4.1. Cell-SELEX (Sistematic Evolution of Ligands by Exponential Enrichment)
With a synthetic and molecular biology in place for AEGIS systems, our laboratory, in collaboration with the Tan laboratory at the University of Florida, attempted the first AEGIS-based in vitro evolution experiment in 2013, where a line of breast cancer cells, MDA-MB-231, was targeted in a Cell-SELEX experiment [100
The library used in this selection contained a 20-nt long random region composed of ACTGZP nucleotides, and surrounded by primer binding sites made of standard DNA. This library was subjected to 12 rounds of selection that gradually reduced the number of cells and incubation times per cycle in order to increase the selection pressure. The surviving pool of cycle 12 was subjected to AEGIS to standard DNA conversion, and submitted for high-throughput sequencing.
The most interesting results from this first AEGIS-LIVE experiment were that: (1) binders appeared after 9-12 rounds of selection, compared to the 15-20 rounds that have been reported as necessary to observe the growth of binders in previous Cell-SELEX experiments with standard nucleic acid libraries; and (2) the Z and P nucleotides, when present, were necessary for binding. The best aptamer from this selection, ZAP-2012, contained 1 P and 1 Z, which are both necessary for binding, and had a kd of ~30 nM.
The second AEGIS-based Cell-SELEX, which was performed on HepG2 liver cancer cells, gave a little more information [93
The schematic of this AEGIS-based Cell-SELEX is shown in Figure 3
. In this case, the synthetic AEGIS-library contained 25 random nucleotides (ACTGZP), and a negative selection step was added to each cycle to increase specificity to the target cell line. Again, only 13 rounds of selection were needed to observe the growth of binders.
Several aptamers were obtained from deep sequencing, and a quantitative improvement in DNA performance with added nucleotides could definitely be seen. Indeed, the aptamers with the lowest kd
contained Z and/or P residues, and were as low as 14 nM, while ACTG-only aptamers had kd
ranging from 326 nM to more than 1 μM (Figure 4
). Moreover, the aptamers were very specific for the cell line used during selection, and did not bind to other cell lines. This feature was not fully achieved in the first attempt at AEGIS-Cell-SELEX, where some aptamers also bound to some other cell lines.
In a third AEGIS-Cell-SELEX experiment, cell engineering was used to place glypican 3 (GPC3), a possible marker for liver cancer diagnostic, on the surface of a murine cell line that does not express GPC3 (1MEA). This engineered cell line (1MEAhGPC3
) was used in positive selection cycles, while wild-type 1MEA cells were used in counter-selections [101
]. In this case, the library’s randomized AEGIS region was 35 nt long.
This experiment required only 11 rounds to produce populations showing substantial binding during bulk reactions. Many of the selected aptamers bound with various levels of specificity to GPC3; others appeared to bind to different proteins on the cell surface, perhaps newly emerging on the host cells as a consequence of their being engineered to express GPC3. Interestingly, the kd for these aptamers were in the low to high nM range when tested on a hGPC3-positive human liver cancer cell line, HepG2. One aptamer in particular, LG5, which carried one Z residue but no Ps, showed a high level of specificity for the intended target regardless of the cell line expressing it, and had an apparent kd (for HepG2) of 6 nM. Again, AEGIS nucleotides were found to be essential for binding.
4.2. AEGIS-LIVE on Anthrax Protective Antigen
The first attempt to select for an AEGIS aptamer against an isolated, specific protein target was performed in 2016. Here, the AEGIS-LIVE experiment sought to obtain binders to the Bacillus anthracis
63-KDa version of protective antigen (PA63), the cleaved, heavier subunit of its precursor PA83 [102
In this experiment, the six-letter ACTGZP AEGIS library contained only 25 randomized positions flanked by 15 nucleotide-long primer binding sites composed of standard DNA, with the aim of selecting for the smallest binders available in the ~1014 sampled different sequences (out of ~1019 total possible sequences). The target, PA63, was presented to the library immobilized to magnetic beads; binding oligonucleotides were recovered magnetically; and AEGIS-PCR with a single biotinylated primer was performed directly on survivors bound to the bead-coupled PA63. After 14 cycles of selection, the binder population had grown by ~30%, and the library from this cycle was subjected to transliteration and deep sequencing. One sequence named PA-Apt1, or PA1, dominated the surviving pool, with 96% of the total reads. This contained two P residues and, similar to the other five most represented sequences, no Z nucleotides, despite Z being present at 17% frequency in the starting library.
In filter-binding assays, PA1 was found to be strictly specific for the 63-KDa version of protective antigen, and unable to bind to PA83, indicating that the PA1 epitope might at least in part reside on the portion of PA83 that becomes exposed after cleavage of the 20-KDa subunit. Extensive mutational and enzymatic analysis revealed that the two P were essential for binding to PA63, and that the folding of the molecule included a four base-pairs stem surmounted by a large highly compact and nuclease resistant loop (Figure 5
A). While the parent molecule that was originally selected had a kd
of ~2.3 μM, subsequent truncations of the molecule reduced the binding constant to ~50 nM (PA1T4). Most interestingly, although expected from the structural studies mentioned above, when two base pairs in the stem were substituted with two P–Z pairs (PA1T4PPZZ), the molecule’s kd
lowered further to ~35 nM (Figure 5
B). This is one example of the effectiveness of post-selection design, where a molecule’s features can be improved by the addition of strategically placed non-natural elements.
In a stepwise type of analysis, AEGIS–PA63 aptamers were also tested against anthrax protective antigen biological activities. First, aptamers were found to bind to PA63 when this is already associated with its natural cell-receptor CMG2 (capillary morphogenesis protein 2, also known as ANTXR2), further narrowing the possible epitope of PA1 on its target surface, excluding the PA63 binding site to CMG2. Further, electrophysiology assays showed that PA1T4 competed with the lethal factor (LF) for binding to PA63 when this is assembled in its channel state across an in vitro membrane system (Figure 5
This molecule is the first example of an aptamer that has evolved towards a specific protein target from an expanded genetic alphabet using the molecular biology and analytical chemistry that has been developed to support it, but without the escamotages used in other systems to circumvent the limitations of extant biology.