Several candidate TSG from the 3p21.3 AP20 and LUCA sub-regions were examined in the gene inactivation test, GIT [2, 12, 13, 16, 34, 42, 43]. The test is based on the functional inactivation of analyzed genes that can be achieved in different ways: by mutation, deletion, methylation etc. According to these results, at least three genes can now be considered as bona fide lung TSG: NPRL2 and RASSF1A from LUCA and RBSP3 from AP20 sub-regions [12, 13, 16].
Earlier the decrease of RBSP3 expression was shown in SCLC, NSCLC, cervical, renal, breast, ovary, leukemia cell lines and primary tumors by Northern blot analysis, RT-PCR and qPCR [12, 44–46]. The decrease or absence of NPRL2/G21 expression was detected in some SCLC, NSCLC and renal cancer cell lines using Northern blot analysis [6, 16]. The decrease or absence of RASSF1A expression was found in SCLC, NSCLC and many other tumors and cancer cell lines [see [24, 27, 28]].
It was reported that promoter methylation was the main mechanism of RASSF1A loss of expression in lung cancer (see Introduction). Homozygous deletion of 3'-part of NPRL2 gene and rare mutations were found in NSCLC and SCLC cell lines [6, 16]. There are no methylation data explaining the loss of RBSP3 expression in lung cancer. However frequent deletions and mutations were reported [10, 12, 15]. In some leukemia cell lines (up to 98%) and acute leukemia lymphoma blood samples (24%) methylation of the promoter region of RBSP3 was reported . Methylation (up to 26%), deletions and decreased expression of RBSP3 were significantly associated with poor prognosis of cervical cancer . Thus, inactivation of RBSP3 might be one of the early events in cervical carcinogenesis.
Loss of heterozygosity and quantitative real-time PCR demonstrated that aberrations in both LUCA and AP20 sub-regions occurred simultaneously in the same tumor with high probability. Thus, it was suggested that aberrations in both LUCA and AP20 sub-regions could be linked [9, 10]. Indeed, homozygous deletions in both regions often occur in the same tumor (P < 3 × 10-7). The estimation of possible interdependency between all aberrations in the loci NLJ-003 (AP20) and NL3-001 (LUCA) as different events was carried out using a permutation test for four types of cancers: lung, renal, breast and ovarian. This test also revealed a significant correlation between different aberrations in these two loci (P < 10-6). The same results were obtained using Pearson correlation for numeric values of copy number changes of these loci. Indeed, proteins RBSP3 and RASSF1A could collaborate in cell cycle arrest: RASSF1A by inhibiting cyclin D1  and RBSP3 by dephosphorylating pRB . Thus functional collaboration of these two genes could result in activation of the RB1 gene.
In this study we tested the hypothesis that TSG in AP20 and LUCA regions were not only deleted but their expression could also be simultaneously down-regulated in NSCLC. This suggestion was indirectly supported by other studies that showed that genes over large chromosomal regions could be regulated in a coordinated fashion [33, 47–49].
First we found that expression of all three genes is rather uniform in lung samples isolated from healthy donors and from normally looking lung samples obtained from NSCLC patients ("normal" matched control samples). Thus adjacent morphologically normal tissues from the patients can be used as paired reference controls to tumor samples. In the study two parameters were analyzed - the level of mRNA decrease and frequency of mRNA decrease in two major NSCLC histological subtypes (AC and SCC) and their subgroups with different characteristics such as clinical Stage, grade, tumor localization, presence of metastases and others. Although both parameters reflect deregulation of gene expression, they are not randomly but rather functionally related.
Expression analysis of the three genes revealed the following main features.
1. Expression of the three studied TSG was significantly decreased in NSCLC: 85% for RBSP3, 67% for RASSF1A and 73% for NPRL2 (P < 0.001). It was statistically valid both for SCC and AC.
2. Down-regulation of the three genes was already evident at Stage I of NSCLC samples. Statistically significant down-regulation of both NPRL2 and RBSP3 was seen in 100% cases at Stage I of SCC.
3. The degree and frequency of the expression decrease for all three genes was more strongly pronounced in SCC than in AC samples (see Table 3). This difference was statistically valid in the case of NPRL2 (P = 0.002).
4. All studied genes were involved in progression of AC. The tendency of more severe expression down-regulation of the RBSP3 was evident during tumor progression of AC with respect to FD and LD (70% and 3-fold in cases without metastases in contrast to 88% and 6-fold decrease in cases with metastases, P = 0.13). For NPRL2, this tendency was also seen only for AC - 83% of cases had decreased expression at Stage III compared to 14% at Stage I and in 75% of AC cases with metastases vs. 20% of cases with metastases, P = 0.08, see Table 3). Expression of RASSF1A revealed the most strongly pronounced correlation between decrease of expression (FD and LD) and tumor progression both in SCC and AC. For example, difference in FD values was obvious between cases with and without metastases. For SCC cases this difference was 33% vs. 81% (P = 0.196) and for AC it was even more sharp, 29% vs. 100% (P < 0.05).
5. Expression of RBSP3 and RASSF1A was most seriously affected in respect to FD. For RBSPS3 it was detected in 85% of all NSCLC cases and for RASSF1A in 67%. However, regarding LD, the expression of NPRL2 was most strongly inhibited (LDav = 8), while RBSP3 showed weaker inhibition (LDav = 5).
6. Preliminary data suggested that no statistically significant difference was observed in cases with relation to age, smoking history and other cytological and pathological characteristics.
7. NotI microarrays and qPCR on genomic DNA we tested for possible mechanisms of the declined expression of RBSP3 in NSCLC. The data suggested that both genetic and epigenetic mechanisms were important for transcriptional inactivation of RBSP3 in NSCLC. Altogether deletions were detected in 25% of AC samples and in 30% of SCC patients. Methylation of RBSP3 was detected in 38% of AC and in 80% of SCC cases. With NotI microarrays we also tested how often LUCA (NPRL2 and RASSF1A) and AP20 (RBSP3) regions were deleted or methylated in the same tumor and found that this occurred in 58% of all studied cases (18 of 26). Thus, most likely both genetic and epigenetic mechanisms are responsible for simultaneous down-regulation of expression of these three TSG.