A little more straightforwardly:
foreach (@array) {
s{/(MSS.*?)/}
{ my $m=$1; $m=~s/ +([^*])/$1/g; $m}ge;
}
Note that I'm using {} as delimiters for the main s///, and that what looks like an inner block of code is the replacement portion. The e option causes it to be eval'd.
Within the eval section, I operate on the matched string, and replace any spaces that are followed by something other than a * with whatever they are followed by.
Without the +, it would squeeze multiple spaces into one space, when followed by a *, and if not followed by a *, it would (of course) remove them all.