In the present study we examined the morphological characteristics and functional role of the TSC tumor suppressor complex as a major regulator of mTOR activity. We could show that the TSC tumor suppressor complex is substantially expressed in lung cancer cell lines and that hamartin and p-mTOR were inversely correlated in three of the five lung cancer cell lines. Furthermore, slightly more than 50% of the NSCLC specimens showed hamartin expression, compared to nearly one third of SCLC with valuable hamartin expression. These findings demonstrate that hamartin expression is a frequent finding in lung cancer. A major challenge is to assess whether accumulation or loss of hamartin (and p-tuberin) reflects primary or secondary events. Both, accumulation or loss of hamartin, could be pathogenically relevant for carcinogenesis. As in normal tissue hamartin is only expressed in bronchial respiratory epithelia but not in alveolar epithelial cells, we cannot conclude if hamartin expression reflects a gain or loss of function in tumor specimens. In NSCLC and SCLC cell lines, high protein levels of hamartin were associated with low p-mTOR and vice versa. This inverse correlation between hamartin and mTOR levels supports an interaction between TSC and mTOR in NSCLC and SCLC. All cell lines used for the present study revealed detectable hamartin and p-TSC2 protein levels indicating that expression differences are rather due to a loss of hamartin expression. This interpretation is also reasonable in the light of previous studies showing a loss of heterozygosity (LOH) of the TSC1-locus on chromosome 9q34 in AC and precursor lesions . Another study also reported LOH for hamartin or TSC2 in 22% of 86 specimens, but none of the 80 lung cancer lines studied .
In SCLC, we found that hamartin expression correlates with p-TSC2 and may point towards a disruption of the hamartin-tuberin complex, which is accompanied by phosphorylation of tuberin and activation (i.e. phosphorylation) of mTOR. Moreover, the expression of hamartin correlated with that of nuclear p-mTOR (CC = 0.405; p = 0.007) suggesting that hamartin may be an interesting surrogate marker for mTOR related signaling. The immunohistochemical characterization of signaling pathways during the routine histological workup of specimens would greatly facilitate the selection of individualized therapeutic regimens that are currently arising by the availability of new molecular targets such as mTOR inhibitors .
Growing evidence supports abnormally activated mTOR to play an important pathogenic role in lung cancer associated with both KRAS and EGFR mutations and may provide a mechanism of resistance to treatment with EGFR inhibitors [26, 27]. The EGFR can be autophosphorylated on various tyrosine sites and distinct downstream signaling cascades are initiated by the EGFR depending on its phosphorylation pattern [28, 29]. As EGFR signaling is partially mediated via KRAS and both KRAS and EGFR can activate PI3K, a potential link with TSC is reasonable (Figure 1). A potential interaction between TSC and KRAS has been postulated in mice . Tumors of animals harboring hamartin loss and KRAS (G12D) expression in lung epithelial cells revealed 1) reduced tumor latency, 2) an activation of mTOR and 3) a response to treatment with rapamycin with improved survival compared to KRAS alone mutant mice . These observations suggest that the TSC complex may be a critical regulator of KRAS-related signaling cascades that are focusing on mTOR. Overall, these data support a rather complex, interdependent regulation of the TSC complex and the EGFR/KRAS signaling. We have therefore used immunohistochemical data of a recent study and speculated if abnormal activation of mTOR is due to pathogenic events upstream of the hamartin-tuberin-complex. Based on this approach, we found a significant correlation between hamartin and p-EGFR (Tyr-1068) resp. p-EGFR (Tyr-992) expression in AC specimens . P-mTOR was also closely correlated with p-EGFR Tyr-1173. These findings indicate that expression or even accumulation of hamartin may also be secondary to EGFR phosphorylation. In contrast, an inverse correlation was found between hamartin and p-EGFR Tyr-992 in SCC specimens indicating different molecular fingerprints in different cancer subtypes.
We have also hypothesized that hamartin, p-tuberin and p-mTOR expression may be dependant of the EGFR mutation status. In 12 cases with already established EGFR mutation status for therapeutic purposes, hamartin accumulation was found both in EGFR-mutated and EGFR wild type tumors (Table 2). Also p-tuberin expression was detected both in EGFR + and EGFR- cases. Nuclear expression of p-mTOR was slightly more frequent in patients harboring EGFR mutations, however it was also detectable in EGFR-wild type cases. Therefore, we conclude that the EGFR mutation status does not affect expression of hamartin, p-tuberin and p-mTOR responsive to EGFR mutations. This assumption is also supported by the observation that phosphorylation of tyrosine residues 922 and 1173, but not phosphorylation of tyrosine residues 1068, have been associated with activating EGFR mutations .
We have also raised the question, if accumulation of hamartin may be secondary to mutational alterations. Notably, LOH for the TSC1 or TSC2 locus has been described in 22% of 86 human lung cancer specimens . In another study more than one third of atypical adenomatous hyperplasia precursor lesions and 53% of concomitant adenocarcinomas displayed LOH on 9q. A substantial proportion of these harbored LOH at loci adjacent to the TSC1 gene . However, only one cell lines used in the present study revealed a TSC1 sequence alteration, i.e. the HCC-827 cell line (AC harboring an acquired mutation in the EGFR tyrosine kinase domain). This sequence alteration was represented by complex alterations in the close vicinity of the stop codon in exon 23 of TSC1. We suggest that this alteration is not functionally relevant as it is located beyond the critical tuberin interaction domain and as it does not result in a truncation of the protein.
Regardless of its possible pathogenic role, hamartin expression may provide as a prognostic marker. Clinical follow up data were available in nearly two third of the cases. As expected the histological tumor type strongly influenced the survival rates. The longest mean survival was observed in AC patients and the shortest in SCLC patients. Regarding TSC-related expression, SCC and SCLC patients revealed a poorer overall survival in hamartin positive cases. In contrast, no prognostic effect of hamartin expression could be observed for AC specimens. Differences regarding its prognostic value may also reflect different molecular pathways involved during carcinogenesis resp. therapeutic strategies. Nevertheless, other independent factors with a potential influence on survival that have recently been discussed could be considered, e.g. an overexpression of MTA3 gene in NSCLC as a risk factor on survival  or an overexpression of IMP3 as a predictor of aggressive tumor behavior . Further investigations in this direction should follow.
In contrast, we could not reveal a prognostic influence of p-tuberin and p-mTOR. In another study focusing on NSCLC, positive cytoplasmic mTOR staining was associated with shorter survival . Furthermore, high mTOR expression has been claimed to be associated with a worse outcome in laryngeal squamous cell carcinomas that have been subjected to postoperative radiotherapy . The discrepancy between our results and the aforementioned studies may be due to the fact, that we recruited an antibody directed only against phosphorylated (i.e. activated) mTOR, implicating a limited comparability of these results. Furthermore, we have observed p-mTOR expression both in the cytoplasm and nucleus and the functional relevance of nuclear mTOR has yet to be elucidated. Classically, mTOR acts in the cytoplasm, but recent findings have supported compartment-specific mTOR functions in other subcellular compartments including the nucleus . The existence of a nuclear shuttling signal in mTOR has been postulated being essential for nuclear mTOR import . These findings fit well with our observations that p-mTOR was detected not only in the cytoplasm but also in the nucleus of tumor cells in immunohistochemical observations. Nuclear p-mTOR staining was found in 22.8% of AC, 35.5% of SCC and 16.3% of SCLC specimens. Thus, the different staining frequencies in the different tumor types may also reflect the compartment-dependant diversity of mTOR signaling. P-mTOR has also been assigned to aggressive histological variants of papillary thyroid carcinoma and, thus nuclear labeling of p-mTOR has been discussed to serve as a diagnostic and prognostic marker as well as a potential therapeutic target .