You can use cheap standard CMOS circuits, e.g. 74HC595 or 74HC4094 as 8 bit output and 74HC165 as 8 bit input. You need only 3 lines or 4, if you want both inputs and outputs: data in or (and) data out, clock and strobe (update input or (and ) output latch.

This method was also known as SPI.
To expand to 16, 24, ... lines you connect the output to the input of the next IC and then transmit or (and) receive 2, 3, ... bytes.
Some 8051 derivates (e.g. AT89C8252) support SPI byte wise by hardware. But its no problem to build it bit wise in software.


Peter