Enamel

Denisova 3 - and Harbin / longi - has been called out as containing a "super archaic" introgression. It had been suspected from Erectus if circumstantially. More circumstantial evidence came in a couple days back. Here's a release mostly from Kirsty Penkman of York; here's the article from China.

This is from proteins, such as Linus Pauling might have approved; not full DNA. These teeth are simply too old and rare to risk mulching for DNA, as we might for Neanders and Denisovans which are everywhere now. One protein is AMBN-M273V found in the tooth-enamel. And in some (mostly later) Denisovan DNA; but not in the rest of us. Another is AMBN-253G which isn't anywhere else, so is now available to tag a tooth as specific to Erectus.

The Neanders perhaps never met any Erectus, as the post-split Denisovans met-n'-matched them. Erectus seems met in southeast Asia, preEemian, maybe when it was Sundaland.

As a humble protein-set rather than DNA, this thesis remains circumstantial: evidence, but not "proof". One reviewer, Massilani, points out that parallel mutations be possible. But: how come these mutations outside Erectus be so rare (M273V) or nonexistent (253G)? Penkman asks similar questions.