of lineage tracing. The same imaging fields were followed during subsequent imaging sessions, using the imaging coordinates of the first imaging session.

Postprocessing and analysis of IVM data

For analysis of three-dimensional tile scans, all intravital images were stitched and processed using Leica Application Suite X software (Leica Microsystems) and ImageJ software (https://imagej.net). Clones were manually annotated in the images, and clone size (number of cells) counted in three dimensions throughout the z-stacks. By following the same clones in our intravital images, we quantified individual clone sizes over time. The fraction of clones that behaved in a synchronized manner with phases of growth, when clones become larger, followed by a phase of regression during which the clones become smaller, was denoted as alternating. To distinguish synchronized clones from non-synchronized in quantitative terms, we identified as synchronized clones those that showed clear, regular alternating phases of growth and regression with the same periodicity, similar to a sinusoidal wave, over multiple consecutive cycles throughout the imaging sessions. By contrast, non-synchronized clones did not exhibit this regular pattern.

Histochemistry and immunohistopathology

Mouse tissues were formalin fixed and paraffin embedded. Haematoxylin and eosin staining was performed on 2 µm sections and IHC staining on 4 µm sections using routine procedures. For IHC staining, antigen retrieval was performed with Tris/EDTA (pH 9.0) (Tris: Sigma, catalogue no. 252859; EDTA: Sigma, catalogue no. EDS) for ER-α (Invitrogen, clone 6F11, catalogue no. MA1-27107, 1:100), CD4 (Invitrogen, clone 4SM95, catalogue no. 14-9766-82, 1:1,000), CD8 (Invitrogen, clone 4SM15, catalogue no. 14-0808-82, 1:2,000), CD31 (Cell Signaling, catalogue no. 77699S, 1:100), F4/80 (Cell Signaling, catalogue no. 70076S, 1:1,000) or citrate buffer (pH 6.0) for progesterone receptor (Fisher Scientific, clone SP2, catalogue no. RM-9102-S1, 1:300). Sections were incubated with primary antibodies overnight at 4 °C. For CD4 and CD8 staining, additional labelling was performed by incubation of goat anti-rat biotin (SouthernBiotech, catalogue no. 3052-08, 1:150) antibody for 30 min at room temperature. Following washing, binding of primary antibody was visualized dependent on species using either the EnVision+ HRP Labelled Polymer anti-rabbit/mouse system (Dako, catalogue nos. K400, K4003 and K4001) or Streptavidin/HRP (Dako, catalogue no. P0397) and the Liquid DAB+ Substrate Chromogen System (Dako, catalogue no. K3468); counterstaining with haaematoxylin was then performed. Slides were digitally processed using a PANNORAMIC 1000 whole-slide scanner (3DHISTECH) and images captured with SlideViewer 2.7 (3DHISTECH). For determination of positive cells, IHC staining was quantified using QuPath 0.4.4 (GitHub) with atomized classifiers.

Immunostaining

Tumour samples were fixed in periodate/lysine/4% paraformaldehyde buffer overnight at 4 °C, incubated in 30% sucrose overnight at 4 °C and embedded in Tissue Freezing medium (Leica Biosystems). Tumours were cryosectioned and immunostaining performed on 10 µm sections. For this, sections were hydrated in PBS for 10 min at room temperature and subsequently blocked and permeabilized for 1 h with 0.5% TritonX/5% normal goat serum (NGS) in PBS. The primary antibody used was anti-PHH3 (Millipore, catalogue no. 06-570, 1:500); the antibody was diluted in 0.1% TritonX/5% NGS in PBS, and tissues were sectioned and stained overnight at 4 °C. Following three washes in PBS, the secondary antibody (donkey anti-rabbit AF568; Invitrogen, catalogue no. A10042, 1:1,000) was incubated for 1 h at room temperature and nuclei stained with TO-PRO-3 (Invitrogen, catalogue no. T3605, 1:5,000). Following three 10 min washes in PBS, stained sections were mounted using VectaShield. All stainings were imaged with an inverted Leica TCS SP8 confocal microscope. Fluorophores were excited as follows: AF594 at 561 nm, YFP at 514 nm and TO-PRO-3 at 633 nm. YFP was collected at 519–555 nm, Alexa-568 at 575–630 nm and TO-PRO-3 at 644–694 nm. All images were collected in 12-bit with a ×25 water-immersion objective (HC FLUOTAR L, numerical aperture 0.95, working distance 2.4 mm). For determination of positive cells, staining was quantified using Fiji/ImageJ v.1.49k and Excel 2016.

EdU detection

EdU was injected intraperitoneally 4 h before mice were killed (1 mg per animal, 5 mg ml^−1 stock in PBS; Sigma, catalogue no. 900584). EdU staining was performed on paraffin sections of thickness 10 µm; sections were deparaffinized, and hot target retrieval performed in citrate buffer pH 6.0. EdU was detected by incubation with 100 mM Tris pH 8.5, 1 mM CuSO₄, 100 mM ascorbic acid and 10 μM AlexaFluor-488 azide (Invitrogen, catalogue no. A10266) for 30 min at room temperature. Subsequently, DNA counterstaining was performed with TO-PRO-3 (Invitrogen, catalogue no. T3605, 1:5,000) in 0.1% TritonX/5% NGS in PBS for 1 h at room temperature. Following three 10 min washes in PBS, stained sections were mounted using VectaShield. All staining was imaged with an inverted Leica TCS SP8 confocal microscopes. Fluorophores were excited as follows: AF488 at 488 nm and TO-PRO-3 at 633 nm; AF488 was collected at 492–530 nm and TO-PRO-3 at 650–700 nm. All images were collected in 12-bit with a ×25 water-immersion objective (HC FLUOTAR L numerical aperture 0.95 W, VISIR 0.17, FWD 2.4 mm). For determination of positive cells, staining was quantified using semiautomated macros in Fiji/ImageJ v.1.49k and Excel 2016.

Flow-cytometric analysis of dying cells

Mammary tumours were collected separately and minced on ice using sterile scalpels, followed by digestion for depletion experiments in 25 μg ml^−1 DNase I (Roche) and 5 Wünsch units TH Liberase ml^−1 (Roche) in PBS at 37 C for 35 min, or in 25 μg ml^−1 DNase I (Roche) and 3 mg ml^−1 collagenase A (Sigma) in PBS at 37 °C for 1 h. The digestion mix was filtered through a 70 µm filter (BD Falcon) while adding DMEM/F12 + GlutaMAX, followed by spinning down for 4 min at 500 relative centrifugal force at 4 °C and resuspension of pellets in 5 mM EDTA/PBS. Cells were washed once in 5 mM EDTA/PBS and centrifuged (4 min at 500 relative centrifugal force at room temperature) before proceeding with antibody labelling. Tumour cells were blocked in fluorescent activated cell-sorting buffer supplied with 20% normal goat serum (Gibco) for 10 min on ice, before labelling with one of the following antibody combinations for depletion experiments: (1) E-cad-eFluor660 (catalogue no. DECMA-1, eBioscience, 1:200), biotin-conjugated anti-mouse CD41 clone eBioMWReg30 (eBioscience, catalogue no. 13-0411-821, 1:200) and anti-mouse CD45 clone 30-F11 (eBioscience, catalogue no. 13-0451-85, 1:200); (2) E-cad-eFluor660 (catalogue no. DECMA-1, eBioscience, 1:200); and (3) anti-mouse CD45-BUV395 clone 30-F11 (BD Bioscience, catalogue no. 564279, 1:200). Secondary labelling was performed using streptavidin-conjugated PerCP (BioLegend, catalogue no. 405213, 1:200). Dead cells were stained using the Apoptosis detection kit PE (Invitrogen, catalogue no. 88-8102-74) according to the manufacturer’s protocol. For depletion experiments no secondary labelling was performed and, following washing, cells were immediately stained with PI and analysed on a Symphony A5 (BD Biosciences) then washed once in 5 nM EDTA/PBS and analysed on a Symphony A5 (BD Biosciences). A broad forward scatter/side scatter (FSC/SSC) gate was followed by gates excluding doublets. Immune cells and megakaryocytes were then excluded, based on staining for CD41 and CD45, in a dump channel. Tumour cells were selected according to YFP positivity and further stringently gated for the presence of the cell death marker annexin and PI, or for PI only for depletion experiments. Data were manually analysed with FlowJo v.10.6.2. Analysis of staining and fluorescent activated cell-sorted samples was pseudomized by number coding.

RNA isolation, complementary DNA preparation and qPCR

RNA was isolated using Trizol reagent (Invitrogen Life Technologies) according to the manufacturer’s protocol. Purity and amount of RNA isolated were analysed using a Nanodrop spectrophotometer. Complementary DNA was prepared using the High-Capacity cDNA Reverse Transcription Kit (Applied Biosystems) according to the manufacturer’s protocol. Sequences of used primers can be found below. Quantitative PCR (qPCR) was performed using Power SYBR Green PCR Master Mix (Applied Biosystems). Thermal cycle conditions used for all qPCR reactions were as follows: 5 min at 95 °C, followed by 40 cycles of denaturation for 30 s at 95 °C, annealing for 30 s at 60 °C and extension for 1 min at 72 °C. PCR reactions were concluded with incubation for 10 min at 72 °C to complete the extension of all synthesized products.

Progesterone assay

To 250 µl serum or calibrator samples was added 10 µl of deuterated internal standard (progesterone-d9, CDN Isotopes). Subsequently, progesterone was extracted with 1 ml of methyl tert-butyl ether and dried using a SpeedVac concentrator. Dried samples were reconstituted in 100 µl of injection solution (methanol:water 2:3, v:v). Thereafter, samples were centrifuged for 5.0 min at 18,213g and 50 µl was injected into a Nexera SIL30ACMP (Shimadzu) autosampler. Chromatographic separation was achieved using a Kinetex EVO 1.7 µm C18 column (2.1 mm, internal diameter 50 mm) (Phenomenex). A gradient protocol of two mobile phases, containing water with 0.1% formic acid and 2 mM ammonium acetate, and methanol, was applied at a flow rate of 0.6 ml min^−1. The gradient started at 50% methanol, which was gradually increased to 75% over 2.3 min. Next, the column was washed with 100% methanol for 0.5 min before returning to the starting condition of 50% methanol, for 0.7 min. The assay had a total run time of 3.5 min. Tandem mass spectrometry analysis was performed with a QTRAP6500+ instrument (Sciex) operated in positive electrospray ionization mode (600 °C) and multiple-reaction monitoring mode. For quantitation of progesterone, mass:charge ratio, 315.1 → 97.0 (progesterone) and 324.2 → 100.0 (progesterone-d9) were monitored. The assay is standardized against the NIST SRM 971 standard. In two serum pools containing 1.1 and 27.5 nmol l^−1, total coefficient of variation was 9.5 and 7.5%, respectively. The lower limit of quantitation was determined at 0.96 nmol l^−1 (coefficient of variation 8.3%) in a serum pool containing low progesterone levels. Analysis of patient samples (sera) was performed blinded, and linked to tumour volume measurements and other clinical parameters only at a later stage.

Statistics and reproducibility

Statistical analyses were performed using R v.4.4.2 (R Development Core Team and the R Foundation for Statistical Computing) by integration of software from open-source packages. For further details on statistics see Supplementary File 1. The level of statistical significance was set at ^#P < 0.1, *P < 0.05, **P < 0.01, ***P < 0.001. For all violin plots, thicker solid centre lines represent the median, thinner solid lines the 25th and 75th percentiles. Statistical analyses of longitudinal mouse data were performed using R v.4.4.2 (R Development Core Team and the R Foundation for Statistical Computing) by integrating software from open-source packages, including nlme⁶⁷, JMbayes2 (ref. ⁶⁸) and packages from tidyverse⁶⁹, including dplyr, tidyr and ggplot2, and were analysed as follows. Both tumour volume over time and time to event (either end of study or death) were recorded for each individual animal. To test for survival, survival regression (Cox) models were used to test whether survival probability between groups was different. For longitudinal tumour measurements, the mixed-effects model⁷⁰ was used. Regression splines (natural cubic splines with 2 or 3 degrees of freedom, depending on the case) were used to fit a nonlinear tumour growth function per animal along time, taking into account all separate tumours for each individual animal separately. This model takes the correlated structure of the data into consideration by inclusion of individual animals as a grouping factor, and tumour (within the animal) as a subgrouping factor. Furthermore, the effects of time and treatment were considered as random and fixed effects, respectively. A joint model was used to combine survival and tumour volume measurements, to determine whether the responses together were different between treatment groups. Splines were again used to represent the effect of time. For this model, multiple tumours per animal had their growth curves averaged. For further details on these statistics, see Supplementary File 2.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.