A string in Java is always in Unicode (UTF-16, effectively). Conversions are only necessary when you're trying to go from text to a binary encoding or vice versa.
What's the character involved? Are you sure it's even present in ISO Latin 1? If it is, I'd expect that character to be stored by your database without any problem. There's no such thing as a "UTF-8 trademark sign". You could have "the bytes representing the trademark sign UTF-8 encoded" but that would be a byte array, not a string.
EDIT: If you mean the Unicode trademark character U+2122, that's outside the range of ISO-Latin-1. There's the registered trademark character U+00AE, which isn't the same thing (either in appearance or in legal meaning, IIRC) but may be better than nothing - if you want to use that then just use:
string replaced = original.replace('\u2122', '\u00ae');