If you don't mind losing the tab within the quotes, pre-process the string to remove those tabs. Here I replaced the embedded tabs with spaces, then just split on tab:

my $var='474627    asidase ta    sidase ala,"lpha-D- ctoside gtohydrol
+ase","razyme","arazyme    (enz Corp)","Melie","lagal","idase bta",   
+ rug    00103';

my $tmp;
$var =~ s{ ("[^"]+") }{ ($tmp = $1) =~ s/\t/ /g; $tmp }xge;

my @each=split(/\t/,$var);

for my $eachvar(@each)
{
    print "$eachvar\n";
}
[download]

Update 1: Oops, I made a mistake in the pattern. The quotes belong on the inside of the capture. (Was: "([^"]+)", Now: ("[^"]+").

Update 2: In response to a private message, here's a little better explanation of the pattern:

# Using s{}{} form of substitute.
# Substitute supports using several different separator formats
# which helps one avoid having to escape things (like '/') within the 
+pattern.
# The 'x' option which means ignore whitespace so that comments can be
+ easily inserted.
# The 'g' option is global obviously.
# The 'e' option says that the replacement part of the pattern is a pe
+rl expression.

$var =~
    s{
        ("[^"]+")       # Matches two quotes and content between them.
                        # Capture the match for use in the replacement
+.
                        #
                        # Disection of pattern:
                        # ("[^"]+")  = full pattern
                        # (       )  = capture everything between pare
+ntheses.
                        #  "     "   = quotes at start and end of patt
+ern.
                        #   [^"]+    = one or more non-quote character
+s
    }
    {                       # The replacement part is a perl expressio
+n.
                            # Original: ($tmp = $1) =~ s/\t/ /g; 
                            # is same as next 2 lines:
        $tmp = $1;          # Make a copy of the captured match.
        $tmp =~ s/\t/ /g;   # Replace tabs with spaces throughout the 
+match.
        $tmp;               # Use resultant value for replacement.
    }xge;
    # x = ignore white space and comments
    # g = global
    # e = expression
[download]