Native tRNAs often contain post-transcriptional modifications to the wobble position to expand the capacity of reading the genetic code. Some of these modifications, due to the ability to confer imperfect codon-anticodon pairing at the wobble position, can induce a high propensity for tRNA to shift into alternative reading frames. An example is the native UGG isoacceptor of E. coli tRNA(Pro) whose wobble nucleotide U34 is post-transcriptionally modified to cmo(5)U34 to read all four proline codons (5'-CCA, 5'-CCC, 5'-CCG, and 5'-CCU). Because the pairing of the modified anticodon to CCC codon is particularly weak relative to CCA and CCG codons, this tRNA can readily shift into both the +1 and +2-frame on the slippery mRNA sequence CCC-CG. We show that the shift to the +2-frame is more dominant, driven by the higher stability of the codon-anticodon pairing at the wobble position. Kinetic analysis suggests that both types of shifts can occur during stalling of the tRNA in a post-translocation complex or during translocation from the A to the P-site. Importantly, while the +1-frame post complex is active for peptidyl transfer, the +2-frame complex is a poor peptidyl donor. Together with our recent work, we draw a mechanistic distinction between +1 and +2-frameshifts, showing that while the +1-shifts are suppressed by the additional post-transcriptionally modified m(1)G37 nucleotide in the anticodon loop, the +2-shifts are suppressed by the ribosome, supporting a role of the ribosome in the overall quality control of reading-frame maintenance.