Are you sure the SCL driver is failing? I do'nt think 340 ohms to ground with the chip powered off is problematic. With no power, the gates of the differential driver transistors are floating and could very easily float to a resistance like that.
What is the failure mode you are trying to debug? Lack of communications? Have you used an oscilloscope to ensure that the SCL line is truly not working?
EDIT: (Since I can't comment yet :( ) What value resistors are you using for biasing the differential bus?