Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

Cancer invasion and metastasis are features of malignant tumor progression, and are the major cause of morbidity in cancer patients (1). Both are active processes that require the breakdown, penetration, and movement by tumor cells through extracellular matrix (ECM) proteins (2- 4). Although it is established that active cellular motility is a necessary component of metastasis, the regulation of this motility is still incompletely understood. Tumor cells may respond to motility stimuli that include ECM components such as laminin and type IV collagen (5, 6), and to stromal/ tumor cell-derived factors such as platelet-derived growth factor, hepatocyte growth factor/scatter factor (HGF/SF), autotaxin (ATX), and autocrine motility factors (7).

ATX was initially isolated as a 125-kD glycoprotein with autocrine motility activity from the culture supernatant of a human melanoma cell line (9). Pretreatment of the cells with pertussis toxin abolished the locomotive responses to ATX, indicating that a G protein is involved in its signal-transduction pathway. The nucleotide sequences of complementary DNA (cDNA) for ATX revealed 45% homology with surface protein PC-1 of plasma cells (10, 11). Two isoforms of ATX, putatively generated by alternative splicing, were subsequently isolated. The two isoforms are brain-type phosphodiesterase I (PD-I), initially isolated from rat brain (12, 13), and ATX^ter, isolated from Ntera2D1 human teratocarcinoma cells (14). PD-I cDNA differs by an insertion of 25 amino acids from both the cDNA for ATX^ter and that for the ATX of melanoma (ATX^mel) (14). Additionally, ATX^mel is distinguishable from both ATX^ter and PD-I by the presence of 52 (156 bp) extra amino acids (14).

ATX is synthesized as a transmembrane propeptide with a short intracellular tail. The extracellular peptide includes four potential N-linked glycosylation sites, a proteolytic cleavage site, two adjacent somatomedin B-like domains of vitronectin with a putative plasminogen activator inhibitor (PAI) binding site, a phosphodiesterase catalytic domain, arginine-glycine-aspartic acid (RGD) tripeptide motifs that are potential sites of recognition by _v integrins, and a loop region of a calcium-binding EF-hand domain. The phosphodiesterase catalytic activity of ATX and dephosphorylation of Thr²¹⁰ appear essential for the motility function of ATX (15).

Northern blot analyses have shown ubiquitous ATX messenger RNA (mRNA) expression in normal human tissues (14). The most abundant expression was found in the brain, placenta, ovary, and small intestine, with intermediate expression found in kidney, prostate, testis, pancreas, colon, and lung. The expression of ATX in human tumors has not been extensively studied. Kawagoe and colleagues (16) recently found high levels of expression of ATX in the neuroblastoma cell line SMS-KAN and in primary tumors. ATX showed an autocrine motility activity in SMS-KAN cells. Kawagoe and colleagues also indicated an absence of detectable ATX expression in a variety of other human tumor cell lines that they screened, including glioma, myeloid and lymphoid leukemia, rhabdomyosarcoma, Wilms' tumor, and Ewing's sarcoma lines. We report here that ATX mRNA is expressed in cultured human bronchial epithelial cells (HBECs) and non-small-cell lung cancer (NSCLC) cell lines, and that in situ hybridization studies demonstrated differentiation-restricted ATX mRNA expression in normal and neoplastic human lung epithelial cells.

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

NSCLC and Normal HBEC Lines

Twelve NSCLC cell lines were routinely cultured in RPMI-1640 supplemented with 10% fetal calf serum at 37°C in a 5% CO₂ atmosphere. The cell lines included six adenocarcinoma (ADC) (MGH-8, MGH-13, MGH-24, NCI-H358, A549, and RVH-6849), three squamous-cell carcinoma (SQCC) (MGH-7, NCI-H157, and NCI-H520), two adenosquamous (ADSQ) carcinoma (MGH-30 and NCI-H125), and one large-cell undifferentiated carcinoma (LCUC) (NCI-H661) line. The genotype and phenotype of these cell lines have been reported previously (17). The primary cultures of normal HBECs (HBE-158 and HBE-154) were done as previously reported (20). These HBECs were cultured in keratinocyte serum free (KSF) medium (GIBCO-BRL, Gaithersburg, MD), with or without supplements of epidermal growth factor (EGF) (5 ng/ml) and bovine pituitary extract (BPE) (50 µg/ml). The immortalization of HBE-135 cells by HPV16-E6E7 genes was reported previously (21).

Normal Lung and NSCLC Tissues

Human normal lung and tumor tissues were obtained from lobectomy or pneumonectomy specimens. NSCLC tissues were harvested and snap frozen in liquid nitrogen, or were fixed in 10% buffered formaldehyde as soon as, but usually within 30 min after, their surgical removal. The corresponding normal lung tissues of the tumor specimens were also available in seven of these cases. The snap-frozen tissues were stored at 80°C prior to subsequent isolation of their RNA. The histopathologic diagnoses of NSCLC were based on routinely stained slides of surgical pathology tissue blocks. The stages of the tumors were derived from the pathologic reports in these cases.

Reverse Transcription-Polymerase Chain Reaction and Cloning

Total cellular RNA was isolated from cultured cell lines according to the method of Chomczynski and Sacchi (22). Two micrograms of total RNA were reverse transcribed through use of a kit purchased from Pharmacia Biotech (Baie d'Urfe, PQ, Canada). The efficiency of the reverse transcription (RT) reaction was first checked by amplification through the polymerase chain reaction (PCR) of -actin, using the primer set listed in Table 1. The cDNA was then subjected to PCR amplification with the primer sets A and B designed from published ATX^mel GenBank sequences (accession No. L35594; GenBank, Bethesda, MD). PCRs were conducted with Taq DNA polymerase (Perkin-Elmer, Foster City, CA) under the following conditions: an initial 94°C denaturation for 4 min, followed by 25 cycles of 94°C for 30 s each, annealing at 50°C for 45 s, and termination at 72°C for 1.5 min. The last cycle was followed by a 10-min reaction at 72°C. The PCR products were electrophoretically separated in a 1.5% agarose gel, denatured in 0.5 M NaOH solution, and transferred to a Hybond-N membrane (Amesham Canada, Oakville, ON, Canada). Initial experiments showed that cDNA amplification was linear within 20 to 30 cycles of PCR for ATX and -actin. RT- PCR detection of HGF/SF was done as previously described (23).

The amplified ATX cDNA fragments were also isolated and subcloned into the pGEM-T vector (Promega, Madison, WI). The cloned ATX sequences were then verified by sequencing. These cDNAs were subsequently used as probes to hybridize the Southern blots of PCR products, and the signals were estimated densitometrically using Computing Densitometer model V3.1 (Molecular Dynamics, Sunnyvale, CA).

ATX Isoform Analysis

The differential expression of the three ATX/PD-I isoforms in lung cancer cells was analyzed with RT-PCR, using primer sets designed from published sequences for ATX^ter (GenBank accession no. L46720). PD-I can be distinguished from both ATX^mel and ATX^ter by an insertion of 75 bp at nucleotide 1,833 of ATX^ter cDNA sequences. With use of the ATX-C primer set (Table 1), RT-PCR is predicted to yield a 446-bp product for ATX and a 521-bp product for PD-I. Furthermore, ATX^mel mRNA expression can be distinguished from ATX^ter by an insertion of 156 bp at nucleotide 1,026 of ATX^ter cDNA sequences. The ATX-D primers would distinguish these two forms of ATX by yielding a 451-bp product for ATX^mel and a 295-bp product for ATX^ter or PD-I.

Northern Blot Analysis

Total RNA was isolated from 38 tumor and seven normal tissues. ATX mRNA expression was then analyzed with the Northern blot technique, as previously described (20). To control the RNA loading, all blots were rehybridized with an 18S cDNA probe (American Type Culture Collection, Rockville, MD).

In Situ Hybridization

Three-micrometer-thick sections of formaldehyde-fixed and paraffin-embedded tissues were mounted on Superfrost-plus slides (Fisher Scientific, Pittsburgh, PA) and then heated for overnight at 45°C. Riboprobes were prepared from PCR-cloned 3' ATX cDNA, using the ATX-B primers as described earlier. Antisense RNA probe was generated from SalI-linearized pGEMT-ATX, using T7 polymerase, and sense RNA probe was generated with SP6 polymerase from the NcoI-linearized plasmid. In vitro transcriptions were performed with the DIG RNA labeling kit (Boehringer-Mannheim Canada, Dorval, PQ, Canada). The sections were dewaxed twice in xylene for 10 min each, rehydrated through graded ethanol solution, and digested for 10 min with 20 µg/ml of proteinase K (Boehringer-Mannheim) in 0.1 M Tris-HCl buffer containing ethylenediaminetetraacetic acid, pH 8.0, at room temperature. The sections were then postfixed with freshly prepared 4% paraformaldehyde for 30 min at room temperature. After rinsing with phosphate-buffered saline, the sections were acetylated for 10 min in 0.1 M triethanolamine, pH 8.0, with 0.25% acetic anhydride. Sections were prehybridized at 50°C for 2-4 h in an aqueous solution containing 2× standard saline citrate (SSC), 1× Denhardt's solution, and 50% formamide, and then hybridized at 50°C overnight in the same reaction buffer, containing ATX digoxigenin-labeled riboprobe (3 ng/µl) and yeast RNA (20 µg/ml) (Boehringer Mannheim). Following hybridization, sections were treated at 37°C for 30 min with ribonuclease (RNase A) (20 µg/ml) (Boehringer Mannheim) in 10 mM Tris-HCl solution, pH 7.6, containing 500 mM NaCl. Slides were then washed in 2× SSC and 0.5× SSC at 50°C for 30 min each, and were then immersed at room temperature for 30 min in a blocking buffer (buffer A) of 100 mM Tris-HCl, pH 7.5, and containing 150 mM NaCl and 1% bovine serum albumin (Boehringer Mannheim) and 0.2% normal goat serum. After an incubation with alkaline phosphatase-conjugated antidigoxigenin F(ab) antibody (Boehringer Mannheim) at 1:500 dilution in or buffer A for 4 h at room temperature, the sections were washed twice respectively with buffer A and buffer B (100 mM Tris-HCl, pH 9.5, containing 100 mM NaCl and 50 mM MgCl). The hybridization result was revealed by development in buffer B, containing 17.5 mg/ml nitroblue tetrazalium salt, 33.7 mg/ml bromo-4-chloro-3-indoyl phosphate, and 1 mM levamisole. The slides were then counterstained with 0.1% methyl green.

Immunohistochemistry

The streptavidin-biotin immunoperoxidase method was used to study the expression of CD20 and CD45RO. Anti-CD20 antibody (Dako Canada, Mississauga, ON, Canada) was used at 1:50 dilution, and anti-CD45RO clone A6 antibody (Zymed, South San Fransisco, CA) was used at 1:200 dilution. Immunoreactivity was revealed with 3-amino-9-ethylcarbazole, which yielded a red precipitate.

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

ATX mRNA Expression In Vitro

Using the RT-PCR technique and consensus primer sets A and B, ATX/PD-I mRNA expression was detected in primarily cultured (HBE-158 and HBE-154) and immortalized (HBE-135/E6E7) bronchial epithelial and NSCLC cell lines (Figure 1A). HBECs cultured in medium supplemented with EGF and BPE showed higher ATX/PD-I expression levels than those cultured in basal medium (Figure 1B). HBECs also expressed low levels of HGF/SF, but only when cultured in supplemented KSF medium. ATX/PD-I mRNA expression was also detected in all NSCLC cell lines tested, and in seven cell lines the expression levels were significantly higher than in HBECs (Figure 1A). Overexpression of ATX/PD-I mRNA was found in cell lines of all histologic subtypes (one of two SQCC, four of six ADC, one of two ADSQ, and one LCUC). It is worth noting that all NSCLC cell lines originated from or formed poorly differentiated tumors, and that their ATX/ PD-I mRNA expression levels could not be correlated with the degree of differentiation in vitro. We have recently reported the spontaneous scattering activities of these NSCLC cell lines (24), and they were not correlated with the levels of ATX/PD-I expression (Figure 1A).

View larger version (49K): [in this window] [in a new window]   Figure 1.   Expression of ATX mRNA in primary normal HBECs and NSCLC cells. (A) RT-PCR demonstrates the expression of ATX/PD-I mRNA in normal bronchial epithelial and various subtypes of NSCLC cell lines, and the lack of a consistent correlation of this expression with the spontaneous scatter activity (SSA) of these cells. Upper panel: The 550-bp ATX/PD-I product obtained when 5'-region-specific ATX-A primers were used. Middle panel: Amplification of the same RT products using -actin primers and hybridization to 32P-labeled human -actin cDNA probe shows almost uniform concentration of mRNA in the samples. Lower panel: SSA of the cells as previously reported (25) (+ = presence of spontaneous scatter activity;  = absence of scatter activity; N = normal human bronchial epithelial cell line; SQC = squamous cell carcinoma; ADC = adenocarcinoma; ASC = adenosquamous carcinomas; LC = large cell carcinoma). (B) Human ATX/PD-I and HGF/SF mRNA expression in HBE 135/E6E7 cell line cultured in KSF medium with (S) or without (SF) growth factor supplements. Amplification of ATX/PD-I and HGF/SF with their specific primers (ATX-B and HGF/SF) gave their respectively expected products, which specifically hybridized to cDNA probes for ATX and HGF/SF. (C ) and (D) Representative Northern blot results demonstrating ATX/PD-I expression in normal lung (L45N) and primary NSCLC tissues. Twenty micrograms of total RNA from tumor and normal tissues were hybridized with an ATX 5'-region cDNA probe. A single 3.2-kb transcript was detected in all samples except L58T, whose RNA was degraded. The membranes were subsequently rehybridized with an 18S cDNA to demonstrate relatively equal RNA loading (N = normal lung tissue; ADC = adenocarcinoma; SQC = squamous-cell carcinoma).

ATX Isoform Expression

To delineate more precisely the type of ATX isoform that is expressed by NSCLC cells, we performed RT-PCR analysis with primers ATX-C and ATX-D. Figure 2 shows that in all four NSCLC cell lines that expressed high levels of ATX/PD-I, ATX^ter was the predominant isoform expressed. Faint bands corresponding to PD-I (521 bp, primers ATX-C) and ATX^mel (451 bp, primers ATX-D) were also noted, indicating low levels of expression of PD-I and ATX^mel.

View larger version (18K): [in this window] [in a new window]   Figure 2.   Differential analysis of ATX/PD-I isoform expression in NSCLC cell lines. (A) RT-PCR done with primer set C (see Table 1), showed predominant expression of ATX (446 bp) in most cell lines, with a much smaller amount of putative PD-I mRNA (521 bp) expressed in H661 cells. (B) RT-PCR done with primer set D further demonstrated the predominant expression of ATXter (295 bp) in these NSCLC cell lines. ATXmel (451 bp) appears to have also been expressed, but at much lower levels in these cell lines.

ATX Expression In Vivo

We further investigated the mRNA expression of the human ATX/PD-I gene in 38 primary NSCLC tissues. These included 20 ADCs, 11 SQCCs, and seven LCUCs. ATX/PD-I mRNA was detected by Northern blot analysis in 31 of 37 (83.8 %) tumors from which intact mRNA was isolated (Figures 1C and 1D). We also used RT-PCR to compare ATX/PD-I gene expression in seven paired-normal/tumor samples. Increased ATX/PD-I mRNA expression in tumor as compared with normal tissue was detected in only two (28.6%) of these cases (data not shown).

The expression of ATX/PD-I mRNA in NSCLC cells was further investigated with in situ hybridization. This technique offers an advantage over comparison of mRNA levels in whole-tissue extracts, since it allows expression levels of ATX/PD-I in tumor cells to be evaluated against ATX/PD-I expression in specific, adjacent normal cells. This study was done on 27 primary NSCLC tissues, including nine SQCC, 16 ADC, and two LCUC tissues.

In normal lung, the bronchial and bronchiolar epithelia showed significant ATX/PD-I mRNA expression, localized predominantly to the basal cells (Figure 3A). Hybridization with the sense probe for ATX/PD-I mRNA showed a complete lack of staining, indicating the specificity of the technique (Figure 3D). Positive but weak staining was also noted in type II pneumocytes, chondrocytes of bronchial cartilage, and vascular endothelial cells. Reactive type II pneumocytes showed markedly enhanced expression. The smooth-muscle cells of bronchial or vascular walls did not show expression of ATX/PD-I mRNA (data not shown). Subsets of peribronchial lymphocytes, especially those forming lymphoid aggregates, showed a strong hybridization signal for ATX/PD-I mRNA (Figure 4A). Using lineage-specific antibodies to CD20 (B cells) and CD45RO (T cells), we found that the majority of lymphocytes in these aggregates stained positively for CD20 (Figures 4B and 4C).

View larger version (191K): [in this window] [in a new window]   Figure 3.   In situ hybridization on formalin-fixed lung tissue sections, done with digoxigenin-labeled riboprobes for the 3' region of ATX/PD-I. Positive signal is revealed as a purplish blue color. (A) Normal bronchiolar epithelium shows ATX/PD-I expression localized to the basal cells. (B) Strong expression in a poorly differentiated squamous-cell carcinoma (T). Note that the staining levels were similar to those of B lymphocytes in the adjacent stroma (asterisk). (C ) Strong expression in a poorly differentiated adenocarcinoma. (D) Negative staining when tumor section (B) was hybridized to the sense probe. (E ) Less homogeneous expression in a moderately differentiated adenocarcinoma. Note that the area in which tumor cells form glandular structures (T) showed decreased staining as compared with adjacent lymphocytes in the stroma (>). (F ) A squamous-cell carcinoma, showing decreased staining in differentiating tumor cells (>). Original magnifications: A and C: ×400; B, D, E, and F: ×100.

View larger version (46K): [in this window] [in a new window]   Figure 4.   A representative lymphoid aggregate that showed extensive hybridization signals for ATX mRNA (A) also predominantly demonstrated positive immunostaining with anti-CD20 antibody (B). Only occasional lymphocytes in these aggregates stained positively with T-cell marker CD45RO (C ).

We compared the ATX/PD-I expression in tumor cells in sections of NSCLC tissue with that in adjacent normal bronchial/bronchiolar epithelia and/or stromal lymphocytes (Table 2). Although cells of all subtypes of NSCLC expressed variable levels of ATX/PD-I, high levels of expression were clearly restricted to poorly differentiated tumors (Figures 3B and 3C), with better (well or moderately) differentiated tumors showing lower levels of expression (Figure 3E). This was also obvious within a single tumor, in which differentiated tumor cells showed loss of staining (Figures 3E and 3F). The difference between the frequency of ATX/PD-I overexpression in tumor cells as compared with normal bronchial/bronchiolar epithelium, and equivalent expression in tumor cells and stromal lymphocytes in poorly differentiated or undifferentiated NSCLCs (10 of 14), as compared with that in well or moderately differentiated tumors (0 of 13), was striking.

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

In this report, we provide evidence that ATX/PD-I is commonly expressed by NSCLC cells both in vitro and in vivo. We also show that ATX^ter is the predominant isoform expressed in these lung epithelial cells, and that in normal human lung, ATX expression is predominantly localized to the basal cells of the bronchial and bronchiolar epithelia. The finding that overexpression of ATX is significantly associated with poorly differentiated or undifferentiated cells and/or tumors suggest that it may be functionally more important in less differentiated epithelial cells. Aside from the human melanoma cells from which ATX was initially isolated, its expression has been found in teratocarcinoma and neuroblastoma cell lines. These also represent cells that are blastic, and hence primitive or undifferentiated. It is possible that ATX^ter plays important roles or functions in the biology of undifferentiated or stemlike epithelial cells.

ATX was initially cloned as an autocrine motility factor from a human melanoma cell line, but a closely homologous isoform was also isolated from a teratocarcinoma cell line. These peptides differ by 52 amino acids, but these amino acids are located outside the known functional domains of ATX/PD-I. It is unclear whether the two ATX isoforms or the brain-derived PD-I produce similar or different biologic effects. Although the riboprobe for our in situ hybridization studies was generated from the sequences that represent the consensus region for all isoforms of ATX/PD-I, we have shown that in vitro, ATX^ter mRNA is by far the most predominant ATX isoform mRNA expressed in these lung epithelial cells. Northern blot analysis of a panel of human tissue mRNA has previously demonstrated high levels of ATX/PD-I expression in the brain, placenta, ovary, and small bowel, with lower expression found in lung, kidney, and testis. The cells that express ATX/PD-I in these tissues remain largely unknown.

Lee and coworkers (14) did not find ATX/PD-I expression in the spleen or thymus, suggesting that it is not expressed in lymphoid tissues. This is contrary to our observation that some lymphocyte populations in peribronchial or tumor stromal tissues showed very intense expression of ATX/PD-I, and these lymphocytes predominantly appeared to demonstrate a B-cell immunophenotype. The PC-1 gene, which is 44% homologous to that for ATX/PD-I, was initially isolated as an activated B (plasma)-cell membrane-specific glycoprotein. PC-1 sequences show homology with ATX/PD-I throughout the extracellular portion of these two proteins, but two proteins' intracellular and transmembrane domains differ significantly. The ATX sequences we isolated and used as in situ hybridization probes showed no significant homology with PC-1 cDNA, indicating that B lymphocytes also express high levels of ATX/PD-I. PC-1 expression has also been demonstrated in nonlymphoid cells, including those of the distal convoluted tubular epithelium of the kidney, epididymis, and chondrocytes.

Besides effecting motility in A2058 melanoma cells, ATX/PD-I was shown to be an autocrine motility factor for neuroblastoma (SMS-KAN) cells (16). In our NSCLC cells, correlation between ATX expression levels and their spontaneous scattering activity was poor, suggesting that the motility activity of ATX is not its predominant role in these cells. ATX may potentially have an autocrine motility function in HBECs. When HBECs are cultured in KSF medium supplemented with BPE and EGF, they show high levels of spontaneous motility (23), but this activity is markedly reduced when HBECs are conditioned to grow in basal KSF medium without supplements. A very low level of HGF/SF is also expressed when HBE cells are cultured in supplemented medium, but not in basal medium. Because purified ATX and anti-ATX antibody are currently not widely available, we have not yet been able to evaluate the pathobiologic roles of ATX in HBECs and NSCLC cells. Nevertheless, the preferential expression of ATX in undifferentiated cells suggests that it may play important roles during neoplastic transformation and tumor progression. The lack of access to clinical follow-up data, and the small number of samples we studied, precluded a correlative analysis with prognosis in cases of NSCLC. Further studies are required to determine the clinicopathologic significance of ATX expression in NSCLC.