there is no other choice than making use of garnet s.s.s. found besides omphacite in eclogite samples. In case of eclogite-amphibolites using hornblende is feasible but you have to keep in mind that this lithology already attests to a retrograde process and it is not identical with the formation of the eclogites, proper. Pure rutile does not offer any possibility to accommodate tetravalent U. There is only one mineral reaction (Pronto reaction) which leads to brannerite (U,Ca) (Ti,Fe)2O6 .This mineral has , however, no meaning for this environment of metamorphism which you are going to study. Brannerite forms from TiO2 and UO2 in the course of ilmenite alteration according to the succeeding reaction: Brannerite formed according to the so-called Pronto reaction between 200 and 300°C from epigenetic alteration of ilmenite 2 FeTiO3 + 4 H2S + UO2 = UTi2O6 + FeS2 +H2O + 2H2 (Ramdohr 1975). This U titanate occurs in a wide range of environments, from paleoplacers such as Witwatersrand/South Africa through Variscan vein-type deposits to Cenozoic Fe deposits (Hüttenberg). I have to note that even this U compound is often a hard nut to crack in case of dating, because of the presence of another sort of alteration products (U "leucoxene"). This is the only mineral aggregate where you can expect U oxide together with TiO2. We made some attempts but the results were not very much promising with a high error bar.
Apart from the inherent solubility of U in certain minerals, the other important parameters are the amount of U in the whole rock (typically low in basic rocks), and the presence/absence of other minerals that can accommodate the element of interest. I found, for instance, that apatite from an U-ore had a low U content, likely because other minerals formed that had a far greater affinity for U. Similarly, a colleague told me that phlogopite from a calcio-carbonatite was excellent for Ar-Ar dating, because it contained no calcium whatsoever, as all the calcium had partitioned in coexisting diopside.