SARS-CoV-2 induces robust germinal center CD4 T follicular helper cell responses in rhesus macaques

CD4 T follicular helper (Tfh) cells are important for the generation of durable and specific humoral protection against viral infections. The degree to which SARS-CoV-2 infection generates Tfh cells and stimulates the germinal center (GC) response is an important question as we investigate vaccine induced immunity against COVID-19. Here, we report that SARS-CoV-2 infection in rhesus macaques, either infused with convalescent plasma, normal plasma, or receiving no infusion, resulted in transient accumulation of pro-inflammatory monocytes and proliferating Tfh cells with a Th1 profile in peripheral blood. CD4 helper cell responses skewed predominantly toward a Th1 response in blood, lung, and lymph nodes. SARS-CoV-2 Infection induced GC Tfh cells specific for the SARS-CoV-2 spike and nucleocapsid proteins, and a corresponding early appearance of antiviral serum IgG antibodies. Collectively, the data show induction of GC responses in a rhesus model of mild COVID-19.

. Were no lung samples collected in this study? It is not clear in the figure and needs to be made clearer. Again, although the premise of the manuscript is Tfh response and antibody development, authors have not discussed much on the high pseudovirus neutralizing antibody titer observed in these macaques compared to other published paper (Chandrashekhar et al. Science 2020). Discussing it would strenghthen the paper. Figure 1B: Viral load data I+CP (n=2) vs I and I+NP (n=6): although, n=2 is low, there seems to be higher loads (at least one log difference) in this group. Is this because of antibody dependent enhancement of infection. More animals for I+CP would have ruled out this option. However, what if these two macaques are complicating the overall interpretation of the results? Additionally, immune complexes formed by binding antibody (even if neutralizing antibody levels fell below the detection limit) and virus, can also impact the generation of immune responses, as these complexes can be recognized by APCs.
Line 176-178 : "At the apex of the effector response, Ki-67+ CD4 T cells, specifically the Th1 but not the Tfh subset was strongly associated with proliferating CD8 T cells (Fig. 1I). In turn, we observed strong antigen-dependent induction of CD8 T cells evidenced by the association between SARS-CoV-2 vRNA from nasal washes and proliferating (Ki67+) CD8 T cells" This is direct correlation suggesting that infection is activating the development of CD8 T cell response. Since there is no decrease in viral loads, it would be difficult to consider this observation in the context of correlates of protection resulting in mild disease outcome.
Reviewer #2: Remarks to the Author: The authors study the immune response to concurrent intranasal, intratracheal, and ocular SARS-CoV-2 infection in rhesus macaques, with a focus on the CD4+ T cell responses. In this model of mild disease, they find induction of Th1-type Tfh cells (expressing CXCR3) in the mediastinal lymph node along with an IgG-dominated antibody response 7-10 days post infection. They also characterize systemic responses in the blood and the spleen. This is an important study, notably for its characterization of the CD4 T cells at the draining lymph node of the site of infection, but the authors should address the points below to strengthen the conclusions made in the manuscript.
Major comments: 1. The authors argue that Th1-type Tfh cells are likely responsible for the anti-SARS-CoV-2 antibodies that appear by day 7. However, it should be emphasized that this is an association and the necessity of this T cell population for the antibody response cannot be addressed in the current study. Furthermore, the authors should support the claim of Th1-like Tfh cells in the mediastinal lymph node by performing the same intracellular IFN-γ/IL-21 staining they use for splenic T cells ( Figure 2C) -but a naïve control will be needed to interpret cytokine induction unless antigen-specific stimulation is used. Although CXCR3 is used to phenotype Th1-polarized cells in peripheral blood and Th1 effector cells in lymph nodes, it is unclear what the combined expression of CXCR3 and CXCR5 would do to follicular localization of presumed Tfh cells and therefore the ability to help B cells.
2. The point of the convalescent plasma inclusion in this study is unclear. It is not mentioned in the abstract nor in any of the figure titles. In Figure 1B, the authors find no effect of convalescent human plasma therapy on viral load, which they argue is due to dilution of the transfused plasma to levels too low to neutralize virus. The authors should show the data supporting this statement. In addition, other studies using similar dosing of convalescent human plasma in patients find that neutralizing titers in the serum increase following plasma transfusion (Duan, 2020, PNAS; Shen, 2020, JAMA). Why do neutralizing antibodies not increase in macaques post-transfusion? It seems this aspect of the study was not well designed or even incorporated into the ultimate findings. It is also unclear whether the study is powered to make any conclusion about convalescent plasma, given only 2 macaques were infused.
3. In Figures 2C-D and S2E, the authors argue that SARS-CoV-2 infection induces polyfunctional CD4 T cells by showing their production, and sometimes co-production, of multiple cytokines following PMA/ionomycin stimulation. It is essential to have an uninfected macaque for comparison in these figures (like Fig. 3G), as it is unclear whether this is truly infection-induced phenotype. Furthermore, PMA/ionomycin is a supraphysiologic stimulus, so the authors should consider demonstrating cytokine production following antigen-specific stimulation. Finally, it is unclear what the "Unstim" label in Figure  S2E refers to, as the corresponding figure legend indicates that this sample is stimulated. 4.Given the observed time course of antibody production in humans after infection, measuring titers only out to day 10 seems inappropriate (ref 26 and 27 do not support the claim made by the authors on line 268 for antibody kinetics). Day 14 is usually the first timepoint considered acceptable for robust antibody detection in humans and would be more consistent with the timing of a Tfh-induced antibody response. Day 7 is quite an early timepoint, and it is possible that the antibody responses at this point may be Tfh-independent. The authors should demonstrate anti-SARS-CoV-2 titers at a later time point, at least 21 days later. 5. Although the authors state that they have identified SARS-CoV-2-specific Tfh cells the data to support this claim is extremely limited (just Fig. 3G). The data presented in fig. 2A does not seem to indicate antigen specificity.
Minor comments: 1. Please cite the AIM assay (line 217). Figures 1J and S2A? Usually t-SNE plots are used to depict high dimensional data such as scRNA-seq or CyTOF, but it does not seem that such methods were performed. 3 In Figure S3A, it is confusing that the key indicates circles as infected and triangles as infected + infused, but then the graph has circles in the uninfected group. 4. In Figure 3C, it is confusing that there are two blue boxes (FDCs and PD1+ CXCR5+ CD4 T cells) and two green boxes (GC B and PD1-CXCR5+ CD4 T cells). 5. Presumably "P/I" refers to pma/ionomycin in Fig. 3G? This information should be added to the figure legend. 6. Recent work has highlighted that SLAM is a Tfh-associated rather than Th1-assocaited molecule. Similarly, the authors state "effector molecule CX3CR1, a marker potentially for newly generated memory CD4 T cell subsets", but is used in this study (and more typically in most studies) to identify Th1-skewed cells. The authors should support their rationale for multiple aspects of defining populations throughout the study.

What data are used to construct the t-SNE plots in
Reviewer #3: Remarks to the Author: In the manuscript "SARS-COV-2 infection induces robust germinal center CD4 T follicular helper cell responses in rhesus macaques", the authors report that SARS-CoV-2 infection results in transient accumulation of activated and proliferating Tfh cells of Th1 phenotype in the blood and MLN.
The infusion of convalescent human plasma is a great control, unfortunately the neutralization titer fell below detectable limits. Why didn't the authors give more plasma to get a better response? I would guess the reason is just the availability of sufficient amounts of plasma and the known overall low neutralization titers in patie4nts. With all the concepts going on to use human plasma to treat patients it had been so interesting to see whether higher doses had resulted in blunting viral load.

I wonder why the infected animals did not show any clinical symptoms of illness upon infection?
The presence of PD-1 CXCR5 expressing cells is a good indicator for ongoing GC reactions. I wonder how a B cell staining looks like in these animals. CD95, Peanut agglutinin or GL-7 in combination with CD38? Germinal center are easy to detect in immunofluorescent analysis of frozen tissue sections. Were any of these analysis done?
It is interesting to see that responses to S, N, M proteins were also detected in non infected animals. As the authors state this was also observed in humans and discussed as cross-reactive T cells to endemic coronaviruses. I wonder how big the chance is for those colony-bread animals housed in an animal research center to get contact with other coronaviruses.
The data presented convincingly show that S and N specific CD4 Tfh cells are induced upon SARS-CoV-2 infection In figure 3C the authors show that 23.2% of B cells (CD20+) express Bcl6. CD95 is also present in the staining cocktail Can the authors gate CD95+ B cells (=germinal center B cells) and show that frequency of Bcl6 expression is higher in the CD95+ B cell compartment. This would help to nail the point of GC induction (though without information about specificity for Sars-CoV-2).
Figure 4 would benefit from IgG isotype data. Do these animals display IgG1 and IgG2? This would help the understanding whether the immune response is exclusively Th1 or alsoTh2.
Overall I think. This study would highly benefit from showing the presence of Germinal center reactions e.g. in the MLN by any type of microscopy (preferentially immunofluorescence).
Germinal center reactions are the source of mutated antibodies. Is there any evidence of hypermutation of anibodies?