We have modified endogenous mouse DLL1 by homologous recombination in one step to tag DLL1 for observation in living cells and tissues, for sorting of DLL1 expressing cells, or for affinity purification to identify DLL1 protein complexes. The employed tags were functional in CHO cells but impinged on DLL1 function in vivo such that somite patterning, the process most sensitive to reduced DLL1 function [25] was affected. In addition, endogenous DLL1 tagged by AcGFP was not detected by fluorescence.
In a previous study DLL1 was C-terminally fused with three different luciferase proteins. Fusion with red luciferase was fully functional, whereas for unknown reasons a DLL1 firefly luciferase fusion was slightly hypomorphic and DLL1 fused to emerald luciferase was non-functional [1]. Our C-terminal fusions behave as hypomorphic alleles similar to but more severe than firefly luciferase.
Based on the published data [1] and our results (this paper) 4 out of 5 C-terminal tags added to DLL1 affected DLL1 function to varying degrees, although in principle C-terminal tagging is possible without impinging on DLL1 activity. Reduction of DLL1 activity does not appear to depend on the length of the tag because the long red luciferase had no effect whereas the short SF tag (plus the peptide encoded by the loxP site) affected DLL1 function more strongly than our longer AcGFPHA tag. Removal of a 16bp fragment in the 3´UTR might affect RNA stability and could be responsible for reduced DLL1 activity in our transgenic mice. However, this appears unlikely since mice tagged with firefly or emerald luciferase had a complete 3´ UTR [1] and also showed hypomorphic Dll1 phenotypes. AcGFP was described as a monomeric protein [22,29]. Thus, in case of the AcGFP tag non-physiological clustering of DLL1AcGFPHA is an unlikely reason for reduced DLL1 function, although abnormal clustering and trafficking of C-terminally tagged DLL1 cannot be excluded. DLL1 carries a PDZ binding domain at its C-terminus which interacts with Acvrinp1, a MAGUK family member [30], and ARIP2, which has been implicated in stabilizing DLL1 and DLL4 [31]. For C-terminal PDZ binding domains a free C-terminus is important for interactions with PDZ proteins in many cases [reviewed in 32,33]. C-terminal extension of DLL1 with tags might interfere with such interactions but this seems unlikely to play a major role as the C-terminal extension by red luciferase had no effect on DLL1 function. It appears that sequences of the C-terminal fusion are an important contributing factor and might affect protein stability or trafficking or processing in vivo and thereby impinge on DLL1 protein function leading to somite patterning defects.
AcGFP fused to DLL1 in DLL1AcGFPHA over expressing CHO cells was detected by direct fluorescence indicating that AcGFP in the context of the fusion protein is functional. However, even in homozygous mouse embryos we did not detect AcGFP fluorescence in any tissue. A plausible explanation could be that expression levels in CHO cells were much higher than low levels of DLL1 expressed from the endogenous locus. Thus, absence of detectable fluorescence signals likely reflect DLL1AcGFPHA levels that are below the limit of detection of our set up. Whether fluorescing proteins with other activation and excitation properties and a better quantum yield (for example ZsGreen1 [29]) fused to DLL1 are sufficient to detect expression of a DLL1 fusion protein in vivo remains to be addressed.
Whereas DLL1AcGFPHA in our mice turned out to be insufficient for direct DLL1 detection by fluorescence our pilot study using the Dll1SF allele demonstrated its usefulness for the purification of DLL1-containing protein complexes (Table S3Additional File 6) from endogenous sources, in this case early embryos. In our pilot study components of the secretory pathway and vesicle transport were identified, which can be expected for DLL1, a transmembrane protein that undergoes endocytotic processing [34]. In addition, enzymes involved in ubiquitination copurified with tagged DLL1. Since DLL1 is modified by ubiquitin [34] also these potential interaction partners support that specific DLL1 protein complexes were affinity-purified. GO term analysis [35,36] showed a surprising enrichment of other identified proteins implicated in metabolic processes, nucleotide-binding and catalytic activity. As far as we know, these proteins have not been implicated in or related to DLL1 function as yet and their significance for DLL1 activity will require further analyses. Given that our tag impinged on DLL1 function it is possible that the tag also prevents the isolation of a subset of DLL1 complexes.
In conclusion, DLL1 activity appears to be highly sensitive to sequences added to the C-terminus of the protein. Which sequences are tolerated by DLL1 are currently not predictable and might only be determined empirically by comprehensive studies in vitro and in vivo.