Introduction

The identification of tumor-specific molecular markers for cancer diagnostics and the targeting of tumor-specific moieties/pathways for developing nontoxic and effective anticancer therapies are important goals. Efficient targeting of cancer cells will depend upon development of molecular probes suitable for in vivo applications, probes that are endowed with the required affinity, specificity, and favorable pharmacokinetic properties.

Aptamers are highly structured oligonucleotides (DNA or RNA) that can bind to targets with affinities comparable to antibodies. They are identified through an in vitro selection process called systematic evolution of ligands by exponential enrichment (SELEX; (1,2) to recognize a wide variety of targets, from small molecules to proteins, and from cultured cells to in vivo imaging targets (3,4,5,6,7,8,9,10,11). These oligonucleotides have properties that are well suited for in vivo diagnostic and/or therapeutic applications: besides good specificity and affinity, they are poorly immunogenic, and their relatively small size can result in facile penetration of tissues. The SELEX technology can now accept chemically modified nucleotides for improved stability in biological fluids (12). Less than two decades after the first applications of the technique, several lead compounds are currently in clinical trials, and the first aptamer drug, Macugen, was approved by the U.S. Food and Drug Administration (FDA) in 2004 (see Reference (13).

However, aptamers that are identified through the standard SELEX process usually comprise ∼70–80 nt, since they are typically selected from nucleic acid libraries with ∼30–40-nt-long randomized regions plus fixed primer sites of ∼20 nt on each side. There are substantive problems with these primer sequences (which generally comprise ∼50% of the library sequences) in that they may comprise a portion of the selected binding sequences, and/or they may base-pair with the central random fragments to form structures that are selected as sites for target binding, which often compromises the significance of results from standard SELEX protocols. To address these problems using RNA libraries, primer sequences have been blocked by complementary oligonucleotides or switched to different sequences midway during the rounds of SELEX (14), or have been trimmed to only 6 and 4 nt (15), or to 9 and 7 nt (16). Wen and Gray (17) designed a primer-free genomic SELEX method, in which the primer sequences were completely removed from the library before selection and were then regenerated to allow amplification of the selected genomic fragments. However, to employ the technique, a unique genomic library had to be constructed, a very elaborate procedure, and there is presumably very limited diversity in the genomic library. While this protocol might be adaptable for use with a random region (in place of the genomic segments), the reliance on a linear extension step to regenerate primers would likely prove problematic. Alternatively, efforts to circumvent problems caused by fixed primer sequences using high efficiency partitioning are met with problems regarding PCR amplification (18).

Here we have developed two methods, minimal primer (MP) selection (termed primer-bridge PCR) and primer-free (PF) selection (termed self-bridge RT-PCR), which significantly simplify SELEX procedures and effectively eliminate primer-interference problems. Here we show that the protocols (the MP protocol and two closely related PF protocols) work in a straightforward manner. The central random regions of the libraries were purified without extraneous flanking sequences, and were bound to a melanoma cell line. The bound sequences were obtained, reunited with flanking sequences, and reamplified to generate a selected sublibrary. Successive rounds of selection increased library binding and increased selectivity for melanoma cells versus binding to nontarget fibroblasts.

Materials and Methods

Oligonucleotides for MP Selection

The random library for the MP selection protocol was (L7-N27) = GAA CAG GAT TAG CGG CCG C-(N)₂₇-TGA TTC GAC TCT AGA GCG, which contains 4²⁷ (1.80 × 10¹⁶) different sequences. The amplification primers were as follows: 5′-end primer (L7-P1): CGC TCT AGA GTC GAA TCA; 3′-end primer (L7-P2): GAA CAG GAT TAG CGG CC; 5′-end bridge pair (5′-end top-strand, 5TS = CGC TCT AGA GTC GAA T and 5′-end bottom-strand, 5BS = TGA TTC GAC TCT AGA GCG); and 3′-end bridge pair (3′-end top-strand, 3TS = pGGC CGC TAA TCC TGT TC and 3′-end bottom-strand, 3BS = GAA CAG GAT TAG CGG CCG C).