This allows a user to find the best matches for the V gene in your query sequence among additional non-germline databases (i.e., nr, genome, etc). This option has NO effect on search against germline gene databases (see explanation below).
A typical rearranged query sequence includes a leader, the V, D, J gene (sometimes the C region is also included). When a sequence is submitted for blast search, the similarity matches will be performed over the entire query sequence. Unlike the germline V gene database which only contains the V gene sequences, other databases such as nr contain many rearranged sequences that also include a leader, the V, D, J and C genes. As a result, the best hit from these databases does not necessarily have the best match to the query V gene; Rather, it has the best match over the entire query sequence (For example, it may have very high similarity to the leader, D, J or C genes in a query sequence but only a low match to the V gene). This is not a problem if the goal is trying to find the best overall matches to a query sequence. However, if the goal is to find best matches to the V gene of a query sequence, then one needs to isolate the V gene part manually from a query sequence and then use it for a search.
With this option on, the V gene part from a query sequence is automatically isolated (based on comparison to hits from the germline V gene database) and then used for search against additional databases like nr. This option should be disabled, however, if the search intention is to find best hits based on overall matches.