**3.3 Connection of two adjacent BACs in BGM**

Itaya et al. (2008) documented that the integration of two partially overlapping fragments is possible in BGM. Overlaying the second on the first fragment in BGM resulted in elongation or connection. Each fragment is called a domino. If serial dominos are toppled, or all dominos are connected in BGM, reconstruction of the full-length guest DNA covered by these dominos completes. This concept was first realized by using dominos made in the pBR322-based plasmids, pCISP401(*cat*) and pCISP402(*erm*) in *E. coli*. Up to 31 dominos designed to completely cover the 135-kbp rice chloroplast genome produced a reconstructed full-length guest DNA (Itaya et al., 2008). The use of BACs instead of pBR322 should work in a similar manner. The scenario for connecting two adjacent BACs in a BGM vector is shown in Fig. 6a. However, selection markers present a problem in the preparation of BACdominos for immediate use. pBR-dominos are commonly prepared from PCR products of less than a few dozen kbp. However, BACs normally carry 100-kbp DNA fragments that exceed the limit of PCR-mediated amplification. Therefore at present we are forced to use BAC clones, such as commercially-available mouse genomic BAC libraries. However, their BAC vector does not possess *ab initio* selection markers for *B. subtilis*.

Fig. 6a. Connecting two adjacent BACs in BGM.

Two BAC clones, pKANEG (196 kb) and pKANEH (220 kb), cover the mouse genomic *jmj*  region; there is a 60-kb overlap sequence. Each BAC was individually integrated into BGM. Transformation of the BGM carrying the 220-kb BAC2-DNA by using purified genomic DNA from another BGM carrying the 196-kb BAC1-DNA leads to homologous recombination between the 60-kb overlapping region and the sequence shared with the *B. subtilis* genome portion (see the splice boxes). This results in the production of a reconstructed mouse genomic *jmj* region (355kb). BAC insertion after I-*Ppo*I digestion was confirmed by agarose gel electrophoresis (right panel). The I-*Ppo*I recognition sequence is indicated by I.

Fig. 6b. Summary of the current achievements made with the mouse a *jmj* gene.

As shown in Fig. 6b, the connection of two sequential guest BACs was first demonstrated for the two BACs covering the mouse gene *jumonji* locus (Kaneko et al., 2009). The two inserts, 196 kbp- and the 220 kbp guests of BGM, shared DNA approximately 60 kbp in size. While the sketch illustrating the connection looks simple, there are difficulties in using marker genes for *B. subtilis*. The combinatorial use of antibiotic markers, already reported by Kaneko et al. (2009) is omitted in Fig. 6a. Instead, the difference from the pBR322-based domino-connection is clearly shown: total genomic DNA isolated from one domino BGM was added to competent cells of the second domino BGM to force double homologous recombination. The transformants obtained by rational antibiotics selection produced a 355 kb-long connected BAC (Kaneko et al., 2009). A summary of our current achievements with the mouse *jmj* gene is presented in Fig. 6b; the most time- and labor-consuming step is the integration of unmarked BACs into the BAC-BGM. Besides these two elaborated experimental works using regular BACs, we have prepared BAC vectors that are designed to accomplish versatile aims in the BGM system. The two new BAC vectors, p108BGMC(*cat*) and p108BGME(*erm*), feature the *B. subtilis* markers shown in Fig. 3. Their presence should facilitate the domino-mediated elongation/reconstruction of BAC-based guest DNA (our unpublished data).

### **3.4 Implementation of sequence engineering in BACs (inversion)**

What can we do with guest DNA? Figure 4 presents a list of possible modifications. Among them, techniques to induce the inversion of guest DNA appears as important as elongation. It is difficult to regulate the orientation of the inserted DNA in BACs. As this difficulty is frequently encountered at the construction of random BAC libraries, tools are needed to invert the insert present in BGM. The method and timing for the induction of large regions of the host *B. subtilis* genome have been described (Toda et al., 1996; Kuroki et al., 2007); they were applied to reverse the orientation of BAC inserts in BGM. Two partially overlapping

Fig. 6b. Summary of the current achievements made with the mouse a *jmj* gene.

elongation/reconstruction of BAC-based guest DNA (our unpublished data).

**3.4 Implementation of sequence engineering in BACs (inversion)** 

As shown in Fig. 6b, the connection of two sequential guest BACs was first demonstrated for the two BACs covering the mouse gene *jumonji* locus (Kaneko et al., 2009). The two inserts, 196 kbp- and the 220 kbp guests of BGM, shared DNA approximately 60 kbp in size. While the sketch illustrating the connection looks simple, there are difficulties in using marker genes for *B. subtilis*. The combinatorial use of antibiotic markers, already reported by Kaneko et al. (2009) is omitted in Fig. 6a. Instead, the difference from the pBR322-based domino-connection is clearly shown: total genomic DNA isolated from one domino BGM was added to competent cells of the second domino BGM to force double homologous recombination. The transformants obtained by rational antibiotics selection produced a 355 kb-long connected BAC (Kaneko et al., 2009). A summary of our current achievements with the mouse *jmj* gene is presented in Fig. 6b; the most time- and labor-consuming step is the integration of unmarked BACs into the BAC-BGM. Besides these two elaborated experimental works using regular BACs, we have prepared BAC vectors that are designed to accomplish versatile aims in the BGM system. The two new BAC vectors, p108BGMC(*cat*) and p108BGME(*erm*), feature the *B. subtilis* markers shown in Fig. 3. Their presence should facilitate the domino-mediated

What can we do with guest DNA? Figure 4 presents a list of possible modifications. Among them, techniques to induce the inversion of guest DNA appears as important as elongation. It is difficult to regulate the orientation of the inserted DNA in BACs. As this difficulty is frequently encountered at the construction of random BAC libraries, tools are needed to invert the insert present in BGM. The method and timing for the induction of large regions of the host *B. subtilis* genome have been described (Toda et al., 1996; Kuroki et al., 2007); they were applied to reverse the orientation of BAC inserts in BGM. Two partially overlapping fragments [ne] and [eo] derived from a neomycin resistant gene [*neo*] play an essential role; [ne] and [eo] are inserted at the terminus of the insert to be inverted in the BGM. Because these two fragments have an identical region designated [e], homologous recombination here produces two segments, [neo] and [e], that accompany the inversion of the intermediate insert between [ne] and [eo] (Toda et al., 1996). The inversion formation is always associated with the formation of [neo] and can be monitored by resistance to neomycin. This manipulation, theoretically simple but complex in its application, is one of the key technologies for BGM. We have already applied this tool to BACs covering other mouse genomic regions (unpublished findings).
