This study examined the characteristics of postmortem tissue quality in two human postmortem brain collections with the goal of identifying optimal tissue quality characteristics to use when selecting cases for human molecular studies. Furthermore, we examined the extent to which molecular measures from human postmortem tissue can be taken to represent such measures from fresh tissue and, therefore, the extent to which data from human postmortem brain tissue can represent the molecular conditions of the live brain. This study assumed optimal clinical diagnosis of cases, an essential feature of the process (Deep-Soboslay et al., 2005
). The answers to these questions serve to suggest additional questions for the area of postmortem tissue quality and provide support for the use of postmortem brain tissue in studies of human brain research (Lewis, 2002
; Tomita et al., 2004
; Torrey et al., 2000
These data show that chemical markers of tissue quality exist and suggest that they are useful in validating tissue quality. The traditional markers of tissue quality have been descriptive and included PMI, agonal condition, case age and health. In our hands, PMI was not predictive for RNA stability (within the limited quality range of our cases), confirming findings from a previous report (Trotter et al., 2002
). Gradually, the field has acquired chemical markers of tissue quality, starting with pH and 28S/18S ratio, and now including direct measures of RNA quality, like RIN. The most useful human postmortem tissue quality marker in our hands is RIN. It correlates with the pH of the tissue and integrates the extent of RNA degradation with the presence of the 28S/18S peaks. However, caution is necessary: RIN is undoubtedly a sensitive indicator of total RNA integrity and a good predictor of overall mRNA integrity. However, since it has been shown that under certain circumstances some specific mRNA species are preferentially degraded (Barrachina et al., 2006
; Buesa et al., 2004
), it is advisable to verify individual mRNA integrity within an experiment. Our finding that RIN is the optimal marker for representing quality agrees with previously published studies that found RNA quality measures in assessing gene expression analysis (Colangelo et al., 2002
; Jones et al., 2006
; Ross et al., 1992
pH appears to be a good marker of peri-mortem tissue quality, but does not appear to be sensitive to freezer degradation, as shown in our tissue-thaw illustration. Cerebellar tissue appears to be an adequate representative for other brain regions when testing for quality markers. Although we did not screen everywhere, we tested several representative regions from brain surface and from deep brain structures and saw pH values correlate highly. The data suggest that measures of tissue quality from a single brain region (like cerebellum) are likely to be representative of whole brain.
Overall, we found that the cases with the best tissue quality were more likely to come from the medical examiner’s office than from other collection sources. It is reasonable to assume that cases where the cause of death is fast and unexpected might demonstrate good pre-mortem conditions with low agonal stress. It must be stressed, however, that before analyzing tissue quality for the collection cases, we had already screened and excluded many ME cases who failed to meet collection criteria (e.g., drug abuse, infectious diseases, head injury) so that the cases that qualified for our collection were already pre-screened for potentially confounding conditions. Therefore, we can only claim that it is for the subset of ME cases where case confounds have already been excluded, that tissue quality is high. In contrast, the tissue collected from the WB program was from cases with advanced age and anticipated deaths, where these and other confounds might have biased tissue quality.
Somewhat unexpectedly, in a preliminary analysis, we showed that postmortem tissue has tissue quality markers that are at least as good as surgical biopsy tissue. We are in the process of examining this question further to determine the source of RNA degradation in the surgical tissue. Several possibilities are reasonable: still active RNAases in the surgical tissue, the surgical anesthesia or exposure to air while fresh. However, these results provide preliminary support for a renewed confidence in molecular measures derived from postmortem tissue (Bahn et al., 2001
; Leonard et al., 1993
; Preece et al., 2003
Furthermore, the results raise the possibility that variability in molecular outcomes across CNS regions and disease groups may more likely be the result of tissue heterogeneity than postmortem degradation.
The data shown here document remarkable stability of the protein we selected in postmortem tissue. However, until further data accrue, documentation of protein stability should still be individually demonstrated for specific experiments. Nonetheless, the proteins that we selected for analysis were varied by size and tissue compartment and are, to some extent, representative. However, others have found variable protein degradation in human and animal postmortem brain using immunoblotting and immunohistochemistry techniques (Irving et al., 1997
; Li et al., 1996
; Liu and Brun, 1995
). Even so, in most cases, protein degradation occurred at high PMIs (> 40 hours). Moreover, we do not exclude the possibility of early degradation for specific proteins and we pursue the question of the stability of protein species using fresh human surgical tissue.
In conclusion, we find evidence for the proposal that human high quality postmortem tissue is a generally reliable laboratory resource for exploring molecular characteristics of healthy and diseased human brain. For any RNA or protein species, its degradation characteristics in human postmortem tissue still need to be documented. Nonetheless, case screening and tissue quality assessments (including RIN) are two indispensable tools to selecting postmortem cases representative of human brain tissue.