Coexpression of Insulin Receptor-Related Receptor and Insulin-Like Growth Factor 1 Receptor Correlates with Enhanced Apoptosis and Dedifferentiation in Human Neuroblastomas

Neuroblastoma is the third most common pediatric cancer and is responsible for ∼15% of all childhood cancer deaths. It is an embryonal tumor of the postganglionic sympathetic nervous system, which most commonly arises in para- and prevertebral ganglia and in the adrenal gland. The 6-year survival probability of all patients with neuroblastoma is ∼60% (own data based on 2030 patients registered from 1980 to 2003). In addition to clinical markers such as stage and age at diagnosis, biological tumor markers have gained increasing importance in determining the prognostic outcome for patients with neuroblastoma. The most extensively studied biological marker for an unfavorable prognostic outcome in patients with neuroblastoma is the amplification of the MYCN³ oncogene (1, 2, 3).

MYCN is differentially expressed in neuroblastomas with or without MYCN gene amplification and in different clinical stages (4). However, the prognostic impact of MYCN expression is controversial and has not been shown to be of prognostic relevance in most studies (5, 6). Among other biological markers for a favorable prognostic outcome in neuroblastoma are expression of the high-affinity nerve growth factor receptor TrkA (7) and the low-affinity nerve growth factor receptor p75NTR (8, 9, 10).

The IGF-1R is of prognostic relevance in a variety of tumor entities and influences apoptosis, differentiation, and proliferation in coexpression with other RTKs (11, 12). In neuroblastoma cells, expression of IGF-1R has been implicated in growth-promoting and antiapoptotic signaling (13). In contrast to its implication in cell survival, proapoptotic action by IGF-1R has been seen in cells that have been triggered to undergo apoptosis through receptors of the death receptor family, such as TNF receptors or p75NTR (14).

Recently, the IRR has been identified as an additional member of the insulin RTK family (15, 16). However, its ligand and biological functions are still unknown. Various ligands activating the IR or the IGF-1R, such as proinsulin, insulin, IGF-1, IGF-2, and relaxin, are not able to bind the extracellular domain and activate IRR (17, 18). The functional potential of the tyrosine kinase domain of the IRR was demonstrated by use of receptor hybrids of a known extracellular domain and the intracellular domain of the IRR. Thus, IR substrate-1 and -2 have been identified as receptor substrates of the IRR, implicating a role for the IRR in influencing downstream signaling pathways such the phosphatidylinositol 3′-kinase and the mitogen-activate protein kinase pathway (17).

In contrast to the widespread patterns of expression of the homologous IR and IGF-1R, IRR demonstrates a very restricted cellular distribution in a subset of tissues of neuronal origin, where its appearance is closely associated to that of IR, IGF-1R, and the nerve growth factor receptor TrkA (19, 20). IRR and TrkA appear early in the embryonal development of dorsal root and trigeminal neurons and later, near the time of birth, in sympathetic neurons. This association is highly selective: TrkA mRNA is not detected anywhere else in the developing nervous system in the absence of coordinate IRR expression (20, 21). Furthermore, it was shown recently that IRR is also coexpressed with IGF-1R in neuroblastoma cell lines (22). Coexpression of receptors of the IR family has been shown to influence activation of downstream signaling pathways and cellular biological functions, which might be explained by heterodimerization of the expressed receptor monomers on the cell surface, resulting in different ligand-binding affinities and autophosphorylation status (23, 24, 25, 26).

These observations prompted us to study the possible role of IRR expression in primary neuroblastomas. We examined the relationship of IGF-1R, TrkA, and p75NTR mRNA expression with the expression of IRR and correlated this expression pattern to biological effects such as proliferation, apoptosis, and morphological differentiation. We also asked whether MYCN amplification/expression status correlates with the IRR expression pattern and whether the IRR might have any prognostic impact on patients with neuroblastoma.

We studied tumor specimens from 49 children with neuroblastoma who had been diagnosed in Germany from 1989 to 1997. All neuroblastoma diagnoses were confirmed by histological assessment of a tumor specimen obtained at surgery. The tumors were classified according to the INSS criteria (27). We studied 8 stage 1 (16%), 7 stage 2 (14%), 9 stage 3 (18%), 14 stage 4 (28%), and 11 stage 4S (22%) tumors. The relative incidence rate in our study reflects the relative incidence rate of neuroblastomas in Germany within the investigated time period (1990–1997): stage 1 (19%), stage 2 (12%), stage 3 (20%), stage 4 (39%), and stage 4S (10%).

The median age of the patients at diagnosis was 12.9 months (range, 0.2–131.6 months). Thirteen (26.5%) tumors had a DSR histology, six (12.2%) were SP-D, and 30 (61.2%) were SP-U (28). All patients were treated in a stage-specific manner according to previously described protocols (29). The survival probability of our study population calculated with Kaplan-Meier analysis was 70.5%.

The neuroblastoma specimens were frozen in liquid nitrogen and stored at −80°C for PCR and Southern and Western blot analyses. Portions of the samples (50–60 mg) were fixed in 10% buffered formalin. Serial sections 4-μm in thickness were cut from the same blocks, mounted on poly-l-lysine-coated slides, and used for H&E staining and in situ end-labeling of DNA (TUNEL assay).

Tumor samples (100 mg each) were homogenized in 1000 μl of RNAzol-B (Biotecx Laboratories Inc., Houston, TX). RNA was isolated after the addition of chloroform according to the guanidinium-isothiocyanate method (30) and finally precipitated with isopropanol. After centrifugation (14000 × g for 20 min), the pellet was washed twice with 70% ethanol, vacuum-dried, and dissolved in 50–80 μl of RNase-free distilled water.

We incubated 7500 ng of total RNA at 70°C for 10 min with 1.5 μl of RNase inhibitor (40 units/μl; Promega), 2.5 μl of oligo-primer-p(dT)15 (0.1 nmol/μl; Boehringer-Mannheim) and distilled water in a volume of 42 μl; we then cooled the mixture to 4°C. First-strand cDNA was synthesized with 1.5 μl of Moloney murine leukemia virus reverse transcriptase (SuperScript-II; Invitrogen) in a total volume of 75 μl [15 μl of 5× first-strand buffer (Invitrogen), 7.5 μl of dithiothreitol (0.1 m; Invitrogen), 3 μl of BSA, 1.5 μl of each deoxynucleotide triphosphate (0.01 μmol/μl; Perkin-Elmer), and 0.5 μl of RNasin]. Reverse transcription was performed for 10 min at 25°C and 60 min at 42°C, and finally for 15 min at 75°C to inactivate the enzyme.

Specific primers for the amplification of IRR, IGF-1R, p75NTR, TrkA, MYCN, and β-actin were searched with OLIGO 6.0 (Medprobe); the specificity was controlled with BLAST 2.0 (http://www.ncbi.nlm.nih.gov/blast/blast.cgi). To avoid coamplification of contaminating DNA, primers were located in two subsequent exons. All PCR products were sequenced by the dye-termination procedure on a ALF-express with 5′-Cy5-labeled primers according to the manufacturer’s protocol (Pharmacia Biotech); the results were compared with the respective published sequence data (http://www.ncbi.nlm.nih.gov/GenBank/index.html; Table 1).

PCR was performed in a total volume of 50 μl containing 5 μl of 10× PCR buffer, 2.5 units of Taq-polymerase-k (MoBiTec), 1 μl of each deoxynucleotide triphosphate (Perkin-Elmer), 50 pmol of each primer, 200 ng of total cDNA, 1 μl of formamide (Sigma), and 33 μl of distilled water.

The primer sequences and individual PCR conditions (annealing and elongation temperature, time, and cycle number) for each PCR are listed in Table 1. In every PCR, an initial denaturation step at 95°C for 3 min was performed, denaturation during cycling was performed at 95°C for 30 s, and the last cycle was followed by an elongation step at 72°C for 7 min. The reaction was then cooled to 4°C.

In the semiquantitative PCR of MYCN relative to β-actin (M/A), the first 5 cycles were done only with the MYCN primers. After the reaction was cooled to 4°C and the β-actin primers were added, the reaction was continued for 28 cycles, as described by Kinoshita et al. (31). PCR conditions have been established for both gene fragments to be in the logarithmic phase at the time of analysis. After agarose gel electrophoresis, the level of MYCN expression in each tumor was determined by densitometry and was expressed as the M/A ratio. MYCN expression was defined as high for M/A at or above the median and as low for M/A below the median expression level.

Relative to common housekeeping genes (β-actin and GAPDH), the investigated receptors were found to be expressed at lower levels. Thus, semiquantification of receptor RNA expression was difficult and inaccurate, particularly in comparison with the MYCN/β-actin PCR. We decided to use a common cDNA-specific PCR with higher cycle number (up to 40) to avoid false-negative results. For a loading control, we equalized the amount of cDNA from each tumor with a GAPDH PCR for 26 cycles, which gave comparable specific bands (data not shown). The β-actin band in the MYCN/β-actin PCR was then used as an additional control. Each specific band was regarded as positive expression. Each tumor that was PCR negative was controlled in a second PCR for 50 cycles. None of the controlled tumors turned positive (Fig. 1).

The amplified PCR fragments (10 μl of the PCR products) were separated on a 2% agarose gel by electrophoresis, identified by ethidium bromide staining, and documented by the Image Master VDS software (Pharmacia).

For the detection of MYCN amplification, the pNB-1 probe (MYCN is localized on 2p24.1; ATCC 41011) was used; control hybridization for a single-copy signal was performed with HuCK (IGKC is the immunoglobulin κ constant region, localized on 2p12; ATCC 59172).

High-molecular-weight DNA was isolated by proteinase K digestion and phenol-chloroform extraction according to standard procedures. We digested 10 μg of DNA from each sample with EcoRI, and fragments were separated by 0.8% agarose gel electrophoresis. The DNA was then transferred to Gene Screen Plus membranes (DuPont) by vacuum-blotting (Pharmacia) according to the specifications of the manufacturer. Prehybridization and hybridization were performed according to standard procedures. Filters were washed twice in 0.1× SSC and 0.1% SDS at room temperature for 20 min and three times at 65°C for 30 min. After a brief rinse in 0.1× SSC, autoradiography was performed at −70°C with Kodak XAR-5 film (Fig. 1).

Cellular protein was extracted by lysing 30 mg of tumor tissue with 300 μl of lysis buffer [NP-40 + 4: 150 mm NaCl, 1% (v/v) NP-40, 50 mm Tris-buffer (pH 8), 5 mg/ml aprotinin, 5 mg/ml leupeptin, 1 mg/ml pepstatin, 0.2 m phenylmethylsulfonyl fluoride]. We separated 30 μg of protein in a 12% gradient SDS–polyacrylamide gel and electroblotted the bands to Hybond-Enhanced Chemiluminescence (ECL) nitrocellulose membrane (Amersham Corp., Arlington Heights, IL). After the membranes were blocked with 5% nonfat dry milk and 0.1% Tween 20 in Tris-buffered saline, they were incubated at room temperature for 2 h with a 1:400 dilution of a rabbit polyclonal anti-cyclin A antibody (Santa Cruz Biotechnology, Santa Cruz, CA), then with an antirabbit peroxidase-conjugated secondary antibody (Amersham). The blot was finally probed by the ECL Western blotting detection system (Amersham). Equal loading of protein was confirmed by rabbit polyclonal cdk2 antibody (Santa Cruz Biotechnology; Fig. 1).

The chromatin in apoptotic cells is cleaved at internucleosomal sites. To detect this fragmented DNA, we performed the TUNEL assay, using the In Situ Cell Death Detection kit (Boehringer Mannheim). Paraffin-embedded sections (4 μm) were deparaffinized and then incubated with proteinase K (20 μg/ml) in PBS for 15 min at 25°C. Methanol containing 0.3% H₂O₂ was applied to neutralize endogenous peroxidase. The sections were incubated with Tdt enzyme in the presence of a nucleotide mixture. After washing in PBS, the sections were stained with 4′,6-diamidino-2-phenylindole.

Apoptosis was quantitated by determining the percentage of positively stained cells within a field at a magnification of ×400. At least 1000 tumor cells were counted from 10–20 randomly chosen fields per slides assayed, and the counts were averaged to obtain the AI. For every tumor, two slides were evaluated. The median AI of all tumors examined was 0.82% (Fig. 2).

Fisher’s exact test was used to examine possible correlations between IRR expression and expression of other receptors (PCR positive versus negative), histology (DSR versus SP-D/SP-U), AI [low (≤0.82%) versus high (>0.82%)], proliferation [low versus enhanced/high (high)], MYCN amplification versus nonamplification, MYCN expression [low (M/A ≤0.220) versus high (M/A >0.220)], and clinical data (Table 2). Reported Ps are not corrected for multiple testing.

The 6-year survival probability data for IRR, IGF-1R, TrkA, and p75NTR expression and MYCN amplification/expression were evaluated by the Kaplan-Meier method, and differences between survival probability curves were calculated using the log-rank test (Table 3). Multivariate analysis of IRR expression with respect to MYCN amplification, stage, age at diagnosis, MYCN amplification, MYCN expression, expression of other RTKs, and histology was calculated using Cox regression analyses (Table 4). P < 0.05 was regarded as significant. Statistical analyses was performed with the SPSS version 10.0 software (SPSS Inc.).

Representative experimental results are shown in Figs. 1 and 2 for the detection of MYCN amplification by Southern blot (Fig. 1,A), RNA expression by reverse transcription-PCR (Fig. 1,B), cyclin A expression by Western blot (Fig. 1,C), and the TUNEL assay (Fig. 2) to determine the AI.

IRR mRNA expression was found in 25 of 49 (51%) primary human neuroblastomas. IRR was expressed in 63% of stage 1, 57% of stage 2, 56% of stage 3, and only 14% of the metastatic disease stage 4 neuroblastomas, but in 82% of disseminated stage 4S neuroblastomas.

Comparison of stages 1, 2, and 3 with stage 4 or stages 1, 2, and 4S with stages 3 and 4 IRR revealed that expression was significantly correlated with stages 1, 2, and 3 (P = 0.016) and stages 1, 2, and 4S (P = 0.002; Fig. 3).

IRR expression was seen in 70% of neuroblastomas diagnosed in infancy, but in only 35% of tumors diagnosed after the first year of life (P = 0.022).

We found a statistically significant coexpression of IRR with IGF-1R (P < 0.001), TrkA (P < 0.001), and p75NTR (P = 0.013; Table 2).

IGF-1R expression was found in 69% of the investigated neuroblastoma tumors, expression of TrkA in 53%, and expression of p75NTR in 71%, respectively. Similar to the IRR, TrkA was significantly correlated with an age of ≤12 month at diagnosis (P = 0.021). There was a trend for preferential expression of the IGF-1R in infants compared with older patients (P = 0.071), but p75NTR was not expressed in an age-dependent manner (P = 0.761).

IRR expression was correlated with an undifferentiated histology (P = 0.0036), as was TrkA (P = 0.022). In contrast to these findings, expression of IGF-1R and p75NTR was not associated with an undifferentiated histology. As expected, deduced from the already known functions of p75NTR, expression of this receptor was correlated with a high AI (P = 0.003) as was coexpressed TrkA (P = 0.003) but not IRR or IGF-1R. Surprisingly, there was no significant correlation between expression of any of the receptors studied and the proliferation rate.

In the subgroup of IGF-1R-expressing tumors (n = 34), IRR coexpression was significantly associated with an undifferentiated histology (P = 0.0017) and a high AI (P = 0.017). There was no correlation of IRR/IGF-1R coexpression with the proliferation rate (Fig. 4).

IRR was expressed in 61% of neuroblastomas without MYCN amplification, but in only 18% of MYCN-amplified neuroblastomas (P = 0.018). In neuroblastomas without MYCN amplification, IRR expression was correlated with high MYCN expression (P = 0.02).

IRR mRNA expression is a new prognostic marker in neuroblastoma. A 6-year survival probability of 91.8% was seen in neuroblastoma patients with IRR expression in the tumor as compared with 49.7% in patients without IRR expression in the tumor (P = 0.003; Table 3; Fig. 5). In a multivariate Cox regression analysis, the prognostic impact of IRR expression was independent of MYCN amplification (P = 0.037), age at diagnosis (older than 12 months versus 12 months or younger; P = 0.045) and histological differentiation (DSR and SP-D versus SP-U; P = 0.014), but it was not independent of stage (1, 2, and 3 versus 4; Table 4).

Because the IRR is a heterodimerization partner for IGF-1R, we were interested in the prognostic value of IRR expression in the subgroup of IGF-1R-expressing tumors (n = 34). In this subgroup we found a 6-year survival probability of 90.9% for tumors that coexpressed IRR/IGF-1R-versus 51.9% in tumors expressing only IGF-1R (P = 0.0108; Fig. 5). In multivariate analysis, IRR expression was prognostically independent of IGF-1R expression (P = 0.014; Table 4). However, we found no correlation with apoptosis, proliferation, and differentiation for IRR expression in the subgroup of IGF-1R-nonexpressing tumors.

TrkA expression was of prognostic significance, and p75NTR approached significance in the 49 neuroblastomas investigated in the present study (P = 0.017 and 0.08, respectively). IGF-1R expression was not of prognostic significance (P = 0.122; Table 3).

The MYCN gene was amplified in 11 (22%) of 49 investigated neuroblastomas. MYCN mRNA expression was strongly correlated with MYCN amplification; thus, 10 (91%) of the 11 neuroblastomas with MYCN gene amplification but only 14 (37%) of 38 tumors without MYCN gene amplification expressed MYCN mRNA at a high level (P = 0.002). MYCN amplification significantly differentiates an unfavorable prognostic subgroup in this study (P < 0.0001). Expression of MYCN RNA was of no prognostic relevance in the group of 49 neuroblastomas (P = 0.6).

IGFs and their receptors regulate cell proliferation, differentiation, and death of normal and neoplastic cells. They have been implicated in the development of the nervous system and in the pathology of neuroblastoma (32, 33). Our study indicates that IRR, a new member of the IR-/IGF receptor family is differentially expressed in primary human neuroblastomas, which is suggestive for a role of the IRR in neuroblastoma biology. We also examined the expression pattern of IGF-1R, TrkA, and p75NTR, which have been previously shown to be expressed in neuroblastoma cells, to investigate the possible interaction of these receptors in concert with IRR on biological effects as proliferation, apoptosis, and differentiation.

In the present study, IRR mRNA expression strongly correlated with a favorable outcome in neuroblastoma, which can be attributed to the close association of IRR expression with favorable tumor stages and a predominant expression in neuroblastomas diagnosed in infancy. Expression of TrkA also correlated significantly with a favorable prognosis, whereas expression of p75NTR and IGF-1R showed only a statistical trend toward a better survival probability (Table 3).

In multivariate analysis, the prognostic impact of IRR expression was statistically independent of MYCN amplification status and age at diagnosis, but not independent of tumor stage (Table 4).

Independency of MYCN amplification and age makes IRR expression interesting for further therapy stratification because at present, in stage 1, 2, 3, and 4S disease, age and MYCN amplification are the most commonly used parameters for stratification.

Dimerization of receptor monomers at the cell surface was shown to be one early event in the activation of the RTKs, leading to autophosphorylation of intracellular domains and further activation of signaling cascades. Physiologically, dimerization is initiated by ligand binding but also by overexpression of receptor monomers exceeding a specific cell surface density threshold for autodimerization (34). We therefore examined the expression pattern of the investigated RTKs themselves and in combination with potential heterodimerization partners that might be important with respect to biological functions. Receptor expression was further compared with proliferation rate (examined by cyclin A protein expression), apoptosis (examined by the TUNEL assay), and morphological differentiation (examined by histology). We showed that IRR expression was strongly correlated with neuroblastomas with an undifferentiated histology, independent of AI and proliferation rate.

Known biological effects of IGF-1R expression in neuroblastoma cells include the induction of proliferation and down-regulation of apoptosis (35, 36, 37, 38). In contrast to these functions of IGF-1R, the expression pattern of this receptor was not associated with differentiation, proliferation rate, or apoptosis in tumor cells of primary neuroblastomas of our cohort. Missing associations between IGF-1R expression and biological functions might in part be explained by coexpression of IRR and IGF-1R. Recently, Kovacina and Roth (24) showed that these two receptors, when coexpressed, form heterodimeric receptors composed of an αβ-subunit of IGF-1R and another αβ-subunit of IRR and that these heterodimeric receptors are functional RTKs. Our correlative observations of enhanced apoptosis and undifferentiated histology in tumors expressing both receptors in contrast to tumors expressing only IGF-1R indicates an important role of IRR as a coexpression partner for IGF-1R in neuroblastoma cells. This observation is particularly important because an own ligand for IRR has to date not been found. Furthermore, the prognostic outcome of patients with IRR/IGF-1R coexpression was more favorable than that of patients expressing only IGF-1R (P = 0.018; Fig. 4).

In addition to IRR/IGF-1R coexpression, we found a significant coexpression of IRR and TrkA. In multivariate analysis, we showed a dependency of the prognostic impact of IRR expression on TrkA expression. This expression pattern has shown to be important for developing sympathetic and sensory neurons (20). Heterodimerization, however, has only been described for IGF-1R and not for TrkA (24).

Expression of the p75NTR was strongly correlated with a high AI but not to the proliferation rate and, in contrast to IRR, not to an undifferentiated histology; thus, it also exerts its function of inducing apoptosis in more differentiated neuroblastoma cells. Recently, it was shown that p75NTR as a member of the TNF receptor family induces apoptosis in neuroblastoma cells (39, 40, 41), which is in accordance with our results. The significant coexpression of p75NTR with the investigated RTKs might explain the correlative trend of RTK expression linked to a higher AI. These results are in accordance with the findings of Niesler et al. (14), who observed potentiation of TNF-induced apoptosis by activated IGF-1R.

In the present study, MYCN-amplified neuroblastomas had higher median MYCN expression compared with neuroblastomas without MYCN amplification, which is in accordance with the published literature (4, 5). We found that IRR mRNA expression strongly correlated with neuroblastomas without MYCN amplification and that within MYCN-nonamplified tumors, IRR was associated with high MYCN expression. This is in accordance with the findings of Cohn et al. (6), who described a trend toward a better prognostic outcome for neuroblastomas diagnosed in infancy without MYCN amplification and higher MYCN mRNA expression. Finally, these data reflect the complexity of influences on cellular functions by MYC transcription factors, because forced expression of MYC oncogenes induces proliferation and prevents differentiation on the one hand, but induces apoptosis on the other (42). The low frequency of IRR expression in MYCN-amplified tumors might reflect the developmental stage from which these MYCN-amplified neuroblastoma cells are derived.

We conclude that IRR might have a role in the biology and pathogenesis of human neuroblastomas. In agreement with the current literature, we hypothesize that IRR might work as a regulatory partner of IGF-1R in IGF signaling. IRR influences the physiological development of the embryonal nervous system and thus might also play a role in tumorigenesis of malignancies originating from premature neuronal cells. Furthermore, our data indicate that the assessment of IRR expression in neuroblastoma tumors at the time of diagnosis provides confirmative prognostic information, which helps in determining the most appropriate duration and intensity of treatment. Future therapeutic approaches may be aimed at influencing IRR expression in human neuroblastomas and, thus, RTK signaling pathways contributing to the activation of apoptosis.

We thank our colleagues from ∼100 German pediatric centers for providing us with neuroblastoma tumor material since 1980. We are grateful to Prof. Dr. Fritz Lampert, former head of the Department of Pediatric Oncology at the Justus-Liebig University of Giessen, for support of our laboratory.