Gene expression analysis of skin grafts and cultured keratinocytes using synthetic RNA normalization reveals insights into differentiation and growth control

Katayama et al. BMC Genomics (2015) 16:476 DOI 10.1186/s12864-015-1671-5 RESEARCH ARTICLE Open Access Gene expression analysis of skin grafts and cultured keratinocytes using synthetic RNA normalization reveals insights into differentiation and growth control Shintaro Katayama1, Tiina Skoog1, Eeva-Mari Jouhilahti1, H. Annika Siitonen2,3, Kristo Nuutila4, Mari H Tervaniemi2,3, Jyrki Vuola5, Anna Johnsson6, Peter Lönnerberg6, Sten Linnarsson6, Outi Elomaa2,3, Esko Kankuri4* and Juha Kere1,3,7* Abstract Background: Keratinocytes (KCs) are the most frequent cells in the epidermis, and they are often isolated and cultured in vitro to study the molecular biology of the skin. Cultured primary cells and various immortalized cells have been frequently used as skin models but their comparability to intact skin has been questioned. Moreover, when analyzing KC transcriptomes, fluctuation of polyA+ RNA content during the KCs’ lifecycle has been omitted. Results: We performed STRT RNA sequencing on 10 ng samples of total RNA from three different sample types: i) epidermal tissue (split-thickness skin grafts), ii) cultured primary KCs, and iii) HaCaT cell line. We observed significant variation in cellular polyA+ RNA content between tissue and cell culture samples of KCs. The use of synthetic RNAs and SAMstrt in normalization enabled comparison of gene expression levels in the highly heterogenous samples and facilitated discovery of differences between the tissue samples and cultured cells. The transcriptome analysis sensitively revealed genes involved in KC differentiation in skin grafts and cell cycle regulation related genes in cultured KCs and emphasized the fluctuation of transcription factors and non-coding RNAs associated to sample types. Conclusions: The epidermal keratinocytes derived from tissue and cell culture samples showed highly different polyA+ RNA contents. The use of SAMstrt and synthetic RNA based normalization allowed the comparison between tissue and cell culture samples and thus proved to be valuable tools for RNA-seq analysis with translational approach. Transciptomics revealed clear difference both between tissue and cell culture samples and between primary KCs and immortalized HaCaT cells. Background Skin is a multi-layered tissue that is composed of continuously renewing epidermis – with keratinocytes (KCs) as a predominant cell type – and underlying dermis populated mostly by fibroblasts. The life span of epidermal keratinocytes is controlled by two alternative pathways: differentiation as their normal function or activation as an altered function in wound healing or skin diseases [1]. Epidermal KCs residing in the basal layer of the epidermis * Correspondence: ; 4 Department of Pharmacology, Faculty of Medicine, University of Helsinki, Helsinki, Finland 1 Department of Biosciences and Nutrition, Karolinska Institute and Center for Innovative Medicine, Huddinge, Sweden Full list of author information is available at the end of the article differentiate through multiple layers and finally shed as cornified dead cells from the skin surface [2, 3]. The relatively noninvasive sampling together with the methods that allow culturing of pure KCs have greatly facilitated research on skin and KCs. In cell culture, KCs are uncoupled from their tissue environment that naturally provides a network of homeostatic control signals; they are induced to either retain an active proliferative state or to differentiate. However, the prolonged KC culturing leads to the induction of cellular senescence [4] and therefore not only primary KCs but also immortalized KC lines, such as HaCaT (a spontaneously immortalized cell line) [5], have been widely studied to understand various normal and altered functions of the skin. HaCaT cells represent a highly popular model system since despite some © 2015 Katayama et al. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly credited. The Creative Commons Public Domain Dedication waiver (http:// creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated. Katayama et al. BMC Genomics (2015) 16:476 UV-inducible mutations in TP53 alleles [6, 7] they are non-tumorigenic and have retained their capacity to differentiate [5, 7, 8]. The comparability of each of the models to intact skin has often been questioned. In the current study, we address the question of how representative models cultured KCs and HaCaTs are for studying human epidermis. Genome-wide expression profiling is an useful approach to screen key genes with respect to different cellular statuses and to further model the regulatory networks [9]. Microarray technology provides a traditional profiling method to measure thousands of known genes simultaneously, but it has recently been replaced by RNA-seq technology that has proven to give more detailed insights into transcriptome. Both technologies have been previously applied to study the gene expression in skin [10–12]. However, an important fact has been largely omitted: when KCs undergo their complex lifecycle, they change not only cell size and cell cycle kinetics, but also the actively transcribed RNA content, with the largest RNA content in fresh, actively growing cultured KCs [13]. In microarray, RNAseq and even qRT-PCR, the same amount of total RNA is loaded for each sample, although yields of the polyA+ RNAs purified from total RNAs may differ. Moreover, normalization for the differential expression test expects equivalent expression levels for several co-expressed genes [14]. Therefore, the genome-wide expression profiling in the previous studies might have underestimated the complexity of the KC transcriptome during their lifecycle. In this study, we revisit the skin and KC transcriptome with respect to fluctuation of polyA+ RNA content by the keratinocyte statuses; differentiated, activated, senescent and immortalized. Four types of human keratinocyte samples represented these cell statuses: epidermal tissue (split-thickness skin grafts; SGs), cultured primary KCs in early and late passages, and HaCaT cell line. To reduce the sample size and sequencing costs and to control the fluctuation of mRNA concentration, we applied single-cell tagged reverse transcription (STRT) sequencing method for expression profiling using 10 ng of total RNA per sample which is ten times less than required for a conventional RNA-seq method [15]. For accurate expression profiling and statistical tests, we employed STRT RNA-seq with synthetic polyA+ spike-in RNA [16], and SAMstrt statistical package with spike-in based normalization [14]. We first evaluated the improvements of our approach on the genome-wide expression profiling and confirmed the accuracy of the improv (...truncated)