Comparison of 35S and biotin as labels for in situ hybridization: use of an HPV model system.

Colorimetric in situ hybridization is a method of potential importance in diagnosis and research. The largest criticism of the method has been a perceived loss of sensitivity compared with autoradiographic techniques. Our more positive experience with automation of colorimetric in situ hybridization led us to undertake a direct comparison of the sensitivity of 35S- and biotin-labeled probes. Serial sections of formalin-fixed, paraffin-embedded cell pellets from four human cervical carcinoma cell lines with known copies of HPV (CaSki, 400-600 copies HPV 16; HeLa, 10-50 copies HPV 18; SiHa, 1-2 copies HPV 16; HTB31, no known copies HPV) were hybridized with protocols optimized for autoradiographic or colorimetric detection. Both methods gave comparable results, with differences in each technique seen at the limits of sensitivity. The 1-2 copies of HPV 16 per SiHa cell can be detected with both methods; however, grain counting is required for interpretation of the autoradiographic result. This degree of sensitivity for colorimetric in situ hybridization in formalin-fixed, paraffin-embedded material is achieved through careful optimization of probe size and labeling, adequate tissue digestion, and removal of background. Autoradiography may be preferred in situations where quantitation is required, but colorimetric detection retains the advantages of speed, potential for automation, and improved localization of signal with comparable sensitivity.