Back in October, Chen et al. reported in Cell the use of an optimized CRISPR guide RNA and endonuclease-deficient/EGFP-fused Cas9 to mark specific loci in live cell nuclei. Their first experiments were to mark telomeric sequences, and in so doing they discovered that the guide RNA was a rate-limiting factor for efficient targeting. They then made 2 improvements to the same "long-form", chimeric guide RNA that's been widely used; these changes to the guide RNA are as follows:
1. U to A change at position #5 downstream of the DNA-binding portion. Purpose: Disrupts a 4-U stretch that may cause premature termination of guide RNA transcripts.
2. Extension of the proximal stem-loop by 5 base-pairs. Purpose: probably stabilizes interaction with Cas9.
This enhanced specific target labeling even while EGFP-Cas9 levels were reduced, which in turn lowered nonspecific background fluorescence. It's likely that these guide RNA modifications will generally increase Cas9 targeting efficiency in other context. Thanks Ian for bringing this to my attention.