Abstract
Gene trapping is a random insertional mutagenesis strategy that aims
to identify novel genes and analyse their function. It usually involves the
introduction into embryonic stem (ES) cells of promoterless reporter/selector
gene constructs whose expression can be activated only after integration
downstream of a gene's regulatory elements. Gene trap insertions result in
production of fusion transcripts consisting of the reporter and endogenous
sequences and the mutated genes can be readily identified using PCR-based
methods such as RACE. Furthermore the biological consequences of the
integration event can be assessed after germ-line transmission. One
limitation of conventional gene trapping is that it can only target genes
expressed in ES cells since selection of insertional events relies on the
endogenous promoter's activity to drive expression of the selectable marker
and to circumvent this problem a new class of gene trap vectors called
poly(A) trap vectors was developed. These constructs contain a 3' selectable
marker whose expression is driven by a constitutively active internal
promoter relaxing the requirement for endogenous gene expression. The
selectable marker lacks a polyA signal but incorporates a splice donor (SD)
signal so only integrations upstream of an endogenous gene's splice acceptor
(SA) and polyA sequences can be selected thus eliminating intergenic
background insertions. However, it has been recently demonstrated that
poly(A) trap vectors are biased towards integrations into the 3'most-intron of
their target genes due to the action of an mRNA-surveillance mechanism
called nonsense-mediated mRNA decay (NMD).
The aim of the study presented here was to assess the efficiency of a
series of gene trap vectors that incorporate two novel features in their design:
(i) the presence of an ATG-less, 5' triple fusion between egfp, P-galactosidase
and neomycin/hygromycin resistance genes to function as a reporter/selector
of the trapped gene's expression state and (ii) a 3' poly(A) trap cassette that
contains the previously uncharacterized rabbit (3-globin exon 2/intron 2 SD
junction and an AU-rich element (ARE) derived from the human GM-CSF
gene. Our results provide evidence that the triple fusion functions properly
and can be potentially used as a reporter of trapped locus activity. We also
show that the presence of the ARE appears to improve the performance of
the rabbit (3-globin SD sequence in the context of poly(A) trapping. More
importantly, preliminary data suggest that our vectors may be resistant to
NMD and thus potentially unbiased in their insertional preference.