The human lung cancer cell lines (KNS62 and Ben) have been described previously [8]. Non-genetically engineered cells were routinely maintained in RMPI 1640 supplemented with 10% FCS, 2 mM glutamine and 1 mM sodium pyruvate (Life Technologies, Inc., Eggenstein, Germany) without penicillin or streptomycin. All cells were kept in a humidified atmosphere containing 5% CO₂ at 37°C.

Resected orthotopically growing tumors were immediately frozen in liquid nitrogen. Five μm thick frozen tissue sections were cut from the tumor and fixed for 10 min in acetone. Immunohistochemical assessment of tumor proliferation was performed by the using monoclonal mouse antibody MIB-1 (Dianova, Hamburg, Germany) against the nuclear antigen Ki-67 and the monoclonal mouse antibody Ki-S4 against topoisomerase IIα (kindly provided by Prof. Parwaresch, Dept. of Hematopathology and Lymph Node Registry, Kiel, Germany). Immunolabeling with the specific antibody was evaluated by counting 200 tumor cells in 3 different "hot spots" in each cryosection at high-power magnification (× 400). Counting was done by 2 independent observers. The labeling indices were calculated as percentage of positive tumor cells. The mean values and standard deviations are based on 3 animals from each group. From each tumor bearing animal 3 cryosections were taken for analysis.

For staining of intratumoral vascular endothelium, cryosections were stained bwith they monoclonal rat anti-mouse MEC13.3 (BD Pharmingen, Hamburg, Germany) against CD31 (PECAM-1). The APAAP method was used for detection [20]. Microvessel density (MVD) was calculated according to Weidner et al. [21]. Briefly, areas of elevated vascular density were identified and subsequently the microvessel entities per optical field (× 200 magnification) were counted in 5 different areas of each tumor. Statistical mean values, SD and p values (t test) were calculated.

Mouse anti-caspase 8 antibodies (clone 5F7) were obtained from Upstate Biotechnology (Eching, Germany). Anti-PARP (clone C-2-10) was obtained from Calbiochem (Bad Solden, Germany), anti-β-actin (clone AC-15) from Sigma, and anti-cytochrome c (clone 7H8.2C12) from Pharmingen (Hamburg, Germany). Rabbit polyclonal antibodies against Bcl-xL were obtained from Pharmingen, anti-Bid from R&D Systems (Wiesbaden, Germany), anti-caspase 9 from Cell Signaling (Berlin, Germany), antibodies to JNK, phosho-JNK, c-Jun and phosphor-c-Jun from Cell Signaling. Peroxidase-conjugated anti-rabbit IgG and anti-mouse IgG were obtained from Amersham (Brunswick, Germany). Rabbit polyclonal anti-cIAP1 H-83, and rabbit polyclonal anti-cIAP2 H-85 antibodies were purchased from Santa Cruz (Heidelberg, Germany). Rabbit monoclonal anti-survivin and anti-XIAP were obtained from Cell Signaling.

The NSCLC cancer cell line KNS 62 was seeded at a density of 1 × 10⁴ cells/well into 96-well flat-bottom microtiter plates, allowed to adhere overnight and labeled with ³H-thymidine (370 kBq/μl; Amersham, Brunswick, Germany) for 3 h. Subsequently, the cells were washed with phosphate-buffered saline (PBS) and incubated with various concentrations of gemcitabine (Lilly), phenylbutyrate or a combination of the two in normal growth medium for up to 72 h. The cells were lysed in 0.05% SDS for 30 min at 37°C to ensure complete release of genomic DNA and harvested by vacuum aspiration on glass-fiber filters. Dried filters were counted using a liquid scintillation counter (Wallac, Switzerland). The percentage of specific DNA fragmentation, indicative of apoptosis, was calculated as: percentage viability = (E/S) × 100, where E (experimental) is the counts per minute (cpm) of retained DNA in the presence of chemotherapy and S (spontaneous) is the cpm of retained DNA in the absence of chemotherapy [9]. Caspase 3 and caspase 8 activity was measured by immunoblotting of total cellular proteins and subsequent detection of caspase 3 and caspase 8 and cleavage of their substrates PARP and Bid. The broad caspase inhibitor zVAD-fmk was obtained from Biomol, Ltd. (Hamburg, Germany). The following inhibitors of different Mitogen-Activated Protein Kinases (MAPK) were employed: 1 μmol/L of SP600125 a JNK-specific inhibitor, 10 μmol/L of SB203580 a p38-specific inhibitor and 0.5 μmol/L of MEK1/2-inhibitor, all from Calbiochem (Bad Soden, Germany).

The cells were washed twice with PBS, trypsinized, pelleted, resuspended in PBS containing 5 mM EDTA and fixed by adding one volume of ethanol (Merck, Darmstadt, Germany). After Rnas -treatment (40 ng RNaseA/μl, Sigma-Aldrich, Taufkirchen, Germany) cells were pelleted, resuspended in PBS containing propidium iodide (200 μg/ml) and subjected to FACS analysis. Cell cytometry was conducted using a FACScan cell analyzer (Becton-Dickinson Bioscience, San Jose, CA, USA). WinMDI2.8 [22] was used for analyzing FACS data.

Mitochondrial integrity was determined by assessing the loss of the mitochondrial membrane potential Δψ_m using an ApoAlert Mitochondrial Membrane Sensor Kit (Clontech, Heidelberg, Germany) followed by FACScan (Becton Dickinson) analysis.

Female SCID-bg (C.B-17/IcrHsd-Prkdc^scid Lyst^bg) mice (Harlan Winkelmann, Borchen, Germany) weighing 14 to 19 g were allowed to become acclimatized for 1 week in a sterile environment where bedding, food and water were autoclaved. Food and water were given ad libitum. Animal experiments and care were performed in accordance with the guidelines of the institutional authorities. The mice were anaesthetized by i.p. injection of a mixture of Midazolam 5.0 mg/kg (Roche, Grenzach-Wehyen, Germany), Fentanyl 0.05 mg/kg (Janssen-Cilag, Neuss; Germany) and Medetomidin 5.0 mg/kg (Pfizer, Karlsruhe, Germany). The orthotopic animal model was performed as previously published [23]. Briefly, after right lateral thoracotomy the lung was carefully exposed and a tumor cell suspension (1 × 10⁶ cells in 30 μl Matrigel, Biosciences) was cautiously injected into the lung tissue. The thoracic wall and the skin were closed with a 6-0 running absorbable suture (Ethicon, Norderstedt, Germany). After completion of the surgical procedure anaesthesia was antagonized by s.c. injection of a mixture of Flumazenil 0.5 mg/kg (Roche, Grenzach-Wehyen, Germany), Naloxon 1.2 mg/kg (Curamed, Neuss; Germany) and Atipamezol 2.5 mg/kg (Pfizer, Karlsruhe, Germany). All mice were inspected daily for complications. Once orthotopic KNS62 and Ben tumors were established (7 d after orthotopic tumor cell injection), the mice were treated with 50 mg/kg GEM i.p. twice a week for 28 days, 300 mg/kg PB by subcutaneous infusion with Alzet osmotic minipumps (model 2002; Alzet, Cupertino, CA, USA) or by combination therapy. The mMinipumps were exchanged after 2 weeks. In the control group NaCl was administered instead of chemotherapy according to the gemcitabine scheme. The animals were sacrificed after 35 days and the tumors were resected. Tumor weight and tumor volume according to the formula of a rotational ellipsoid (V = length × height × width × 0.5236) were calculated. Resected tumors were bisected and cryo- and formalin-fixed for further investigations. The data were analyzed using SPSS for Windows (V 11.0, Chicago, IL, USA). The results are given as means ± SD. Differences in tumor volume between relevant subgroups were analysed and p values were calculated by Mann-Whitney U test. A global p value of less then 0.05 was considered to be statistically significant.

We analyzed the sensitivity of two different NSCLC cell lines (KNS62 and Ben) to increasing doses of GEM and PB. The cell lines underwent apoptosis in a dose-dependent manner, showing fragmentation of cellular DNA, though KNS62 was less sensitive than Ben to GEM and PB (Fig. 1, upper and middle panel). When GEM and PB were combined, either in high dosage (10 μg/ml GEM and 5 mM PB) or in low dosage (1 μg/ml GEM and 2 mM PB), the rate of viable cells was significantly decreased compared to single substance treatment. Remarkably, an effect exceeding the sum of single agent treatment was detectable in the KNS62 low dosage treatment group (Fig. 1, lower panel).

Various indicators of apoptotic cell death were investigated in KNS62 and Ben cells after treatment with GEM and PB in combination. PI-FACS analyses of the PI-stained cells focused especially on the sub-G1 cellular DNA fraction. The combination treatment revealed a significant increase in DNA in the sub-G1 fraction compared to gemcitabine treatment alone (Fig. 2, upper panel). After 72 h of combination therapy 46% of KNS62 cells and 54% of Ben cells were detectable in the sub-G1 cellular (apoptotic) fraction, compared to only 19% of KNS62 and 24% of Ben after treatment with gemcitabine alone. To quantify the early apoptotic phenotype Annexin-V/PI-FACS analyses were performed (Fig. 2, lower panel). As early apoptotic events Annexin-V-positive cells (right lower quadrant in the density dot plot) as well as PI-positive and Annexin-V-positive cells (right upper quadrant) were summarized. After combination therapy significantly more cells revealed early morphologic events of apoptosis (KNS62: 34%; Ben: 39%) than cells treated with gemcitabine alone (KNS62: 11%; Ben: 14%).

The activation of key apoptotic proteins was investigated to evaluate the influence of GEM, PB and combination chemotherapy on apoptosis at the molecular level. In death receptor-mediated apoptosis, receptor activation is followed by cleavage of caspase-8 and its substrate BID, a BH3 domain-containing pro-apoptotic protein that subsequently becomes activated. Cleavage of caspase-8 and Bid was low in KNS62 cells after GEM and PB treatment alone, but significantly increased with combination therapy (Fig. 3a). Similarly, cells subjected to combination therapy exhibited clearly increased cleavage of caspase-9 in KNS62.

The dependence of chemotherapy-induced cell death on caspase-mediated apoptotic pathways was confirmed by the observation that the broad caspase-inhibitor zVAD prevented apoptosis-related DNA fragmentation and PARP cleavage in treated cells (Fig. 3b+c). However, DNA fragmentation was only partially inhibited, suggesting further mechanisms besides caspase-dependent apoptosis.

The involvement of mitochondria in chemotherapy-mediated apoptosis was determined by assessing mitochondrial integrity. After 24 h of combination chemotherapy only 11% of KNS62 cells exhibited a loss of Δψm, compared to 7% in the gemcitabine group and 8% in the phenylbutyrate group. This difference increased over time from 29% (48 h) and 44% (72 h) of cells with defective Δψm in the combination group compared with 12%/16% (48/72 h) for gemcitabine and 14%/19% (48/72 h) for phenylbutyrate (Fig. 4a).

These results were confirmed by the demonstration of cytochrome-c, Smac/Diabolo and AIF release from mitochondria into the cytosol, as detected by Western blot analyses of cytosolic proteins. In the cytosolic fractions of combination chemotherapy-exposed KNS62 cells there was a significantly higher release of cytochrome-c, Smac/Diabolo and AIF into the cytosole compared to single agent (gemcitabine or phenylbutyrate) therapy (Fig. 4b). An analysis of protein expression levels of various family members of the inhibitors of apoptosis proteins (IAP) revealed that especially c-IAP1 and c-IAP2 were significantly down regulated by PB and combination therapy, whereas XIAP remained stable and survivin showed only moderate regulation (Fig. 4c).

Since mitogen-activated protein kinases (MAPK) have been determined to be substantially involved in controlling chemotherapy-induced apoptosis [24,25], we investigated the involvement of MAPK in GEM and PB combination therapy-induced apoptosis. While treatment of KNS62 with either GEM or PB induces phosphorylation of ERK1/2, p38, JNK and its target c-Jun, combination therapy amplifies this effect substantially (Figure 5a). The overall level of these proteins (ERK1/2, p38, JNK and c-Jun) remaiedn unchanged (data not shown). The impact of activation of different MAP-Kinases on apoptosis was tested by co-incubation of specific inhibitors. Only specific blocking of p-JNK significantly inhibited the induction of apoptosis by chemotherapy (Figure 5b), whereas the level of phosphorylated c-Jun as the target of activated JNK was effectively decreased by the JNK inhibitor SP600125 (Fig. 5c).

The effect of gemcitabine and phenylbutyrate on in vivo tumor growth was investigated in an orthotopic SCID mouse model. Each group comprised six animals. In untreated animals, KNS62 the mean tumor size was 110 mm³ (SD: 32.6) compared to 92.5 mm³ (SD: 42.2) in the GEM group, 79.3 mm³ (SD: 37.0) in the PB group and 33.8 mm³ (SD: 24.3) in the combination group. The tumor size was significantly smaller in the combination group compared to GEM (p = 0.008) or PB (p = 0.028) chemotherapy alone (Fig. 6a).

In orthotopically growing Ben tumors the mean tumor size in the untreated group was 95 mm³ (SD: 30), in the GEM group 36.6 mm³ (SD: 21.6), in the PB group 29.7 mm³ (SD: 18.8) and in the combination therapy group 16.2 mm³ (SD: 10.3). Like in the KNS62 orthotopic model in the Ben tumors were significantly smaller in the combination therapy group compared to GEM (p = 0.035) and PB (p = 0.026) (Fig. 6a).

The analysis of the proliferation activity of orthotopically growing tumors by means of Ki-67 and topoisomerase IIα staining indices revealed significant inhibition of proliferation in both combination therapy groups (KI-67: KNS62: GEM = 25.7% vs. combi = 15.3%; Ben: GEM = 18.2% vs. combi = 13.8%; topoisomerase IIα: KNS62: GEM= 34.7% vs. combi = 17.3%; Ben: GEM = 27% vs. combi = 19%) compared to untreated animals or animals with single agent therapy (Fig. 6b+c).

The rate of apoptotic cells (staining of cytokeratin 18 fragments with M30 antibody and phosphorylation of histone H2B with anti p-H2B) was only slightly elevated (data not shown). The microvessel density (MVD; anti-CD31 positivity) was also only slightly lower in the combination group (data not shown).