Looking at your routing...
When the "S/RF" meter climbs, that may be more of one of the many Xtals in the Crystal Filter units sliding off frequency - simply because the Xtals inside are getting out of tolerance.
Ok, for your area to look for USB / LSB - you getting power from that Diode D207, D208 and D209 in the various modes?
Well, AM/FM are shared, looks like D207 handles that. D208 is your USB Diode to feed the Oscillator Q187-EDIT - but if LSB is ok, then you may have more of an issue with RX toggling over to SSB mode...
LSB may be fine due to it's "shared" close frequency but if AM is not knocking down (as in latched) - when you check the "mode" by bumping the switch to move the mode selection on the display - look at PIN7 of the 4558 chip - the TX toggle stuff. IF Pin 7 goes High that's one thing - but if it does not change states as the PLL tries to "lock in" - you have more of an UNLOCKED state when you go to USB.
Verify Pin 7 for that - the other modes would show one voltage while the USB mode would be another...
IF that's the result - so it may be more of a failed component because it's not toggling on and the Oscillator Q187 isn't sending anything to the board because of several components not sending their "influence" to Q187 when in USB. LSB is the only other mode to have a condition to work so check and see if it's sending power to the Xtal - for is USB isn't - the Xtal at Q187 "dies".
See D208 - and you may need to MEASURE AT THE XTAL X103 for a tickle voltage - that would drop in USB mode - meaning failed part - but if AM/FM forces power due to a blown shorted diode - then that region needs to be checked - again depending on condition of 4558 Pin 7...
Even AGC would "Stay engaged" showing carrier - not just due to a blown diode but lack of power to local Q187 oscillator from the other mode to power Q187 and to tell it (CPU) "I'm working" - Out Of Lock (OOL) condition from the CPU.
What I'm trying to tell you from the above is, it may not be from a "Shared line" due to a shorted diode, more like a blown part and Q187 can't ring anything to the rest of the board because X103 isn't getting a "tickle voltage" to make it ring. Verify that X103 receives a voltage from that D207, D208 (USB) and D209 (LSB) as you toggle the modes - any failures show up as a change in the voltage for that mode.
An easy way I do this...
Look for and take a marker to "Dot" D208 on the foil side of the board. Note which side of D208 goes TOWARDS X103 - that should be the BANDED side of the diode - keep your test leads' POSITIVE on that diodes BANDED pad trace - then clip a NEGATIVE lead to the common foil ground or even the case the damn things a messy noisy ground anyways - NEGATIVE GROUND ONLY radios do that... Just bump the mode switch and look for that X103 receiving voltage.
Now, if you have a shorted diode - the power at the Positive lead will rise because there is more sections of the board sending power to the X103.
A blown diode will show up right away and that means you also have proper isolation of all the toggling functions so you're ok. Just back track until you find the USB power feed. This is why you need to look at Pin 7 of the 4558 chip - tells on the bad section because it won't TX right.
