Skip to main content

Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

Specific gene expression profiles distinguish among functional allelic variants of the mouse Pthlh gene in transfected human cancer cells

Abstract

The mouse parathyroid hormone-like hormone (Pthlh) gene encodes three allelic variants characterized by amino acid substitutions that are associated with susceptibility (PthlhPro) or resistance (PthlhThr and PthlhSerAspTyr) to two-stage skin carcinogenesis and to modulation of cell migration in vitro in transfected human cancer cells. cDNA microarray hybridization analysis of 8473 transcript clones revealed a similar gene expression profile for the PthlhThr and PthlhSerAspTyr alleles but a distinct pattern for the PthlhPro allele, suggesting an association between a specific gene expression profile and biological function of the Pthlh alleles. Some of the genes modulated by the Pthlh alleles, e.g., ANXA1, CCL2, FN1 and TFF3, play a role in cell migration and may represent candidate targets for this Pthlh function. Our study demonstrates the potential usefulness of gene expression profiling of genetic variants for the functional characterization of candidate cancer modifier genes.

Introduction

About 100 cancer modifier genes, whose allelic variants control genetic predisposition or resistance to different types of cancer, have been mapped in the mouse genome (Dragani, 2003). Identification of these genes would contribute considerably to the understanding of the mechanisms of tumorigenesis and thus provide new tools for prevention and treatment of cancer. However, difficulties in the functional analysis of several dozens of candidate genes mapping in restricted linkage regions have thus far prevented large-scale positional cloning of candidate cancer modifier genes.

We describe here a rapid method for the functional analysis of allelic variants of candidate cancer modifier genes containing amino-acid substitutions. The method involves analysis of the gene expression profile induced in vitro by these allelic variants in transfected human cancer cells, since functional biological differences between allelic variants are likely associated with alterations in gene expression. Such analyses might also provide insight on the biochemical mechanisms and downstream regulation associated with the allelic effects.

As a model cancer modifier gene, we tested the allelic variants of the mouse parathyroid hormone-like hormone (Pthlh) gene (Strewler, 2000). Two of the alleles, PthlhPro and PthlhThr, are linked, respectively, with genetic predisposition and resistance to two-stage skin tumorigenesis in phenotypically selected mice derived from laboratory inbred strains (Manenti et al., 2000). The evolutionarily distant mouse species Mus spretus carries an additional allele (PthlhSerAspTyr) and these mice are genetically resistant to skin carcinogenesis (Nagase et al., 2001). In vitro effects of Pthlh alleles on cellular morphology, adhesion efficiency and migration capability, as well as growth of the transfected human squamous cell carcinoma cell line NCI-H520 in nude mice, correlate with in vivo susceptibility/resistance to skin tumorigenesis of the mouse strains and lines carrying these alleles (Manenti et al., 2000; Benelli et al., 2003).

Here, we show that the allelic variants of the Pthlh gene transfected into NCI-H520 cells are associated with distinct gene expression profiles, pointing to the potential for broader application of such analyses in assessing the functional activity of cancer modifier genes.

Pthlh alleles induce specific gene expression profiles in transfected cells

Sequence analysis of about 1000 spotted clones of our IMAGE clone collection revealed false insert in about 10% of the clones (not shown), that is, slightly less than the reported 15–20% misidentified cDNA clones or contaminants representing more than one gene in the same cDNA collection (Ross et al., 2000).

For the comparison of PthlhPro-PthlhThr-transfected cells, genes whose expression differed >0.5-fold (measured as log2 ratio) in both experiments were used for further analysis. In the gene profiling experiment, gene expression matrices containing log2 ratios of Cy-5/Cy-3 of the three experiments were constructed using a mean normalization. Data for genes with more than 5% missing (invalid) values were excluded for cluster analysis.

Comparison of the gene expression profiles of PthlhThr and PthlhPro alleles from a typical dye-swap experiment using customized glass slides containing about 8500 cDNA genes identified 205 genes that were differentially expressed at a log2 ratio of ±0.5 in cells carrying either allele. Specifically, 142 genes were upregulated and 63 genes were downregulated.

Expression of these genes was investigated in a gene profiling experiment in which cDNAs from cells transfected with the PthlhThr, PthlhPro, or PthlhSerAspTyr allele were compared to reference cDNAs from cells transfected with empty vector. This analysis confirmed the differential expression of almost all the genes previously evaluated by dye-swap analysis of PthlhPro and PthlhThr. Indeed, only 19 (9%) of the 205 genes selected from the dye-swap experiment showed a different profile. From the gene profiling experiment, genes showing a standard deviation of >0.3 among the three Pthlh alleles were selected and 244 clones identified for use in an unsupervised hierarchical clustering analysis. The dendrogram in the Figure 1 shows that the Pthlh allele derived from the M. spretus strain (PthlhSerAspTyr) clusters with the PthlhThr allele, the two alleles showing a highly correlated expression pattern (r=0.68, P< 0.0001), whereas the PthlhThr and PthlhPro alleles, which show functional biological differences, had a poor correlation coefficient (r=0.15, P=0.02).

Figure 1
figure1

Gene expression profiles of PthlhThr and PthlhSerAspTyr alleles with each other correlate but differ from that of the PthlhPro allele. Unsupervised hierarchical clustering analysis was performed on the expression profile of 25 selected genes in NCI-H520 cells transfected with the mouse PthlhThr, PthlhPro, or PthlhSerAspTyr allele. The normalized expression index for each gene transcript (rows) in each allele (columns) is indicated by color gradations, with green and red representing low and high mRNA levels, respectively. NCI-H520 cells stably transfected with the mammalian expression vector pCR3.1 or recombinant vectors carrying the PthlhThr, PthlhPro, or PthlhSerAspTyr variant or left non-transfected were maintained as described (Manenti et al., 2000; Benelli et al., 2003). Three independent preparations of each transfectant were grown and used for microarray analysis. Two different cDNA microarray experiments were performed. In the first, 8 μg of total RNA [extracted using the RNAfastTM–II kit (M-Medical-Molecular System)] from PthlhThr- and PthlhPro-NCI-H520 cells was reverse-transcribed using oligo(dT)20 and SuperScript reverse transcriptase (Invitrogen, Calrsbad, CA,USA) and hybridized on the same slide. Samples were labeled with the fluorescent dyes Cy-3 and Cy-5 (Amersham Bioscience, Uppsala, Sweden). As a control, a fluor-reverse approach was used in which each sample was labeled twice by inverting the Cy-3 and Cy-5 dyes. In the second experiment, 6 μg of total RNA from PthlhThr-, PthlhPro-, or PthlhSerAspTyr-transfected NCI-H520 cells and vector-transfected NCI-H520 cells was hybridized. cDNA from cells transfected with empty vector was labeled with Cy-3, while PthlhThr , PthlhPro, PthlhSerAspTyr cDNAs were labeled with Cy-5. Samples were hybridized onto two different cDNA arrays (prepared in-house according to (De Cecco et al., 2004) containing 4318 and 4155 unique clones spotted in triplicate and belonging to the human sequence verified IMAGE clone collection (Research Genetics/Invitrogen, Carlsbad, CA,USA). All experiments were performed as described (De Cecco et al., 2004). The Pearson metric and average linkage methods were applied in hierarchical clustering. Clustering and visual analyses were performed using J-Express Pro software (Molmine, Berger, Norway) (Dysvik and Jonassen, 2001).

Microarray results correlate with kRT–PCR analysis

Genes showing the greatest differences in expression in the dye-swap experiment were selected for analysis by gene profiling, of 32 clones (isolated from the original IMAGE collection and re-sequenced) that agreed in both the experiments, three clones contained multiple inserts and were not considered for further analysis, two clones did not grow sufficiently to allow DNA preparation for sequencing and were also discarded, and three clones annotated with different accession numbers corresponded to the same gene (FXYD3). Overall, 25 independent clones were evaluated further.

To determine whether the gene expression patterns observed by cDNA hybridization on microarrays represented differences related to the different biological activities of PthlhThr, PthlhPro, PthlhSerAspTyr alleles, we performed kinetically monitored RT-PCR (kRT–PCR) experiments for the 25 genes showing the greatest difference in expression levels. Overall, the correlation coefficient between the microarray and kRT–PCR values (both expressed as log2) was 0.51 (P<0.0001) (Figure 2), confirming that microarray analysis detected real changes in mRNA expression levels. The kRT–PCR results (log2 values) for the 25 genes (Table 1) confirmed that PthlhThr and PthlhSerAspTyr mRNA levels were highly correlated (r=0.79, P< 0.0001), whereas PthlhPro levels were not correlated with those of the two other alleles. Comparison of PthlhPro with PthlhThr-transfected cells by kRT–PCR, revealed the highest levels of upregulated expression (11- to 50-fold) in the IGFBP3, IGFBP5, FN1 and CCL2 genes, whereas the ANXA1, CEACAM6 and TRIM29 genes showed the most prominent downregulation (13- to 120-fold) (Table 1).

Figure 2
figure2

Correlation between cDNA microarray and kRT–PCR measurements of mRNA expression levels. Data for 25 gene transcripts in stable PthlhThr-, PthlhPro-, or PthlhSerAspTyr- transfectants of NCI-H520 cells are expressed as log2 of mRNA levels with respect to the levels of vector-transfected NCI-H520 cells (reference sample). For kRT–PCR, total RNA (1.5 μg) from PthlhThr-, PthlhPro-, PthlhSerAspTyr- or vector-transfected NCI-H520 cells were reverse-transcribed using oligo(dT)20 and SuperScript reverse transcriptase (Invitrogen, Carlsbad, CA,USA). Gene-specific PCR primers were designed using the Unigene mRNAs or GenBank reference sequences to amplify fragments 100–150 bp in length. kRT–PCR amplification mixtures contained 3 μl template cDNA, 12.5 μl 2 × QuantiTect SYBRGreen PCR Master Mix (Qiagen, Valencia, CA,USA), and 0.3 μ M specific PCR primers. Reactions were run in triplicate using an ABI GeneAmp 5700 sequence detection system (Applied Biosystems, Foster City, CA,USA). The endogenous human GAPDH gene (NM_002046) was used as a control for differences in the amounts of cDNA used. Correlations between expression patterns obtained by cDNA microarray and kRT–PCR analyses were calculated by the non-parametric Spearman rho correlation coefficient.

Table 1 kRT–PCR value of 25 selected genes whose mRNA expression is modulated by Pthlh alleles in NCI-H520 cells.

Gene expression profiles point to functional allelic differences

The functions of the three alleles of the candidate mouse cancer modifier gene Pthlh (Manenti et al., 2000) are known, with PthlhPro allele linked to genetic susceptibility to two-stage skin carcinogenesis and increased cell migration in vitro, whereas PthlhThr and PthlhSerAspTyr are linked with genetic resistance to skin carcinogenesis and with reduced cell migration of transfected cells in vitro (Manenti et al., 2000; Nagase et al., 2001; Benelli et al., 2003). We found that specific mRNA expression profiles are associated with the three Pthlh alleles, with a clustering that is associated with the biological activities of the alleles. Indeed, the PthlhThr and PthlhSerAspTyr alleles showed highly correlated expression profiles that were clearly different from that of the PthlhPro allele (Figure 1, Table 1).

The genes whose expression correlated with the different biological activity of PthlhThr, PthlhPro and PthlhSerAspTyr alleles belong to different gene families. No apparent clustering of biochemical pathways was evident by analysis of gene ontology. Interestingly, four of the 25 selected genes, that is, ANXA1, (annexin 1 (Parente and Solito, 2004)), CCL2 (or MCP-1) (monocyte chemoattractant protein (Maus et al., 2003)), FN1 (extracellular protein fibronectin (Leeb et al., 2004)), and TFF3 (trefoil factor 3 (Hoffmann et al., 2001; Oertel et al., 2001)), are involved in cell adhesion and/or migration. We found overexpression of FN1 and CCL2 genes and downregulation of TFF3 and ANXA1 genes associated with the migration-prone PthlhPro allele (Table 1, Figure 1). Consistent with our results, overexpression of CCL2 or ANXA1 in other systems promotes or inhibits cell migration, respectively (Cambien et al., 2001; Parente and Solito, 2004). Thus, alteration of ANXA1, CCL2, FN1, and TFF3 gene expression may reflect a mechanism responsible for the allele-specific cell adhesion and migration effects of the mouse Pthlh gene (Benelli et al., 2003). ANXA1 (Parente and Solito, 2004), CCL2 (Maus et al., 2003), and TFF3 (Hoffmann et al., 2001) are also involved in the modulation of the inflammatory response and may represent targets of the parathyroid hormone-related protein (PTHrP), which can modulate the inflammatory response in several conditions (Funk, 2001).

Three genes, IGFBP3, encoding the binding and transport protein for insulin-like growth factors (Zhang et al., 2004), CEACAM6, encoding the carcinoembryonic antigen-related cell adhesion molecule 6 (Duxbury et al., 2005) and TFF3 (Yio et al., 2005), are associated with poor prognosis of human cancers. Since high serum PTHrP levels are correlated with poor prognosis in several cancers, including lung cancer (Hiraki et al., 2002; Deans et al., 2005), these genes whose expression is modulated by Pthlh alleles may also represent targets of the PTHrP pathway.

In conclusion, our study demonstrates that the functional alleles of the candidate mouse cancer modifier gene Pthlh can be detected by gene expression profile analysis of transfected human cancer cells. Such an approach may provide a useful tool for the functional characterization of alleles represented by amino acid changes in protein products of other candidate cancer modifier genes.

Accession codes

Accessions

GenBank/EMBL/DDBJ

References

  1. Benelli R, Peissel B, Manenti G, Gariboldi M, Vanzetto C, Albini A et al. (2003). Oncogene 22: 7711–7715.

  2. Cambien B, Pomeranz M, Millet MA, Rossi B, Schmid-Alliana A . (2001). Blood 97: 359–366.

  3. De Cecco L, Marchionni L, Gariboldi M, Reid JF, Lagonigro MS, Caramuta S et al. (2004). Oncogene 23: 8171–8183.

  4. Deans C, Wigmore S, Paterson-Brown S, Black J, Ross J, Fearon KC . (2005). Cancer 103: 1810–1818.

  5. Dragani TA . (2003). Cancer Res 63: 3011–3018.

  6. Duxbury MS, Matros E, Clancy T, Bailey G, Doff M, Zinner MJ et al. (2005). Ann Surg 241: 491–496.

  7. Dysvik B, Jonassen I . (2001). Bioinformatics 17: 369–370.

  8. Funk JL . (2001). Int J Immunopharmaco 1: 1101–1121.

  9. Hiraki A, Ueoka H, Bessho A, Segawa Y, Takigawa N, Kiura K et al. (2002). Cancer 95: 1706–1713.

  10. Hoffmann W, Jagla W, Wiede A . (2001). Histol Histopathol 16: 319–334.

  11. Leeb SN, Vogl D, Grossmann J, Falk W, Scholmerich J, Rogler G et al. (2004). Am J Gastroenterol 99: 335–340.

  12. Manenti G, Peissel B, Gariboldi M, Falvella FS, Zaffaroni D, Allaria B et al. (2000). Oncogene 19: 5324–5328.

  13. Maus UA, Waelsch K, Kuziel WA, Delbeck T, Mack M, Blackwell TS et al. (2003). J Immunol 170: 3273–3278.

  14. Nagase H, Mao JH, de Koning JP, Minami T, Balmain A . (2001). Cancer Res 61: 1305–1308.

  15. Oertel M, Graness A, Thim L, Buhling F, Kalbacher H, Hoffmann W . (2001). Am J RespCell Mol 25: 418–424.

  16. Parente L, Solito E . (2004). Inflamm Res 53: 125–132.

  17. Ross DT, Scherf U, Eisen MB, Perou CM, Rees C, Spellman P et al. (2000). Nat Genet 24: 227–235.

  18. Strewler GJ . (2000). N Engl J Med 342: 177–185.

  19. Yio X, Zhang JY, Babyatsky M, Chen A, Lin J, Fan QX et al. (2005). Clin Exp Metastasis 22: 157–165.

  20. Zhang ZW, Newcomb PV, Moorghen M, Gupta J, Feakins R, Savage P et al. (2004). Cancer Causes Control 15: 211–218.

Download references

Acknowledgements

This work was funded in part by Grants from Associazione and Fondazione Italiana Ricerca Cancro (AIRC and FIRC), Italy and from EC RISC-RAD project.

Author information

Affiliations

Authors

Corresponding author

Correspondence to T A Dragani.

Rights and permissions

Reprints and Permissions

About this article

Cite this article

Gianni-Barrera, R., Gariboldi, M., De Cecco, L. et al. Specific gene expression profiles distinguish among functional allelic variants of the mouse Pthlh gene in transfected human cancer cells. Oncogene 25, 4501–4504 (2006). https://doi.org/10.1038/sj.onc.1209478

Download citation

Keywords

  • parathyroid hormone-like hormone
  • parathyroid hormone-related protein
  • PTHrP
  • NCI-H520
  • cancer modifiers

Search

Quick links