The present study was undertaken as part of our ongoing efforts to understand the biological and clinical significance of LEDGF/p75 expression in human cancer. Our results indicated significant upregulation of both LEDGF/p75 transcript and protein in prostate, colon, and thyroid tumors, inferred by the analysis of transcript expression in the Oncomine cancer gene microarray database (when data available) and the TissueScan Cancer Q-PCR array, and analysis of protein expression by IHC in TMAs. The observed upregulation of LEDGF/p75 in prostate tumors is consistent with our previous reports that this protein is the target of autoantibodies in some patients with PCa, is upregulated in both PCa cell lines and tissues, and promotes chemoresistance in PCa cell lines when ectopically overexpressed 
Significant upregulation of LEDGF/PSIP1 transcript was also observed in some Oncomine datasets for cancers of the breast, cervix, esophagus, kidney, head and neck, liver, lung, lymphoma, ovary, pancreas, salivary gland, skin, and stomach. However, only 4 out of 8 tumor types (prostate, colon, breast and thyroid) exhibited significant upregulation of the LEDGF/p75 transcript in the TissueScan Cancer Q-PCR array. Although other tumor types (kidney, lung, and ovarian) exhibited >1.02 fold LEDGF/p75 transcript elevation in this array, the upregulation was not statistically significant. These differences between the Oncomine database and the TissueScan Cancer Q-PCR array could be attributed to the small sample size in the latter, methodological/platform variations, analysis of LEDGF/PSIP1 vs LEDGF/p75, and the use of unrelated tumor data sets.
In addition to prostate, colon and thyroid cancers, the LEDGF/p75 protein levels were significantly elevated, as assessed by IHC analysis of TMAs, in liver and uterine cancers. To control for false positives, which are expected when numerous independent statistical tests are performed, FDR analysis was done. The FDR adjusted p-values suggested that the LEDGF/p75 protein overexpression in prostate, thyroid, and liver tumors was not by chance. Elevated expression of LEDGF/p75 in liver and thyroid tumors was associated with younger age, while none of the five tumor types displayed a significant correlation between increased LEDGF/p75 expression and increased tumor stage, most likely due to the relatively small number of tumor samples in stages pT1 and pT4.
An interesting observation in our IHC analysis was that LEDGF/p75 was not exclusively confined to the nucleus, but often appeared in the cytoplasm of tumor cells. We cannot rule out the possibility that this could be due to non-specific reactivity of the Scripps-Ab5087 anti-LEDGF/p75 antibody. However, this would seem unlikely given that the antibody reacted specifically with LEDGF/p75 in immunoblots, and the cytoplasmic staining was not observed in all tissues, particularly in normal tissues. It could be speculated that depending on the particular tumor microenvironment, LEDGF/p75 could shuttle between the extracellular milieu, the cytoplasm, and the nucleus, as observed for other stress or survival proteins 
. While there is compelling evidence that LEDGF/p75 exhibits a nuclear localization in cultured cell lines, early studies suggested that the protein exists as both endogenous and secreted forms 
. It would be important to determine in future studies whether LEDGF/p75 is secreted from tumor cells, and if this release provides a survival advantage to cancer cells in a stressful tumor microenvironment.
We did not observe a statistically significant overexpression of LEDGF/p75 protein in bladder tumors, which is inconsistent with a previous report by Daugaard et al. 
showing upregulation of LEDGF/p75 mRNA in bladder tumors from Affymetrix microarray data. None of the 5 Oncomine bladder cancer datasets analyzed for LEDGF/PSIP1 transcript expression showed significant upregulation. However, our data on elevated transcript expression of LEDGF/p75 in breast tumors was in agreement with that of Daugaard et al. 
. Our analysis showed significant upregulation of LEDGF/PSIP1 transcript levels in breast tumors in 2 of the 17 Oncomine breast cancer gene microarray datasets, and robust elevation of LEDGF/p75 transcript (2.26 fold, P
0.037) in the TissueScan Cancer Q-PCR array. Likewise, our evaluation of LEDGF/p75 protein levels in the breast tumor tissues by IHC came very close to being statistical significant (P
0.087). Daugaard et al. 
also reported lack of LEDGF/p75 transcript upregulation in colon cancer, whereas we found significant transcript upregulation in 3 of 10 Oncomine colorectal cancer datasets, and in the TissueScan Cancer Q-PCR array. Our IHC analysis also indicated LEDGF/p75 overexpression in colon cancer. These discrepancies between our results and those of Daugaard et al. 
might be related to differences in the patient cohorts or in the methodologies/platforms used.
A limitation of the present study is that the analyses of LEDGF/p75 transcript and protein expression were done in unrelated tumor datasets. It should be emphasized, however, that this study was designed to determine if LEDGF/p75 is a commonly upregulated protein in major human cancers. Our results indicate that LEDGF/p75 upregulation in human cancer is selective, with statistically significant elevation of both transcript and protein in prostate, colon, and thyroid tumors, as determined by TissueScan Cancer Q-PCR arrays and IHC analysis of TMAs. However, we observed a lack of transcript-protein correlation for the other cancer types. While this is likely due to the use of unrelated tumor samples in the transcript and protein analyses, we cannot rule out that in certain tumors the expression of LEDGF/p75 transcript does not correlate with protein expression. There are several reports documenting this phenomenon for other cancer-associated proteins 
. For instance, integrative studies of transcript and protein expression profiles in tumors and cancer cell lines showed discordant protein and mRNA expression, revealing a 20%–65% concordance depending on the tumor type 
. As a specific example, transcript expression of the Pdc4 tumor suppressor gene was found suppressed in tumors compared to normal tissues, but the protein levels were elevated 
. Likewise, it was observed that in nasopharyngeal carcinomas, the LMP1 mRNA and protein expression levels did not correlate 
. Factors leading to discordant transcript-protein correlation in tumor cells include abnormal protein stabilization via protein-protein interactions or post-translational modifications, differential lifetimes of the mRNA and protein species, mRNA regulation by miRNAs, differential regulation or processing of protein isoforms, and other mechanisms not yet completely understood 
Our results highlight the need for analyzing simultaneously LEDGF/p75 mRNA and protein expression in the same tumor samples in future studies aimed at investigating the role of this pro-survival protein in the progression of a specific cancer type, and its potential value as a biomarker for tumor aggressiveness or chemoresistance. Such studies should involve a large patient sample size with complete annotated clinical data, as well as inclusion of several disease-free normal control tissues. Indeed, a limiting factor in our studies, and in most studies comparing gene expression in tumor vs normal tissues, is that the control tissues analyzed are often histologically/morphologically “normal” tissues adjacent to the tumors. These “normal adjacent” tissues are susceptible to “field cancerization”, an effect that has been documented in many human tumor types and that involves molecular abnormalities in the normal adjacent tissue such as increased expression of cancer-associated genes or proteins, changes in gene methylation patterns, microsatellite alteration, increased oxidative DNA damage and angiogenesis, and TMPRSS2-ERG-fusions 
. Field cancerization associated with increased oxidative stress and inflammation in areas surrounding the tumors could lead to upregulation of stress response proteins such as LEDGF/p75 in the normal tissue adjacent to the tumor, resulting in underestimation of the extent of its overexpression in the tumors. Consistent with this notion, we reported recently that the expression levels of the stress/antioxidant proteins peroxiredoxins 3 and 4 in normal tissues adjacent to prostate tumors are intermediate between tumor tissues and non-diseased normal prostate tissues 
To our knowledge, this is the first comprehensive survey of the expression of LEDGF/p75 transcript and protein in human cancers. In summary, our results revealed a significant upregulation of LEDGF/p75 (transcript or protein) in several cancer types, particularly in prostate, breast, colon, liver, thyroid and uterine malignancies (summarized in ). Because our study was limited by the lack of complete clinical and follow-up patient data associated with the TMAs, the prognostic significance of LEDGF/p75 overexpression in human cancer still remains unclear. Further studies analyzing simultaneously both LEDGF/p75 transcript and protein in the same tissues, using large patient cohorts with complete clinical and follow-up data, are necessary to determine if LEDGF/p75 upregulation correlates with tumor progression and aggressiveness in a specific cancer type. Taken together, the results presented here further establish LEDGF/p75 as a cancer-related protein, and strengthen the rationale for investigating the biological and clinical significance of its upregulation in human tumors.