From briefly skimming the paper, it looks like the only use of lemma 4 is in lemma 5, where the author explicitly states that it is possible to prove without the Riemann hypothesis, if one is willing to work with a bound of about 10^6000000000000000000.