Targeting of proteins to the endoplasmic reticulum (ER) usually requires N-terminal signal peptides (SP) of approximately 22 amino acids in length. However, a substantial number of proteins contain exceptionally long SPs of 40 amino acids and more, an example being protein shrew-1/AJAP1. Using shrew-1's SP as example, the NtraC model has been developed by dissecting long SPs into two functionally distinct subdomains ("N" and "C") separated by a β-turn rich transition area ("tra"). Further proteins have been identified by computational analysis complying with the NtraC model. Here we used the SPs of two of these proteins, DCBLD2 and RGMa (including three isoforms), to show that the NtraC model applies to a growing group of SPs. We demonstrate that the full-length SPs of RGMa and DCBLD2 are functional and furthermore that the C-domains are sufficient and essential for ER targeting, whereas the N-domains are dispensable. Thus, the N-domains are available for additional functions.