o-stems vs. u-stems

Good morning or evening, everybody!

My name is Petr Strossa. I have observed this forum for some
months already, whithout having posted anything, but having had
great pleasure from many contributions. I am interested mainly in
PIE roots and morphology, just as an amateur, but have thoroughly
read one magnificent book on this topic and looked very many
times into various etymological dictionaries. (I have quite
a nice private collection of dictionaries, grammars and books
about languages, this is another of my hobbies...)

Now, I came to a question when reading Cyril Babaev's "*ano-"
Word-A-Week entry. It is presented as an o-stem, and I don't see
any reason why it should not originally (or: "generally") be that
(although several given forms, as Greek "annis" and Germanic
"anen", look differently) - but, what I am now especially
interested in: why does it work like a u-stem (anus, -u:s) in
Latin? While thinking about this, I found another similar case:

supposed *snusos (`daughter-in-law')
-> Greek "nyos"
- but Latin "nurus, -u:s", again looking like a u-stem (from
hypothetical variant *snusus?)

And a third example:
Latin "domus" mixing u-stem and o-stem forms (e.g., gen.sg.
"domu:s", but acc.pl. "domo:s")
vs.
Greek "domos" (a regular o-stem, I hope).
BTW, the corresponding Old Slavonic "domU" was a u-stem.

What was most probably the real relation between PIE o-stems and
u-stems, especially in such words as those given here?

1. (?) All or most of such words were originally quite
conservative consonant stems and later developed differently into
o-stems or u-stems in different PIE dialects.

2. (??) All or most of such words developed into o-stems in
common IE, but later changed quite regularly into u-stems in some
branches (as Italic). But why whould it be so - when u-stems do
not at all look much numerous and productive at almost any
place and period (except a relatively late case of some
adjectives in Lithuanian as I have read)? Could the reason be,
e.g., that feminine o-stems were so very uncommon in these
languages? (But again, Latin humus is a relatively "perfect"
feminine o-stem!)

3. (???) The u-stems were in fact generally older, and they were
slowly disappearing in all the PIE dialects, mostly by conversion
into o-stems.

Any suggestions?
P.S.