CA19.9 AND TYPE 1 CHAIN LEWIS ANTIGENS: UNRAVELING THE MOLECULAR BASIS OF EXPRESSION IN GASTROINTESTINAL TISSUES TO IMPROVE THE CLINICAL EFFECTIVENESS AS TUMOR MARKERS

Background CA19.9 is one of the more diffusely used tumor markers in both research and clinical practice. It consists of a heterogeneous group of molecules that share the terminal sialyl Lewis a (sLea) structure, one of the so-called type 1 carbohydrate-chain Lewis antigens, which also include Lewis a (Lea), Lewis b (Leb) and disialyl-Lewis a. Lea, Leb and disialyl-Lea have been considered products of the epithelia of the normal gastrointestinal tract, while CA19.9 has been assumed as an abundant product of cancer cells, due to the reactivity found by immunohistochemical staining of cancer tissues with the anti CA19.9 antibody (recognising sLea). B3GALT5 is one of the glycosyltransferase enzymes required for synthesizing such antigens, whose gene transcription is under the control of multiple promoters able to drive synthesis of transcripts with different 5’ UTRs but identical coding sequence. Paradoxically for the biosynthesis of a tumor marker, all transcripts appear silenced in colon cancer. Moreover, CA19.9 was scarcely detectable by dot-blot staining of colon cancer lysates and we recently found a false reactivity with anti-CA19.9 antibody of mouse tissues stained by immunohistochemistry, since sLea is not synthesized in the rodents due to the lack of any 1,4 fucosyltransferase activity. Aim Regarding the current concept of CA19.9 as a tumor marker, our study aims at revisiting the histological, biochemical and molecular aspects of Lewis antigen expression in vivo and to assess their actual relevance in different cancers. III Methods Expression and biosynthesis of type 1 chain Lewis antigens in the colon and the pancreas were studied by immunodetection in tissue sections and lysates, quantification of glycosyltransferase transcripts, bisulfite sequencing, and chromatin immunoprecipitation assays. Results CA19.9 was poorly detectable in normal colon mucosa and almost undetectable in colon cancer, while it was easily detected in the pancreatic ducts, together with Lewis b antigen, under both normal and cancer conditions. B3GALT5 transcripts were down-regulated in colon cancer, while they remained expressed in pancreatic cancer. Even ST3GAL3 transcript appeared well expressed in the pancreas but poorly in the colon, irrespective of normal or cancer conditions. CpG islands flanking B3GALT5 native promoter presented an extremely low degree of methylation in pancreatic cancer with respect to colon cancer. In a DNA region about 1 kb away from the B3GALT5 retroviral promoter, a stretch of CG dinucleotides presented a methylation pattern potentially associated with transcription. Such a DNA region and the transcription factor-binding site provided overlapping results by chromatin immunoprecipitation assays, corroborating the hypothesis. Conclusions Our results suggest that CA19.9 is a physiological product whose synthesis strongly depends on the tissue specific and epigenetically regulated expression of B3GALT5 and ST3GAL3. It acquires tumor marker properties in the pancreas due to duct obstruction and/or to inverted polarity of transformed ductal cells, which in turn give rise to reabsorption into vessels and elevation of circulating levels. The data also suggest that other Lewis antigens may share the same properties.