You might need to provide more details about your experimental goal and methods to get useful feedback on the question. Are you doing a screen to identify modulators of RNA Pol II activity (like a dropout screen), or looking to specifically disrupt function of genes related to RNA Pol II? For the gene selection bit, relying on GO categories only might be a blunt instrument that'll give you both false positives and negatives depending on what your experimental goal is. Are you interested in targeting RNA Pol II regulators or associated genes that might well be in different GO categories than the ones you list? There are commercial providers of sgRNAs and/or sgRNA libraries (e.g. Thermo, Synthego) from which you can order pre-prepared libraries. Different guide sequence selection strategies can apparently lead to better or worse editing performance, so paying attention to the design approach used (for commercial libraries) or selecting the best bioinformatics design approach (for doing it yourself) would also be important. You can of course also check out the literature to see what others have done.