Serum exosomal-annexin A2 is associated with African-American triple-negative breast cancer and promotes angiogenesis

Archived serum samples of breast cancer patients (n = 169) and non-cancer females (n = 68) were collected from the Tissue Management Shared Resource, Simmons Comprehensive Cancer Center, University of Texas Southwestern Medical Center, Dallas, Texas. The samples were stored at − 80 °C, were thawed at room temperature (RT), and immediately placed on ice prior to use. All the archived serum samples were acquired under Institutional Review Board (IRB)-approved protocols at the site of collection and the University of North Texas Health Science Center, Fort Worth, Texas. The samples were analyzed in a double-blinded study where the clinicopathological reports of the sample were not revealed to the investigator until after completion of analysis.

Exosomes from breast cancer serum samples were isolated by using a total exosome isolation reagent (Life Technologies, USA) according to the manufacturer’s protocol. Briefly, the serum samples were thawed at RT and centrifuged at 2000×g for 30 min at 4 °C to remove cells and debris. This clarified serum sample (100 μL) was mixed with 20 μL of the reagent and vigorously mixed with a vortex and pipetting up and down until there was a homogenous solution. This mixture was incubated at 4 °C for 30 min. After incubation, the sample was centrifuged at 10,000×g for 10 min at RT. The supernatant was discarded and the exosomal pellet was resuspended in PBS for analysis. Average sizes of the exosomes were determined by a Malvern Zetasizer particle size analyzer (Malvern Instruments Ltd., Malvern, UK). The exosomal pellet was resuspended in PBS, and the size distribution was analyzed. The results were reported as the average of five runs with triplicates in each run.

AnxA2 levels in serum exosomes were analyzed by an ELISA kit (R&D systems, MN) according to the manufacturer’s protocol. Briefly, a 96-well microplate was coated with capture antibody overnight at 4 °C, washed three times, and blocked with blocking buffer for 2 h at RT. Next, the plates were incubated with serum exosomes and diluted in buffer for 2 h at RT. The plates were washed and coated with detection antibody for 2 h at RT and washed again. The plates were incubated with Streptavidin-HRP for 20 min at RT, washed, and further incubated with 3,3′,5,5′-tetramethylbenzidine (TMB) peroxidase substrate. The reaction was stopped using 2N H₂SO₄ and the optical density was read at 450 nm with wavelength correction at 540 nm. Samples were run in triplicate (n = 3).

The Matrigel plug assay was performed as described previously with slight modifications [21, 25]. Briefly, 500 μL of Matrigel (BD Biosciences, CA) mixed with either PBS (negative control) or serum-derived exosomes (100 μg; pooled from 5 random patients to eliminate bias) treated with or without control peptide (LGKLSL) or AnxA2 inhibitory peptide (LCKLSL) were injected subcutaneously at the left or right lower abdominal wall of the 4- to 6-week-old athymic nude mice (Harlan Laboratories, WI). Three mice were injected for each control and experimental group. Mice were sacrificed 18–20 days after the Matrigel injections and the Matrigel plugs were recovered and photographed. Matrigels were snap frozen in liquid nitrogen for hemoglobin estimation using Drabkin’s reagent.

Hemoglobin estimation from the Matrigel was performed by Drabkin’s method [26]. To quantify the formation of functional vasculature in the Matrigel plug, the amount of hemoglobin was measured using a Drabkin reagent kit 525 (Sigma, MO) following the Drabkin and Austin methods [27]. Briefly, the Matrigel plugs were homogenized in a Dounce homogenizer on ice in the presence of 0.5 mL deionized water and allowed to stand overnight at 4 °C. The lysate was centrifuged at 5000×g for 10 min and the supernatant was collected. Next, 0.3 mL of each sample was mixed with 0.5 mL of Drabkin’s reagent and allowed to stand for 15 min at room temperature. The absorbance was read at 540 nm by using Drabkin’s reagent solution as blank. A standard curve was constructed by using known concentrations of hemoglobin, and the concentration of the samples was obtained from the standard curve.

GraphPad Prism 8 (GraphPad Software, CA) and SPSS software (SPSS Inc., IL) were used for all statistical analysis. Scatter plots were used to plot the serum exo-AnxA2 levels, and the results were presented as mean ± SEM. Comparison of mean between two groups was conducted using Student’s t test, while the comparison for more than two groups was conducted using one-way ANOVA. Data that did not satisfy parametric assumptions were analyzed using non-parametric test. Survival data of serum exo-AnxA2 were derived from clinical information for each breast cancer patient. The cutoff values for AnxA2 expression for “low” and “high” were determined using the median of the serum exo-AnxA2 concentrations in breast cancer patients. Overall survival (OS) was defined as the interval between the date of surgery and date of death from any cause. Disease-free survival (DFS) was defined as the interval from the date of surgery or treatment to the date of recurrence diagnosis. Kaplan–Meier estimation and log-rank tests were used to analyze differences in survival durations (reported using hazard ratios and corresponding 95% confidence intervals (CI)) [28]. These analyses determined the impacts of the serum exo-AnxA2 on OS and DFS. To determine whether serum exo-AnxA2 could be a potential diagnostic tool for aggressive breast cancer, receiver operating characteristic (ROC) curves were used to compare the serum exo-AnxA2 levels of cancer patients and non-cancer patients. Statistical significance was two-tailed and considered significant if P value was at least ≤ 0.05: (*), P < 0.05; (**), P < 0.01; (***), P < 0.001; (****), P < 0.0001.

We previously showed that AnxA2 is present in exosomes derived from the breast cancer cells [21]. Therefore, we investigated whether AnxA2 is expressed in exosomes isolated from the serum samples of breast cancer patients. As the first step, exosomes isolated from serum samples of breast cancer patients and non-cancer females were characterized for the expression of exosomal marker proteins and size analysis. The average size of isolated exosomes indicated that the vesicles were approximately 87.85 ± 21.30 nm in diameter (Fig. 1a). Furthermore, Western blot analysis revealed that the serum exosomes were positive for the expression of exosomal protein markers CD9, TSG101, and flotillin-1 while calnexin, an endoplasmic reticulum marker not expressed in exosomes, is negative (Fig. 1b). Our results show that AnxA2 is also present in exosomes isolated from serum samples of non-cancer females and ER⁺, HER2⁺, and TNBC breast cancer patients (Fig. 1b). To demonstrate that AnxA2 present in serum exosomes originate from epithelial cells, we performed immunoprecipitation of exosomes using anti-AnxA2 or anti-EpCAM antibodies to purify AnxA2- and EpCAM-positive exosomes, respectively. Western blot analysis show that EpCAM-positive breast cancer exosomes express AnxA2 and exosomal markers CD9, TSG101, and flotillin-1 (Fig. 1c). Similarly, AnxA2-positive exosomes also contain epithelial cell marker EpCAM and exosomal marker CD9 (Fig. 1c). In addition, EpCAM- and AnxA2-positive exosomes are negative for the expression of calnexin and GM130, respectively, and show the purity of exosomes isolated from breast cancer serum samples. These findings demonstrate that exosomes isolated from serum samples of non-cancer females and subtypes of breast cancer patients carry AnxA2 and are derived from epithelial cells.

Having shown that AnxA2 is present in the serum samples of breast cancer patients and predominantly localized at the surface of the exosomes [19, 21, 22], we measured the circulating concentrations of exo-AnxA2 in serum samples of breast cancer patients and non-cancer females. Serum samples collected from breast cancer patients (n = 169) and non-cancer females (n = 68) were analyzed in a double-blind study for exo-AnxA2 protein levels. The demographics and health information of breast cancer patients and non-cancer females are listed in Table 1. ELISA analysis shows that exo-AnxA2 concentration was significantly elevated in serum samples of breast cancer patients (n = 169, 83.33 ± 2.040 ng/mL, P < 0.0001) in comparison to non-cancer (n = 68, 34.21 ± 2.238 ng/mL) females (Fig. 2a). We have previously shown that exo-AnxA2 collected from cell culture supernatants is a promoter of angiogenesis [21]. To further confirm our findings, we performed a Matrigel plug assay in athymic nude mice with exosomes collected from serum samples of breast cancer patients and non-cancer females. Gross examination of Matrigel plugs showed abundant vessel formation in the plugs containing serum exosomes and few vessels in plugs with PBS alone (Fig. 2b). However, the extent of vessel growth was significantly higher in Matrigel plugs with exosomes derived from serum samples of breast cancer patients than plugs containing serum exosomes of non-cancer females. When we incubated the serum exosomes with LCKLSL AnxA2 inhibitory peptide, we found drastic decrease in degree of vessel formation than incubation with LGKLSL control peptide (Fig. 2b); this confirms both groups of exosomes induced new vessel formation in an AnxA2-dependent manner. We further confirmed our findings by analyzing the amount of hemoglobin present in the Matrigel plugs via Drabkin’s method (Fig. 2c). The hemoglobin concentration in the plugs containing breast cancer exosomes was significantly higher (~ 3.2 fold) than those with non-cancer serum exosomes. As evident from Matrigel plug images, incubation of the plugs with breast cancer serum exosomes resulted in approximately 5.8-fold decrease in hemoglobin concentration with LCKLSL AnxA2 inhibitory peptide treatment, which did not occur with LGKLSL control peptide treatment (Fig. 2c). In addition, injection of Matrigel with non-cancer serum exosomes plus LCKLSL had similar effects (~ 2.2-fold decrease) compared to LGKLSL control peptide treatment. These findings suggest that high concentration of serum exo-AnxA2 is a potent inducer of angiogenesis in breast cancer patients.

The clinicopathological features such as tumor size, grade, lymph node metastasis, and TNM stage and its association with serum exo-AnxA2 relative expression status were examined in breast cancer patients. Significant association between breast tumor grades and serum exo-AnxA2 relative expression levels were observed (Fig. 3a) and tumor size, lymph node metastasis, and TNM stage showed no significant association with circulating levels of serum exo-AnxA2 with the progression of the disease. As shown in Fig. 3a, the mean concentration of serum exo-AnxA2 in non-cancer patients was 34.21 ± 2.238 ng/mL (n = 68), whereas that in patients with grade I, II, and III breast tumor was 63.49 ± 2.372 ng/mL (n = 16, P < 0.0001), 71.27 ± 2.548 ng/mL (n = 49, P < 0.0001), and 91.37 ± 2.852 ng/mL (n = 94, P < 0.0001), respectively. The concentration of serum exo-AnxA2 in grade III breast patients was significantly higher than that of grade I and II breast tumor patients (P < 0.0001). However, no significant difference was observed between grade I and II breast tumor patients.

Emerging evidence shows that AnxA2 is upregulated and correlated to poor prognosis in patients with breast cancer [29, 30]. In our study, the Kaplan–Meier method was performed to analyze the relationship between serum exo-AnxA2 levels and OS of patients with breast cancer. We used the median expression value for serum exo-AnxA2 in breast cancer patients (n = 169) to stratify into a high exo-AnxA2 (> 77.87 ng/mL) and a low exo-AnxA2 groups (< 77.87 ng/mL). The results demonstrated that breast cancer patients with high levels of serum exo-AnxA2 (n = 85) had significantly shorter OS (hazard ratio 2.802; 95% CI = 1.030–7.620; log-rank P = 0.0353) than those with low levels of serum exo-AnxA2 (n = 84; Fig. 3b). In addition, we also determined the correlation between exo-AnxA2 levels in the serum and DFS of the breast cancer patients. The median expression value of serum exo-AnxA2 in breast cancer patients (n = 107) was used for DFS evaluation and stratified into high exo-AnxA2 (> 70.87 ng/mL) and low exo-AnxA2 groups (< 70.87 ng/mL). We found that high serum exo-AnxA2 levels (n = 54) were associated with worse DFS (hazard ratio 7.934; 95% CI = 1.778–35.398; log-rank P = 0.0301) in breast cancer patients (n = 53; Fig. 3c). Taken together, our survival analysis confirms that high expression of circulating exo-AnxA2 in serum results in a poor survival of the breast cancer patients and suggests that the circulating exo-AnxA2 could predict prognosis of breast cancer patients.

In our previous studies, we have shown that AnxA2 is overexpressed in TNBC in comparison to other subtypes of breast cancer [20, 31]. Therefore, we measured the expression of exo-AnxA2 levels in serum samples of ER⁺, HER2⁺, and TNBC breast cancer patients (Table 1). Our ELISA analysis clearly suggests that the relative expression of exo-AnxA2 levels were significantly elevated in TNBC (n = 68, 109.1 ± 2.905 ng/mL) in comparison to ER⁺ (n = 50, 57.35 ± 1.545 ng/mL, P < 0.0001), HER2⁺ (n = 59, 78.25 ± 1.146 ng/mL, P < 0.0001), and non-cancer (n = 68, 34.21 ± 2.238 ng/mL, P < 0.0001) serum samples (Fig. 4). These observations show that the expression of exo-AnxA2 is predominantly associated with TNBC subtype. Our previous study indicates a strong association of AnxA2 expression with AA women with breast cancer and implicating AnxA2 as a contributor to the aggressive biology of TNBC [30]. Here, we further compared the expression of exo-AnxA2 levels in sera of AA and CA with breast cancer and non-cancer females. ELISA analysis revealed that AnxA2 expression is significantly elevated in serum exosomes isolated from AA TNBC (n = 29, 118.9 ± 4.086 ng/mL, P < 0.0001) patients in comparison to CA TNBC (n = 27, 97.60 ± 3.298 ng/mL) patients (Fig. 5a). In contrast, the concentration of serum exo-AnxA2 levels in ER⁺ patients were significantly high in CA ER⁺ (n = 25, 64.70 ± 0.561 ng/mL, P < 0.0153) compared to AA ER⁺ (n = 25, 50.01 ± 2.223 ng/mL) patients (Fig. 5a). However, no significant differences were observed in AA and CA patients of HER2⁺ breast cancer and non-cancer females. Data presented in Figs. 3a and 5a show that serum exo-AnxA2 levels in breast cancer patients increases with the tumor grade progression and exo-AnxA2 expression is high in serum samples of TNBC patients, respectively. Therefore, comparison analysis of relative expression of exo-AnxA2 levels in serum samples of different breast cancer subtypes with tumor grade progression were performed (Table 1 and Fig. 5b). Our analysis (one-way ANOVA followed by Dunnett’s multiple comparison test) clearly suggests that the relative expression of exo-AnxA2 levels were significantly elevated in different tumor grades of ER⁺ (grade I: n = 12, 60.38 ± 2.276 ng/mL, P < 0.0001; grade II: n = 25, 58.83 ± 1.720 ng/mL, P < 0.0001; or grade III: n = 12, 50.33 ± 4.364 ng/mL, P < 0.01), HER2⁺ (grade I: n = 4, 72.83 ± 4.099 ng/mL, P < 0.0001; grade II: n = 15, 75.78 ± 1.887 ng/mL, P < 0.0001; or grade III: n = 36, 79.79 ± 1.540 ng/mL, P < 0.0001), and TNBC (grade II: n = 9, 98.33 ± 5.249 ng/mL, P < 0.0001; or grade III: n = 46, 111.1 ± 3.304 ng/mL, P < 0.0001) patients in comparison to non-cancer (n = 68, 34.21 ± 2.238 ng/mL) female serum samples. In addition, our study showed that exo-AnxA2 expressions were significantly high in the serum samples of grade III TNBC patients compared with grade II TNBC patients (P < 0.029; Fig. 5b). In contrast, no significant difference in exo-AnxA2 levels were observed in ER⁺ or HER2⁺ breast cancer patients with the progression of tumor grades. To ensure that the relative expression of serum exo-AnxA2 is significantly higher in AA TNBC patients, the expression of serum exo-AnxA2 levels in CA and AA women were further analyzed after adjusting the tumor grades in TNBC population (Additional file 1: Figure S1). Our analysis clearly suggest that the relative expression of exo-AnxA2 is significantly high in serum samples of grade III AA TNBC (n = 24, 120.2 ± 4.455 ng/mL, P < 0.01) patients compared to grade III CA TNBC (n = 20, 99.16 ± 4.155 ng/mL) patients. However, no statistical difference in serum exo-AnxA2 levels were observed between grade II AA TNBC (n = 2; 116.6 ± 10.95 ng/mL) patients and grade II CA (n = 7; 93.11 ± 4.592 ng/mL) TNBC patients. Our observation of exo-AnxA2 preferential association with TNBC strongly suggests a role of serum exo-AnxA2 levels in predicting aggressiveness of TNBC in AA women.

The results presented in Fig. 2 suggest that the high expression of exo-AnxA2 levels is a potent inducer of angiogenesis in breast cancer patients. To further confirm that high expression of exo-AnxA2 levels in serum correlated with aggressive metastasis in TNBC, an in vivo Matrigel plug assay was performed using exosomes derived from ER⁺, HER2⁺, and TNBC breast cancer subtypes and non-cancer females. Data shown in Fig. 6a demonstrate a visible increase in vessel formation in plugs containing TNBC exosomes in comparison to other breast cancer subtypes and non-cancer exosomes. We further confirmed our observation through quantification of new blood vessel formation within these Matrigel plugs through hemoglobin estimation by Drabkin’s method [26]. Our results show that hemoglobin concentration in Matrigel plugs containing TNBC exosomes is approximately fourfold higher compared to plugs containing ER⁺, HER2⁺, or non-cancer exosomes (Fig. 6b). To further examine that aggressive metastases in AA women with TNBC is correlated with the high levels of circulating exo-AnxA2 present in TNBC patients, an in vivo Matrigel plug assay was further performed in female nude mice using exosomes derived from serum samples of AA and CA TNBC patients with or without LGKLSL or LCKLSL peptides. We visually observed attenuation of angiogenesis in Matrigel plugs containing LCKLSL AnxA2 inhibitory peptide in both CA and AA TNBC exosomes and increased angiogenesis in exosomes alone or exosomes containing LGKLSL control peptide (Fig. 6c). Hemoglobin analysis of Matrigel plugs further confirmed that CA and AA TNBC exosomes containing LCKLSL inhibitory peptide inhibits approximately 5-fold and 7.5-fold blood vessel formation, respectively (Fig. 6d), compared to their respective CA and AA serum exosomes. However, the exosomes treated with LGKLSL control peptide did not show any significant reduction in hemoglobin concentration and blood vessel formation compared to their respective exosomes alone (Fig. 6c, d). In addition, the hemoglobin content is significantly high (~ 1.2-fold) in plugs containing AA TNBC exosomes in comparison to plugs with CA TNBC exosomes. Our observation of exo-AnxA2 preferential association with AA TNBC patients suggests a potential role for exo-AnxA2 as a contributor to the aggressiveness of TNBC in AA women.

To determine whether serum exo-AnxA2 could be a potential diagnostic tool for aggressive breast cancer, receiver operating characteristic (ROC) curves were used to compare the serum exo-AnxA2 levels from 169 breast cancer patients and 68 non-cancer patients. The area under the curve (AUC) for the ROC curve of the test with exo-AnxA2 levels in serum samples of breast cancer patients as the disease indicator was 0.9484 ± 0.01327 (95% CI = 0.9223–0.9744, P < 0.0001; Fig. 7a). The diagnostic ability of serum exo-AnxA2 was also evaluated in ER⁺ (n = 50), HER2⁺ (n = 59), and TNBC (n = 58) patients compared to non-cancer (n = 68) patients. ROC curves of serum exo-AnxA2 in breast cancer subtypes showed that AUC values of ER⁺, HER2⁺, and TNBC were 0.8304 ± 0.03843 (95% CI 0.7551–0.9058, P < 0.0001), 0.9958 ± 0.0029 (95% CI 0.9899–1.000, P < 0.0001) and 1.000 ± 0.000 (95% CI 1.000–1.000, P < 0.0001), respectively (Fig. 7b). These results indicate that serum exo-AnxA2 levels might be an appropriate diagnostic tool for aggressive breast cancer specifically in TNBC patients.