Enzymes in general have substrate binding sites that are specific to a particular substrate. As stated in the question, the NdeI restriction enzyme recognizes the palindromic sequence 5’ CATATG 3’. This implies that the enzyme must have a region that interacts with the bases in that sequence, most likely involving hydrogen bonding (since these are the strongest type of intermolecular interactions, all of the nucleotide bases are capable of hydrogen bonding, and amino acids, which make up the enzyme, also hydrogen bond).
The selection of palindromic sequences is interesting. The sequence is a palindrome in that the complementary strand of DNA, read in the other direction/backwards, is the same. That would matter if the enzyme is able to interact with and recognize both strands of DNA simultaneously, looking for the same sequence on both.
The article I have linked below describes type II restriction endonucleases, of which NdeI appears to be one, as homodimers. This means they are made of two identical polypeptides (same amino acid sequence and same folded shape). Such a dimer would have two identical binding sites and would be capable of recognizing, and binding to, the same base sequence occurring on both strands.
If quaternary structure has been covered in your course, you might feel you could infer the dimeric structure of type II restriction endonucleases. Otherwise, you might want to limit your answer to discussing how the enzyme probably interacts with the DNA.
The linked article, although technical in its language, includes other interesting information about the function of these enzymes: for example, that non-specific binding to the phosphate “backbone” of the DNA is part of the process of aligning with the nucleotide bases and eventually recognizing the correct sequence.