Avoiding HTMLEncoding and URIEncoding is not important for SEO reasons. Maybe there are search engines out there that doesn't understand those encodings, but such half-assed search engines aren't going to impact on your market success. Just worry about professional search engines; all professional search engines (and I'm not just talking about the big ones, all of them) understand how web pages work.
Beyond that, I'd make sure that every object the string is met by understands appropriate encodings (UTF-16 in terms of the first level of .NET processing, UTF-8 in terms of input and output to and from the web). However, it may be a good idea to have links based on the string output URI encoded (the %xx type of encoding, not the &#xXXX; encoding) as some user-agents out there have less issues with such URIs than with IURIs (they're both understood perfectly, and as equivalent by modern search engines) while more modern user-agents will hide the encoding from users, so you get backwards-compatibility without hurting users of modern browsers, or search engines.