I am designing a small board with a 915Mhz transceiver IC breakout (based on the CC1101 + CC1190 chipsets), controlled by an STM32. I have used the popular nrf24l01 chipset in the past, but was disappointed by the limited range and poor penetration through walls, so I decided to try using an amplified sub 1Ghz option.
The board has an nRF52 co-processor on board, to allow me to do higher bandwidth wireless debugging over BLE to a PC, as well as send over-the air firmware updates.
My first pass had both radios laid out next to each other in parallel, as shown in the image below. After routing, I started thinking that this is a poor choice in layout, since it places the two SMA antenna connectors parallel and in very close proximity to each other, which might cause RF interference (maybe even near field effects? harmonics?) that could significantly reduce the performance of each radio. I also thought that there might be interference with the SPI bus shared between the nRF52 and STM32 introduced by the parallel placement of the 915Mhz radio.
I then decided to modify the layout so that the antennae were at 90 degrees from one another, like this. My reasoning is that making the 'direction' of the RF circuits orthogonal to each other will minimize interference (with the added benefit of shorter and more isolated SPI traces). I also have some vague impression that cutting the board up with the radio sections in jutted-out strips like this will help prevent the copper on my board from sinking radiated emissions from the transmitters. 
I'm inexperienced with RF electronics and am unsure whether my logic is sound. Is one of these layouts better than the other? Does it not matter? (because the SMA antennae are hinged and manually adjustable, or because the large difference in operating frequency between the two radios means they're unlikely to interfere?)
