Hello YeBin,
Modeller states that you have more aminoacids in the alignment than in the pdb-file (353/352), and if you compare the fasta-sequence and the pdb-file you will find a lysine (K) at the C-terminus that is not present in the pdb-file. Delete the K from your alignment or change it to a "-" and Modeller should work fine.
Remember that there might no be all aminoacids present in a pdb-file for structural reasons: loops at the surface may be highly mobile, and can therefore often not be solved by crystallography, and that means you will have a gap in the structure. Before modelling, check the sequence.
You can use DeepView to write out the aminoacids present in the pdb-file by loading the protein and then saving the sequence in FASTA-format.
Best wishes, Thorsten