Metagenomic Next-Generation Sequencing of Nasopharyngeal Specimens Collected from Confirmed and Suspect COVID-19 Patients

Abstract

Metagenomic next-generation sequencing (mNGS) offers an agnostic approach for emerging pathogen detection directly from clinical specimens. In contrast to targeted methods, mNGS also provides valuable information on the composition of the microbiome and might uncover coinfections that may associate with disease progression and impact prognosis. To evaluate the use of mNGS for detecting severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) and/or other infecting pathogens, we applied direct Oxford Nanopore long-read third-generation metatranscriptomic and metagenomic sequencing. Nasopharyngeal (NP) swab specimens from 50 patients under investigation for CoV disease 2019 (COVID-19) were sequenced, and the data were analyzed by the CosmosID bioinformatics platform. Further, we characterized coinfections and the microbiome associated with a four-point severity index. SARS-CoV-2 was identified in 77.5% (31/40) of samples positive by RT-PCR, correlating with lower cycle threshold (Ct) values and fewer days from symptom onset. At the time of sampling, possible bacterial or viral coinfections were detected in 12.5% of SARS-CoV-2-positive specimens. A decrease in microbial diversity was observed among COVID-19-confirmed patients (Shannon diversity index, P = 0.0082; Chao richness estimate, P = 0.0097; Simpson diversity index, P = 0.018), and differences in microbial communities were linked to disease severity (P = 0.022). Furthermore, statistically significant shifts in the microbiome were identified among SARS-CoV-2-positive and -negative patients, in the latter of whom a higher abundance of Propionibacteriaceae (P = 0.028) and a reduction in the abundance of Corynebacterium accolens (P = 0.025) were observed. Our study corroborates the growing evidence that increased SARS-CoV-2 RNA detection from NP swabs is associated with the early stages rather than the severity of COVID-19. Further, we demonstrate that SARS-CoV-2 causes a significant change in the respiratory microbiome. This work illustrates the utility of mNGS for the detection of SARS-CoV-2, for diagnosing coinfections without viral target enrichment or amplification, and for the analysis of the respiratory microbiome.IMPORTANCE SARS-CoV-2 has presented a rapidly accelerating global public health crisis. The ability to detect and analyze viral RNA from minimally invasive patient specimens is critical to the public health response. Metagenomic next-generation sequencing (mNGS) offers an opportunity to detect SARS-CoV-2 from nasopharyngeal (NP) swabs. This approach also provides information on the composition of the respiratory microbiome and its relationship to coinfections or the presence of other organisms that may impact SARS-CoV-2 disease progression and prognosis. Here, using direct Oxford Nanopore long-read third-generation metatranscriptomic and metagenomic sequencing of NP swab specimens from 50 patients under investigation for COVID-19, we detected SARS-CoV-2 sequences by applying the CosmosID bioinformatics platform. Further, we characterized coinfections and detected a decrease in the diversity of the microbiomes in these patients. Statistically significant shifts in the microbiome were identified among COVID-19-positive and -negative patients, in the latter of whom a higher abundance of Propionibacteriaceae and a reduction in the abundance of Corynebacterium accolens were observed. Our study also corroborates the growing evidence that increased SARS-CoV-2 RNA detection from NP swabs is associated with the early stages of disease rather than with severity of disease. This work illustrates the utility of mNGS for the detection and analysis of SARS-CoV-2 from NP swabs without viral target enrichment or amplification and for the analysis of the respiratory microbiome.

Keywords: COVID-19; SARS-CoV-2; metagenomic next-generation sequencing; metagenomics; nasopharyngeal.

FIG 1

Relationships of metatranscriptomic sequencing sample characteristics to sequencing results. (a) Relationship between the proportion of reads mapped to SARS-CoV-2 and the number of reads matched to human sequences. (b to i) Plots of sequencing results against the Ct values of metatranscriptomic sequencing samples. Panels a to f are color coded by severity index values, and panels g to i are color coded by days from onset of symptoms. The severity index was defined on a scale of 1 to 4, as follows: 4, not admitted; 3, admitted; 2, intensive care unit; and 1, required ventilator. Black dots represent samples with unknown onset or severity index values. (b) Relationship of Ct values determined by LDT-RT-PCR to the number of days from symptom onset. (c) Relationship between Ct values and the proportion of total matches of SARS-CoV-2 by CosmosID. (d, g) Relationship between Ct values and the sequencing coverage across the SARS-CoV-2 strain Hu-1 genome. (e, h) Relationship between Ct values and the number of reads analyzed by CosmosID until the first SARS-CoV-2 read was detected. (f, i) Relationship of Ct values to the proportion of SARS-CoV-2 reads present in the sample. Simple linear regression analysis was performed for each set, and the null hypothesis is rejected for panels b to i (P < 0.01), and for each set, R² is reported.

FIG 2

Bacterial diversity analysis of metagenomic sequencing results. (a, b, c) Alpha diversity analysis of metagenomic sequencing results. SARS-CoV-2 positivity was determined by LDT-RT-PCR. (a, b) Shannon diversity plot and Chao diversity plot of SARS-CoV-2-negative and -positive samples at the species level. (c) Shannon diversity plot of SARS-CoV-2-negative samples and SARS-CoV-2-positive samples at different periods post-onset of symptoms. (d, e, f) Beta diversity principal-coordinate analysis of metagenomic sequencing results at the species level. (d) Bray-Curtis analysis of bacterial community composition diversity between SARS-CoV-2-negative and SARS-CoV-2-positive samples. (e) Bray-Curtis analysis of bacterial community composition diversity grouped by PCR and sequencing positivity. (f) Bray-Curtis analysis of bacterial diversity in different disease severity groups. (a, b) Wilcoxon rank sum tests were performed between positive and negative SARS-CoV-2 groups for the Shannon diversity index (P = 0.0097) and Chao diversity (P = 0.0082). (c) Wilcoxon rank sum tests were performed between disease onset groups (no significance). (d, f) PERMANOVA tests were performed on Bray-Curtis distance matrices for SARS-CoV-2-positive and -negative groups (P = 0.027) (d) and groups of disease severity (P = 0.022) (f). (e) Pairwise PERMANOVA tests were performed on Bray-Curtis distance matrices between groups defined by SARS-CoV-2 positivity by RT-PCR and sequencing. RT-PCR+/sequencing+ (PCR+/Seq+) versus RT-PCR–/sequencing– (PCR–/Seq–) (P = 0.007); PC1 and -2, principal components 1 and 2. The severity index was defined on a scale of 1 to 4, as follows: 4, not admitted; 3, admitted; 2, intensive care unit; and 1, required ventilator.

FIG 3

Changes in bacterial microbiome composition in COVID-19 patients. (a, b) Relative abundances of bacteria at the species level in COVID-19-positive and COVID-19-negative samples. (a) Overall relative abundances of bacterial species in COVID-19-positive and COVID-19-negative patients. (b) Relative abundances of bacterial species in individual COVID-19-positive and COVID-19-negative patients, with samples ordered by Ct value determined by LDT-RT-PCR. A Wilcoxon rank sum test was performed on overall relative abundance values for Propionibacteriaceae (*, P = 0.028) and for Corynebacterium accolens (*, P = 0.025).