I'm trying to use the pdfcomment package in XeLaTeX. I can't seem to be able to use Unicode characters such as Czech ň or ě properly in \pdfmarkupcomment. Usually any such trailing characters of the comment disappear. Here's a minimal example:
\documentclass{article}
\usepackage{fontspec}
\usepackage{polyglossia}
\setdefaultlanguage{czech}
\usepackage{pdfcomment}
\hypersetup{unicode}
\begin{document}
Běžně. Běžně.
\pdfmarkupcomment{Běžně}{}. Běžně. % This would be typeset as "Běžn. Běžně" leaving out "ě"
\end{document}
The pdfcomment package documentation says:
Internally, the argument ⟨comment⟩ needs to be converted to PDFDocEncoding/PDFUnicode.
and then it just says \hypersetup{unicode} in a footnote to this, which is what I put in my LaTeX document, but it does not seem to solve the problem. (I have no idea what "needs to be converted to PDFDocEncoding/PDFUnicode" means.)
Alternatively, if there is a different way to highlight arbitrary parts of text (i.e. without creating a box around it), that would work OK with Unicode, I will be happy to use it.

\pdfmarkupcommentwants two mandatory arguments. The problem seems to be a bug insoulpos. – egreg Mar 08 '15 at 14:50