I do fraction collection on four analytes from a (treated) biological extract. AKA, my analytes are disgusting and have a pretty large range of polarities (for RP). I mainly focus on prep work, but do some analytical work when it is needed as well, but this response will be focused on primarily prep experiments and phenomena that I've observed.
I frequently inject samples of 20-100 mg/inj (200ul)-- the samples are dark prior to injection. For my injection solvent I actually use an almost purely organic solution (90% ACN + 1% acid), while my method is mainly isocratic (25% ACN/75% H2O 0.15% Formic Acid). When I first experimented with this method and the high-org injection solvent I had started with much higher concentrations (500mg/ml), but I would run into an issue, as you mentioned, in which my analyte would not interact with either my mobile phase or stationary phase and would elute at the dead-time like an immiscible bubble. I suspect some tinkering will have to occur on your methods, but I would attempt something similar overall. I'm sure that depending on the solvent and your analytes this affect could either be pronounced or reduced-- for example I still use ACN as my ORG modifier, but DMSO isn't your MP solvent. Additionally, my analytical lab frequently injects samples onto our RP column in DMSO of a pure, proprietary compound, as the analyte is not soluble in water or ACN, but always are able to get the sample back at it's exact retention and with barely any sample loss. I would say set-up some experimental conditions and give it a go.