You need to use a force field that describes such a species, which is not common. pdb2gmx is only able to understand the residues in the force field's .rtp file; if you have (or add) a deprotonated tyrosine there (which also requires a new residue name), then you can accomplish this.
I'm using Gromos force field . I tried to add my residue to it using GROMACS instructure but I faced a problem . The problem is the paramaterization of the residue .
First, I tried to assign charges manually but I didn't find charges of O-1 . Then I tried to use PRODRG, which failed to paramaterize the charged group . Eventually, I tried to use ATB server, which the job was killed and failed to parameterize the residue .
You'll have to parametrize phenolate manually. That would probably involve some QM calculations to get suitable starting charges that can be refined empirically based on experimental target data. The problem will be finding suitable target data. Normally, Gromos96 parametrization (and you haven't said which parameter set, which is important) is based on condensed phase free energies of solvation. I'm not sure if such data exist for phenolate. Parametrizing molecules under Gromos96 can require a great deal of empirical adjustment and can require considerable effort.
That's probably a reasonable start, but parametrizing full residues is not how the force field is generally adapted. Amino acid side chain analogs are used to parametrize new residues. That is, the backbone parameters are unchanged for all of the amino acids. The compound shown there requires changes to the backbone, which are not likely acceptable within the context of the normal procedure. Parametrization of just phenolate would suffice. Parameters from ATB are generally pretty good, but everything should always be validated as thoroughly as available experimental data allow.
Consider general recommendations here: http://pubs.acs.org/doi/abs/10.1021/ci100335w