Reliability of MD simulation

GROMACS version:
GROMACS modification: Yes/No
I have sent my manuscript to a journal and reviewer commented this:

Only a single, relatively short (100 ns) simulation was performed for the beta-water and AMX-bound beta complex. This is understandable, as MD simulations are time-consuming and often prevent longer simulations, or multiple simulations, to be performed in a timely manner. Nonetheless, the authors should comment on, and acknowledge, the possible limitations of such a short simulation and whether the predicted results are reliable.**
How I can respond to this?
Note: The RMSD of both systems have reached with the equilibrium state at 65 ns.

The required time scale of any MD simulation depends on the dynamics you want to investigate. If the reviewer thinks that 100 ns is too short to assess the system, there is no real getting around this. The argument is basically that the simulation is too short and you’ve only done one simulation. Experiments require repetition, and so do simulations. You have made one observation and have not determined the reproducibility of your observation. That’s the problem. Single MD trajectories are generally not considered publishable on their own. In terms of statistics, what if the outcome of your one simulation was the outlier? You have no way of knowing.

RMSD does not tell you equilibrium. And if it did, it tells you that you have 35 ns of usable data, not 100 ns. This is even more problematic.

My recommendation is to run more replicate simulations of suitable length. Convergence should be assessed more robustly since RMSD does not tell you this.