I have not used the pde functions myself. I have, however, used the ODE functions, and knowing something about those and reading the documentation https://www.mathworks.com/help/matlab/ref/pdepe.html it looks to me as if you should have time-dependent boundary conditions in a single call.
Instead, what you should do is make two calls with two different tspan. The first one would be out to 10 hours, and the second one would start from 10 hours and run to whenever. You would use the outputs of the first call as inputs to the second call. You would have arranged 1000 mg N/L as the value for the first call, but 0 as the value for the second call.