[wtbob]

FWIW, this would be a literal s-expression translation:

    (mrow (mi x)
          (mo =)
          (mfrac (mrow (mo -) (mi b) (mo ±)
                       (sqrt (mrow (mi b) (msup (mi 2)) (mo -) (mi 4ac))))
                 (mrow (mi 2a))))

And this would be a saner one, where mrow is implied:

    ((mi x) (mo =) (mfrac ((mo -) (mi b) (mo ±)
                           (sqrt (mi b) (msup (mi 2)) (mo -) (mi 4ac)))
                          (mi 2a)))

I think either of those is clearly and inarguably superior.