Blue-White Selection Step Enhances the Yield of SAGE Concatemers

In recent years, serial analysis of gene expression (SAGE) has been widely employed to examine mRNA transcription profiles of various cell lines and tissues (6,8,10–12). SAGE is based on the generation of a unique 10–11-bp tag following the most 3′CATG of each mRNA transcript in the tissue or cell sample. The tags are then linked together to form long concatemers and amplified in bacteria. A random sampling of bacterial colonies is selected for sequencing, and the number of times a particular tag sequence arises provides a measure of the relative abundance of the transcript in the mRNA population. Although theoretically simple, preparation of SAGE libraries through the multiple-step protocol (10) can be difficult. Accordingly, numerous modifications have been made to the original protocol to improve the yield of constructed SAGE libraries (1,5,7,9). During the successful completion and sequencing of five different SAGE libraries from rat pheochromocytoma cells [(2) and unpublished findings], we found that 13%–50% of the bacterial colonies selected for sequencing contained no inserts. In addition, a number of colonies contained small concatemer inserts (less than 15 tags). This significantly lengthened the time required to sequence our SAGE libraries. Here, we describe a simple improvement in the SAGE protocol that favors the selection of bacterial colonies containing larger concatemers for DNA sequencing and that greatly reduces the number of colonies selected that do not contain concatemer inserts. This modification can be used for all versions of the SAGE method that use the pZErO Vector System (Invitrogen, Carlsbad, CA, USA) and substantially increases the sequencing throughput of SAGE tags. The two main features of the pZErO vector are a Zeocin cDNA insert that provides resistance to the antibiotic Zeocin and a LacZα-ccdB fusion cDNA insert that, when expressed, kills vector-containing bacteria by inhibiting DNA gyrase activity (3,4). The insertion of cDNA into the multiple cloning site of the pZErO vector disrupts the inframe induction of LacZα-ccdB cDNA expression, thereby permitting the growth of bacterial colonies. However, we and others (Chang and Moore, personal communication, Columbia University) have found while analyzing our SAGE libraries that the pZErO vector can in some instances re-ligate without inserts and allow for subsequent growth of bacterial colonies even in the presence of the Lac promoter activator IPTG. We hypothesized that such growth can occur either because a subpopulation of the DH10B bacteria is resistant to ccdB or because religation of the empty vector results in a frameshift that abolishes expression of the LacZαccdB fusion protein. To distinguish between these two possibilities, concatemers were first subcloned into pZErO as directed in step 13 of the SAGE manual (version 1.0e; www.sagenet. org), and the ligation mixture was transformed into DH10B bacteria. However, rather than growing the bacteria on LB/Zeocin agar plates as usual, the plates were first treated with 40 μg/mL X-Gal and 1 mM IPTG. We reasoned that if a frame shift occurred during the ligation, then colonies without inserts would be white because of the absence of LacZα-ccdB expression. In contrast, if ligation were in frame, then bacterial colonies without inserts would be blue because of expression of the LacZα-ccdB fusion protein. For the library analyzed in Table 1, analysis of both white and blue colonies revealed that empty vectors were recovered only from blue colonies. This indicates that a subpopulation of the bacteria is resistant to the levels of LacZα-ccdB expression achieved in our cultures. Such resistance was apparently not due to genetic drift or adaptation of the bacteria in our laboratory because colonies with insert-free vector were recovered with at least three independent batches of bacteria received directly from the supplier (Invitrogen). These findings indicate that blue-white selection can eliminate or substantially decrease the prevalence of empty vectors encountered during the sequencing steps of the SAGE protocol. Analysis of white and blue colonies revealed that aside from empty vectors, many of the blue colonies also contained inserts (Table 1). Moreover, the presence of blue colonies was not uncommon. When we examined 392 random colonies, 54% were blue. A comparison of inserts present in blue and white colonies revealed an important difference, namely that the white colonies contained on average larger concatemers (Table 1). Analysis of concatemers from 38 white and 38 blue colonies showed that the average number of tags per concatemer is typically more than 50% greater in the white colonies than in blue colonies when colonies with empty vectors are considered in the calculation, and 40% higher when empty vectors are excluded. The reason for the difference in size of inBenchmarks