EpCAM-based MACS of T-EV from human serum samples

Since we aimed to establish an isolation protocol for T-MV as well as T-Exo from peripheral blood of cancer patients, we selected EpCAM as marker for T-EV since it was the only protein that was expressed on both vesicle populations. To separate T-EV from other EV populations present in blood e.g. EV derived from platelets, leukocytes or red blood cells, we chose an EpCAM-based sorting approach via MACS (see 2.2.4.3) because this method was already described for the sorting of Exo from peripheral blood of ovarian cancer patients (Taylor & Gercel-Taylor, 2008). Moreover, EpCAM has been previously used to identify circulating tumor cells in metastatic breast cancer (Riethdorf et al, 2007).

First, we tested the method in vitro by spiking 5 ml of human AB serum samples with a defined amount of T-MVS and titrated the amount of EpCAM-MicroBeads which allows sorting of these vesicles with the highest quality and quantity. Using MACS with different concentrations of EpCAM-MicroBeads, around 30% of the input T-MVS could be sorted on LD columns as determined by Lowry assay (Fig. 34A). There was no significant difference in MV yields between 50, 10 or 5 µl EpCAM-MicroBeads used for labeling of T-MVS. In order to assess if the MicroBeads also lead to non-specific selection of benign blood-derived MV, human AB serum was incubated with 10 µl EpCAM-MicroBeads alone. This resulted in selection of residual MV showing that to some extent unspecific binding occurs, however, MV counts were considerably higher in the presence of T-MVS. These results could be further confirmed by flow cytometric counting of MV yields as described in 2.2.4.1.4. Again, there was no significant difference in the addition of 5 or 10 µl EpCAM-MicroBeads for sorting and the number of sorted MV was clearly higher when T-MV_S had been added to the serum samples compared to EpCAM-MicroBeads alone (Fig. 34B).

Fig. 34: MACS of T-MVS spiked into human AB serum samples

Human AB serum samples were spiked with different concentrations of T-MV_S and subsequently sorted by MACS. A+B, For quantification the amount of column-bound EpCAM+ MV was measured by Lowry assay (A) and MV counts were determined by flow cytometry (B). The indicated lines represent the means of the samples.

C+D, Sorted MV were compared to standard samples with known concentrations by Western Blotting using the established T-MV markers (C) and the results of three independent experiments were averaged (D) (means±SD, n=3, n.d. = not detectable). E+F, The column-bound sorted MV (E) as well as the column flow-through (F) were characterized for expression of tumor and MV markers by flow cytometry. Histogram overlays show the specific fluorescent signal in green and the isotype control in grey.

The vesicles which had been positively selected on LD columns were characterized to confirm the specific binding of T-MV_S. Western Blot analysis of established T-MV markers (see 3.1.4) demonstrated expression of Tubulin, EMMPRIN as well as Her2 on the sorted vesicles suggesting that they are indeed T-MVS (Fig. 34C). Moreover, the band densities of the samples from the sorted MV were further compared to a dilution series of T-MVS with known concentrations and thereby allowed quantification of the MV yields by Western Blotting which again confirmed that addition of 5 or 10 µl EpCAM-MicroBeads for labeling led to equal MV yields (Fig. 34D). In contrast, putative MV which had been sorted by EpCAM-MicroBeads in the absence of T-MVS, were negative for all markers and only a single band of 55 kDa was detected in these samples which corresponds to an unspecific background signal of the MicroBeads (Fig. 34C).

Further characterization of the sorted MV by flow cytometry confirmed a high expression of EpCAM, EMMPRIN and Her2 on the column-bound vesicles while they were negative for the Exo marker CD63 (Fig. 34E). In contrast, none of the three tumor markers was found in the flow-through (Fig. 34F). Addition of 50 µl EpCAM-MicroBeads masked most epitopes on the sorted MV and did not allow subsequent characterization by flow cytometry (data not shown) which indicates that as low concentrations of the MicroBeads as possible should be used for MACS. Taken together, the results suggest that indeed the added T-MV_S are selected on LD columns and can be eluted from them, although around 70% of T-MV_S are lost during the sorting process, probably due to insufficient elution of the vesicles from the LD columns.

Fig. 35: LD, rather than LS, columns lead to higher MV yields in MACS of T-MV

Human AB serum samples were spiked with equal amounts of T-MV_S, labeled with 5 µl EpCAM-MicroBeads and sorted on LS or LD columns by MACS. A, The yields of column-bound sorted MV were determined by Lowry assay. The lines represent the means of the samples. B, MV yields in both samples were further determined by comparing the band densities of three MV markers (EMMPRIN, Her2, Tubulin) in Western Blots analyses with a dilution series of T-MV_S with known concentrations (as shown in Fig. 34C). The results for all three markers were averaged and are displayed as percentage of input T-MV_S (means±SD, n=3, *p<0,05).

To test if the yields of sorted T-MV_S can be increased by using LS columns, which are recommended for positive selection by MACS, instead of LD columns which were originally engineered for negative selection, we labeled T-MVS with 5 µl EpCAM-MicroBeads and sorted them from human AB serum with LS columns. Surprisingly, yields of positively-selected MV were considerably lower in Lowry assays (Fig. 35A) as well as flow cytometric MV counting (Fig. 35B) when LS columns were used for selection. This might be explained by the fact that LS columns are specifically designed for positive selection of strongly labeled cells and since MV are much smaller and therefore display a weaker magnetic signal, they possibly pass through the columns.

Fig. 36: MACS of T-ExoS spiked into human AB serum samples

Human AB serum samples were spiked with T-ExoS, incubated with 5 or 10 µl EpCAM-MicroBeads and sorted on LD columns by MACS. A, The amount of input (w/o MACS) and column-bound Exo after MACS was determined by Lowry assay (n=2). The indicated lines represent the means of the samples. B, Western Blot analysis: Expression of Exo markers on input T-Exo_S before (w/o) MACS and column-bound Exo after sorting of T-MV_S with 5 or 10 µl EpCAM-MicroBeads (=B.). A dilution series of unsorted T-Exo_S is shown on the left as positive control.

After having established the type of column as well as the concentration of EpCAM-MicroBeads which are most applicable for sorting of EpCAM-positive vesicles, we tested the method for its suitability to sort Exo. Again, human AB serum samples were spiked with T-ExoS, labeled with 5 µl EpCAM-MicroBeads and sorted on LD columns. However, neither in Lowry assays nor in Western Blots we detected any signals in the sorted samples which had been eluted from the columns (Fig. 36). This indicates that EpCAM-based MACS is not suitable for the selection of T-Exo from human serum samples, probably due to the low expression of EpCAM on T-Exo (see Fig. 33).

Fig. 37: MACS for MV in serum samples from cancer patients

A, Serum samples of five metastatic cancer patients were labeled with 5 or 10 µl EpCAM-MicroBeads and sorted for EpCAM-positive MV on LD columns by MACS. The amount of column-bound sorted MV was determined by Lowry assay. The indicated line represents the mean of the samples. B, Flow cytometry: Sorted MV from three of the patients were characterized for tumor and blood cell markers. n.d. = not detectable

Nonetheless, since the established approach seemed to be appropriate for EpCAM-based MACS of T-MV, we investigated whether T-MV can also be sorted in vivo from serum samples of cancer patients. All patients were in an advanced disease stage with multiple metastases and were therefore expected to carry a high number of T-EV in their blood.

Furthermore, we only included patients who did not receive chemotherapy at the time of sample acquisition to prevent the release of apoptotic bodies from dying tumor cells.

Serum samples (5-10 ml) of five metastatic cancer patients including one patient with primary breast cancer and four patients with primary lung cancer were incubated with up to 10 µl of EpCAM-MicroBeads followed by MACS on LD columns. Column-bound vesicles were eluted, subjected to MV isolation and MV yields quantified by Lowry assays (Fig. 37A). In two of the samples from lung cancer patients high MV counts (408 µg and 205 µg) were measured, whereas the amount of sorted MV was comparably lower in the other three samples with only 14,6 µg, 9,3 µg or 3,8 µg, respectively. The three samples which contained sufficient protein for further analysis were additionally characterized by flow cytometry (Fig.

37B). Although the sorted MV in all samples were positive for EMMPRIN, EpCAM could only be detected in one sample. Moreover, a considerable amount of the vesicles was positive for the platelet marker CD62P as well as the leukocyte marker CD45 and the endothelial cell marker CD62E which suggests that the MV sorted by EpCAM-MicroBeads are not derived from circulating tumor cells, but were false-positively sorted benign blood cell-derived MV.

3.3.3 Establishment of a protocol for the isolation of MV from peripheral blood