Most commonly used RVDs include NI, HD, NG and NN which recognize:
Four repeat variable diresidues commonly used in TALEN assembly define simple cipher: NI comprising asparagine isoleucine preferentially binds adenine forming hydrogen bond between Asn and N7, HD histidine aspartate specifies cytosine via Asp carboxyl hydrogen bond to cytosine exocyclic amine and salt bridge, NG asparagine glycine recognizes thymine via van der Waals contacts between glycine carbonyl and methyl group at 5 position, NN asparagine asparagine tolerates guanine and adenine due to bulky purine accommodation but later improved by NK asparagine lysine or NH asparagine histidine more specific for guanine. Thus to target sequence A C T G one would arrange NI HD NG NN respectively. Target design typically requires preceding T at position 0 due to N-terminal cryptic repeat. Golden Gate assembly uses RVD modules as building blocks for custom TALE genes 18 repeats long yielding specificity approaching single locus in genome. Variants include N* recognizing 5-methylcytosine enabling methylation-sensitive editing, and combination of RVDs allow recognition of virtually any DNA sequence except problematic homopolymeric runs that reduce affinity. This RVD-nucleotide dictionary transformed programmable DNA binding beyond zinc fingers.
Ref: Cermak et al. Nucleic Acids Res 2011 39:e82 RVD NI HD NG NN; Bogdanove Voytas Science 2011.