Cytokine production of lung cancer cell lines: Correlation between their production and the inflammatory/immunological responses both in vivo and in vitro

Abstract

Cytokines produced by tumor cells may have various effects on antitumor immune responses and tumor growth. In the present study, the cytokine production of 31 lung cancer cell lines was evaluated, while any correlation with the histological type, the induction of tumor‐specific cytotoxic T lymphocytes (CTL) in vitro, and angiogenesis and the infiltration of inflammatory cells in tumor tissues were also examined. Production of interleukin (IL)‐1α, IL‐1β, IL‐4, IL‐6, IL‐8, IL‐10, tumor necrosis factor (TNF)‐α, granulocyte macrophage colony stimulating factor (GM‐CSF), granulocyte colony stimulating factor, transforming growth factor (TGF)‐β and vascular endothelial growth factor (VEGF) in the culture supernatant was measured using enzyme‐linked immunosorbent assay. Each cytokine was produced in a substantial number of the tumor cell lines. In particular, IL‐6, IL‐8, TGF‐β and VEGF were produced in 18 (55%), 29 (94%), 31 (100%) and 28 (90%) of 31 cell lines, respectively. However, neither IL‐4 nor TNF‐α was produced at all by any tumor cell line. TGF‐β production was significantly higher in adenocarcinoma than in squamous cell carcinoma (P = 0.03). Immunohistochemical staining revealed the magnitude of macrophage infiltration, and angiogenesis in surgically resected tumor tissue specimens correlated well with GM‐CSF and IL‐8 production from the corresponding cell lines. Among six lung cancer cell lines, CTL were induced in the three lung cancer cell lines that produced a lower amount of TGF‐β (<100 pg/mL). These findings suggested that TGF‐β produced by tumor cells could inhibit the induction of CTL in vitro. The present results suggest that the production of various cytokines from tumor cells might exert various paracrine effects both in vivo and in vitro. (Cancer Sci 2007; 98: 1048–1054)

Abbreviations

AT

autologous tumor

CM

culture medium

CTL

cytotoxic T lymphocytes

ELISA

enzyme‐linked immunosorbent assay

G‐CSF

granulocyte colony stimulating factor

GM‐CSF

granulocyte macrophage colony stimulating factor

IFN

interferon

IL

interleukin

NK

natural killer; mAb, monoclonal antibody

MLTC

mixed lymphocyte–tumor cell culture

MVC

microvessel count

NSCLC

non‐small‐cell lung carcinoma

RLNL

regional lymph node lymphocytes

RPMI

Roswell Park Memorial Institute

TGF

transforming growth factor

Th

T helper

TNF

tumor necrosis factor

VEGF

vascular endothelial growth factor.

Recent studies have evaluated the cytokine network involved in the local inflammatory and immune responses against tumors.^{( 1} , ² , ³ , ^{4 )} In addition to infiltrating inflammatory cells and immune cells, tumor cells have also produced cytokines that may alter tumor growth, tumor immunogenicity and the host defense mechanisms against cancer. A variety of cytokines, such as IL‐4, IL‐6, IL‐8, IL‐10, TGF‐β and VEGF, may either regulate tumor growth or modify the antitumor immune responses.^{( 5} , ⁶ , ⁷ , ^{8 )}

Inflammatory cytokines have been broadly categorized into two types: Th‐1 cytokines, such as IL‐2, IL‐12, IFN‐γ and TNF‐α, which are responsible for cellular immunity; and Th‐2 cytokines, such as IL‐4, IL‐5, IL‐10, and IL‐13, which are responsible for the induction of humoral immunity. Recent studies have indicated the presence of Th2 type cytokines in the microenvironment of tumors, suggesting that these cytokines may play a role in the suppression of cellular immunity against cancer.^{( 3} , ⁵ , ^{9 )}

Among the cytokines detected in the tumor microenvironments, TGF‐β may play various roles in the progression of tumors. The role of TGF‐β has been reported to be responsible in the growth^{( 10 )} and angiogenic properties^{( 11 )} of tumors, as well as in the ability to augment the invasive and metastatic potential of neoplastic cells.^{( 12} , ^{13 )} Moreover, TGF‐β has been shown to inhibit expansion, the cytotoxic activity of CD‐8 T cells, and the cytokine production of NK cells and IL‐2‐ or IL‐12‐activated killer cells.^{( 14} , ^{15 )} Some investigators have reported the plasma concentration of TGF‐β to be elevated in patients with lung and colorectal cancers, and it has been correlated with the progression of tumors.^{( 16} , ¹⁷ , ^{18 )}

In the present study, the amount of various cytokines produced in the culture supernatant of lung cancer cell lines was measured and the histopathological effect of these cytokines in the corresponding tumor tissues as paracrine effects was analyzed. Moreover, the authors tried to induce CTL from RLNL using stimulations with autologous tumor cells, and the effect of cytokines produced by tumor cells on cellular immunity was also examined.

Materials and Methods

The study protocol was approved by the Human and Animal Ethics Review Committee of the University of Occupational and Environmental Health, Japan and a signed consent form was obtained from each subject before obtaining the tissue samples used in this study.

Culture medium.  The CM consisted of RPMI 1640 (Gibco BRL, Grand Island, NY, USA) supplemented with 10% heat‐inactivated fetal calf serum (Equitech‐bio, Ingram, TX, USA), 10 mM HEPES, 100 IU/mL penicillin G and 100 mg/mL streptomycin sulfate.^{( 19 )}

Cell lines.  The lung cancer cell lines for 28 NSCLC and three small‐cell lung carcinomas were analyzed in the present study. The cell lines of NSCLC were histologically classified according to the World Health Organization nomenclature.^{( 20 )} These included 15 adenocarcinoma, seven squamous cell carcinoma, four large cell carcinoma, one adenosquamous carcinoma and one unclassified carcinoma. Of the 31 cell lines of lung cancer, 21 cell lines were established from surgically resected specimens by the authors, as indicated in Table 1. The procedure used for the establishment of the tumor cell line has been reported previously.^{( 21 )} Briefly, fresh tumor tissue specimens were excised from surgical specimens and were then minced into small pieces with scissors. The minced tissue specimens were then shaken with a mixture of 0.1 mg/mL of deoxyribonuclease type I, 1 mg/mL collagenase type IV and 0.5 mg/mL of hyaluronidase type V (Sigma, St Louis, MO, USA) in CM at 150 shakes/min at 37°C for 1 h. After washing three times with Hanks’ balanced solution, the cells were placed in a flask and then were maintained for 12 months as a monolayer culture by serial passages in CM. PC9, A549, 1–87, Sq‐1 and RERF‐LC‐AI were kindly donated by Dr Kyogo Itoh, Kurume University, Kurume, Japan. PC13, YM21, PC10, PC1, and PC6 were provided from the Medical Institute of Bioregulation, Kyushu University, Fukuoka, Japan.

Table 1.

Cytokine production of 31 lung cancer cell lines

Cell line	Histology	IL‐1‐α	IL‐1‐β	IL‐4	IL‐6	IL‐8	IL‐10	TNF‐α	GM‐CSF	G‐CSF	TGF‐β	VEGF
A110L †	Ad.	0	0	0	8630	207	0	0	0	1073	79	184
A129L †	Ad.	0	0	0	0	229	0	0	0	0	141	273
A925L †	Ad.	13	0	0	0	549	0	0	351	0	79	273
B203L †	Ad.	0	0	0	403	462	0	0	447	0	109	1394
B514L †	Ad.	0	0	0	290	1286	24	0	46	0	205	65
B901L †	Ad.	0	0	0	529	435	0	0	261	0	62	1987
C422L †	Ad.	0	0	0	0	866	0	0	22	142	17	213
D611L †	Ad.	0	15	0	250	77	0	0	94	0	167	3855
E522L †	Ad.	0	10	0	250	0	0	0	0	0	303	4077
F1121L †	Ad.	0	14	0	250	44	0	0	42	0	97	1885
PC9	Ad.	0	0	0	270	591	0	0	0	0	167	166
1–87	Ad.	0	0	0	34	2355	0	0	0	59	101	166
A549	Ad.	0	0	0	0	276	0	0	0	0	260	55
803YT †	Ad.	0	0	0	910	396	0	0	0	1716	141	786
PC13	Ad.	0	0	0	0	20	0	0	0	0	146	157
B1203L †	Sq.	0	0	0	0	568	0	0	0	0	84	133
C1026L †	Sq.	56	13	0	49	107	0	0	0	0	222	326
Sq‐1	Sq.	0	0	0	0	846	0	0	0	150	27	215
RERF‐LC‐AI	Sq.	0	0	0	14	75	0	0	0	0	18	106
QG56 †	Sq.	0	0	0	0	1134	0	0	0	141	38	202
YM21	Sq.	0	0	0	100	6277	0	0	0	497	22	76
PC10	Sq.	0	0	0	0	533	0	0	0	0	35	0
A904L †	La.	0	0	0	14 720	316	24	0	0	2406	246	187
C311L †	La.	0	11	0	250	9	0	0	0	0	313	784
C831L †	La.	0	11	0	250	44	0	0	0	0	255	1016
G603L †	La.	0	20	0	250	37	0	0	127	0	294	1085
A529L †	Ad‐sq.	0	0	0	14 360	7743	0	0	562	179	242	92
F1012L †	Unclass.	0	13	0	0	0	0	0	0	0	280	317
QG90 †	Small	0	0	0	0	759	0	0	0	97	22	234
PC1	Small	0	0	0	0	994	0	0	0	52	8	0
PC6	Small	0	0	0	0	807	0	0	0	0	39	0

^† Cell line established in the authors’ laboratory. The unit quantity of each cytokine is pg/mL. Ad., adenocarcinoma; Ad‐sq., adenosquamous cell carcinoma; G‐CSF, granulocyte colony stimulating factor; GM‐CSF, granulocyte macrophage colony stimulating factor; IL, interleukin; La., large cell carcinoma; Small, small cell carcinoma; Sq., squamous cell carcinoma; TGF, transforming growth factor; TNF, tumor necrosis factor; Unclass., unclassified carcinoma; VEGF, vascular endothelial growth factor.

Collection of culture supernatants of the cell lines.  The cells were plated at 5 × 10⁵ cells in a 25‐cm² flask (Falcon; Becton Dickinson, Oxnard, CA, USA) in CM and incubated at 37°C with 5% CO₂. At 24 h after plating, almost all the tumor cells were adherent to the bottom of the flask and then the CM was completely replaced. Subsequently, the culture supernatants were collected after another 24 h of incubation, centrifuged to remove cellular debris, and then frozen at –80°C until the measurement of cytokines was performed.

ELISA.  The levels of immunoreactive IL‐1‐α, IL‐1‐β, IL‐4, IL‐6, IL‐8, IL‐10, TNF‐α, GM‐CSF, G‐CSF, TGF‐β and VEGF were measured using ELISA. The following commercially available ELISA kits were used: IL‐1‐α, IL‐4, IL‐6, IL‐10 (Endogen, Boston, MA, USA), IL‐1‐β, GM‐CSF and TGF‐β (Amersham Lifescience, Braunschweig, Germany), TNF‐α, G‐CSF and VEGF (BioSource International, Camarillo, CA, USA) and IL‐8 (Toray, Tokyo, Japan). The procedures of these cytokine assays were done according to the instruction manual for each cytokine.

Immunohistochemical analysis of tumor tissues.  The 3–5 µm sections of the paraffin‐embedded tumor tissue specimens were applied to an indirect antiperoxidase immunohistochemical assay using an Envision kit (Dako, Carpentaria, CA, USA). IL‐8 and TGF‐β expression were examined with antihuman IL‐8 (BioSource International, Camarillo, CA, USA) and antihuman TGF‐β1 (Santa Cruz Biotechnology, Santa Cruz, CA, USA). For determination of the degree of staining intensity, stained sections were evaluated by two of the authors (TF and KY) using a blind protocol design (the observers had no information on the data of cytokine production in the corresponding cell lines). Staining intensities consisted of: faint staining = ±, moderate staining = + and strong staining = ++. The representative staining intensities are illustrated in Fig. 1b and c. Detection of macrophages was carried out using antihuman CD‐68 antibody (Nichirei, Tokyo, Japan). An analysis of infiltration into the tumor tissues of the leukocytes, macrophages and neutrophils was carried out using HE staining. On each section, 10 microscopic high‐power fields that had the most accumulated positive staining (hot spots) at magnitude ×400 were selected. The percentage of stained cells in these areas was evaluated for CD‐68‐positive cells as macrophages. For further analysis, the value of each sample was then categorized into three groups (negative, 0–20%; positive, 20–50%; double positive, >50%). To evaluate tumor angiogenesis, a microvessel count was performed using CD‐31 antibody (Dako). Large vessels with thick muscular walls were excluded from the counts. The numbers of microvessels were counted in a high‐power field at magnitude ×400, and then the sum of the counts in four fields was recorded as the MVC score.

Figure 1.

Representative photos of immunohistochemical staining of lung cancer tissues. Surgically resected tumor tissues were stained with (a) anti‐CD‐68, (b) interleukin (IL)‐8, (c) transforming growth factor (TGF)‐β or (d) CD‐31 antibody. (a) CD‐68 staining if macrophages in tumor tissues of patient G603 (left panel) and C311 (right panel). Each corresponding cell line produced 127 and 0 pg/mL of granulocyte macrophage colony stimulating factor (GM‐CSF), respectively, as shown below the image. In the tumor tissue of patient G603, macrophages were highly infiltrated. In patient C311, only a few macrophages could be detected. (b) IL‐8 staining of tumor tissues of patients A529 and C311. Each corresponding cell line produced 7743 and 9 pg/mL of IL‐8, respectively. The tumor cells of patient A529 were strongly positive (++), whereas those of patient C311 were weakly positive (±). (c) TGF‐β staining in tumor tissues of patient C311 and A110. Each corresponding cell line produced 313 and 79 pg/mL of TGF‐β, respectively. The tumor cells of patient C311 were strongly stained (++), whereas those of patient A110 were moderately stained (+). (d) CD‐31 staining of microvessels in patients B514 and G603. In the tumor tissue of patient B514 (left panel), the microvessel count (MVC) was high (8 counts in the field) and the corresponding cell line produced 1286 pg/mL of IL‐8 and 65 pg/mL of vascular endothelial growth factor (VEGF). In the tumor tissue of patient G603 (right panel), the MVC score was low (4 counts in the field) and the corresponding cell line produced 37 pg/mL of IL‐8 and 1085 pg/mL of VEGF. Arrowheads show microvessels stained by anti‐CD‐31 monoclonal antibody.

CTL induction and evaluation of CTL activity.  Regional lymph nodes were obtained at the time of surgery. Each lymph node was divided into two parts for the histological diagnosis and for this study. The latter part of each lymph node was mixed and prepared for RLNL as described previously.^{( 19 )} RLNL were frozen in a deep freezer at –130°C until use. They were rapidly thawed and stimulated with irradiated (100 Gy) either AT cells or CD‐80‐transfected AT cells weekly at a tumor‐to‐lymphocyte ratio of 1:10 in CM as described previously.^{( 22 )} In order to evaluate the inhibitory effect of TGF‐β on CTL induction, autologous MLTC was performed in the presence or absence of 100 or 400 pg/mL of TGF‐β1 (R&D Systems, Minneapolis, MN, USA). RLNL (9 × 10⁶ cells) were seeded to nine wells (1 × 10⁶/each well in 2 mL of CM). The nine wells were prepared for three groups as follows: no addition of TGF‐β (Group A), addition of 100 pg/mL of TGF‐β (Group B) and 400 pg/mL of TGF‐β (Group C). 10 µL of TGF‐β at each concentration (0, 20 and 80 ng/mL) was added in the MLTC of Group A, B and C every day. The tumor‐specific CTL induction was considered to be successful if CTL lysed more than 10% of AT at an effector/target ratio of 30:1 and did not lyse autologous Epstein–Barr virus transformed B cells in the 4‐h standard ⁵¹Cr release assay at day 28 of the MLTC.

Statistical analysis.  The Mann–Whitney U‐test was used to determine the continuous variables between the two groups. The findings were considered to be significant if the P‐value was less than 0.05. Correlation analysis between IL‐8 and MVC score was performed with Spearman's correlation coefficient by rank test.

Results

Cytokine production of tumor cell lines.  IL‐1‐α, IL‐1‐β, IL‐6, IL‐8, IL‐10, GM‐CSF, G‐CSF, TGF‐β and VEGF were secreted by various lung cancer cell lines (Table 1). IL‐8, TGF‐β and VEGF were detected in 29 (94%), 31 (100%) and 28 (90%) of 31 cell lines, respectively. IL‐1‐β, IL‐6, GM‐CSF and G‐CSF were detected in eight (26%), 17 (55%), nine (29%) and 11 (48%) cell lines, respectively. IL‐1‐α or IL‐10 was produced by two independent cell lines. None of the cell lines produced either IL‐4 or TNF‐α (Table 2).

Table 2.

Frequency and quantity of production of each cytokine in 31 lung cancer cell lines

Cytokine	Detection limit (pg/mL)	No. detected/ tested (%)	Range (pg/mL)
IL‐1‐α	4	2/31 (6)	13–56
IL‐1‐β	1	8/31 (26)	11–20
IL‐4	4	0/31 (0)	–
IL‐6	4	17/31 (55)	14–14720
IL‐8	4	29/31 (94)	9–7743
IL‐10	4	2/31 (6)	24–24
TNF‐α	4	0/31 (0)	–
GM‐CSF	1	9/31 (29)	22–562
G‐CSF	4	11/31 (48)	52–2406
TGF‐β	4	31/31 (100)	8–313
VEGF	24	28/31 (90)	55–4077

G‐CSF, granulocyte colony stimulating factor; GM‐CSF, granulocyte macrophage colony stimulating factor; IL, interleukin; TGF, transforming growth factor; TNF, tumor necrosis factor; VEGF, vascular endothelial growth factor.

Correlation between cytokine production and histologic cell type.  The production of TGF‐β was compared between adenocarcinoma (n = 15) and squamous cell carcinoma (n = 7). The quantity of TGF‐β produced was significantly higher in adenocarcinoma than in squamous cell carcinoma (P = 0.03; Table 3). In the adenocarcinoma cell lines, GM‐CSF was detected in seven of 15, while it was not detected at all in the squamous cell carcinoma cell lines.

Table 3.

Comparison of each cytokine production between adenocarcinoma and squamous cell carcinoma cell lines

	Adenocarcinoma (n = 15) mean ± SD (pg/mL)	Squamous cell carcinoma (n = 7) mean ± SD (pg/mL)	P‐value
IL‐1‐α	1 ± 3	8 ± 21	0.75
IL‐1‐β	3 ± 5	2 ± 5	0.81
IL‐6	788 ± 2184	23 ± 38	0.07
IL‐8	520 ± 615	1363 ± 2199	0.28
IL‐10	2 ± 6	0	–
GM‐CSF	84 ± 146	0	–
G‐CSF	199 ± 502	113 ± 183	0.60
TGF‐β	138 ± 75	64 ± 73	0.03
VEGF	1036 ± 1357	151 ± 107	0.08

G‐CSF, granulocyte colony stimulating factor; GM‐CSF, granulocyte macrophage colony stimulating factor; IL, interleukin; TGF, transforming growth factor; VEGF, vascular endothelial growth factor.

In vitro functional analysis of TGF‐β on induction of tumor‐specific CTL.  CTL induction was performed using autologous RLNL against six lung cancer cell lines (Table 4). CTL could be induced in only three of six cases by stimulation with the irradiated autologous lung cancer cell lines. All three cell lines in which CTL could be induced produced a lower quantity of TGF‐β (97 pg/mL) than the other three lines in which CTL could not be induced (109 pg/mL), as shown in Table 4. However, CTL could be induced in all six cases, when the irradiated autologous cancer cells transfected with costimulatory molecule CD‐80 were used as stimulators. Furthermore, we tried to induce CTL from autologous RLNL against F1121L in the absence or presence of TGF‐β. At day 28, CTL activity was induced in the wells in which TGF‐β was not added (control wells), as shown in Fig. 2A, In contrast, no CTL could be induced when TGF‐β was added at either 100 or 400 pg/mL (Fig. 2B,C).

Table 4.

Correlation between CTL induction and amount of TGF‐β produced from tumor cell lines

Cell line	Histology	Success of CTL induction		TGF‐β production (pg/mL)
		without CD80	with CD80
A110L	Ad.	+	+	79
B1203L	Sq.	+	+	84
F1121L	Ad.	+	+	97
A529L	Ad‐sq.	–	+	242
A904L	La.	–	+	246
B203L	Ad.	–	+	109

Ad., adenocarcinoma; Ad‐sq., adenosquamous cell carcinoma; CTL, cytotoxic T lymphocytes; La., large cell carcinoma; Sq., squamous cell carcinoma, TGF, transforming growth factor.

Figure 2.

The inhibitory effect of transforming growth factor (TGF)‐β on cytotoxic T lymphocyte (CTL) induction. (A) CTL activity, which was induced using autologous mixed lymphocyte–tumor cell culture in the presence of absence of TGF‐β, was assessed by cytolytic activity to autologous cell line, F1121L. CTL activity against F1121L was successfully induced by autologous regional lymph node lymphocytes in the absence of TGF‐β. (B,C) In contrast, the addition of TGF‐β 100 or 400 pg/mL completely inhibits the induction of CTL activity.

In vivo functional analysis of G‐CSF and GM‐CSF on the infiltration of relevant inflammatory cells in tumor tissue specimens.  The correlation between the tumor‐derived cytokine and the infiltration of inflammatory cells in nine surgical specimens was evaluated. The infiltration of neutrophils was assessed using HE staining. In addition, the infiltration of macrophages was assessed using an immunohistochemical analysis with anti‐CD‐68 antibody. No significant relationship was detected between neutrophil infiltration and G‐CSF production. Four of the nine cell lines did not produce GM‐CSF, and in the corresponding tumor tissues, no macrophage infiltration was observed. In contrast, macrophage infiltration was observed in the other five tumor tissues from which corresponding established cell lines produced GM‐CSF (Table 5 and Fig. 1a). However, a correlation between the amount of GM‐CSF produced and the magnitude of macrophage infiltration could not be elucidated.

Table 5.

Correlation between the amount of GM‐CSF produced from tumor cell lines and infiltration of macrophages in the corresponding tumor tissues

	Histology	GM‐CSF (pg/mL)	Infiltration of macrophages in the tumor tissues
A529L	Ad‐sq.	562	+
G603L	La.	127	+
D611L	Ad.	94	++
B514L	Ad.	46	++
F1121L	Ad.	42	++
A110L	Ad.	0	−
C1026L	Sq.	0	−
C311L	La.	0	−
C831L	La.	0	−

–, <20%; +, 20–50%; ++, >50% macrophages in infiltrating inflammatory cells. Ad., adenocarcinoma; Ad‐sq., adenosquamous cell carcinoma; GM‐CSF, granulocyte macrophage colony stimulating factor; La., large cell carcinoma; Sq., squamous cell carcinoma.

Correlation between IL‐8 or VEGF production by tumor cells and angiogenesis of corresponding tumor tissues.  Immunohistochemical staining for IL‐8 and TGF‐β in representative cases of nine surgical specimens is shown in Fig. 1b,c. Positive staining by the antibody against each cytokine was detected among tumor cells and partially among stromal cells. It was observed that the higher the level of IL‐8 production by a tumor cell line, the stronger is the immunohistochemical staining in the corresponding tumor tissue (Table 6). However, such correlation could not be observed in production of TGF‐β. Furthermore, the authors assessed the degree of angiogenesis as the MVC score, which was evaluated using immunohistochemical staining with anti‐CD‐31 mAb in four high‐power fields at magnification ×400 (representative stains are shown in Fig. 1d). As a result, a significant correlation was observed between the MVC score and IL‐8 produced by tumor cells (r = 0.833, P < 0.01; Fig. 3A). In contrast, the MVC score tended to be inversely correlated with VEGF, although this difference was not significant (r = −0.621, P = 0.07; Fig. 3B).

Table 6.

Correlation between the amount of IL‐8 and TGF‐β produced from tumor cell lines and the intensity of immunohistochemical staining in the corresponding tumor tissues

Cell line	IL‐8		TGF‐β
	ELISA †	IHC ‡	ELISA †	IHC ‡
A110L	207	++	79	+
A529L	7743	++	242	+
B514L	1286	++	205	±
C1026L	107	+	222	++
C311L	9	±	313	++
C831L	44	+	255	++
D611L	77	+	167	++
F1121L	44	±	97	++
G603L	37	±	294	+

^† The amount (pg/mL) of interleukin (IL)‐8 or transforming growth factor (TGF)‐β produced by each cell line.

^‡ The intensity of immunohistochemical (IHC) staining of tumor tissues of corresponding cell lines. ±, faint staining; +, moderate staining; ++, strong staining; ELISA, enzyme‐linked immunosorbent assay.

Figure 3.

Correlation between interleukin (IL)‐8 or vascular endothelial growth factor (VEGF) production by tumor cells and microvessel count (MVC) score in the corresponding tumor tissues. (A) IL‐8 production from tumor cells in vitro correlated well with the MVC score in the corresponding tumor tissues. Pearson's correlation coefficient (r) is 0.833 (P < 0.01). (B) In contrast, VEGF production from tumor cells showed a tendency toward an inverse correlation with MVC score in the corresponding tumor tissues (P = 0.07).

The correlation between inflammatory cytokines (such as IL‐1, IL‐6 and IL‐8), C‐reactive protein and the white blood cell count in the peripheral blood was also assessed; however, no correlation was observed between these factors.

Discussion

In the present study, the levels of production of various cytokines were evaluated in 31 lung cancer cell lines using ELISA. Proliferation or growth of tumor cells by cytokines may be controlled either by their direct or indirect effects on angiogenesis or host immunity. IL‐8, which is known to recruit inflammatory neutrophils and promote the interaction between tumor cells and inflammatory cells,^{( 23 )} has been reported to be produced in NSCLC. In the present study, all but two lung cancer cell lines produced various amounts of IL‐8. Other reports have suggested IL‐8 to be a potent angiogenic and growth factor for malignant tumors.^{( 24} , ^{25 )} The present results also indicated that the angiogenic function in tumor tissues evaluated using MVC was correlated with amount of IL‐8 produced in vitro (Fig. 3A). Moreover, a correlation between in vitro production of IL‐8 and in vivo immunohistochemical staining with anti‐IL‐8 antibody was observed, as shown in Table 6 and Fig. 1b. VEGF was expressed in 28 of 31 (90%) lung cancer cell lines. VEGF is known as a potent mitogen for vascular endothelial cells, and plays a key role in the growth and metastasis of tumor by promoting angiogenesis.^{( 26 )} However, the present results showed an inverse correlation between the amount of VEGF produced in cell lines and angiogenesis in the tumor tissues (Fig. 3B). Recent studies have showed that VEGF regulates angiogenesis independently from angiogenesis of IL‐8, and that VEGF inhibits angiogenesis under the existence of TGF‐β.^{( 25} , ^{26 )} IL‐6 may have a pleiotropic function, involving malignant transformation and progression against tumor cells and the promotion of an immune response by inhibition of regulatory T cells.^{( 27} , ^{28 )} In the present study, 16 of 31 lung cancer cell lines (55%) produced various amounts of IL‐6. IL‐6 may provide an autocrine mechanism that is involved in the growth of tumor cells and a paracrine mechanism that is involved in immunological response enhancement.

Various amounts of TGF‐β were produced in all of the cancer cell lines in the present study. TGF‐β has been reported to have angiogenic properties;^{( 11 )} however, the authors could not detect any correlation between the amount of TGF‐β produced in vitro and in vivo using the MVC score. TGF‐β has a potent suppressive effect on immune responses exerted by CD‐8 T cells.^{( 29} , ³⁰ , ³¹ , ^{32 )} The present data clearly demonstrate that the addition of TGF‐β suppresses CTL induction (Fig. 2), and may suggest that the higher the amount of TGF‐β produced by a tumor cell line, the lower the success rate of CTL induction against it (Table 4 and Fig. 2). However, when MLTC was performed using tumor cells transfected with CD‐80, such a suppressive effect of TGF‐β was abrogated, as shown in Table 4. The interaction of CD‐80–CD‐28 might thus overcome the inhibitory effect of TGF‐β on CTL induction. 4‐1BB, a member of the TNF receptor family, has the ability to both prevent apoptosis of CD‐8⁺ CD‐28⁻ T cells and to enhance the long‐term response of effector and memory CTL.^{( 33 )} As a result, the costimulation of 4‐1BB blocks the immunosuppressive function of TGF‐β.^{( 31} , ^{34 )}

IL‐10 also has been known to have an immune suppressive function by means of induction of regulatory T cells.^{( 35 )} However, only two of 31 cell lines in the present study produced IL‐10.

GM‐CSF and G‐CSF were produced from nine and 11 of the lung cancer cell lines tested, respectively, in the present study. However, the amount of GM‐CSF produced in vitro did not correlate with the degree of macrophage infiltration in the corresponding tumor tissue specimens. No correlation was observed between the G‐CSF production in vitro and the degree of neutrophil infiltration in the corresponding tumor tissue. These CSF stimulate the proliferation and differentiation of myeloid stem cells while also promoting the activation and survival of leukocytes at the inflammatory site.^{( 36} , ^{37 )} CSF extend the life span of mature leukocytes by inhibiting their apoptotic death; this function may not be confined only to hematopoietic cells, but may extend also to cancer cells.^{( 38} , ^{39 )}

In the present study, it was observed that almost all lung cancer cell lines produced various kinds of cytokines. These cytokines seemed to have such functions as tumor progression, angiogenesis and an inhibition of the immune response. It may thus be a potentially useful strategy to manipulate tumor‐derived cytokines to augment the effectiveness of anticancer treatments.