In silico selection approach to develop DNA aptamers for a stem-like cell subpopulation of non-small lung cancer adenocarcinoma cell line A549

Human lung carcinoma cell line A549 (ATCC® CCL-185™, ATCC, Manassas, VA) at passage 60 was cultured in DMEM (Millipore Sigma, Burlington, USA) supplemented with 10% foetal bovine serum until they reached about 80% confluence. Cells were washed to remove residuals of medium and then detached from the bottles with 2mM EDTA solution (Millipore Sigma, Burlington, MA). The cell line authentication was performed with IdentiCell STR allele protocol and showed a 100% match with A549 cells (IdentiCell, Department of Molecular Medicine, Aarhus, Denmark).

A random library (5’-FITC- GCC TGT TGT GAG CCT CCT-N34-CGC TTA TTC TTG TCT CCC-3’) containing 34 random bases flanked by constant regions for the binding of primers during the PCR amplification reactions was used.15 The forward primer was 5’-BBB-GCC TGT TGT GAG CCT CCT-3’, where BBB indicates 3 subsequent biotin moieties, while the reverse primer was labelled with a 6-FAM or Fluorescein Isothiocyanate (FITC- 5’-GGG AGA CAA GAA TAA GCG-3’). Before the first selection cycle, the library was amplified by a PCR reaction (PCR Conditions described in Supplementary material), subjected to denaturation and the fluorescence single-stranded DNA was purified by denaturation PAGE. At the beginning of the SELEX cycle, the aptamer pool in selection buffer was denatured at 95°C for 10 min and then placed immediately on ice for 10 min followed by 20 min incubation at room temperature.

The incubation of cells with the aptamers was followed by the BRAZIL technique with a centrifugation of a mixture of cells and oligonucleotides through a dibutyl phthalate: cyclohexane (9:1 [v: v]; d=1.03 g.ml-1) layer16.

Briefly, the cells with bound aptamers were collected from the pellet of the organic phase and separated by centrifugation (13,000 g for 10 min (Hettich Universal 32R centrifuge, HETTICH Instruments LP, Beverly, MA). The pellet containing aptamers was extracted by phenol/chloroform (1:1) for purification of obtained DNA followed by PCR as detailed above and strand separation for the next cycle of SELEX. To perform selection of ligands with higher affinity, the stringency of the selection was gradually increased in each cycle by adding 0.3 – 3 mg/ml tRNA for reducing nonspecific binding and increasing the ratio of DNA molecules over cells (10⁶ – 10⁵ cells).

In the seventh cycle the first negative selection with human blood cells (erythrocytes, leukocytes, and thrombocytes in the amount of 10⁷ cells) was performed, followed by a positive selection step against 10⁵ A549 cells. Blood cells were separated from plasma by density gradient separation.17 A selection step against a CD90⁺ A549 cells was done following cell sorting purification of this subpopulation.

These assays were performed with live A549 cells, which not been treated with fixation agents according to Nascimento et al. (2016).18 Pools were incubated in binding buffer (1.25mM HEPES, 0.27mM KCl, 0.14mM CaCl₂, 0.06mM MgCl2, 7.19mM NaCl, 0.9% glucose) for 10 min at 95°C and then immediately placed on ice for 10 min and maintained for 20 min at room temperature. For the interaction between single-stranded DNA molecules and cells, the solution containing the aptamers was gently mixed with cells (10⁶ cells /microtube) to a final volume of 100 μl and 200 pM concentration of aptamers. The mixture was incubated for 30 min at room temperature with gentle shaking. The mixture was centrifuged at 200 X g and the supernatant was discarded. A wash was performed with binding buffer, the mixture was again centrifuged and the supernatant discarded. Then 500 μl of binding buffer was added, followed by flow cytometry analysis. For analysis of double-labelled cells, the anti-CD90 antibody (1: 1,000) was added to the cells along with the oligonucleotide pools. For cell sorting the same procedure was followed, and a gate was determined to collect the double-labelled cells for CD90 expression and aptamer oligonucleotides.

The pool from the seventh cycle contained aptamers with enhanced binding to the subpopulation of A549 cells, expressing CD90, as was detected by increased fluorescence signals. Since the identified subpopulation demonstrated stem-like cells characteristics13, cell sorting was used to isolate the identified fraction of cells. A fraction of 2,500 cells was successfully collected and aptamers were purified for PCR amplification (data not shown).

Selection pool aptamers obtained in each cycle were tested with restriction analysis (RFLP – Restriction Fragment Length Polymorphism) for the presence of conserved restriction sites. Different bands were obtained representing fragments of DNA with conserved restriction sites, thus showing selective enrichment of specific groups of oligonucleotides during SELEX (Suppl. Figure S1).

Figure 1

The sequences from last three cycles were used for Sanger and Next generation sequencing (Supplementary material). Selected sequences were further analysed with different bioinformatics approaches.

Three different in silico selection procedures were used for processing the selected sequences: CLC Genomics Workbench software, Shell script in combination with pipelined use programs for processing nucleotide sequences- MEME19, MEGA620 and UnaFold21 and COMPAS software (AptalT GbmH, Planneg, Germany).22

We have compared molecular evolutionary relations (with MEGA6 program20 ) and motif similarity (with MEME suite 19) of chosen four best binding candidates (A155_18, A452_3, A373_4, A218_12) and already published aptamer S6 binding to A549 cells (5’- GTG GCC AGT CAC TCA ATT GGG TGT AGG GGT GGG GAT TGT GGG TTG-3’).11

Here we aimed to develop aptamer probes targeting NSCLC circulating tumour cells due to their potential diagnostic, prognostic and predictive capacity. For the development of NSCLC specific aptamers, we have chosen the most commonly used human lung adenocarcinoma A549 cells as target for cell-SELEX. Human blood cells were adopted as a negative control for cell SELEX to increase the selectivity of generated aptamers to A459 cells.

In the selection process, the cultured A549 cells were first incubated with a 70-base synthetic single stranded DNA library. The DNA sequences that bound to the A549 cells were then eluted and separated with the BRAZIL technique, as described in Material and Methods.

We have compared aptamer pools from the original library, the sixth cycle and the negative selection. Regarding the percentage of labelled cells for each pool, no significant difference was observed. We observed that the population of cells that were positive for the aptamer pool from negative selection had higher fluorescence, as shown by the shift to the right in the density plot and the overlap of the histograms (Figure 1A). Furthermore, we performed flow cytometry using antibody for CD90 detection in A549 cells and we observed a well-defined population that was positive for CD90 and aptamers (Figure 1A). This population of cells was sorted out and bound aptamers were eluted and amplified by PCR. Figure 1B reports the increase in cell labelling of the sixth selection pool following a negative selection against blood cells.

The obtained aptamer pool after the negative selection against blood cells and sorting out aptamers binding to the CD90⁺ subpopulation was used for further sequencing in order for identification aptamer candidates with conserved structural motifs.

The sequencing of the seventh cycle pool by Ion Torrent PGM next generation sequencing technology resulted in 239,713 reads, from which 151,814 reads were unique. In silico selection of aptamers from the NGS data was accomplished using three different procedures as described in Supplementary material (Table S1).

Using CLC Genomics Workbench software, the starting pool of 239,713 reads was narrowed down to 16 representative sequences having the highest number of sequence members, ranging from 1 080 to 142 copies per representative sequence. They were subjected to further selection based on the presence of conserved sequence motifs. Aptamers A786_1, A574_ 2, A278_8 were selected for in vitro validation, each carrying a unique motif.

In the second in silico aptamer selection, using online available nucleotide sequence processing programs, reads having ambigous bases or incorrect or no sequence of the library primers were first filtered out. In the remaining high quality sequence set 80,312 reads were unique and 81.7% of the reads appeared in the set only once (no copies). Twenty sequences with the highest numbers of copies (reads), ranging from 786 to 138 copies, were ranked with regard to stability of the computed folding, using ΔG (kcal/mol) as a measure of stability, and the stability of putative loop in the three most stable computed foldings as given by the UnaFold tool. Aptamers A155_18, A452_3, A373_4, A218_12 were selected for in vitro validation.

In the third in silico selection procedure, that was executed using COMPASS software, the reads were first clustered in 998 families with defining patterns in the loop space region. Each family was represented by a sequence (one with the highest number of repetitions) and we selected the 21 most frequent among them for further analysis. Grouping families into respective clans was depended on loop space similarity - 12 of the 21 most frequent families were positioned in 10 different family clans. Further scoring for structure stability was applied in terms of melting temperature in addition to pattern ranking and similar to the second selection procedure aptamer sequences, A155_18 and A452_3 were again selected as the best binding candidates.

The selected aptamer candidates were additionally compared with aptamer S6. The constructed dendrogram (Supplementary material, Figure S1) shows clusters of aptamer sequence candidates, where aptamer A155_18 and positive control S6 clustered together. Among the selected aptamer candidates, A155_18 and S6 are also the only two sharing two same motifs: GGTGG/CG and GCCAGT; according to the UnaFold-predicted secondary structure, the motifs are placed into the comparable structural sequences between the loops (Supplementary material). The possibility of cross-contamination was excluded as the seventh SELEX pool was sequenced before the positive control S6 was purchased.

The selected and fluorescence-labelled four candidates, A155_18, A452_3, A373_4, A218_12, together with known binder S6 and the negative control sequence (the sequence with an unknown random region with the length of 34 nucleotides, and flanked by known primers) were used in an in vitro binding test to determine the binding characteristics. Binding to A549 cells was demonstrated with flow cytometry analysis. All four selected aptamers (Figure 1A) showed enhanced binding to target the cell line versus the unselected control (CTRL_N34.012), according to the percentage of fluorescently labelled cells as measured by flow cytometry. Kd values of all four selected aptamers are in nanomolar range (Figure 2).

Figure 2

In comparison with control random aptamer, highest specific binding rates of 26% to A549 cells was obtained with A155_18. Flexible binding of aptamers at different temperatures can expand their repertoire of applications. Since the selection was performed at room temperature (25°C), we performed binding assays also at 4°C and 37°C. There was observed no difference in binding of aptamer A155_18 (data not shown). This is particularly important as the clinical application of aptamers will be carried out under physiological conditions.

During the cell selection aptamers are interacting specifically with the outer plasma cell membranes of targeted cells. In our cell-SELEX selection the targeted cell line A549 was enriched by the expression of the stemness marker, the membrane-bound protein CD90. This subpopulation was selected by flow sorting against CD90 in A549 cell population. The latter indicates aptamer ability of recognizing CD90-positive cells, which may be part of the tumor-initiating circulating cells.