The primary objective of this communication is to report the development of a preparative flow sorting method based on immunofluorescent staining of intracellular antigens that can also deliver RNA suitable for quantitative analysis of global gene expression. While these objectives have been achieved individually (FACS based on intracellular antigens, high quality RNA prepared from intact flow sorted cells), to our knowledge they have not been combined in a carefully explored, robust protocol.
The main issues are that the cells have to be permeabilized to expose the intracellular antigens to the antibodies (Figure ), which however also allows some of the intracelluar RNA to diffuse out from the cells. Thus, a fixation step becomes necessary. The permeabilization step can also release endogenous RNases, as well as allow extracellular contaminating RNases to access the RNA. Formaldehyde is a very effective fixative that covalently crosslinks proteins and nucleic acids, which essentially eliminates the losses of RNA, and also works well to inactivate endogenous RNases. However, this also creates a problem of releasing the RNA from the crosslinks, as the crosslink reversal requires heat and salt, to which RNA is intrinsically susceptible due to chemical fragmentation.
This creates a compromise situation, where the formaldehyde fixation and crosslink reversal need to be balanced against each other. For example, in the laser capture microdissection methodologies, the extensive formalin fixation of the tissues necessitates aggressive crosslink reversal, which in turn typically results in extensive RNA fragmentation. The protocol we present here works well for cultured HDF, but may need to be further optimized for other cell types, and almost certainly for tissues. Attention to the parameters discussed above should facilitate this optimization. Control of endogenous and exogenous RNases is of obvious importance, and while we have not found it necessary, the RNAlater reagent (Qiagen) has been found to be compatible with antibody staining [27
], and could be used in the initial steps of our protocol.
Heat-induced fragmentation of RNA introduces mostly random breaks into the phosphodiester backbone. Relatively mild fragmentation will thus rapidly obscure the rRNA bands when visualized by electrophoresis, and drop the RIN values on the Bioanalyzer instrument (Agilent) into the 3-5 range. Such samples will display as broad smears in the high mw (200-4000 nt) range. However, because of the relatively large size of these fragments, in our hands such preparations perform very well in applications based on short range sequence analysis, such qPCR, hybridization microarrays, or next generation sequencing. More extensive fragmentation, typically due to RNase contamination, results in loss of the high mw migrating material, and accumulation of signal in the low mw (<200 nt) range. Such preparations do not perform well in downstream applications.
We processed RNA samples obtained by our procedure all the way through Affymetrix Gene Chip expression profiling, and compared the gene expression patterns to those obtained using standard RNA preparation (direct lysis of intact cells in Trizol reagent). The very high correlation coefficients between the data sets (Figure ), which were generated from the same starting material (exponentially growing LF1/TERT cells), indicate that none of the steps in our protocol introduce systematic changes (losses or gains) of individual mRNAs across the whole transcriptome.
The biological observation that stimulated the development of these protocols was the surprising discovery that fully immortalized HDF cultures continue to generate senescent, p16-positive cells, with no evidence of DDR. The p16-positive cells do not cycle and are continuously diluted as the cultures proliferate, while new p16-positive cells are generated in the cultures from proliferating cells, thus maintaining a steady-state frequency of p16-positive cells of approximately 20%. We still do not understand the upstream signaling that triggers these spontaneous and apparently stochastic senescence events, and the ability to obtain these rare cells in amounts sufficient for molecular studies will greatly aid these studies. The methods presented in this communication should be of interest to many investigators working in diverse model systems.