Deconvolution of Bulk Transcriptomics Data Suggests Changes in Conjunctival Immune Cell Populations in Trachoma

doi:10.21203/rs.3.rs-1949261/v1

Download PDF

Short Report

Deconvolution of Bulk Transcriptomics Data Suggests Changes in Conjunctival Immune Cell Populations in Trachoma

https://doi.org/10.21203/rs.3.rs-1949261/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted

You are reading this latest preprint version

Objective: Trachoma, a neglected tropical disease caused by ocular infection with Chlamydia trachomatis, progresses from repeated infections and conjunctivitis to scarring, trichiasis and blindness. It is yet unknown which immune cells are responsible for mediating pathogenesis. Our objective was to deconvolute bulk transcriptomics data in order to estimate how conjunctival immune cell populations change in active and scarring trachoma. Results: Six deconvolution methods (quanTIseq, TIMER, MCP-counter, xCell, EPIC, and CIBERSORT) were used to estimate the relative abundance of cell populations in five bulk transcriptomic datasets. Cell types from each deconvolution method were mapped to one another using the Immunedeconv R package. In active disease, B cells, neutrophils, natural killer cells, CD4+ T cells and CD8+ T cells were raised (direction of change the same for at least 75% of methods, p<0.05 for at least two methods by logistic regression). B cells and neutrophils were raised in trachomatous scarring and inflammation, while the relative abundance of CD8+ T cells was decreased. In those with scarring and no inflammation, monocyte/macrophages, neutrophils and CD8+ T cells were decreased compared with controls.

Trachoma

Deconvolution

Chlamydia Trachomatis

Transcriptomics

Conjunctiva

B cells

neutrophils

NK cells

T cells

scarring

Onset of neglected tropical disease trachoma typically occurs in childhood, beginning with repeated ocular infections with Chlamydia trachomatis and conjunctivitis (active trachoma) that may progress to conjunctival scarring, in-turning of the eyelashes (trichiasis) and corneal damage later in life (1). While conjunctival scarring is thought to be mediated by inflammatory immune responses rather than C. trachomatis itself, it is unknown which immune cells are responsible for pathogenesis (2).

Immunostaining of biopsy specimens have previously given clues as to changes in leukocyte populations during disease. For active trachoma, the number of immunophenotyping studies is very limited. These studies described conjunctival lymphoid follicle organisation with infiltration of polymorphonuclear leucocytes, macrophages and lymphocytes (3, 4). One also observed a small number of mast cells and eosinophils (4). In the later stages of disease, two further studies found marked increases in both CD4 + and CD8 + lymphocytes, as well as moderate increases in B cells (5, 6). While the CD8 + lymphocytes have previously been considered to be cytotoxic T cells, a more recent paper found that in trachomatous scarring CD8 did not co-localise with CD3, suggesting these CD8 + lymphocytes may represent innate natural killer (NK) cells (7).

While these studies point towards humoral, cell-mediated and innate immune responses playing an interlinked role in trachoma pathogenesis, immunostaining biopsy samples does have a number of limitations. Due to the invasiveness of biopsy collection, surgery in active trachoma is no longer considered acceptable and obtaining appropriate healthy tissue for controls is not warranted. Samples from older patients with later stages of disease can be obtained when undergoing corrective surgery but obtaining sufficient and appropriate healthy control tissue is limiting. Furthermore, immunohistochemistry is usually limited to single antibody markers, and subsequent analysis is generally semi-quantitative.

Therefore we overcome some of these limitations by performing in silico deconvolution of bulk transcriptomics data. As transcriptional profiling was performed on conjunctival swabs rather than invasive biopsies, samples were collected from a larger number of participants (total n = 193) in several trachoma endemic locations, with inclusion of healthy controls. Furthermore, transcriptional profiling allows simultaneous quantitative measurement of thousands of genes, rather than a handful of cell-surface markers.

Our analysis pipeline is summarised in Fig. 1. Gene expression matrices from five datasets, three characterising transcriptional changes in active disease (GSE20436, GSE20430 (8) and GSE106961 (9)), and two characterising changes in trachomatous scarring (GSE24383 and GSE23705 (10)) were used. Raw gene expression matrices were quantile normalised and collapsed to genes using the collapseRows() function in the WGCNA R package, taking the probe with the maximum mean as being representative (11). Expression matrices were then deconvoluted by quanTIseq, TIMER, MCP-counter, xCell, EPIC and CIBERSORT using the Immunedeconv R package (12). Cell types from each deconvolution method were mapped to one another by the Immunedeconv R package. A logistic regression model was then used to assess, for each method, whether each cell type was significantly changed in active disease or trachomatous scarring. Source dataset was included in the model as a covariate. A genuine change in a cell population was considered likely if the direction of change was the same for at least 75% of methods, and if this change was significant (p < 0.05) for at least two methods (Table 1). A logistic regression model was also used to assess changes in individual source datasets (Fig. 2).

Table 1

Cell types considered to be changed (direction of change the same for at least 75% methods, change significant (p < 0.05) for at least two methods) in active trachoma, trachomatous scarring with inflammation, and trachomatous scarring without inflammation
Active trachoma vs no active trachoma
Cell	Direction	p < 0.05
B cell	6/6 ↑	5/6
Neutrophil	6/6 ↑	4/6
NK cell	4/5 ↑	2/4
T cell CD4+	4/5 ↑	3/4
T cell CD8+	5/6 ↑	2/5
Scarring and inflammation vs no scarring
Cell	Direction	p < 0.05
B cell	6/6 ↑	5/6
Neutrophil	6/6 ↑	6/6
T cell CD8+	5/6 ↓	3/5
Scarring vs no scarring
Cell	Direction	p < 0.05
Monocyte/macrophage	6/6 ↓	5/6
Neutrophil	6/6 ↓	6/6
T cell CD8+	6/6 ↓	3/5

In active disease, B cells, neutrophils, NK cells, CD4 + T cells and CD8 + T cells were all considered raised, consistent with previous literature (Table 1). The two datasets from the Gambia gave more similar results to one another, with no increase in CD8 + T cells, NK cells, neutrophils or B cells seen in the Solomon Islands dataset (Fig. 2a). This is in line with the hypothesis that active trachoma in the pacific islands has a distinct pathogenesis to that in Sub-Saharan Africa (9). The proportion of mast cells was only estimated by a small number of deconvolution methods, and did not reach statistical significance. However, mast cells were increased in both Gambian datasets.

In those with scarring and inflammation, B cells and neutrophils appeared to be increased relative to those without scarring (Table 1). This change was consistent between the Ethiopian and Tanzanian datasets (Fig. 2b). Surprisingly, CD8 + T cells were decreased. In those with scarring and no inflammation, monocyte/macrophages, neutrophils and CD8 + T cells were decreased compared with controls (Table 1). While the changes in monocyte/macrophages and CD8 + T cells were consistent between datasets, the decrease in neutrophils only occurred in the Ethiopian dataset (Fig. 2c).

One limitation with these data sets is that the deconvolution methods applied estimate the relative rather than absolute numbers of immune cells. If the number of one cell type greatly increases, another cell type may appear by comparison to be decreased, even if the absolute number of the second cell type remained the same or increased to a lesser degree. This likely accounts for the discrepancy between the literature and our findings: the increase in squamous epithelial cells seen in trachomatous scarring (4, 13) could drive the observed decrease in monocyte/macrophages, neutrophils and CD8 + T cells.

A second limitation is that deconvolution methods heavily rely on reference cell profiles, but the gene expression of certain cell types could markedly deviate from references in response to environmental changes. Furthermore, it can be difficult to distinguish closely related cells with similar gene expression profiles, and unlike flow cytometry or single-cell transcriptomics it is not possible to analyse co-expression.

Despite these limitations, we provide evidence that natural killer cells may play a role in active trachoma, and that neutrophils and B cells are raised both in the inflammation that occurs during active disease, as well as inflammation in scarring trachoma. These data will be useful a comparator for conjunctival immunophenotyping in the experimental models of trachomatous disease and infection required to advance vaccine development against Chlamydia trachomatis.

C. trachomatis

Chlamydia trachomatis

NK Cells

Natural Killer Cells

Ethics approval and consent to participate: This research note analyses publicly available data from already published studies.

Consent for publication: Not applicable

Availability of data and materials: This research note analyses data already publicly available from the NCBI Gene Expression Omnibus (GSE20436, GSE20430, GSE106961, GSE24383 and GSE23705).

Competing interests: The authors declare that they have no competing interests

Funding: Both authors would like to acknowledge funding support by the European Commission through the Horizon 2020 TRACVAC project.

Authors' contributions: AB: Conceptualisation, analysis, data visualisation, writing (original draft). MH: Conceptualisation, writing (review and editing).

Acknowledgements: Not applicable

Anthony W. Solomon, Burton MJ, Gower EW, Harding-Esch EM, Oldenburg CE, Taylor HR, et al. Trachoma. Nat Rev Dis Prim. 2022;8(32).
Hu VH, Holland MJ, Burton MJ. Trachoma: Protective and Pathogenic Ocular Immune Responses to Chlamydia trachomatis. PLoS Negl Trop Dis [Internet]. 2013;7(2). Available from: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3573101/pdf/pntd.0002020.pdf
El-Asrar AMA, Van den Oord JJ, Geboes K, Missotten L, Emarah MH, Desmet V. Immunopathology of trachomatous conjunctivitis. Br J Ophthalmol [Internet]. 1989;73(4):276–82. Available from: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1041713/pdf/brjopthal00596-0038.pdf
Guzey M, Ozardali I, Basar E, Aslan G, Satici A, Karadede S. A survey of trachoma: The histopathology and the mechanism of progressive cicatrization of eyelid tissues. Ophthalmologica [Internet]. 2000;214(4):277–84. Available from: https://www.karger.com/Article/PDF/27504?casa_token=RUZ7HJfyRX4AAAAA:4PaSROvtyWe6dvp5w7b64DDd5ArElwY-n-FhJh5n9oJ60pMCszAyoIkQYrA4_D7BywtUCmzRuvxX
Reacher MH, Pe’er J, Rapoza PA, Whittum-Hudson JA, Taylor HR. T Cells and Trachoma: Their Role in Cicatricial Disease. Ophthalmology. 1991;98(3):334–41.
Burd EM, Tabbara KF, Nasr AM, Taylor PB. Conjunctival lymphocyte subsets in trachoma. Int Ophthalmol. 1988;12(1):53–7.
Hu VH, Luthert PJ, Derrick T, Pullin J, Weiss HA, Massae P, et al. Immunohistochemical Analysis of Scarring Trachoma Indicates Infiltration by Natural Killer and Undefined CD45 Negative Cells. PLoS Negl Trop Dis. 2016;10(5):1–15.
Natividad A, Freeman TC, Jeffries D, Burton MJ, Mabey DCW, Bailey RL, et al. Human conjunctival transcriptome analysis reveals the prominence of innate defense in Chlamydia trachomatis infection. Infect Immun. 2010;78(11):4895–911.
Vasileva H, Butcher R, Pickering H, Sokana O, Jack K, Solomon AW, et al. Conjunctival transcriptome profiling of Solomon Islanders with active trachoma in the absence of Chlamydia trachomatis infection. Parasites and Vectors. 2018;11(1):1–8.
Hu VH, Weiss HA, Ramadhani AM, Tolbert SB, Massae P, Mabey DCW, et al. Innate immune responses and modified extracellular matrix regulation characterize bacterial infection and cellular/connective tissue changes in scarring trachoma. Infect Immun. 2012;80(1):121–30.
Langfelder P, Horvath S. WGCNA: An R package for weighted correlation network analysis. BMC Bioinformatics [Internet]. 2008;9. Available from: https://bmcbioinformatics.biomedcentral.com/track/pdf/10.1186/1471-2105-9-559.pdf
Sturm G, Finotello F, List M. Immunedeconv: An R package for unified access to computational methods for estimating immune cell fractions from bulk RNA-sequencing data. Methods Mol Biol. 2020;2120:223–32.
Burton MJ, Rajak SN, Bauer J, Weiss HA, Tolbert SB, Shoo A, et al. Conjunctival transcriptome in scarring trachoma. Infect Immun [Internet]. 2011;79(1):499–511. Available from: https://iai.asm.org/content/iai/79/1/499.full.pdf

No competing interests reported.

Download PDF

Version 1

posted

You are reading this latest preprint version

Deconvolution of Bulk Transcriptomics Data Suggests Changes in Conjunctival Immune Cell Populations in Trachoma

Status:

Version 1

Abstract

Figures

Introduction

Results

Limitations

Abbreviations

Declarations

References

Additional Declarations

Status:

Version 1