Selective depletion of metastatic stem cells as therapy for human colorectal cancer

Abstract Selective elimination of metastatic stem cells (MetSCs) promises to block metastatic dissemination. Colorectal cancer (CRC) cells overexpressing CXCR4 display trafficking functions and metastasis‐initiating capacity. We assessed the antimetastatic activity of a nanoconjugate (T22‐GFP‐H6‐FdU) that selectively delivers Floxuridine to CXCR4+ cells. In contrast to free oligo‐FdU, intravenous T22‐GFP‐H6‐FdU selectively accumulates and internalizes in CXCR4+ cancer cells, triggering DNA damage and apoptosis, which leads to their selective elimination and to reduced tumor re‐initiation capacity. Repeated T22‐GFP‐H6‐FdU administration in cell line and patient‐derived CRC models blocks intravasation and completely prevents metastases development in 38–83% of mice, while showing CXCR4 expression‐dependent and site‐dependent reduction in foci number and size in liver, peritoneal, or lung metastases in the rest of mice, compared to free oligo‐FdU. T22‐GFP‐H6‐FdU induces also higher regression of established metastases than free oligo‐FdU, with negligible distribution or toxicity in normal tissues. This targeted drug delivery approach yields potent antimetastatic effect, through selective depletion of metastatic CXCR4+ cancer cells, and validates metastatic stem cells (MetSCs) as targets for clinical therapy.

To synthetize the control 5'-(FdU) 5 -3' pentamer, we used a controlled pore glass (CPG) support functionalized with DMT-FdU prepared as described above. Then, the control pentamer sequence was assembled on a DNA synthesizer (392 Applied Biosystems, Foster City, CA, USA) using a 1 µmol synthesis cycle by successive additions of DMT-protected FdU phosphoramidite. After assembling of the sequence, oligonucleotide support was treated with aqueous ammonia (32%) for 2 hrs at room temperature and the resulting product was purified by HPLC. HPLC conditions: column X-bridgeTM OST C18 (10x50 mm, 2.5 µm); 20 min linear gradient from 0 % to 40%, flow rate 2 mL/min; solution A was 5% acetonitrile in 0.1 M aqueous triethylammonium acetate(TEAA) and solution B 70% acetonitrile in 0.1 M aqueous TEAA. Pentamer was characterized by mass spectrometry (MALDI-TOF).
Several batches of pentamer FdU oligonucleotide were synthesized in 1 µmol scale on an automated RNA/DNA synthesizer using β-cyanoethylphosphoramidite chemistry and following standard protocols. 3'-Thiol-Modifier C3 solid support (Link Technologies, Lanarkshire, Scotland) was used for the introduction of the thiol group at the 3'-end, then hexaethyleneglycol phosphoramidite (Glen Research) was used as spacer. Finally, the synthesis was completed by repetitive additions of the DMT-protected FdU phosphoramidite. After the assembly of the sequence, oligonucleotide support was treated with aqueous ammonia (32%) with 0.1 M 1,4-dithiothreitol (DTT) for 2h at room temperature. The ammonia solution was concentrated to dryness and the product was desalted on NAP-10 (Sephadex G-25) columns eluted with water prior to use. The purity of the pentamer FdU-HEG-SH was analyzed by HPLC using the conditions described above (see appendix Fig. S1).
Pentamer was quantified by absorption at 260 nm and confirmed by MALDI mass spectrometry (MALDI-TOF).

determination of Drug/Nanoparticle Ratio
The products obtained after the T22-GFP-H6-FdU synthesis reaction were characterized using mass spectroscopy to measure their molecular mass. The volume and size distribution of the nanoparticles was determined by dynamic light scattering at 633 nm (Zetasizer Nano ZS, Malvern Instruments Limited, Malvern, Worcestershire, UK). Nanoconjugate size was also measured using purified samples, diluted to 0.2 mg/mL and contrasted by evaporation of 1 nm platinum layer in carbon-coated grids, before being visualized in a Hitachi H-7000 transmission electron microscope (TEM), as described (Unzueta 2012b). The drug to nanoparticle ratio was obtained by analyzing the UV spectra of T22-GFP-H6 and T22-GFP-H6-FdU nanoconjugate and calculating the number of FdU molecules per T22-GFP-H6 nanoparticle.

T22-GFP-H6-FdU internalization, CXCR4 specificity and cytotoxicity in CXCR4 + HeLa cells
We assessed T22-GFP-H6-FdU capacity to internalize in a second cell type, the CXCR4 We also studied T22-GFP-H6-FdU cytotoxic activity using the MTT metabolic test (Roche, Basel, Switzerland). To that purpose, we exposed CXCR4 + HeLa cells to T22-GFP-H6-FdU at 1-1,000nM concentration range and measured their viability at 48 hours as compared to equimolecular concentrations or free oligo-FdU. Afterwards, we constructed a graphic displaying the linearized T22-GFP-H6-FdU dose-response trend line representation to compare cell viability for both compounds.
Reduction of cell viability was also determined by optical microscope images of HeLa cells exposed to 1µM T22-GFP-H6-FdU for 48 h, as compared to T22-GFP-H6 or free FdU. We generated an efficient metastatic model, in NOD/SCID mice, that received an intracecal microinjection (ORT) of SW1417-luc CRC cells disaggregated from SC tumours previously generated in a different cohort of NOD/SCID mice. Briefly, when SC tumors reached a volume of 700 mm 3 , mice were sacrificed by cervical dislocation and tumours were excised, discarding the necrotic areas, and three hundred mg of viable tumour tissue was then cut into pieces and disaggregated in a mix of 0.05% trypsin (Invitrogen) and 100 mg/ml DNase (Sigma-Aldrich). The mix was pipetted 30 times, using a 10 ml pipette, and incubated for 10 minutes at 37°C with shaking. It was then re-pipetted 30 times, using 10 ml, 3 ml and 1 ml pipettes, and re-incubated for 5 minutes at 37°C with shaking. The obtained SW1417 single-cell suspensions were filtered through a cell strainer and centrifuged at 1,000g for 10 minutes before counting the cells. We then microinjected 2×10 6 cells, previously grown in culture and resuspended in 50 µl of media, in the cecum of each mouse.

T22-GFP-H6-FdU induction of DSBs and proteolyzed PARP in the subcutaneous
We used the SC+ORT metastatic SW1417 CRC model developed in NOD/SCID to evaluate the capacity of the nanoconjugate for metastasis prevention. We ramdomized mice into three groups: buffer (n=11), T22-GFP-H6-FdU (n=12) and free oligo-FdU (n=11) and administered repeated i.v.
boluses at equimolecular doses, as follows: T22-GFP-H6-FdU 20ug, free oligo-FdU: 2.6 nmols, or buffer), every three days (q3d) for a total of 12 doses. We initiated the T22-GFP-H6-FdU administration one week after tumor cell implantation before metastatic dissemination had occurred (Appendix Fig S5). The experiment was finished when the first animal of the Buffer-treated group required to be euthanized. At necropsy, we recorded the number and size of visible metastases in all organs per mouse, counted ex vivo the number of metastatic foci that emitted bioluminiscence in the target organs, using the IVIS® 200-Spectrum, and performed the histopathological and immunohistochemical analyses to confirm location and number of metastases. We also evaluated CXCR4 expression by IHC, using the described anti-CXCR4 antibody, in primary tumor and metastatic foci at the different organs affected by metastases (peritoneum, liver, lung and lymph nodes). We quantified the fraction of CXCR4 + cancer cells remaining in tumor tissue (CXCR4 + CCF) after 6 treatment. The obtained results were used to study a possible correlation between CXCR4 + CCF and antimetastatic effect at the different sites.

T22-GFP-H6-FdU biodistribution and toxicity in bone marrow and circulating blood cells
We used mice bearing subcutaneous CXCR4 + SW1417 CRC tumors to assess T22-GFP-H6-FdU uptake (measuring green fluorescence emission) in bone marrow and circulating blood cells 5h after the administration of T22-GFP-H6-FdU at a range of 10-100 µg doses. At necropsy, bone marrow was extracted and registered ex vivo in the IVIS® 200-Spectrum equipment. Mouse blood was collected by punction in the maxillary plexus. The erythrocytes, leucocytes and platelets were isolated by the Ficoll density gradient method using the standard protocol. Further, pellets were obtained by centrifugation of the isolated cell extracts at 600g, 10min and then recorded to measure fluorescence using IVIS® 200-Spectrum equipment. Toxicity was evaluated in H&E stained bone marrow slides.

Statistical analysis
Sample size was defined on the basis of previous preliminary experiments. No animals nor samples were excluded from the analyses. Randomization of animals into control and experimental groups were performed using the SPSS program. Histology and immunohistochemical samples were coded so that the researcher that analysed them did not know to which group they belong to. Normal distribution of the data was tested using the Shapiro-wilk test. The homeogeneity of the variance between groups was tested using the Levene´s test. We used the Fisher's exact test to analyze possible differences between control and experimental groups of affected mice regarding metastatic rates at the different organs. The non-parametric tests, Kruskal-Wallis and post-hoc pairwise Mann-Whitney U two-sided tests were used to compare number and size of metastatic foci in the affected organs among groups. All quantitative values were expressed as mean±s.e.m. All statistical tests were performed using SPSS version 11.0 (IBM, New York, USA). Differences among groups was considered significant at a P <0.05.
Appendix Table S1. Number of mice with undetectable metastases (Mets-free) and number of mice bearing metastatic foci (Mets +) at the end of the prevention of metastasis experiment in the T22-GFP-H6-FdU, free oligo-FdU or Buffer-treated groups, when using both, the sw1417 or M5 colorectal metastatic models. for 2h at room temperature. The ammonia solutions were concentrated to dryness and the product was desalted on NAP-10 (Sephadex G-25) columns eluted with water prior to use. Free oligo-FdU synthesis: Control pentamer oligo-FdU without HEG and thiol groups was prepared as before but using 3'-succinyl-FdU controlled pore glass as solid support. Finally, the oligonucleotide was deprotected with aqueous ammonia (32%) for 2h at room temperature. Note that representative microphotographs of tumor cells IHC-stained for anti-γH2AX 5h after treatment with T22-GFP-H6-FdU, free oligo-FdU or Buffer are depicted in Figure 3A of the main text.