The first would work if you provided an extra set of parens or unary plus to make it parse the way you intended it to in light of the optional paren behavior.

$ perl -MO=Deparse,-q,-p -e 'print +(1,2,3)[1];'
print(((1, 2, 3))[1]);
-e syntax OK
$ perl -MO=Deparse,-q,-p -e 'print( (1,2,3)[1] );'
print(((1, 2, 3))[1]);
-e syntax OK
[download]

@U = keys ('A'=>1,'B'=>2,'Z'=>26);
[download]

@U = keys %{{ map { $_=>1} @NU }};
[download]

Not with parenthesis, which create lists. The %{ { ... } } is the shortest way I know of.

Perl 6 - second systems done right

@U = keys ('A'=>1,'B'=>2,'Z'=>26);

That's the same as getting every second item (starting with the first) and removing duplicates. There's no good shorthand for that, even if you omit the requirement to remove duplicates.

@U = keys %{{ map { $_=>1} @NU }};

If the contents of @NU has no duplicates, then that's the same as

my @U = @NU;
[download]

If your question is how to remove duplicates, then there's an answer in perlfaq4. Some solutions

my @U = uniq @NU;        # From List::Util
[download]

I use the idiomatic

my %seen;
my @U = grep !$seen{$_}++, @NU;
[download]