The distinctions (between all the ways "meme" is used) are addressed pretty well in van Driem's papers and book chapters. In short, Dawkinsian "memes" are a very weakened version, mostly based on rote mimicking and copying. His meme was actually more strongly defined in the prior literature that he borrowed the term from and diluted.

The Leiden hypothesis leans on a harder version, in which memes are quite literally a new strata of life and our minds are co-evolved host organisms that straddle two layers -- a cyborg of biological and semiotic organisms. Our minds might be better conceived as akin to the fertile replication machinery of the biological systems below (ie. DNA polymerase) -- Where the linear packets of information are remixed and recombined after being delivered from the internal and external environment. And if there's some truth there, then it prompts lots of questions (e.g., about what is the most appropriate unit of life at the cultural/linguistic level. Because DNA polymerase is not what we normally consider "alive" -- we consider the aggregate cell alive. etc. etc)