Divergent adaptive immune responses define two types of long COVID

Abstract

Background: The role of adaptive immune responses in long COVID remains poorly understood, with contrasting hypotheses suggesting either an insufficient antiviral response or an excessive immune response associated with inflammatory damage. To address this issue, we set to characterize humoral and CD4+ T cell responses in long COVID patients prior to SARS-CoV-2 vaccination.

Methods: Long COVID patients who were seropositive (LC+, n=28) or seronegative (LC-, n=23) by spike ELISA assay were recruited based on (i) an initial SARS-CoV-2 infection documented by PCR or the conjunction of three major signs of COVID-19 and (ii) the persistence or resurgence of at least 3 symptoms for over 3 months. They were compared to COVID patients with resolved symptoms (RE, n=29) and uninfected control individuals (HD, n=29).

Results: The spectrum of persistent symptoms proved similar in both long COVID groups, with a trend for a higher number of symptoms in the seronegative group (median=6 vs 4.5; P=0.01). The use a highly sensitive S-flow assay enabled the detection of low levels of SARS-CoV-2 spike-specific IgG in 22.7% of ELISA-seronegative long COVID (LC-) patients. In contrast, spike-specific IgG levels were uniformly high in the LC+ and RE groups. Multiplexed antibody analyses to 30 different viral antigens showed that LC- patients had defective antibody responses to all SARS-CoV-2 proteins tested but had in most cases preserved responses to other viruses. A sensitive primary T cell line assay revealed low but detectable SARS-CoV-2-specific CD4 responses in 39.1% of LC- patients, while response frequencies were high in the LC+ and RE groups. Correlation analyses showed overall strong associations between humoral and cellular responses, with exceptions in the LC- group.

Conclusions: These findings provide evidence for two major types of antiviral immune responses in long COVID. Seropositive patients showed coordinated cellular and humoral responses at least as high as those of recovered patients. In contrast, ELISA-seronegative long COVID patients showed overall low antiviral responses, with detectable specific CD4+ T cells and/or antibodies in close to half of patients (52.2%). These divergent findings in patients sharing a comparable spectrum of persistent symptoms raise the possibility of multiple etiologies in long COVID.

Keywords: SARS-CoV-2; T cell; common cold coronavirus; humoral immune response; long COVID; seronegative and seropositive.

Figure 1

Spike-specific antibodies distinguish two groups of long COVID patients. Antibodies specific to the Wuhan SARS-CoV-2 spike were measured in the S-flow assay. (A) Percentage of spike-expressing 293T cells (293T-S) bound by serum IgG. (B) Mean fluorescent intensity (MFI) of IgG bound to 293T-S cells. (C) Percentage of 293T-S bound by serum IgA. (D) MFI of IgA bound to 293T-S cells. HD, healthy donors; LC-, seronegative long COVID patients; LC+, seropositive long COVID patients; RE, recovered patients. Horizontal bars represent medians. Differences between groups were evaluated with the Kruskal-Wallis statistical test with Dunn’s correction for multiple comparisons. ****P<0.0001.

Figure 2

Antibody responses to multiple SARS-CoV-2 proteins distinguish the two groups of long COVID patients. (A-C) IgG responses to the nucleoprotein (NP), the M and E proteins (ME), and the S2 spike subunit (S2) from SARS-CoV-2 were measured based on MFI in a multiplexed Luminex assay. Horizontal bars represent medians. Differences between groups were measured with the Kruskal-Wallis test with Dunn’s correction for multiple comparisons. **P<0.01; ***P<0.001; ****P<0.001. (D-F) Correlation between antibody responses to different SARS-CoV-2 proteins: NP and Spike (D); ME and Spike (E); and S2 and Spike (F). The linear regression on log-transformed MFI values measured in the Luminex assay is indicated by a straight full line. The linear correlation coefficient R and the P value are reported on each graph. The dashed line represents the threshold for seropositivity estimated by spike antibody measurements in the Luminex assay.

Figure 3

Similar antibody reactivity to common cold coronaviruses in the two long COVID groups. (A-D) IgG responses to the nucleoprotein (NP) of the four common cold coronaviruses OC43 (A), HKU1 (B), 229E (C) and NL63 (D) were measured based on MFI in a multiplexed Luminex assay. (E-H) Antibody responses to the spike (S) of the four common cold coronaviruses. Horizontal bars represent medians. Differences between groups were measured with the Kruskal-Wallis test with Dunn’s correction for multiple comparisons. *P<0.05; **P<0.01; **** P<0.0001.

Figure 4

Analysis of SARS-CoV-2 peptide-specific CD4+ T cell responses in primary T cell lines. CD4+ T cell lines restimulated with the matrix M141 peptide (bottom row) or not restimulated (NS, top row), were fixed, permeabilized, and analyzed by flow cytometry in the viable CD3+ CD4+ gate for the intracellular production of IFN-γ (x axis) and TNF-α (y axis). Representative examples of responses are shown for one pre-pandemic healthy donor (HD), one seronegative long COVID patient (LC-), one seropositive long COVID patient (LC+), and one recovered patient (RE).

Figure 5

Divergent CD4+ T cell responses in seronegative and seropositive long COVID patients. The frequency of CD4+ T cells producing TNF-α (A) and IFN-γ (B) in primary T cell lines stimulated with the indicated peptide is reported. Peptides are designated by the viral protein (M: matrix, N: nucleocapsid; S spike) and the position of the first a.a. in that protein. Horizontal bars represent medians. Differences between groups were evaluated with the Kruskal-Wallis statistical test with Dunn’s correction for multiple comparisons. *P<0.05; **P<0.01; ***P<0.001; ****P<0.0001.

Figure 6

Positive correlation between the CD4+ T cell response and the antibody responses to SARS-CoV-2. (A) The percentage of individuals with a CD4+ T cell response to each of the 9 immunodominant peptides tested is reported. A responder is an individual for whom the % TNF-α+ or % IFN-γ+ cells in CD4+ T cells is above the limit of sensitivity (LOS = 2 x SD for the HD group) and has a stimulation index SI >2 for the considered peptide. Peptides derived from the matrix (M), nucleoprotein (N) and spike (S) are color-coded in blue, red, and orange, respectively. (B, C) The summed frequency of CD4+ T cells producing TNF-α (A) and IFN-γ (B) in response to the 9 peptides tested is reported. (A-C) Horizontal bars represent medians. Differences between groups were evaluated with the Kruskal-Wallis statistical test with Dunn’s correction for multiple comparisons. **P<0.01; ***P<0.001; ****P<0.0001. (D, E) Linear regression between the CD4+ T cell response to the M141 peptide, measured by the % of IFN-γ+ cells, and the antibody response, measured in the S-flow assay by the MFI of spike-cells bound by serum IgG (D) or IgA (E). The linear regression coefficient R and the P value for the slope of the regression line being different from zero are reported.