Ligation independent cloning

I was following Nick Oswald's approach for ligation independent cloning (LIC) until I realized that his method can't really work. The reference to the original work that he linked to was okay and should be adhered to. I think Nick's method failed because he was thinking of adding a start codon and several bases just to bring up the number of bases to at least 15 bases of "overlap" (the N-term amino acid MRRGAP). However, when you studied the bases that he used in the example (I'm guessing that the method he showed was just a proof-of-concept, without any empirical value), one big problem arises.

The %GC content is extremely high (at 78.8% GC). This will proof to be a downfall because whenever users want to design primers that incorporate termini that is complementary to the LIC sequence, they will realize that the homo-dimerization will be a big problem (homodimer delta G = -22.78 kcal/mole). That is just for the forward primer with 5' extension of TGCGGCGTGGCGCGCCGCAA. This alone is already a big headache to many people. Another major problem is that there is a hairpin problem. The formation of hairpin is very likely! Imagine what will happen if more gene-specific bases are added to the bases already prone to hairpin formation...

I don't understand why Nick's approach requires incorporation of start codon. If LIC cloning is really versatile, addition of ATG to the sequence after the LIC complementary termini should be a breeze. 

I have made the vector with the LIC sequence, but I'm afraid I'm stuck with the primer design and PCR steps. I wished I had studied Nick's method in detail before embarking on the DNA vector construction. This entry is just to warn others to read Nick's approach with caveat. My advise is that if you want to incorporate LIC sequence, refer to this paper.