RNA-seq normalization for tissues with different Cell Density

[rpeiro]

Hi all:
We are starting a new RNA-seq project of a tissue. We over express a gene in a part of this tissue (representing about 25 % of cells of the sample), but we will collect RNA of the whole sample. We have case and control with 3 replicates each.
We think that the signal of potential differentially expressed genes with a low average expression levels can be at least partially masked due to just this 25 % of cells that over express this gene.
Does someone know a normalization procedure that can we use?
Spike-in normalization, as described in https://www.biostars.org/p/92540/ is useful for cases with transcriptional amplification.
For experimental reasons is not operative to directly count the number of cells over expressing the gene.
Thank you in advance.