Web-based gene expression analysis—paving the way to decode healthy and diseased ocular tissue

RNA sequencing allows the nucleotide sequences of millions of RNA molecules in a sample to be analyzed [24]. By comparing these sequences with the known reference genome, it is possible to identify and quantify different RNA molecules. The RNA serves as a template to produce proteins or can exert regulatory functions in this process. Thus, transcriptome analysis provides unbiased insights into the functional state of a tissue (Fig. 1).

Comparing samples from diseased and healthy tissues can provide detailed insights into the pathophysiology of a disease and can identify novel diagnostic and prognostic biomarkers. Thanks to the Human Genome Project [13] and technological advances, the costs and time of sequencing have significantly decreased in recent years, and it is likely that this trend will lead to an increased use of the technology in clinical routine [8]. In addition to unfixed samples, specialized sequencing methods can analyze the transcriptional profile of archived formalin-fixed and paraffin-embedded samples, greatly facilitating the analysis of rare diseases [2].

Transcriptome analysis has so far been used in particular in oncology [7, 9, 28]. For example, using transcriptome data from lung tumors and control tissue, diagnostic biomarkers were identified that enabled differentiation between tumor and control tissue with an accuracy of 98% in an independent validation dataset consisting of over 1000 tumors [9]. In addition, squamous cell and adenocarcinoma of the lung were differentiated with a classification accuracy of 95% [9]. Another example of an application of RNA sequencing in clinical routine is the estimation of tumor prognosis based on its transcriptional profile. Uhlen et al. analyzed the transcriptome of over 8000 samples of the most common cancer types and identified prognostically relevant biomarkers for each entity, which allowed the prognosis to be estimated [28]. The prediction of treatment response of a tumor also represents an interesting and clinically useful application of transcriptome analysis. The molecular characterization of various cancer types using RNA and DNA sequencing enabled classification across cancer types into four molecular subtypes with subtype-specific response rates to immune checkpoint inhibitor therapy, thus, providing a foundation for personalized cancer therapy [7]. A recently published statement of the German Medical Association (Bundesärztekammer) assumes that in the next few years, molecular tumor classification will become the standard procedure for most patients starting at the initial diagnosis with the aim to provide a precise and personalized treatment strategy [20].

In ophthalmology, RNA sequencing has been comparatively rarely used so far, especially in clinical practice. Recently, a gene-expression-based diagnostic classification of conjunctival squamous cell carcinoma and papilloma was described [3, 15]. In addition, gene expression of specific cell receptors mediating SARS-CoV‑2 infection has been investigated in ocular surface tissues [14] and intraocular tissues [16] using RNA sequencing. Hyalocytes from the vitreous of patients with epiretinal membranes or macular holes were also recently characterized as an active and immunomodulatory cell population using RNA sequencing [4]. A prognostic gene expression signature for ocular tumors was successfully obtained for choroidal and conjunctival melanoma [21, 32]. Based on the transcriptional profile, uveal melanoma was classified into four prognostically relevant molecular subtypes [21]. This classification achieved a higher predictive power for distant metastases 5 years after brachytherapy than the traditional classification according to the American Joint Committee on Cancer Staging Manual (8th Edition) [17]. Likewise for conjunctival melanoma, 20 prognostically relevant biomarkers have been identified to estimate the risk of local recurrence or distant metastases [32]. For neovascular age-related macular degeneration (nAMD), RNA sequencing of choroidal neovascularization (CNV) membranes identified calprotectin (S100A8/S100A9) and secreted phosphoprotein 1 (SPP1) as novel nAMD-associated factors [22, 23, 31]. Intravitreal injection of an SPP1 inhibitor significantly modulated CNV size in the murine laser CNV model, highlighting the role of the factor as a potential new therapeutic target for nAMD [23].

With technological progress leading to a significant increase in transcriptome analyses, large databases containing a variety of publicly available transcriptome datasets of different diseases have emerged in recent years [6, 10]. One of the largest databases is the Cancer Genome Atlas, which to date contains the sequencing data of over 84,000 tumor samples from 67 different entities [6]. The diversity of these data has made it possible to catalog typical genetic and molecular alterations occurring in different tumors, both to increase knowledge of each tumor entity and to improve understanding of cross-entity mechanisms of carcinogenesis [11]. In addition, the raw sequencing data are publicly available and can be used, for example, as a validation dataset [9]. Reference should also be made at this point to the Human Protein Atlas [27], which catalogs human proteins in cells, tissues, and organs using a combination of various “omics” technologies, such as mass spectrometry and antibody-based proteomics. Despite the numerous possibilities mentioned above, the Cancer Genome Atlas does not yet include ocular tissues, with the exception of uveal melanoma. Although efficient algorithms exist to analyze the available raw sequencing data, they require advanced bioinformatics skills and are also relatively time-consuming. For these reasons, there is a need for databases that contain transcriptional profiles of ocular tissues while allowing intuitive data analysis.

Here, we provide an overview of the available ocular transcriptome databases (Table 1).

Transcriptome databases such as the Cancer Genome Atlas [6] so far contain only very few ocular tissues and provide only the sequencing raw data, which require advanced bioinformatics skills to analyze. Therefore, specialized databases with different application focuses have emerged to provide transcriptional profiles of ocular tissues while enabling intuitive data analysis. Regarding the databases summarized in this review, Spectacle, the UCSC Cell Browser, and the Broad Institute’s Single Cell Portal allow intuitive exploration of single-cell RNA sequencing data from retina, choroid, cornea, iris, trabecular meshwork, and scleral tissues. The Human Eye Transcriptome Atlas provides the largest number of different ocular tissue types, contains the highest number of diseased ocular entities, and achieves a high standard of quality through methodological homogeneity. Ocular transcriptome databases provide comprehensive and intuitive insights into the transcriptional profiles of various ocular tissues and diseases, allowing rapid hypothesis testing in the search for new diagnostic and therapeutic targets.