
	Version PDF

Study of housekeeping gene expression in human keratinocytes using OLISA™, a long-oligonucleotide microarray and q RT-PCR

European Journal of Dermatology. Volume 16, Number 2, 136-40, March-April 2006, Investigative report

Summary

Author(s) : M Bonnet-Duquennoy, H Abaibou, M Tailhardat, K Lazou, S Bosset, B Le varlet, P Cleuziat, R Kurfürst , Laboratoires LVMH Recherche, 185 avenue de Verdun, 45804 St Jean de Braye Cedex, France, bioMérieux, 5 rue des Berges, 38000 Grenoble, France, INSERM U503 and IFR 128, Unite d’Immunologie et Allergologie Clinique, Centre Hospitalier Lyon-Sud, F-69495 Pierre-Benite Cedex, France, Authors with equal contribution.

Summary : In recent years, applications of microarray platforms have been extended to different areas of research including cosmetic and pharmaceutical. Although microarray technology is still improving its sensitivity and flexibility, researchers often turn toward quantitative RT-PCR for data validation. Assessment of messenger RNA quantity by these methods is based on comparison with internal standard genes, mainly housekeeping genes, so called because their synthesis occurs normally at a constant level. However, numerous studies showed that expression of these genes could vary in given situations. Here, we report results on four housekeeping genes (GAPDH, β-2 microglobulin, S40 and S26 ribosomal sub-units) with constant expression levels established on OLISA™ microarray using different keratinocyte cultures. Moreover, qRT-PCR validation demonstrates that S26 ribosomal is a good housekeeping gene on keratinocytes and skin studies. Our data indicate that S26 gene can be routinely used to standardize results to investigate differentially expressed genes in a healthy human skin.

Keywords : gene, keratinocytes, microarray, qRT-PCR, skin

Pictures

ARTICLE

Auteur(s) : M Bonnet-Duquennoy^1,4, H Abaibou^2,4, M Tailhardat^1,2,3, K Lazou¹, S Bosset³, B Le varlet¹, P Cleuziat², R Kurfürst¹

¹Laboratoires LVMH Recherche, 185 avenue de Verdun, 45804 St Jean de Braye Cedex, France
²bioMérieux, 5 rue des Berges, 38000 Grenoble, France
³INSERM U503 and IFR 128, Unite d’Immunologie et Allergologie Clinique, Centre Hospitalier Lyon-Sud, F-69495 Pierre-Benite Cedex, France
⁴Authors with equal contribution

accepté le 29 Septembre 2005

Different studies have demonstrated the importance of transcriptomic analysis (e.g., using DNA microarrays) of different physiological pathways and various dermatological diseases [1-5]. However, enthusiasm is sometimes moderated by the recent demonstration of technological caveats in this type of analysis [6]. Interestingly, most of the microarray studies performed to date using 5,000 to 20,000 genes identified less than 100 classifying genes. Therefore, low cost, low density transcriptomic assays that are both quick to perform and sensitive to clinical compatibility and high diagnostic quality can offer significant advantages over high density microarrays. Accordingly, we chose an OLISA™ (OLISA™: OLIgo Sorbent Array) developed by bioMérieux and which can monitor simultaneously the expression of up to 15 different genes per well using a standard lab 96-well microtiter plate. Different studies show that OLISA technology is a sensitive, rapid and quantitative microarray method [7, 8]. The assessment of messenger RNA expression by qRT-PCR and microarray methods are based on comparison with internal standard genes, so-called housekeeping genes because their expression occurs in any tissue and cell at relatively constant levels. The accurate quantification of a true reference gene allows the normalisation of differences in the amount of amplifiable RNA or cDNA in individual samples obtained from (i) different amounts of starting material, (ii) varying quality of starting material and (iii) differences in RNA preparation and cDNA synthesis, since the internal reference gene is exposed to the same preparation steps as the gene of interest. However, numerous studies showed that even these genes see their expression vary in given situations [9, 10]. Consequently, we chose to identify first in our cellular or tissue models at least one housekeeping gene as a precondition to initiate such microarray studies in our laboratory to understand the physiological characterization of healthy skin. Thus, four likely housekeeping genes (GAPDH, S40 ribosomal sub-unit, S26 ribosomal sub-unit and β-2 microglobulin) previously described [10, 11], were selected to be assessed at the expression level on normal human keratinocytes with or without treatments. Because of their well known transcriptional effects on keratinocytes, retinoic acid and hydrogen peroxide were chosen as reference treatments to verify no variation of housekeeping gene expression [12-15]. Finally, the results on S26 ribosomal sub-unit were further confirmed with the quantitative reference technology (qRT-PCR) on normal human keratinocytes and on healthy skin.

Materials and methods

Cell culture

Normal human keratinocytes (NHK) were obtained from plastic surgery of healthy skin. NHK were cultivated in a keratinocyte serum-free medium (K-SFM Gibco, Paisley, GB) supplemented with 25 μg/ml BPE and 2.5 ng/ml EGF (K-SFMc) in a humidified atmosphere (5% CO₂, 37 °C).

Cells were treated with hydrogen peroxide 25 μM for 1 hour at 37 °C and replaced for 24 hours in culture media or with retinoic acid (RA) at 1μM dissolved in culture medium for 24 h.

Skin specimens and tissue preparation

Facial skin was obtained from 10 Caucasian patients undergoing a surgical facelift. Two types of skin were recovered for the study: i) sun-exposed pre-auricular skin which was centred by the preauricular wrinkle. Wrinkles representing the clinical signs of photo-aged skin were marked with Indian ink before sampling, as previously described [16]; ii) homolateral post-auricular skin at a distance from the post-auricular fold. This piece of skin is covered by hair and corresponds to sun-protected skin.

Epidermis and dermis were isolated after incubation in dispase buffer (2 MU/ml in complete medium, Roche Diagnostics, Meylan, France) overnight at 4 °C. After treatment, epidermis was snap-frozen in liquid nitrogen and stored at – 80 °C.

Nucleic acid extraction and purification

Total RNA extraction was performed with 1 ml of the commercially available RNA extraction mixture, RNAplus (Q-BIOgene, Illkirch, France) in accordance with the manufacturer’s instructions. The concentration of RNA and its purity were estimated by optical density at 260 nm and 260/280 nm ratio and by electrophoresis through agarose gels (Sigma Chemical Co, St Louis, USA). The RNA stock solutions were stored at – 80 °C. Total extracted RNA was treated with RNase-free RQ1 DNase (Promega Corp, Madison, USA) at 37 °C for 30 minutes.

Design, synthesis and spotting of probes

All 60-mer probes were optimised based on bioMérieux (Grenoble, France) specific criteria, chemically synthesized for highly efficient coupling oligonucleotides (purity > 95%) then spotted onto a 96-well microtiter plate (OLISA™) at bioMérieux in a circle format as previously described [8].

Synthesis of labelled cDNA

10 μg of total RNA were retro-transcribed with Superscript II Reverse-Transcription kit, (Invitrogen Corporation; Carlsbad, USA) in the presence of dNTP-biotin. cDNA were purified on microcon filters following the manufacturer’s recommendations (Millipore Corp., Canton, USA).

Principle of OLISA^TM technology

Biotin-labelled targets were denatured with 0.2N NaOH for 5 minutes, then hybridised to oligonucleotide probes in the presence of a hybridisation solution (0.1 M Na₂HPO₄/KH₂PO₄, 0.5M NaCl, 0.6% Tween 20, 2% PEG 4000, Salmon DNA sperm, pH 7) for one hour at 37 °C. Wells were washed at room temperature (RT) four times with 200 μl of the washing buffer (0.1 M Na₂HPO₄/KH₂PO₄, Tween20, EDTA) in a fully automated washing system. The detection solution was added to each well and incubated at 25 °C for 15 minutes. The wells were washed a second time as described above. 50 μl of the staining solution BM-purple-Alkaline Phosphatase (Roche, Basel, Switzerland) were added to each well following incubation at 25 °C for 15 to 60 minutes. If necessary, 4 μl of EDTA (0.5 M) can be added to each well to stop the reaction process. Hybridised arrays were then read using Apimager^® (bioMérieux, Grenoble, France), a densitometry reading system and the density ratio measurement was determined by taking into account the normalized diameter and background noise. Spots were considered for analysis if their response was significantly higher than the background noise plus three standard deviations. Data analysis was performed using proprietary software (ApiAnalyser™). For in vivo samples, a uniform scaling factor was applied to each well assuming that the signal, for each housekeeping gene under special conditions, should be unchanged for normalizing the data. Therefore, the signals from genes of interest were normalized per well using the signal corresponding to the chosen housekeeping gene then the inter-wells ratios were calculated independently for each gene.

Reverse transcription and PCR assays

RNA samples (1 μg) were subjected to reverse transcription with MMLV reverse transcriptase (Invitrogen, Netherlands) and non specific random hexamer primers (Amersham Pharmacia Biotech, USA) at 42 °C for 60 minutes.

To amplify and quantify target cDNA of S26 ribosomal sub-unit, reverse transcribed samples were subjected to PCR amplification on the LightCycler™ (Roche Diagnostics, Meylan, France) in 20 μl containing a final concentration of 0.5 μM of each specific primer (Upstream: 5’CGCAGCAGTCAGGGACAT3’ Downstream: 5’AGCACCCGCAGGTCTAAATC3’), MgCl2, and 2 μl of ready-to-use LightCycler™ DNA Master SYBR^® Green I. The reaction conditions were initial denaturation at 95 °C for 8 minutes followed by 40 cycles of denaturation at 95 °C for 15 seconds, annealing at 60 °C for 10 seconds, and extension at 72 °C for 10 seconds. PCR product specificity was checked with melting curve procedure. PCR sensitivity, PCR efficiency and target cDNA quantification were determined with standard curve performed with a 10-fold serial dilution of plasmidic constructions obtained with the PCR cloning system (Qiagen, Courtaboeuf, France) in accordance with the manufacturer’s instructions. Quantitative analysis was performed using the LightCycler™ software using real-time fluorogenic detection. The log-linear portion of the standard’s amplification curve was identified and the crossing point was the intersection of the closest corresponding line through the log-linear region and the noise band. The standard curve is the plot of crossing point versus the log of plasmidic copy number. The standard curve slope represents the overall reaction efficiency.

Results

Gene expression levels of GAPDH, S40, S26 and β-2microglobulin genes with OLISA™

To make sure that the expression levels of these four potential housekeeping genes behave in a similar way over time in a non treated cell culture, hybridisation signals of their messenger RNA were measured in a kinetic mode for up to two hours. Thus, Total RNA was extracted from normal human keratinocyte cells, retro-transcribed and then hybridised to OLISA™ biochips without any amplification process as previously reported [10]. Positive hybridisations were observed through the enzymatic reaction which generates a precipitated substrate. Thus, the volume of this substrate is proportional to the degrees of hybridisation between a probe on the spot and its corresponding target. In order to compute this volume, the Apimager^® reader system extracts, from the images taken in a grey value, the optical density that is proportional to the size and the signal of each spot. Therefore, the Apimager^® reader performs semi-quantitative and quantitative measurements based on the principle of the optical densitometry measurement (Data not shown). Thus, the resulting grey levels under non-saturation conditions are considered for further analysis only if their values are above the background noise value plus three sigma (sigma stands for standard deviation). Moreover, these measured grey levels were compared with the others in a scatter plot approach prior to any kind of cell treatment. As shown in ( figure 1 ), a high positive correlation was established when the GAPDH gene is compared to S26 ribosomal protein (Panel A), S40 ribosomal protein (Panel B) and β-2 microglobulin (Panel C). Furthermore, alignments within these scatter plots of dots corresponding to grey level signals lead to stretching lines with significant correlation factors ranking from 79 to 94% (table 1( Table 1 )). When cells underwent treatments by either 25 μM hydrogen peroxide (H₂O₂) or 10-6 M of Retinoic Acid (RA), no significant change in the expression levels of these four genes was observed as shown in ( figure 2 ). Variations observed, especially on S26 and GAPDH expression after RA treatment, were not statistically significant. Thus, these variations observed on the expression levels of these four genes are due to possible technical fluctuations that probably occur at the levels of the transcriptase efficacy and the sample loading. Taken together, these results clearly demonstrate that these four genes were constantly expressed and not affected by any of the treatments tested, suggesting their use as housekeeping genes in our topic studies. Recently, several controversies were reported on data obtained from the same biological samples on different commercial microarrays [6, 17]. Therefore, the validation of results from biochip platforms more often passes by appeal confirmation from the q-RT-PCR platform. ( Figure 2 ) shows some variations on the S26 ribosomal protein gene after RA treatment. Thus, we chose to validate our OLISA results for this gene using q RT-PCR method.
Table 1 Correlation factors obtained by alignments of dots from grey levels of the 4 genes

	S26 ribosomal protein	GAPDH	S40 ribosomal protein
S26 ribosomal protein	–	–	–
GAPDH	+ 0.89 ^s	–	–
40S ribosomal protein	+ 0.79 ^s	+ 0.94 ^s	–
β2 microglobulin	+ 0.83 ^s	+ 0.88 ^s	+ 0.90 ^s

Messenger RNA levels of S26 gene with q-RT-PCR

First of all, quantitative real time PCR was used to measure the basal level of the S26 transcript in non treated NHK cells versus RA treated NHK cells. Thus, this level in 100 ng of total RNA was 4,016,000 ± 750,206 copies and 4,910,750 ± 1,221,660 copies, respectively for non-treated and RA treated NHK cells. Hence, when both results were compared, no statistical difference was observed (p = 0.3). This result suggests that the RA treatment does not affect the expression of S26 ribosomal protein coding gene in NHK cells. Thus, this result confirmed that the S26 coding gene could be used as a good housekeeping gene for our studies on keratinocytes.

To verify this result on skin, the S26 transcript levels were measured on both biopsies surgically taken from sun-exposed and sun-protected skins. Total RNA was extracted from frozen epidermis obtained from the facial skin of 10 Caucasian patients who underwent surgical facelift. As we can see in ( figure 3 ), both biopsies exhibit a similar pattern for the S26 transcript. Taken together, these results strongly support our hypothesis which stipulates that the gene coding the 26S ribosomal protein is a good housekeeping candidate gene to be used in such a study to assess the gene expression profiling in human skin.

Discussion

Transcriptomic analysis using micro-array technologies provides a powerful tool for investigating differential gene expression in pathological and normal skin, for testing responses to cutaneous stressors, for understanding the physiological process in healthy skin, such as photo-ageing, pigmentation disorders and for identifying new active compounds [1-5].

One major key in gene expression studies is the choice of the right housekeeping gene that allows standardized results. However, the absence of inter-platform reproducibility of the technology was reported by several papers, supporting the need for validation of the data mainly using the reference technology q-RT-PCR [6, 17]. Therefore, the absence of an adequate housekeeping gene leads more often to discrepancies between different pangenomic commercial tools. Moreover, the cost of the technology and difficulties of managing a large amount of data probably explain the second problem which is the absence of clinical validation of the genes set emerging from retrospective studies for prospective dermatological and cosmetic studies. For this, we need a sensitive, reliable and cost competitive microarray platform for reaching all the gene expression requirements in a very short time. For these reasons, we chose OLISA™ as a simple versatile oligonucleotide microarray (OLISA™: OLIgo Sorbent Array) which is based on a robust colorimetric method with a single temperature and easily adaptable on major standard laboratory equipments.

Gene transcription studies using quantitative real time PCR or microarray platforms should start with the right selection of at least one appropriate housekeeping gene that is useful for individual experimental settings and therefore to be used as a reference gene for further analysis such as standardization process. Few studies were performed on the validation of housekeeping genes on skin. We agree with other authors that more than one housekeeping gene should be used in order to obtain the most reliable results in gene transcription analysis. We demonstrated herein that GAPDH, 40S ribosomal sub-unit, S26 ribosomal sub-unit and β-2 microglobulin genes are truly housekeeping genes in such studies, using either normal human keratinocytes or skin biopsies. Moreover, our results obtained on OLISA™ matched those obtained on the qRT-PCR platform for the S26 ribosomal protein coding gene. Our studies of housekeeping genes validation using OLISA™ seemed to be encouraging for more exploration to come in gene expression profiling and therefore to contribute to a significant breakthrough in the cosmetic and dermatological fields which are working to understand and quantify all physiological mechanisms that occur within a healthy skin.

Acknowledgements

We warmly thank Dr S. Schnebert, Dr. P. Courtellemont and all people involved in this project for their constant encouragement, support and fruitful discussions, A. Bernois and D. Pellé de Queral for statistical analysis and A. Challon for technical assistance.

References

1 Hamatani T, Daikoku T, Matsumoto H, Carter MG, Ko MS, Dey SK. Global gene expression analysis identifies molecular pathways distinguishing blastocyte dormancy and activation. Proc Natl Acad Sci USA 2004; 13: 10326-31.

2 Kunz M, Ibrahim SM, Koczan D, Scheid S, Thiesen HJ, Gross G. DNA microarray technology and its applications in dermatology. Exp Dermatol 2004; 13: 593-606.

3 Sellheyer K, Belbin TJ. DNA microarrays: from structural genomics to functional genomics. The applications of gene chips in dermatology and dermato-pathology. J Am Acad Dermatol 2004; 51: 681-92.

4 Urschitz J, Iobst S, Urban Z, Granda C, Souza KA, Lupp C, Schilling K, Scott I, Csiszar K, Boyd CD. A serial analysis of gene expression in sun-damaged human skin. J Invest Dermatol 2002; 119: 3-13.

5 Cole J, Tsou R, Wallace K, Gibran N, Isik F. Early gene expression profile of human skin to injury using high-density cDNA microarrays. Wound Repair Regen 2001; 9: 360-70.

6 Tan PK, Downey TJ, Spitznagel Jr. EL, Xu P, Fu D, Dimitrov DS, Lempicki RA, Raaka BM, Cam MC. Evaluation of gene expression measurements from commercial microarray platforms. Nucleic Acids Res 2003; 19: 5676-84.

7 Cuzin M. DNA chips: a new tool for genetic analysis and diagnostics. Transfus Clin Biol 2001; 8: 291-6.

8 Perrin A, Duracher D, Perret M, Cleuziat P, Mandrand B. A combined oligonucleotide and protein microarray for the co-detection of nucleic acids and antibodies associated with HIV, HBV and HCV viral infections. Anal Biochem 2003; 322: 148-55.

9 Thellin O, Zorzi W, Lakaye B, De Borman B, Coumans B, Hennen G, Grisar T, Igout A, Heinen E. Housekeeping genes as internal standards: use and limits. J Biotechnol 1999; 75: 291-5.

10 Schmittgen TD, Zakrajsek BA. Effect of experimental treatment on housekeeping gene expression: validation by real-time, quantitative RT-PCR. J Biochem Biophys Methods 2000; 46: 69-81.

11 Vincent S, Marty L, Fort P. S26 ribosomal protein RNA: an invariant control for gene regulation experiments in eucaryotic cells and tissues. Nucleic Acids Res 1993; 21: 1498.

12 Mrass P, Rendl M, Mildner M, Gruber F, Lengauer B, Ballaun C, Eckhart L, Tschachler E. Retinoic acid increases the expression of p53 and proapoptotic caspases and sensitizes keratinocytes to apoptosis: a possible explanation for tumor preventive action of retinoids. Cancer Res 2004; 64: 6542-8.

13 Bernard FX, Pedretti N, Rosdy M, Deguercy A. Comparison of gene expression profiles in human keratinocyte mono-layer cultures, reconstituted epidermis and normal human skin; transcriptional effects of retinoid treatments in reconstituted human epidermis. Exp Dermatol 2002; 11: 59-74.

14 Kanda N, Watanabe S. 17beta-estradiol inhibits oxidative stress-induced apoptosis in keratinocytes by promoting Bcl-2 expression. J Invest Dermatol 2003; 121: 1500-9.

15 Rojkind M, Dominguez-Rosales JA, Nieto N, Greenwel P. Role of hydrogen peroxide and oxidative stress in healing responses. Cell Mol Life Sci 2002; 59: 1872-91.

16 Bosset S, Barre P, Bonnet-Duquennoy M, Kurfurst R, Bonte F, Schnebert S, Levarlet B, Nicolas JF. Wrinkles – biological and immunological features. Eur J Dermatol 2002; 12: 247-52.

17 Marshall E. Getting the noise out of gene arrays. Science 2004; 306: 630-1.