The best argument for DNSSEC is that it can be used to enable specific key discovery (and not in any actual security that it's providing per se). But it's not entirely clear that it's a good idea for that auxiliary purpose. Distributing PGP keys arguably would work better if you had the SMTP server report PGP keys when available (obviously, verification would have to come via external means). It certainly would be far faster to deploy this sort of "useful" service in an application-level protocol that all interesting clients already deal with, as opposed to requiring looking up another protocol that can be harder to access with standard APIs.