I did install the Zimbra Beta on my test server.
It seems no problem that the Zimbra supports Japanese.
But I found character encoding of message body is not optimal.
For character encoding you are using utf-8 and quoted-printable for message body with 2-byte character set.
"quoted-printable" is good for messages which are mainly 7bit character set because "quoted-printable" does not change 7bit code. But you are using "Content-Transfer-Encoding: 7bit"(MIME) with 7bit message, so base64 encoding is a better choice because of lower overhead for "quoted-printable".
"quoted-printable" makes 8bit code 3x larger. Since a 2-byte character is represented by 3-bytes or 4-bytes with utf-8, after "quoted-printable" encoding it becomes 9-bytes or 12-bytes. That means 2-byte character set will be up to 12-byte with utf-8 and quoted-printable!
If you can substitute Base64 for quoted-printable, the same higher order utf-8 character can be represented with only 4-bytes to 8-bytes and we can save storage device.
Could you please use Base64 encoding instead of quoted-printable for 2-byte character set for us 2-byte guys?
I guess this is very important for 2-byte character set area.