In theory, you can't edit directly a sequential file. You would need to read the input file, preferably line by line, edit the lines and write them to another file, and either before or after that, do the file renaming if needed (another option is to read the input file into memory, do the changes into memory, close the file and then reopen it in write mode, which will erase everything in the file, and write the stored lines into the newly created file, but this other option may not be used if your file is so big that it will not fit in memory).

So you would need something like this:

# ...
open my $in, "<", "uuu.txt" or die "cannot open the input file$!";
open my $out, ">", "out.txt" or die "cannot open the output file$!";
while (my $line = <$in>) {
     $line =~s/[\s\d]//g;
     print $out "$line\n";
}
close $_ for ($in, $out);
# ... renaming, etc.
[download]

However, this being Perl, you have some command line options which will do the boiler plate code (opening and closing the files, reading input line by line, and file renaming) for you behind the scene, so that this one-liner (or the almost identical one proposed by vinoth.ree) will do everything for you:

perl -i.bak -p -e 's/[0-9\s]//g' uuu.txt
[download]