How can I know my transcriptional fusion strain expression reflect the endogenous gene expression? Is there any paper in c.elegans about this topic?
If when fused to a cDNA for your gene the promoter you use in your transcriptional fusion can rescue the mutant phenotype for loss-of-function of that gene, that would help - though it still could be the case that the particular expression that’s important for rescuing activity is weak or transient and isn’t reflected well in what you see when you look at GFP fluorescence. More generally, site-of-action studies can support the ideas about expression you get from the gfp reporter.
It is possible to make your transgenes especially good - you can knock gfp into the endogenous locus (not trivial to do!), or into a cosmid or fosmid containing your gene, so as to make your reporter include as many of the control elements that regulate the gene as you can. Similarly, in the context of a Fire vector or the like, you could try to include as much upstream sequence as possible, and you can replace the 3’ UTR in the vector with that from your gene. It’s not terribly uncommon for introns to contain transcriptional control elements, especially if they are large and have evolutionarily conserved sequences. It may work to PCR the intron or introns and stick them well upstream of gfp in your transcriptional reporter. Also in terms of the best possible gfp reporter, there is a pretty good argument to be made for single-copy transgenes, low-copy-number transgenes, and complex arrays, for getting as faithful as possible a look at transcriptional regulation.
Antibodies against your protein are another way of looking at your gene’s expression.
There are methods to directly look at message (in situ hybridization, or smFISH, for example), but it can be difficult to identify the cells containing the labeled messages, and I don’t think these methods are in terribly common use.