First, I’ll answer your question to me. Then I’ll comment on the other responses.
Assuming you build a subscriber with a tlm_fifo or analysis_fifo in it (actually, an analysis_fifo may be all you need), then you simply declare a get_port with the appropriate type parameter in your sequence. Then, after creating the instance of your sequence, you assign its get_port to the analysis_fifo.get_port and then your sequence can call get_port.get() to get the transaction out of the fifo.
There is still the question of synchronization. Assuming that each transaction being driven depends on a transaction being received, then you can use a fifo of depth one. Except for the first transaction being driven, of course.
I think a better explanation of why you think you need such a setup would be helpful. There may be better ways of accomplishing what you think you need.