Unexpected pKa values and Coulombic interactions for amino acid residues

[rankgab]

Hello everyone!

Short story:

I am using two very similar (almost identical) PDB files as inputs to determine pKa values for amino acids, but the pKa values I get for some amino acids are very different. The residues that are listed in Coulombic interaction column in the output file are not the same as well.

Long story:

I am using PROPKA (version 3.5.0 or version 3.5.1, but the result is the same) to determine the pKa values for amino acids in PSI Lhca4 complex in pH = 7. I use two different crystal structures of PSI: PDB ID: 4XK8 and PDB ID: 5L8R. In both cases I only take chain 4, so I create two new PDB files, containing only atoms from Lhca4 complexes. Both structures are very similar.

I use simple command for both cases:
propka3 Lhca4.pdb

The difference in pKa that worries me particularly is for GLU153. For 4XK8 structure I get pKa = 7.14; for 5L8R structure I get pKa = -0.48. Here are some snapshots of the outputs:

4XK8

5L8R

What I notice is that not only I get different pKa values for GLU153, but different residues are listed in the Coulombic interaction column as well. I expected the pKa values (and Coulombic interaction section) to be similar since the crystal structures are very similar. Here are the two structures matched on top of each other (the atoms listed in Coulombic interaction column for 5L8R are depicted in yellow):

Question

What might be the reason for such differences in pKa values?

I am also adding the output files for both structures (as .txt files).
Lhca4_4XK8_pka.txt
Lhca4_5L8R_pka.txt