Endometrial biopsies from 161 patients (70 control, 54 type I EC and 37 type II EC) were obtained from patients attending general gynecology clinics or postmenopausal bleeding (PMB) clinics within the Swansea Bay and Cwm Taf Morgannwg University Health Boards (SBUHB and CTMUHB). Postmenopausal patients that presented with bleeding or abnormally thickened endometrium (over 4 mm), identified incidentally in imaging investigations (abdominal ultrasound, MRI) performed for other clinical reasons, were included in the study. All patients with PMB or thickened endometrium underwent transvaginal ultrasound and Pipelle endometrial biopsy and hysteroscopy. Patients with cancer diagnosis at Pipelle biopsy underwent staging MRI and were scheduled for hysterectomy and/or bilateral salpingoophorectomy for type I disease, and hysterectomy, bilateral salpingoophorectomy, omentectomy and pelvic node dissection for type II disease. The control group included postmenopausal patients that underwent hysterectomy for vaginal prolapse and PMB patients with normal Pipelle sample and hysteroscopy findings.

Histological evaluation of endometrial samples, cancer diagnosis and staging was confirmed by the pathology department in SBUHB as part of routine clinical care. For endometrial cancer type I tumors included grade 1 and grade 2 endometrioid adenocarcinoma. Endometrial cancer type II included serous, clear cell and mixed adenocarcinoma tumors and high-grade endometrioid tumors (grade 3).

Follow-up time was up to 60 months. Survival was defined as the date from confirmed histological diagnosis after primary surgery to the date of death. Disease-free time was defined as the date from confirmed histological diagnosis after primary surgery to the date of recurrence or last visit (for those in the study for less than 60 months).

Patients that were peri or premenopausal, presenting with abnormal uterine bleeding (menorrhagia, intermenstrual bleed, postcoital bleed, amenorrhea) were excluded from this study. Patients with infection, chronic inflammation, autoimmune disease, endometritis, endometrial hyperplasia, and other cancers were excluded from the study. Ethical approval for immunohistochemistry analysis of FFPE EC patient samples within the study was obtained through Local Research Ethics Committee (reference 07/WMW02/50) for the collection of biopsies from consented EC patients (prospective analysis). Formal written consent was obtained from all patients at the time of recruitment into the study. Patients in control and study groups were matched with regard to body mass index and smoking status.

Data on age, BMI, parity, smoking status, menopausal status, hormone intake of any type, and comorbities were recorded in the study database. Data were also recorded on surgical procedure, histological type and stage, adjuvant treatment (radiotherapy, chemotherapy), follow up, recurrence free period, post-recurrence treatment and overall survival period.

Monoclonal antibodies against RAGE were produced using standard protocols for monoclonal antibody production [21]. Briefly, mice were immunized with keyhole limpet hemocyanin (KLH)-conjugated RAGE, or KLH-conjugated peptides corresponding to amino acids (aa) 198–217 or 327–344 of the RAGE protein. Clones were selected based on a positive ELISA screen using bovine serum albumin (BSA)-conjugated peptides. Post-fusion, individual clones were selected by limiting dilution and clonal expansion to identify genetically stable, antibody producing cells for subsequent antibody production. One clone with affinity for the full RAGE protein (RBGO1), two clones with affinity for aa198–217 (RBGO2 and RBGO3) and one with affinity for aa327–344 (RBGO4) were selected for antibody production. Antibodies were purified from tissue culture medium using protein G affinity purification.

All procedures were performed in accordance with the Animals (Scientific Procedures) Act 1986, and the guidance issued in ‘Responsibility in the case of Animals in Bioscience research: expectations of the major research council and charitable funding bodies.’

Briefly, 6 week-old female athymic nude mice were subcutaneously inoculated with 5 × 10⁵ HEC1A cells. Mice bearing 5 mm in diameter tumours were distributed into three groups of 5 mice each. Mice were treated with control (PBS), RBGO1 ADC (3 mg/kg) or mcF (45 μg/kg, which is equivalent to the drug dose delivered by the ADC) via intravenous injection. Treatments were performed twice weekly for 4 weeks and tumour volumes were measured twice weekly. Tumours and organs: brain, heart, lungs, stomach, pancreas, liver, kidneys, ovaries, uterus, bowel and spleen, were removed following sacrifice. Preparation of formalin-fixed paraffin-embedded samples was performed as previously described using a Ventana machine (Ventana Biotek Solutions, Tucson, AZ, USA) [22].

Statistical analyses were performed using IBM SPSS Statistics 22 with biological replicate as the experimental unit. Initially the data were tested for homogeneity, and log or square root transformed if appropriate. Parametric data were analyzed by analysis of variance (ANOVA) using Dunnett’s pairwise multiple comparison t-test for individual group comparisons. Non-parametric data were analyzed by Kruskal-Wallis followed by Mann Whitney U tests for multiple comparisons. Associations were analyzed using Factorial Logistic Regression. Overall survival and disease-free period was analyzed using Kaplan-Meier survival analysis and where appropriate, curves compared using the Log Rank (Mantel-Cox) test. Correlation within the patient data was determined using Spearman’s Rank Order Correlation. Co-localisation within the internalization experiments was determined using Pearson’s correlation coefficient (PCC) and Manders co-localisation coefficient. Data are presented as mean with standard deviation (SD), p < 0.05 was considered statistically significant, and the number of independent experiments is stated in the figure legends.

Endometrial biopsies from 161 patients (70 control, 54 type I EC and 37 type II EC) were obtained as described in Methods (Patient demographics are shown in Additional file 2: Table S1). Median age at presentation was 57.5 ± 10.3, 67 ± 14.8, or 72 ± 6.0 years, respectively. Mean body mass index (BMI) at presentation was 31.1 ± 7.1, 35.6 ± 11.6, or 31.0 ± 6.2, respectively. Within the patient cohort age was a significant determining factor for EC (Factorial Logistic Regression = Type I EC: LR χ² = 9.836, p = 0.003; Type II EC: LR χ² = 25.229, p < 0.001), but BMI, smoking, parity and diabetes were not.

RAGE expression was apparent in the stromal cells of the endometrium and was also detected in glandular/luminal epithelium. Expression within control endometrium was limited (Fig. 1a), whilst within endometrial biopsies from type I (Fig. 1b) or type II (Fig. 1c) EC, significant RAGE expression was observed. Semi-quantitative analysis (H-score) of RAGE expression in each of the patient groups showed a significant increase in RAGE expression in type I and type II EC compared to control patients (Fig. 1d; p < 0.001). RAGE expression was also signifncantly greater in the type II EC patient group compared to the type I EC group (Fig. 1d; p < 0.05). Additionally, quantification of RAGE mRNA in patient biopsies using quantitative (q) PCR, confirmed that RAGE mRNA expression was also significantly upregulated in type I and type II EC compared to control patients (Fig. 1e; p < 0.001). Kaplan-Meier overall survival analysis over a 5 y period, using number of months of survival following surgery, indicated a significantly reduced survival for type II EC compared to control patients (Fig. 1f: Log Rank (Mantel-Cox) test; p < 0.0001). Furthermore, correlation analysis showed a significant correlation between increased RAGE expression and reduced survival in the type II EC group (Spearman’s Rank Order Correlation: ρ = − 0.3914, p < 0.05, Additional file 3: Figure S1A). Disease recurrence following initial treatment was also common within the type II EC group (60% by 29 months; Fig. 1g). Correlation analysis again showed a significant correlation between increased RAGE expression and a reduced disease-free period (Spearman’s Rank Order Correlation: ρ = − 0.4915, p < 0.01, Additional file 3: Figure S1B). No correlations between RAGE expression and patient age, BMI, smoking, parity or diabetes were apparent in any of the patient groups.Fig1Fig. 1

The receptor for advanced glycation end products (RAGE) is over expressed in endometrial cancer (EC) and associated with reduced survival. RAGE expression was determined by immunohistochemistry in biopsies (n = 67) from healthy patients (a; n = 25) and patients with type I (b; n = 24) or type II (c; n = 18) EC. Biopsies were formalin fixed and paraffin embedded before sectioning and staining with α-RAGE antibody. Representative images were acquired on a Zeiss Axio Imager 2 microscope and analyzed using the ZEN 2012 image analysis software. Scale bars are 50 μm. RAGE expression (H-score) was conducted blind by three of the authors (NT, LM & DG) independently and the mean score for each slide used (d). Kaplan-Meier survival curves were constructed using Graph Pad PRISM 6 based on survival (months) following surgery (e). Within type II EC patients, time to disease recurrence following surgery (months) was monitored (f) and correlated with RAGE expression (g). Biodistribution studies were perfomed in nude athymic mice, which were dosed intravenously with anti-RAGE antibody conjugated to the fluorophore Alexa-750 (3 mg/kg) and sacrificed after either 24 h or 3 wks. Organs were harvested and homogenized and the fluorescence from the tissue slurry measured using a fluorescence microplate reader (Varioskan LUX, ThermoFisher) at wavelength 750 nM. Fluorescence was normalised using the weight of the tissue and values expressed as Fluorescence Intensity per gram of tissue (h & i). Data points for RAGE expression (H-score) represent individual patients (d). Data were analyzed by ANOVA and Dunnett’s pairwise multiple comparison test; values differ from healthy, ***p < 0.001, *p < 0.05

The association between RAGE and EC led us to consider anti-RAGE antibodies as a potential therapeutic approach. To this end, we examined RAGE expression within four EC cell lines (Ishikawa - type I EC; and HEC1A, HEC1B, HEC50 - type II EC) by western blot, confocal microscopy and qPCR. In agreement with patient biopsies, high expression of RAGE was apparent in all four cell lines but absent in primary, non-malignant endometrial cells (Additional file 4: Figure S2A). Quantification of RAGE mRNA revealed the same pattern of expression, with significantly (p < 0.05) more RAGE mRNA present in EC cell lines compared to non-malignant primary endometrial cells (Additional file 4: Figure S2C). Immunofloursecence analysis revealed RAGE localizes at the cell membrane (Additional file 4: Figure S2B), and that the expression of RAGE in type II EC cell lines is higher than the type I EC cell line tested (Additional file 4: Figure S2D & E). Furthermore, we evaluated the expression of RAGE in a variety of human tissues (Brain, breast, kidney, liver, lung, lymph node, pancreas, spleen and uterus, Additional file 5: Figure S3). Western blot analysis confirmed that RAGE expression was absent or very low in these healthy tissues as previously reported [15]. The only exception to this was the lung tissue, which is known to express higher levels of unique RAGE isoforms not found elsewhere [16].

Next we considered the efficacy of anti-RAGE antibodies as an EC therapeutic by exploring the ability of commercially available anti-RAGE antibodies to effect cell killing in vitro. The EC cell lines Ishikawa, HEC1A, HEC1B and HEC50 were exposed to the following anti-RAGE antibodies (1 μg/ml to 100 μg/ml) for periods up to 96 h: N-16 (Santa Cruz Biotechnology, Cat. No sc-8230), A-9 (Santa Cruz Biotechnlogy, Cat. No. sc-365,154), ab37647 (Abcam, Cat. No. ab37647), MAB 5328 (Merck-Millipore, Cat. No. MAB5328), ab3611 (Abcam, Cat. No. ab3611) and MAB11451 (Bio-techne, Cat. No. MAB11451). None of the anti-RAGE antibodies tested had any effect on cell health (Data not shown).

The abcence of in vitro cell killing with anti-RAGE antibody alone led us to explore ADCs targeting RAGE as a potentially more effective therapeutic strategy. Using a small panel of antibodies (RBGO1–4) previously developed and characterized within our laboratory [23], we explored the suitability of ADCs as a therapeutic approach to the treatment of EC. Initially, we conducted in vivo bio-distribution experiments to demonstrate the feasibility of such an approach (Fig. 1). Anti-RAGE antibody raised against the whole RAGE protein (RBGO1), conjugated to the fluorophore Alexa-750 (3 mg/kg), was administered intravenously to female athymic nude mice and mice were sacrificed after either 24 h or 3 wks. Organs were harvested and homogenized with the fluorescence from the tissue slurry measured using a fluorescence microplate reader (Varioskan LUX, ThermoFisher) at wavelength 750 nM. Fluorescence was normalised using the weight of the tissue and values expressed as Fluorescence Intensity per gram of tissue. After 24 h, accumulation of anti-RAGE antibody was apparent primarily in the uterus, ovary and liver. Lower concentrations of antibody were noted in the spleen, lung and kidney and concentrations within other organs were at the limit of detection (Fig. 1h & i). After 3 wks, antibody concentrations within all organs, with the exception of the liver, were at base levels (Fig. 1h).

As previously described, antibodies within the panel were raised against the whole RAGE protein (RBGO1); the C1 domain peptide, aa198–217 (RBGO2 and RBGO3) and the transmembrane proximal region, aa327–344 (RBGO4) [23]. To identify the binding region of the RBGO1 antibody that was raised against the whole RAGE protein, we conducted epitope mapping using a peptide array of 404, 15aa peptides with a 14aa overlap. Arrays were probed with RBGO1 antibody at 1, 10 or 100 μg/ml for 16 h at 4 °C and spot intensities imaged using a LI-COR Odyssey imaging system. Analysis of the spot intensities indicated that the RBGO1 antibody bound with high affinity to a highly conserved region within the V-domain of the RAGE protein.

Key to the development of an efficacious ADC is the internalization of antibody to facilitate cytotoxin delivery to the cell interior. Initial experiments assessed the internalization of our panel of antibodies in HEC1A cancer cells, which have high RAGE expression, following incubation with 1 μg/ml of each antibody over a 1 h period, using confocal microscopy (Fig. 2). Following fixing and permeabilization, staining with secondary antibody alone caused no non-specific binding or background fluorescence (Fig. 2f). HEC1A cells treated with the RBGO4 antibody (Fig. 2e) had the lowest fluorescence after 1 h, with approximately 2.5-fold more fluorescence in cells treated with RBGO2 (Fig. 2c) or RBGO3 (Fig. 2d) antibody, and approximately 7.5-fold more in cells treated with the RBGO1 antibody (Fig. 2b). Quantification of the mean fluorescence as a function of cell area, showed that the quantity of fluorescence in cells treated with the RBGO1 antibody was significantly more (p < 0.001; Fig. 2g) than for the other 3 antibodies. Although this pattern of internalisation matched our previous cell surface staining data [23], we evaluated antibody binding kinetics to whole RAGE protein via surface plasmon resonance (Fig. 2h). These data confirmed that, as previously, binding affinity between RBGO1 and RAGE was high, whilst binding to the other three antibodies was poor, thereby confirming that binding kinetics profile of this batch of antibodies was as previously described [23].Fig2Fig. 2

RBGO1 antibody, targeting the V-region of RAGE is internalized more rapidly than antibodies targeting other regions of the RAGE protein and binds with higher affinity to whole RAGE protein. Schematic diagram of the relative binding positions on the RAGE protein of each of the 4 antibodies tested (a). HEC1A endometrial cancer cells were treated with control medium or medium containing monoclonal antibodies against RAGE at 37 °C for 1 h. After incubation, the cells were washed, fixed and permeabilized. Internalized antibody: RBGO1 (b), RBGO2 (c), RBGO3 (d) or RBGO4 (e), was imaged via fluorescently labelled secondary antibodies and nuclei stained with DAPI. Cells were also incubated only with the secondary antibody as negative control (f). Images were acquired on a Zeiss LSM 710 confocal microscope and analyzed using the Zen 2012 image analysis software. The quantity of internalized antibody was determined using Image J software as a function of cell area (g). For antibody binding kinetics (h), antibodies were captured to a Sensor Chip CM5 via an amine coupled anti-mouse antibody followed by single-cycle kinetics experiments. RBGO1, RBGO2, RBGO3 or RBGO4 antibodies were exposed to whole RAGE protein (2.5 to 200 nM) and data were fitted using a one-to-one Langmuir binding model. Data are expressed as mean (SD) from 3 independent experiments. Data were analyzed by ANOVA and Dunnett’s multiple comparison test. RBGO1 differs from each of the other antibodies, ***p < 0.001.

To further explore the utility of a RAGE targeted ADC, we conjugated each of the four antibodies to the antimitotic agents: monomethyl auristatin E (MMAE), via a lysosomally cleavable dipeptide valine-citrulline (vc) linker; or monomethyl auristatin F (MMAF), via a non-cleavable maleimido caproyl (mc) linker (Additional file 6: Figure S4).

Drug loading of the conjugates was analyzed using a combination of hydrophobic interaction chromatography (HIC) and reverse phase chromatography - Polymer Laboratories Reverse Phase PLRP (Additional file 7: Figure S5). Due to the complex disulfide structure of an IgG2b antibody and potential conjugation site variability, the PLRP chromatographic patterns for the RBGO1 antibody (Additional file 7: Figure S5A, B) were too complex to accurately determine the average drug-antibody ratio (DAR). They did however indicate a good level of drug conjugation and analysis of the traces (Area Under Curve) suggested an average DAR of 3.5. For the RBGO2 (Additional file 7: Figure S5C, D), RBGO3 (Additional file 7: Figure S5E, F) and RBGO4 (Additional file 7: Figure S5G, H) antibodies, which were IgG1, the PLRP traces for both vcE and mcF were clearly discernible, showing drug loading up to a DAR of 4.

We next compared cytotoxicity following the exposure of normal endometrial, HEC1A or Ishikawa cancer cells to antibodies, auristatins or ADCs (Fig. 5a–f). Cells were cultured in the presence of vcE (Fig. 5a–c; 0.01 to 100 μM), mcF (Fig. 5d–f; 0.01 to 100 μM), RBGO1 (Fig. 5a–f; 0.01 to 100 μg/ml), RBGO1-vcE (Fig. 5a–c; 0.01 to 100 μg/ml) or RBGO1-mcF (Fig. 5d–f; 0.01 to 100 μg/ml) for 96 h and cell viability determined using the RealTime-Glo™ MT Cell Viability Assay. Normal endometrial cells were resistant to killing by any of the treatments, with the lethal dose 50 (LD₅₀) concentrations for all treatments being > 100 μM (Fig. 5a, d). LD₅₀ values for HEC1A cells (Fig. 5b, e) were: vcE = 65 μM, mcF and RBGO1 > 100 μM, RBGO1-vcE = 13 μg/ml (≡ to 0.3 μM vcE) and RBGO1-mcF = 5 μg/ml (≡ 0.09 μM mcF). LD₅₀ values for Ishikawa cells (Fig. 5c, f) were: vcE = 4 μM, mcF = 3 μM, RBGO1 > 100 μM, RBGO1-vcE = 11 μg/ml (≡ to 0.2 μM vcE) and RBGO1-mcF = 7 μg/ml (≡ 0.1 μM mcF). These data suggested that RBGO1-ADCs preferentially kill endometrial cancer cells compared to normal endometrial cells. Additionally, in Ishikawa cells, we observed a 20-fold increase in sensitivity to killing for RBGO1-vcE compared to vcE (LD₅₀: 4 → 0.2 μM; Fig. 5l, c); and a 30-fold increase in sensitivity to killing for RBGO1-mcF compared to mcF (LD₅₀: 3 → 0.1 μM; Fig. 4o, f). In HEC1A cells, we observed a more than 200-fold increase in cell sensitivity to killing for RBGO1-vcE compared to vcE (LD₅₀: 66 → 0.3 μM; Fig. 4k, b), which could be due to higher RAGE expression in HEC1A cells compared to Ishikawa cells.Fig5Fig. 5

a–f RBGO1 ADCs preferentially target endometrial cancer cells and increase drug sensitivity by up to 200-fold. Normal endometrial, HEC1A cancer or Ishikawa cancer cells were incubated with control medium or medium containing vcE (a–c; 0.01 to 100 μM) or mcF (d–f; 0.01 to 100 μM), or RBGO1 (a–f), RBGO1 -vcE (a–c) or RBGO1 -mcF (d–f; 0.01 to 100 μg/ml) for 96 h. Cell viability was determined by RealTime-Glo™ MT Cell Viability Assay and lethal dose 50 (LD₅₀) values determined following curve fitting using a 4-parameter logistic model. Drug equivalencies were calculated based on an average DAR of 3. Data are expressed as mean (SD) from 4 independent experiments and normalized to the untreated control to account for cell growth during the period of the experiment. (g–j) RBGO1 ADC is more efficacious in HEC1A EC cells than HER2 ADC. HEC1A EC cells were treated with RBGO1 ADC (g; 0.1 to 5 μg/ml), HER2 ADC (h; 0.1 to 5 μg/ml), or RBGO1 ADC and HER2 ADC (i; 0.1 to 5 μg/ml) for 96 h and cell viability determined at 0, 24, 48, 72 and 96 h using the RealTime-Glo™ MT Cell Viability Assay. Heat map color intensities were based on percent cell viability compared to the untreated control (g–i; see scale in figure). Relative cell viability plots were fitted using a 4-parameter logistic model (j). Data displayed are means of three independent experiments. Data were analyzed by ANOVA and Dunnett’s multiple comparison test. ADCs differs from each other within the same dose, *p < 0.05, **p < 0.01

Having determined the greater efficacy of the RBGO1 ADC compared to the other RAGE targeting ADCs. We evaluated the effectiveness of the RBGO1 ADC against a vcE conjugated ADC targeting the human epidermal growth factor receptor 2 (HER2), since this antigen is already used as a therapeutic target for the ADC, Kadcyla® (Fig. 5g–j). Peptide growth factors frequently implicated in EC include members of the type I receptor tyrosine kinase family, which includes HER2 [27]. Since over expression of HER2 is typically associated with type II EC [28, 29], we used HEC1A cells for our ADC comparison experiments because they are derived from a type II EC tumor and express high levels of HER2 [30]. Cells were cultured in the presence of RBGO1 ADC (Fig. 5g; 0.1 to 5 μg/ml), HER2 ADC (Fig. 5h; 0.1 to 5 μg/ml), or RBGO1 ADC and HER2 ADC (Fig. 5i; 0.1 to 5 μg/ml) for 96 h and cell viability determined at 0, 24, 48, 72 and 96 h using the RealTime-Glo™ MT Cell Viability Assay. The effectiveness of the RBGO1 ADC (Fig. 5g) in HEC1A cells was confirmed, with dosage and time-responses observed. A dose and time effect was also apparent for the HER2 ADC (Fig. 5h), although far less HEC1A cell killing was observed compared to the RBGO1 ADC. A combination therapy approach (Fig. 5i) demonstrated that after 96 h treatment with both ADCs, the contribution to HEC1A cell killing of HER2 ADC was minimal compared to the effect of RBGO1 ADC. This was further confirmed by statistical analysis of each of the doses tested at the 96 h time point (Fig. 5j), which demonstrated significantly more HEC1A cell killing by RBGO1 ADC compared to HER2 ADC (p < 0.05).

To verify the suitability of RBGO1 ADC for full in vivo evaluation, we administered (intravenously) RBGO1 ADC (at 3 mg/kg or 20 mg/kg) to female, athymic mice. Bodyweight was measured at days 3, 6, 8, 13, 17 and 21 and mice were sacrificed at either 24 h or 3 wks following dosing, after which full blood counts and an aspartate aminotransferase (AST) ELISA were performed.

Although bodyweight in animals treated with the high dose of RBGO1 ADC decreased slightly during the study, no significant changes were apparent (Fig. 6a). Full blood counts (Additional file 10: Table S2) indicated that animals treated with RBGO1 ADC (3 mg/kg) had a reduced white blood cell count compared to control animals. Animals treated with RBGO1 ADC (20 mg/kg) had reduced white blood cells and reticulocytes, and an increased platelet count compared to control animals. Serum AST activity was not elevated in any of the treatment groups 24 h after dosing and only in the RBGO1 ADC (20 mg/kg) treatment group 3 wks after dosing (Fig. 6b). However, no signs of distress or ill health were noted during the study in any of the treatment groups, indicating that any toxicity caused by the RBGO1 ADC was minimal even in the high dose treatment group. Furthermore, histological analysis (Fig. 6jc) demonstrated an absence of toxicity across all the treatment groups. A low level of inflammation was noted in the liver, lungs and kidneys of some animals, but as this was observed in all treatment groups including the control it was not a consequence of treatment with RAGE-ADC (Additional file 11: Table S3). Cross-reactivity of the RBGO1 antibody with murine RAGE was confirmed by western blot analysis using the RBGO1 antibody (Fig. 6d). RAGE expression was absent in brain, kidney, spleen, bladder, bowel, stomach, uterus, ovary and heart, with weak expression in the liver and high expression in the lungs noted.Fig6Fig. 6

RBGO1 ADC is not toxic in a murine in vivo model. PBS (Control) or RBGO1 ADC was administered (intravenously) to female, athymic mice aged 5–7 weeks and weighing approximately 28-35 g, at a dose of either 3 mg/kg or 20 mg/kg. Bodyweight a was measured at days 3, 6, 8, 13, 17 and 21 and mice were sacrificed at either 24 h or 3 wks following dosing, after which full blood counts and an aspartate aminotransferase (AST) ELISA were performed (b). Organs were harvested immediately following sacrifice, and formalin fixed and paraffin embedded before sectioning and staining with hematoxylin and eosin (c). Western blot analysis of mouse tissue was performed using the RBGO1 antibody (d). Representative images were acquired on a Zeiss Axio Imager 2 microscope and analyzed using the ZEN 2012 image analysis software and magnifications are shown on each image. Low level inflammatory cell infiltration is indicated in the ‘Liver’ image (⟶). Data displayed in histograms are means of three animals. Data were analyzed by ANOVA and Dunnett’s multiple comparison test. ADC treatments differ from control, **p < 0.01

To evaluate the efficacy of RBGO1 ADC in vivo we first explored the utility of the ADC within a 3D culture model (Fig. 7a, b). HEC1A cells were cultured in low adherent culture plates to enable the formation of spheroids. Once formed, spheroids were treated with RBGO1 ADC (0.01–100 μg/ml), RBGO1 antibody (100 μg/ml) or mcF (200 nM) for 72 h. After treatment, cell viability was evaluated using the CellTiter 3D Glo Viability Assay. As with the 2D culture cell killing experiments, treatment with RBGO1 antibody had no observable effect on cell viability, whilst treatment with mcF was effective. The LD₅₀ for RBGO1 ADC was 7.4 μg/ml, which was similar to that noted within the 2D culture experiments and confirmed the potential for the RBGO1 ADC to be effective in vivo.Fig7Fig. 7

RBGO1 ADC is effective within a 3D in vitro tumour model and effectively reduces tumour growth in a murine xenograft model of disease. a, b HEC1A cells cultured in low-adherent plates to enable spheroid formation, were incunbated with medium containing RBGO1 ADC (0.01–100 μg/ml), RBGO1 antibody (100 μg/ml) or mcF (200 nM) for 72 h. Cell viability was determined using the CellTiter 3D Glo Viability Assay and luminescence measured using a FLUOstar Omega microplate reader. Representative images of spheroids were acquired on a Zeiss Axio Imager 2 microscope. Relative cell viability plots were fitted using a 4-parameter logistic model (J). Data displayed are means of three independent experiments. c, d RBGO1 ADC (3 mg/kg), mcF (45 μg/kg) or PBS (Control) were adminstered intravenously to female athymic, nude mice bearing 5 mm HEC1A xenograft tumours on a twice weekly basis for 4 weeks. Bodyweights and tumour volumes were measured twice weekly. After 4 weeks, mice were sacrificed and organs harvested for evaluation of any systemic toxicity. Data displayed in are means of five animals with error bars omitted for clarity. Data were analyzed by ANOVA and Dunnett’s multiple comparison test. Treatments differ from control (PBS), *p < 0.05