regexec() trouble

All,

 It seemed that verifying email addresses with Regular Expressions would be
a good learning exercise, but so far all I've accomplished is to waste the
better part of a day. We already have email validation routines in place so
this project is really about learning REs. Would someone be kind enough to
take a look at this before I chuck the whole thing and go back to doing
something productive in RPG?

Background: Running a plain-vanilla V5R3 US QCCSID=37 machine. ScottK's
excellent REDEMO program runs fine so I believe the basic configuration is
correct. The following code works as expected:

RegPattern = 'test';
rc = regcomp(reg: %trimr(RegPattern): REG_BASIC+REG_ICASE+REG_NOSUB);
// regexec works l/c
Chr256 = 'test@xxxxxxxxxxx';
rc = regexec(reg: %trim(Chr256): 0: match: 0);
// regexec works u/c
Chr256 = 'TEST@xxxxxxxxxxx';
rc = regexec(reg: %trim(Chr256): 0: match: 0);
// regexec correctly fails this one
Chr256 = 'SHOULDFAIL@xxxxxxxxxxx';
rc = regexec(reg: %trim(Chr256): 0: match: 0);

Problem: The following pattern works correctly on web-based RE engines and
compiles without error in my test program, but won't match test strings when
called from RPG.

// From: http://www.regular-expressions.info/email.html 
RegPattern = '^[A-Z0-9._%-]+@[A-Z0-9.-]+\.[A-Z]{2,4}$';
rc = regcomp(reg: %trimr(RegPattern): REG_BASIC+REG_ICASE+REG_NOSUB);
// Why won't this work?!?                      
Chr256 = 'test@xxxxxxxxxxx';                   
rc = regexec(reg: %trim(Chr256): 0: match: 0);
 
My suspicions center on RPG mis-interpreting the '^' character in the
pattern, but I'm not sure how to verify it or fix the problem. Does anyone
have suggestions how to make this RE work correctly? Many thanks, JK