Shell Programming and Scripting

I have a file with various syntax that I'm try to run sed on to change over a specific regexp and having problems. I want to swap all instances of a token so long as the token isn't followed by either an alphanumeric character, an underscore, or an exclamation point.

The test file looks like this:
test
test4
testF
test!
test;test

I'm trying to run sed like this
% sed -e 's/test[^[:alnum:]&^_&^\!]/FFF/g' test

What I see is this which is close but not exactly what I'm looking for:
test
test4
testF
test!
FFFtest

Notice the last entry does not have a ";" between FFF and test. Is there something I can use in the replace string to represent that I want to print out the negated matched expression found?

Code:

$ sed 's/^test\([^A-Za-z0-9_!]\)/FFF\1/' testfile 
test
test4
testF
test!
FFF;test

EDIT: To elaborate a bit, yes - the \(...\) in the LHS stores the matched negated regex result, which is printed in the RHS with \1.

Cheers,
ZB