According to Wikipedia, the angular resolution is inversely proportional to the baseline (the separation between the telescopes). The order of magnitude of the baseline of the Event Horizon Telescope is 10,000 km; the Earth-Moon distance is about 40 times larger, and Earth-Mars at its maximum 25,000 times larger. That means we could theoretically reach 1.5 arcmicroseconds (at this wavelength) with an additional telescope at the Moon, and 2.4 arcnanoseconds with Mars.
However, to obtain proper interferometry we need (regardless of the baseline size) to know the telescopes' precise location with the precision of a quarter of the wavelength. For 230 GHz radio waves, the wavelength is 1.3 mm, so the precision needs to be 0.3 mm. I suppose that this is incredibly difficult to reach for bodies moving through space.
I am not well-versed enough in the matter to say what the effect of arranging the telescopes in a line (on Earth-Moon or Earth-Mars scale, it's a line, unless you add space telescopes at e.g. L4 and L5 Lagrange points) vs. a grid like the Event Horizon Telescope is. But interferometry does work for two telescopes already.
Note that angular resolution is not the only metric you should consider; objects become fainter when you enlarge them, so you'll need more telescopes (and/or a longer exposure time) to get the same level of brightness in the pictures.