Upper Airway Epithelial Tissue Transcriptome Analysis Reveals Immune Signatures Associated with COVID-19 Severity in Ghanaians

The immunological signatures driving the severity of coronavirus disease 19 (COVID-19) in Ghanaians remain poorly understood. We performed bulk transcriptome sequencing of nasopharyngeal samples from severe acute respiratory syndrome coronavirus-2 (SARS-CoV-2)-infected Ghanaians with mild and severe COVID-19, as well as healthy controls to characterize immune signatures at the primary SARS-CoV-2 infection site and identify drivers of disease severity. Generally, a heightened antiviral response was observed in SARS-CoV-2-infected Ghanaians compared with uninfected controls. COVID-19 severity was associated with immune suppression, overexpression of proinflammatory cytokines, including CRNN, IL1A, S100A7, and IL23A, and activation of pathways involved in keratinocyte proliferation. SAMD9L was among the differentially regulated interferon-stimulated genes in our mild and severe disease cohorts, suggesting that it may play a critical role in SARS-CoV-2 pathogenesis. By comparing our data with a publicly available dataset from a non-African (Indians) (GSE166530), an elevated expression of antiviral response-related genes was noted in COVID-19-infected Ghanaians. Overall, the study describes immune signatures driving COVID-19 severity in Ghanaians and identifies immune drivers that could serve as potential prognostic markers for future outbreaks or pandemics. It further provides important preliminary evidence suggesting differences in antiviral response at the upper respiratory interface in sub-Saharan Africans (Ghanaians) and non-Africans, which could be contributing to the differences in disease outcomes. Further studies using larger datasets from different populations will expand on these findings.


Introduction
Severe acute respiratory syndrome coronavirus-2 (SARS-CoV-2) emerged to be a significant public health concern driving the ongoing coronavirus disease 19 (COVID-19) pandemic [1].Beyond the conventional health complications, infection with SARS-CoV-2 was also associated with psychological alterations, including heightened levels of anxiety, stress, and depression, even in hospitalized patients, and this was particularly prevalent during the initial wave of the pandemic [2,3].SARS-CoV-2 utilizes the angiotensin-converting enzyme 2 as a receptor for host cell tropism, which is mainly enhanced by the transmembrane protein TMPRSS2 [1,4].SARS-CoV-2 infection occurs primarily through the upper respiratory interface, and airway immunity is essential in determining the fate of SARS-CoV-2 infection [5].COVID-19 is characterized by varying degrees of

Study Population.
The study population (n = 75) included 52 unvaccinated SARS-CoV-2 infected and 23 uninfected Ghanaians from whom NS samples were collected following informed consent at the Ridge Hospital Accra, Ghana.COVID-19-related symptoms accompanied by a positive SAR-CoV-2 polymerase chain reaction (PCR) test were the criteria for inclusion into our COVID-19 disease cohort, while a negative SARS-CoV-2 PCR result and no symptoms of respiratory infection were used as criteria for inclusion as healthy controls.Samples from the SARS-CoV-2-infected individuals were collected at an acute stage of the disease.Clinicians classified COVID-19-infected patients as severe or mild cases according to the disease case definitions.Confirmatory tests for SARS-CoV-2-specific genetic material by real-time reverse transcription-quantitative PCR were performed at the West African Centre for Cell Biology of Infectious Pathogens (WACCBIP), University of Ghana.The clinical record was available only for a few study participants (Supplementary 1).
2.2.RNA Extraction.RNA was extracted from 300 µl of NS samples using the Quick-RNA Miniprep Plus kit (Zymo Research) following the manufacturer's instructions.Briefly, samples were lysed for 30 min, and nucleic acid was precipitated using absolute ethanol.Sample enrichment for RNA was archived by DNAse treatment followed by column purification.Isolated RNA was eluted in nuclease-free water, and only RNA samples with A260/A280 ratio >1.8 and concentrations above 1 ng/µl were considered for library preparation, as previously examined [22].

Library Preparation and mRNA
Sequencing.The NEB-Next ® Ultra II Directional RNA Library Prep Kit (#7760 L) for Illumina (New England Biolabs) was used for sequencing library construction according to manufacturers' instructions.Briefly, oligo dT-bound beads were used to isolate mRNA, followed by fragmentation for 15 min at 94°C and complementary DNA (cDNA) synthesis.Sequencing libraries were then constructed and amplified using the NEBNext multiple oligos, following manufacturers' instructions.Qubit and TapeStation were used to determine library concentration and size using the high-sensitivity DNA kits.Libraries were generated and sequenced pair-end (150 cycles × 2) on the Illumina Novaseq 6000 system at the Scripps Research Institute using the Novaseq SP reagent kit.Output read files were adapter trimmed and demultiplexed using bcl2fastq v2.20.0.422 (Illumina) to generate unique FASTQ files per sample, with near zero mismatches.

2
Journal of Immunology Research 2.4.Differential Gene Expression Analysis.FASTQ files were pseudo-aligned to an indexed genome generated from the human cDNA fasta sequence (GRCh38) using Kallisto v0.48.0 [23].Only samples with >5 million pseudo-aligned human reads (Supplementary 1) were used for downstream analysis in RStudio v4.2.1.To control for gender and age in the analysis, the median age of participants, 46.5 years (17-94 years), and DDX3Y gene (Y-linked) expression were used to infer participant age and gender, respectively, when absent in the metadata.Transcript IDs were mapped to human genes using an annotated human reference genome (hg38) available in biomaRt v1 [24].Transcript counts were normalized, and differences in gene expression between groups, while controlling for gender and age, were examined using the likelihood ratio test (lrt) and Wald test (wt) in Sleuth v0.27.3 [25].The false discovery rate was corrected using the Benjamin-Hochberg test, and gene expression differences with an adjusted p-value <0.05 were considered statistically significant.A heatmap of the top differentially expressed genes (DEGs) was generated using the Bioconductor package, ggplot2 version 3.3.6[26].Volcano plots were generated using EnhancedVolcano package version 1.14.0 [27], and genes with p-value < 0.05 and log2 fold change (log2fc) >1 were reported as upregulated, while those with log2fc <0 were reported as downregulated.GraphPad Prism v9.4.1 was used to construct the violin plots with log-transformed expression values of selected genes, and the significant level was determined using the unpaired t-test.ClusterProfiler package v 4.8.2 [28] was used in R version 4.3.1 software for gene set enrichment (GSE) analysis of DEGs to identify associated biological pathways.Pathways with adjust-value < 0.05 were reported.
The expression of certain ISGs, such as ISG15, IFIT1, and CXCL8, have been reported to be different in males versus females infected with SARS-CoV-2 (p-value < 0.05) [35].By comparing the expression of these genes in our dataset, the difference in their expression in Ghanaian males vs females in our COVID-19 cohort was not statistically significant (Figure 2(d)-2(f )), contrary to a previous report [35].
Taken together, we found that COVID-19 severity in the Ghanaian cohort was associated with dysregulated inflammatory response mediated by MAL, IL1A, IL23A, CRNN, and S100A7 overexpression and suppression of antiviral immune response-related pathways.A similar association has been reported in other populations [8,[11][12][13]38].

Antiviral Genes Are Differentially Expressed in COVID-
19-Infected Ghanaians Compared with Non-Africans.We further sought to determine whether the expression of antiviral response genes in the upper respiratory interface of SARS-CoV-2-infected Ghanaians differs in other populations.Toward this, we compared our data with a publicly available dataset (GSE166530) from Singh et al. [9] studying COVID-19 immune response signatures in a small cohort of SARS-CoV-2-infected Indians (n = 36) within South Telangana, a population characterized by higher COVID-19 severity and mortality [9].The selection of this data was based on the availability of publicly accessible raw FASTQ data files.Additionally, the data were generated from a similar tissue type, specifically upper airway epithelial tissue, which facilitated a direct comparison.We grouped all cases reported by Singh et al. [9] as a SARS-CoV-2-infected Indian cohort and grouped all the cases from our study to form a SARS-CoV-2-infected Ghanaian cohort.Compared with SARS-CoV-2-infected Indians, an overexpression of antiviral responses-related genes, including TMEM265, IFI6, ISG15, IFITM3, IFIT1, BST2, CCL2, LCN2, and OAS1, was observed in Ghanaians infected with SARS-CoV-2 (Figures 4(a Though preliminary, these observed differences in antiviral gene expression at the primary site of SARS-CoV-2 infection may suggest a more robust innate antiviral immune response in SARS-CoV-2-infected Ghanaians compared to their Indian counterparts.This may have contributed to the reduced COVID-19 severity in Ghanaians and likely other sub-Saharan Africans.Most of these upregulated antiviral genes in SARS-CoV-2-infected Ghanaians were also found to be upregulated in Ghanaians with mild COVID-19 compared to those with severe COVID-19 and uninfected controls (Supplementary 2).

Discussion
The immunological signatures driving COVID-19 severity in Ghanaians remain elusive and need to be better understood.This study investigated the transcriptome differences at the upper respiratory interface of SARS-CoV-2-infected Ghanaians with mild and severe clinical phenotypes to characterize immune signatures at the primary SARS-CoV-2 infection site and identify drivers of disease severity.Consistent with earlier studies [8,9,11,12], we report an upregulation of immune response-related genes accompanied by activation of antiviral pathways and suppression of cellular biogenesis pathways in the upper airway epithelial tissue from COVID-19infected Ghanaians compared with uninfected controls.HLA-A and HLA-DR genes were upregulated in the upper airway of SARS-CoV-2-infected Ghanaians (Supplementary 2) and are known mediators of the adaptive immune response by antigen processing and presentation [39,40], suggesting that HLA-A and HLA-DR overexpression may be activating the adaptive immune response vital to virus-infected cell elimination [32].Cytokines are known regulators of immune response via cellto-cell communication.Regulation of adaptive immune response was the top enriched activated pathway in our COVID-19-infected cohort compared to controls (Figure 2(a)), suggesting the involvement of cytokines with immune regulatory potential, including IL-2 [41][42][43].In addition to protein-coding genes, non-protein-coding LGALS17A was found among the top upregulated genes.Considering the role of noncoding genes in regulating the activities of their target protein-coding genes, LGALS17A upregulation may suggest a critical role in SARS-CoV-2 pathophysiology by regulating the activities of a relevant gene(s) involved in SARS-CoV-2 replication.Neuropeptide B/W receptor-1 (NPBWR1) is the receptor for Neuropeptides B (NPB) and is required for the activation of NPB/NPBWR1 signaling, which plays a vital role in physiological processes, including energy homeostasis and metabolism [44].Earlier work has shown that NPBWR1 knockout mice had defective cellular metabolic processes compared to the wild-type [33,34].In this study, NPBWR1 was among the top downregulated protein-coding genes in our SARS-Cov-2-infected cohort, with cellular component biogenesis being one of the suppressed processes (Figure 2(a)).Noting the critical role of NPBWR1 in metabolic processes to provide the energy and building blocks required for cellular component biogenesis, NPBWR1 downregulation may be driving the suppression of cellular component biogenesis pathways.This could present a previously undescribed SARS-CoV-2 pathogenic mechanism.Comparing Ghanaians with mild vs severe COVID-19 reveals a diminished antiviral response in Ghanaians with severe COVID-19 marked by downregulation of antiviral genes OAS1, CCL8, SAMD9L, selected overexpressed proinflammatory cytokines (IL1A, IL23A, and S100A7) in Ghanaians with severe COVID-19.(e-h).Violin plots of selected antiviral-related genes (ISG15, SAMD9L, IFIT1, and CXC11) that were downregulated in severe cases.Log2fc cutoff = 1, −Log10 p-value, * p-value < 0.05; * * p-value < 0.01; * * * * p-value < 0.001.

8
Journal of Immunology Research HLA-A, CXCL11, ISG15, IL32, and IFIT2, and suppression of antiviral immune response pathways.A similar trend was also observed in previous studies in other populations [8,11,13].
Severe COVID-19 has been chiefly associated with inflammatory cytokines such as interleukin 6 (IL-6), IL-8, and IL-10 overexpression [8,11,12,45,46].Though Tapela et al. [47] reported some association between IL-6 and IL-8 cytokine concentration in plasma samples and COVID-19 severity, the expression of these cytokines was not found to be significantly upregulated in our severe COVID-19 cohort.However, in this study, an upregulation of other pro-inflammatory cytokines, including CRNN, IL1A, IL23A, IVL, and S100A7, was associated with severe COVID-19.CRNN was the most upregulated gene, and keratinization was the top-activated process in individuals with severe COVID-19 cohort.Keratinocytes represent the first line of the host defense system, and their hyperproliferation contributes to the pathogenesis by infiltration of inflammatory cells [48,49].CRNN overexpression was previously shown to aberrantly regulate keratinization by activating the Phosphoinositide 3-Kinase/Akt Pathway, leading to inflammatory diseases, such as psoriasis [36].Epithelial cells are directly infected during SARS-CoV-2; thus, CRNN overexpression in our severe COVID-19 cohort may represent a potential pathogenic mechanism employed by SARS-CoV-2 to induce dysregulated inflammatory response via upregulating keratinization at the primary site of infection.In addition, the MAL gene, an important component in NF-κB signaling pathway activation [37], and TMPRSS11B were among the top 10 upregulated genes in Ghanaians with severe COVID-19.TMPRSS11B is implicated as a driver of lung carcinoma [50], and severe COVID-19 is associated with lung abnormalities [51,52].Since SARS-CoV-2 is known to induce pathology in the lung, TMPRSS11B upregulation in individuals with severe COVID-19 may also represent another SARS-CoV-2 pathogenic mechanism.TMPRSS11B also interacts with CRNN (Supplementary 1).Our result on immune signatures mediating COVID-19 severity in Ghanaians agrees substantially with findings from other studies [8,11,12,46,53].The SAMD9L pathway was previously shown to be a critical host barrier that poxviruses subvert most to establish an infection [54] and was among the ISGs found to be significantly downregulated in Ghanaians with severe COVID-19 compared with mild cases.The suppression of SAMD9L in individuals with severe COVID-19 suggests that it may also be a critical host restriction factor that SARS-CoV-2 must antagonize to establish disease.Additionally, MUC21, a gene previously associated with lung adenocarcinoma, was also upregulated in Ghanaians with severe COVID-19 [55].We found an insignificant difference in the expression of previously reported antiviral genes, ISG15, IFIT1, and CXCL8, in males and females Ghanaians infected with SARS-CoV-2, contrary to a previous report [35].However, this observation might be influenced by the sample size used in this study (Table 1).COVID-19 severity is considerably lower in sub-Saharan Africans, particularly West Africans, compared with non-Africans and Black African Americans [16,21,56].We observe an upregulation of genes involved in antiviral response pathways, including OAS1 that mediates RNase L pathway [57,58], IFIT1, and APOE at the upper respiratory airway of COVID-19-infected Ghanaians compared with a relevant publicly available dataset (GSE166530) from an Indian COVID-19 cohort [9] (Figure 4).The upregulation of these antiviral genes in COVID-19-infected Ghanaians may suggest a more robust antiviral response at this critical interface.Though preliminary, this observed difference in antiviral gene expression at primary infection sites may have contributed to the reduced COVID-19 severity in sub-Saharan Africans, particularly Ghanaians.To our knowledge, this is the first direct comparison of immune response-related gene expression in the upper respiratory interface between SARS-CoV-2-infected West Africans and a non-African population and the first COVID-19 bulk host transcriptome dataset from West Africans.

Conclusions
In conclusion, this study describes immune signatures at the primary site of SARS-CoV-2 infection and identifies immune signatures driving COVID-19 severity in SARS-CoV-2-infected Ghanaians.It further provides important preliminary evidence suggesting that antiviral genes are more highly expressed at the primary site of SARS-CoV-2 infection in sub-Saharan Africans (Ghanaians) compared with non-Africans (Indians), which may be driving the differences in antiviral response and clinical outcomes.Our overall report on DEGs in COVID-19-infected Ghanaians corroborates previous reports from similar studies.

TABLE 1 :
Disease characteristics of COVID-19-infected Ghanaians used in this study.
Shortness of breath, chills, and mechanical ventilation were associated with severity.