Should I create user stories as super tasks and then create subtasks with the actual
tasks/requirements?
Yes, same reason as @superM gave. Our team actually puts all the acceptance criteria and requirements in the user story (super task) and then create subtasks to outline what exactly the developers need to do to accomplish that story (eg, create table, add service to do X, tweak server params, etc).
We also use the GreenHopper plugin for JIRA (http://www.atlassian.com/software/greenhopper/overview) to track our Epics and Stories in the backlog. Once the Stories are estimated with Story Points and committed to a Sprint then the team creates sub-tasks under each Story for what they will actually have to do to get the software to implement the feature. The GreenHopper plugin does a great job visualizing the sub-task to User Story relationship. However, even without the plugin, vanilla JIRA will let you specify a main task (call that your story) and then sub-tasks (what the team knocks out during the sprint).
Should I estimate super tasks/user stories, or only the sub tasks?
We estimate story points at the Story level only (the GreenHopper plugin adds a custom field for that value, but if you don't use the plugin you could create a custom field if you wanted). But on our team, as the developers complete their sub-tasks under each story, they log their work (eg, how many hours it took them) so we can report on it later. We never log work (hours) to the story (super task) and we never estimate the hours to complete the sub-tasks or stories going into them, it's always post-mortem in our case.