If you bitwise or (|) an uppercase letter with a space, (assuming latin-1/ASCII files), it will lowercase it:

print 'ACGT' | '    ';;
acgt
[download]

So, if you translate all the 'N's in your mask to spaces and then bitwise or the sequence and the mask, it will achieve your goal very efficiently:

$s = 'GGTACACAGAAGCCAAAGCAGGCTCCAGGCTCTGAGCTGTCAGCACAGAGACCGAT';;
$m = 'GGTNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNT';;

( $mm = $m ) =~ tr[N][\x20];;
print $mm;;
GGT                                                    T

print $s | $mm;;
GGTacacagaagccaaagcaggctccaggctctgagctgtcagcacagagaccgaT
[download]

Which makes your entire program (excluding the unmentioned fact that your files may be in FASTA format):

#! perl -slw
use strict;

open SEQ,  '<', 'data1.dat' or die $!;
open MASK, '<', 'data2.dat' or die $!;

while( my $seq = <SEQ> ) {    ## Read a sequence
    my $mask = <MASK>;        ## And the corresponding mask
    $mask =~ tr[N][ ];        ## Ns => spaces
    print $seq | $mask;       ## bitwise-OR them and print the result 
}
close SEQ;
close MASK;
[download]

Redirect the output to a third file and you're done.