How can I use regex groups in the "Fields with name matching regex"

I use time series:
irate(collectd_tail_exim_total{instance=~"{instance}.+", job="exim", type=~"{exim_types}"}[3m])
Legend:
{{ instance }} - {{ type }}
and I get field name like this:
mx02.mail.lan:9103 - rejected_rcpt
I want to cut this string, because it too long.
I go to the “Overrides” tab and type in the “Fields with name matching regex”:
(mx…).+(- .+)
but in the “Change the field or series name” I have not found how to use regex group, I try \1\2, $1$2, $$1$$2, but nothing work.
It is possible? Or maybe have exists another way, for example “Transform” tab?