In earlier Latin the adjective was gnātus preserving the *ǵ as g, but by the classical Latin period g was regularly lost word-initially before n (cf also Classical Latin nōscō for earlier gnōscō)
As for *n̥ this is indeed the expected reflex in this position. To a limited extent there was a sonority hierarchy determining which consonants vocalised.
Glides, nasals, and resonants all vocalised the same way. Broadly speaking starting from the end of the word and working your way forwards you vocalise any of them that falls between two unvocalised consonants (or, with a few complications, between an unvocalised consonant and a word boundary).
It's only at this point that you determine which (if any) laryngeals should be vocalised (following the same procedure as for glides, nasals, and resonants).
In this instance, that means the n is vocalised (as it lies between the *ǵ and the *h₁), but the laryngeal is not (as it is immediately preceded by a vocalised nasal).
Adding syllable boundaries explicitly, we have *ǵn̥h₁.tós. Whilst *n̥ generally gives Latin en (or in if subject to vowel reduction), an exception is that *n̥H (where H is any laryngeal) gives nā.
Lastly, whilst the usual reflex of *o in Latin is o, it is raised to u when subject to vowel reduction (during the Old Latin stage when stress was still root-initial).
Putting that all together, we see that nātus is indeed the expected regular reflex of PIE *ǵn̥h₁tós.