Thousands of years ago, Northwest Indo-European tribes had a hunting taboo on the word for bear. Northwest Indo-European tribes were the peoples that became the Germanic and Balto-Slavic peoples, who migrated from north of the Caucasus to northwestern Europe. Because they believed it might jinx the hunt, the real name for a bear was never used, and ultimately disappeared.
So while other Indo-European tribes retained the native word for a bear, Germanic and Balto-Slavic people employed euphemisms to avoid saying it. Thus the word was lost forever. English bear came from Old English beron "the brown (one);" Russian is medved is from Proto-Slavic *medvě̀dь "honey-eater." But what if the Germanic people had not lost the native word? What if it survived? What would it look and sound like? Fortunately, the native word for bear is so well-attested in other languages that it is possible to speculate - albeit cautiously.
So to go from Proto-Indo-European to Proto-Germanic, there are two directions one can proceed. The first:
1. Loss of laryngeals, palatovelar merges with velars as *ḱ > *k, epenthetic *u vowel before sonorant *r: *urt-ko-s
2. Grimm's Law *t > *þ but *k is unaffected: *urþ-ko-s. Suffix *-ko- fossilizes sometime after.
3. Nominative suffix vowel shift and voicing of the sibilant: *urþkaz
Now, the thing I'm not positive about is whether *-tk- would undergo metathesis early on and become *-kt-. If I understand Ringe correctly ("From Proto-Indo-European to Proto-Germanic" 2006), he believes that metathesis would have occurred. Ringe points to Greek metathesis arktos < *artkos but Beekes ("Etymological Dictionary" 2008) interpreted Latin ursus as the product of cluster simplification rather than metathesis. And while there are *-tk- words that underwent the change, there are also PIE *-tk- examples that did not metathesize on its journey into Proto-Germanic (cf. *maþka- "maggot;" coincidentally, there was a post-Germanic metathesis in English maggot when *-þok- > *-kot- > *-(g)got). I must admit, I'm out of my comfort zone here so I will attempt a second reconstruction assuming Ringe's metathesis:
1. Loss of laryngeals, palatovelar merges with velars as *ḱ > *k, epenthetic *u vowel before sonorant *r: *urt-ko-s. Suffix *-ko- fossilizes sometimes shortly thereafter.
2. Metathesis of *-tk- to *-kt-: *urkto-s.
3. Grimm's Law *k > *h but *t is unaffected: *urhto-s
4. Nominative suffix vowel shift and voicing of the sibilant: *urhtaz
So there you have it. Two separate reconstructions for a Proto-Germanic word for "bear:" *urþkaz and *urhtaz. Just goes to show you how difficult a reconstruction can be, even among the best-attested etyma!