Well, I'm not using Mule but I am using AMQ. I know it's not the same, but it should give you some ideas.
For sending, I use camel code in my portlets.
For receiving, I use a simple spring MDB to take the message and publish into the LMB; disconnects portlet code from having to deal w/ the JMS, etc.
The big problem for most people is that they really do not think their implementation through. Normally you'll have a portlet on a page and you want that portlet to send/receive. You write all the code and you think it's all good.
But in doing so you overlook the runtime implementation. Your portal admin drops the portlet on a page, and all of a sudden you have X number of end users w/ this portlet on the page and they're all separately trying to post messages and/or receive messages, but more often than not users want a specific message, not just any message.
Things get really bad if the portlet is dropped on a public page that guests can access.
So you really must consider your runtime implementation, not just how you handle the integration from a code perspective.
