I am searching a protein sequences against a nucleotide database ("tblastn") using mmseqs and need nucleotide sequence alignments as an output.

I start by the following (pg 23 of https://mmseqs.com/latest/userguide.pdf)

mmseqs search ecoli_proteins ecoli_genome alnDB tmp

and then get alignments as follows:

mmseqs convertalis ecoli_proteins ecoli_genome alnDB alnDB.html --format-mode 3

However, this gives me protein sequence alignments, when I really want the nucleotide base alignments (i.e. would be aligned codon-by-codon).

What is the proper way to get this output from mmseqs?