HTNV infection of CD8+ T cells is associated with disease progression in HFRS patients

Hantaan viruses (HTNVs) are zoonotic pathogens transmitted mainly by rodents and capable of infecting humans. Increasing knowledge of the human response to HTNV infection can guide the development of new preventative vaccines and therapeutic strategies. Here, we show that HTNV can infect CD8+ T cells in vivo in patients diagnosed with hemorrhagic fever with renal syndrome (HFRS). Electron microscopy-mediated tracking of the life cycle and ultrastructure of HTNV-infected CD8+ T cells in vitro showed an association between notable increases in cytoplasmic multivesicular bodies and virus production. Notably, based on a clinical cohort of 280 patients, we found that circulating HTNV-infected CD8+ T cell numbers in blood were proportional to disease severity. These results demonstrate that viral infected CD8+ T cells may be used as an adjunct marker for monitoring HFRS disease progression and that modulating T cell functions may be explored for new treatment strategies.

3) The scRNAseq experiments and their analysis are not adequately described or interpreted. a. It is not clear how infection status is accounted for in the analysis. Are all the CD8s that are sequenced presumed to be infected and can that be supported by any data? I would suggest the authors assess HTNV gene expression in each cell of their dataset to determine infection. b. It is not clear how the differences in gene expression between CD8 subsets relate to HTNV infection. Relating to the comment above, were any infectrf CD8 cells analysed? Transcriptional differences between Tnaive, Tcm, Te and Tem are expected c. Page 8 line 206 states PCA analysis was used however there are no PCA plots in the manuscript, only tSNE plots in fig 4. d. The whole section on the RNAseq analysis needs to be revisited and clearly explain the samples used (what infected vs uninfected datasets are included) and the analysis needs to be clearly explained in relation to infected and uninfected cells (and then can be further divided in CD8 subsets).
4) The authors demonstrate that the frequency of HTNV-infected CD8 cells correlates with measures of clinical severity and with the acute phase of disease. Given the number of patients in each subgroup of acute (febrile, hypotensive, oliguric) and convalescent (diuretic, convalescent) samples, it we would be of interest to determine how the frequency of HTNV-infected CD8 cells differs between those groups.
Minor comments: Line 250 page 10: 'to quantity' should be corrected to 'to quantify' Page 14 flow cytometry section: the HTNV-NP staining method needs futher information, was this intracellular? What fix/perm methods have been used?
Reviewer #3 (Remarks to the Author): This report by Liu et al. is a well-written investigation into immune responses and potential biomarkers of HTNV infection. As the authors nicely summarize, there is relatively little known regarding immunity and pathogenesis of HTNV infection, and thus, immunological insight as well as predictive biomarkers of disease hold widespread utility. The primary conclusions of this study are that HTNV infects CD8 T cells, and that HTNV-infected CD8 T cells are predictive of disease progression in HFRS patients. The authors should be commended for their relatively large dataset and an impressive amount of work, use of single-cell RNAseq, and some potentially novel and interesting findings. However, the overall robustness of the results suggesting CD8 T cells are infected with HTNV and that this is a relevant biomarker is limited.
1. Based on the evidence presented, the extent to which HTNV infects CD8 T cells in infected patients is unclear and not very convincing.
a. The crux of the authors' conclusion (that HTNV infects CD8 T cells and this is relevant for disease progression) here appears to be mainly derived from Fig. 1B. However, there are several issues of concern with this figure. The authors do not present proper controls. There should be staining from uninfected, healthy controls (in the representative plots in B as well as in the analysis in C to understand the level of background staining) as well as a negative staining control or FMO. It's hard to discern the level of non-specific binding occurring here, and 82.1% of all CD8 T cells in a patient being infected with HTNV seems extraordinarily high. Indeed, the frequency decreases over time in Fig. 1B which is encouraging (although there is no graphical/statistical demonstration of this and appears to be for just one individual), but so does the activation level and phenotype of CD8 T cells which can potentially change expression of nonspecific targets. Confidence in the staining would also be increased by sorting NP+ and NP-CD8 T cells and measure viral mRNA expression in the two populations through qPCR. b. Related to the above point, Fig. 2B provides some confidence that this process is occurring in vivo and the authors have nicely included PBMCs from an uninfected control individual. However, in Fig 2B, the authors show two infected CD8 T cells, which appears to be 100% of the CD8 T (or at least 67%) cells in the field. There's no quantification of this for multiple individuals, so it's hard compare this with Fig. 1 and be fully confident that this is happening in vivo to the level the authors suggest.
c. Regarding the frequency of infected CD8 T cells: The authors report a range of 0.15-82% of HTNVinfected CD8 T cells in Fig. 1D, with a mean of 15.2%. In Fig. 2C-D, at 96h and ex vivo it appears 100% of all CD8 T cells are infected. Overall, it's hard to confidently grasp the level of CD8 T cell infection in vivo and in vitro and many of these frequencies seem very high (especially as the frequency of HTNV in vitro infected monocytes was reported to be ~1% in reference # 19 with nearly 100x MOI).
2. Related to the point above, the authors do a nice job in providing supporting data to their main conclusion in Figures 2&3. Figures 2/3 highlight that CD8 T cells can be infected by HTNV, and support production of infectious virions. These are interesting findings and, for the most part, well done experiments in these figures. However, in vitro infection findings do not necessarily indicate this happens in humans in vivo, and thus strengthening the findings from Fig1 or 2B are essential. These in vitro studies would have also been strengthened by comparing to either a population of cells known to produce infectious virus (i.e. a validated cell target of HTNV) and/or by comparing with infection of CD4 T cells (as the summary sentence in line 138 is that "HTNV preferentially infects CD8 T cells").
3. The main conclusion that HTNV-infected CD8 T cells could be a useful biomarker is also not very convincing as Fig 5B indicates that CD8 T cells are an equally robust marker (which has already been shown to correlate with disease stage), and there are other immunological parameters that correlate better with disease stage. 4. The authors do a nice job with the scRNAseq analysis and this is certainly a technique which could reveal some really interesting biology. However, it appears uninfected cells are not included in the analysis. Therefore, it's hard to determine what the impact of viral infection was. If the authors could distinguish infected and uninfected cells via viral mRNAs in their analysis, and then compare those populations this would be a very interesting piece of data (however it appears that nearly all cells are infected in vitro from Fig. 2). 5. The overall study design is unclear. There are 280 blood samples from 119 patients (line 105 and abstract) but then 122 PBMC samples are used for Fig 1. It's not clear how many blood samples are from the same patient? Also, having blood samples from the same patient throughout the course of infection (as other studies have done) would dramatically enhance the strength of the results as well.

Reviewer #1
Major comments "In the study described in this manuscript by Liu et al., they show that the CD8 primary T cells can be infected with hantaan virus (HTNV) and these cells support the full replication cycle of HTNV. Further, they also did the transcriptomic analysis to identify signature genetic changes in various subsets of infected cells. The manuscript is written fine and experiments mostly support the conclusions. However, there are some major and minor points which can make the manuscript stronger upon addressing them."

"The most important point which is missing is assessing cytokine responses.
Although authors show changes in CD8 T cells, the functionality of the cells is crucial, mainly to deduce correlates. The authors do mention that the material was limited but from past experience and literature, I think cytokine response can be evaluated from the same samples. This is the main drawback of the manuscript which concerns me the most and needs to be verified experimentally."

Answer:
Thank you for your suggestion. Cytokines are known to be important factors in HFRS.
It is generally accepted that the pathogenesis of hantavirus infections is the result of virus-mediated host immune response. Among immune parameters, certain cytokines such as IL-1, IL-6, IL-10, and TNF-a were suggested to be involved in the pathogenesis, since increased levels of these cytokines were found in patients with HFRS. 1 The cytokines detected include TNF-α, IL-2, IL-6, and IFN-γ, which are produced by T cells and may mediate capillary leakage. Exposure to high doses of TNF-α in vivo is known to induce shock, capillary leakage, and mortality, and therapy with high doses of IL-2 causes an increase in vascular permeability 1. As showed in Fig.5A, we have detected CD8 + T cells related to inflammatory cytokines including perforin, granzyme A, and granzyme B. The concentrations 2 median (IQR) in the plasma of the corresponding patient samples is significantly higher at the acute stage than at the convalescent stage (P<0.001).
On this basis, we made further statistics according to the disease stage. In brief, the levels of perforin, and Granzyme B serum concentrations progressively increased from the febrile phase to the hypotensive phase and then gradually decreased during the convalescent phase. These results had been added as Figure.S2 in the revised manuscript.
The description has been added in the results section. For infection, in pilot experiments, we examined various MOI titrations using the MTT assay and discovered that MOI=0.1 is a suitable titer to achieve quantifiable infections, similar to published studies [3][4][5] .
Specifically, we investigated HTNV 76-118 progeny production in primary T cells with three different MOI at various time points. For infection, purified CD8 + T cells were plated at 5×10 6 cells/well in a 6-well plate and then exposed to HTNV at

Answer:
We have analyzed out data and summarized as descriptive statistics in the form of table S1, by examining the clinical and laboratory features of patients with HFRS stratified by five stages or two disease phases (acute and convalescent), and found age or gender do no co-segregate with patients' clinical features. Microbiol. 10, (2020)" has been inserted following the above-mentioned text. (Page 10，Line 293 and Page 10，Line 303).

Introduction line 67 "however, too strong or too weak a T cell response has been
linked to severe disease in both HCPS and HFRS". It should be "however, too strong or too weak of a T cell response has been linked to severe disease in both HCPS and HFRS".

Answer:
This has been modified in the text accordingly. (Page 4，Line 80)

Reviewer #2
Major comments:  3) The scRNAseq experiments and their analysis are not adequately described or interpreted.

a. It is not clear how infection status is accounted for in the analysis. Are all the CD8s
that are sequenced presumed to be infected and can that be supported by any data? I would suggest the authors assess HTNV gene expression in each cell of their dataset to determine infection.

Answer:
To gain insight into the potential functional alterations occurring in HTNV-infected CD8 + T cells, we performed scRNA-seq using CD8 + T cells sorted by positive selection with magnetic beads from PBMCs. We then infected the cells with HTNV (MOI=0.1) for 48 h. Not all of the CD8 + T cells were infected, but a part of them did.
In PBMC of patients, it can be seen that the infected T cells are a cluster of CD8 low cells.
Because scRNA-seq technologies rely on polyadenylated RNA isolation and amplification, current scRNA-seq methods can, in theory, detect these viral RNA

scRNA-seq quantifications and statistical analysis
Raw reads were processed to generate gene expression profiles using an internal pipeline. Briefly, after filtering read one without poly T tails, cell barcode and UMI

4) The authors demonstrate that the frequency of HTNV-infected CD8 cells correlates with measures of clinical severity and with the acute phase of the disease. Given the number of patients in each subgroup of acute (febrile, hypotensive, oliguric) and convalescent (diuretic, convalescent) samples, it would be of interest to determine how the frequency of HTNV-infected CD8 cells differs between those groups.
Answer: "As shown in Figure.S5, the patient samples were counted according to the stages of the disease；febrile, hypotensive, oliguric，diuretic, and convalescent. As the disease progressed, the number of the HTNV + CD8 + cells gradually decreased." The above paragraph has been inserted into the revised results section.(see Page 6， line 147-150) " In a previous study in 1990 10  This phenotype is consistent with our conclusion. " The above paragraph has been inserted into the revised discussion section (page 11-12,

"Based on the evidence presented, the extent to which HTNV infects CD8 T cells in
infected patients is unclear and not very convincing.

1) .The crux of the authors' conclusion (that HTNV infects CD8 T cells and this is
relevant for disease progression) here appears to be mainly derived from Fig. 1B. However, there are several issues of concern with this figure. We have used the same method to test the HTNV-NP staining from uninfected (healthy) donors as negative controls as suggested, an experiment of which is shown as Figure.S4 in the revised submission (Page 5. Line 123-124).

2) "82.1% of all CD8 T cells in a patient being infected with HTNV seems
extraordinarily high. Indeed, the frequency decreases over time in Fig. 1B which is encouraging (although there is no graphical/statistical demonstration of this and appears to be for just one individual), but so does the activation level and phenotype of CD8 T cells which can potentially change the expression of nonspecific targets."

Answer:
As we described in the results (page 6 ， line 148-156), Figure 1D presents a representative of the proportion of HTNV-infected CD8+ T cells and CD4+ T cells in a particular sample. On average, 13.1% (range: 0.15% to 82.1%) of the total CD8+ T cells infected by HTNV were in the acute phase (red), which was still significantly higher than the 1.64 % (range: 0.05% to 15%) in the convalescent phase (P<0.001).

3) "Confidence in the staining would also be increased by sorting NP+ and NP-CD8 T cells and measure viral mRNA expression in the two populations through qPCR."
Thank you for your suggestion. We tried this method on freshly isolated patient H20-2 this year and found a viral load of NP-positive CD8 T cells was higher than that of NP-negative cells (Figure below). We will add this method to future research. The samples from previous studies last year have all been used for flow cytometry and cannot be sorted.
b. "Related to the above point, Fig. 2B provides some confidence that this process is occurring in vivo and the authors have nicely included PBMCs from an uninfected control individual. However, in Fig 2B, Fig. 1 and be fully confident that this is happening in vivo to the level the authors suggest."

the authors show two infected CD8 T cells, which appear to be 100% of the CD8 T (or at least 67%) cells in the field. There's no quantification of this for multiple individuals, so it's hard to compare this with
Answer: Fig. 2B is a representative graph. We have tested another group of patient's PBMC (H24-2)and found the staining was the same ( Figure S6). The degree of infection differs in patients with HFRS. Immunofluorescence results only play a representative role and are not convenient for statistics. However, flow cytometry is more representative. Fig. 1D, with a mean of 15.2%. In Fig.   2C-D, at 96h and ex vivo, it appears 100% of all CD8 T cells are infected. Overall,

it's hard to confidently grasp the level of CD8 T cell infection in vivo and in vitro and
many of these frequencies seem very high (especially as the frequency of HTNV in vitro infected monocytes was reported to be~1% in reference # 19 with nearly 100x MOI).

Answer:
In Figure 2D， we collected culture supernatants of infected primary CD8+ T cells at 0, 48, 72, and 96 h post-infection, added them to Vero E6 cells that are susceptible to HTNV infection, and then stained the cells for the HTNV-NP protein 7 days later.
The results indicating that infectious virions accumulated over time in the culture supernatants of CD8+ T cells. Combined with Figure 2C, the data demonstrate that HTNV can complete its replication cycle in CD8+ T cells. In HFRS patients, T cells are not only primitive, so they cannot be compared.
Very high infection levels were also observed in a previous study in 1990 10  The MOI in reference # 19 is 7.5 for PUUV which is very different from our HTNV. There are many MOI identification methods, and the virus quantification method of our research group has been used in many articles 4,5 . For infection, we did the MOI titration and MTT before, because of the small size of T cells, a virus of 10 7 TCID50/ml, the amount of virus needed by the 5×10 6 cells/well is 71.4ul. Based on reference 3-5 , MOI=0.1 is a suitable titer.
In this experimental study, we investigated HTNV 76-118 progeny production in primary T cells with three different MOI at determined time points. For infection, purified CD4 + and CD8 + T cells were plated at 5×10 6 cells/well in a 6-well plate and then exposed to HTNV at MOI=0.1,0.5,1 for 2 h. The results were analyzed by qPCR ( Figure.S2)

HTNV) and/or by comparing with infection of CD4 T cells (as the summary sentence in line 138 is that "HTNV preferentially infects CD8 T cells").
Answer: These are valid suggestions. After comparing HTNV susceptible cells HUVEC, THP-1, and A549, we found that the number of viruses entering the cell and their ability to replicate is different. We modified our wording and used "preferentially" instead of "specifically" in the revised text. (Page 6. Line 152)

The main conclusion that HTNV-infected CD8 T cells could be a useful biomarker
is also not very convincing as Fig 5B indicates

Answer:
The HTNV-infected CD8 T cells are a part of total CD8 T cells, Thrombocytopenia is a typical clinical manifestation of HFRS, while the increase of GMZB is closely related to CTL function, but the proportion of infected T cells can better reflect the relationship between the course of disease and prognosis. In viral infectious diseases, it is accepted that the proportion of CD8 T cells is related to the progression of the disease, but not all viruses can infect CD8 T cells directly, and it is related to disease course. This is also a new phenomenon that we have discovered. However, we appreciate the reviewer's suggestion and therefore change the papers' title from "predictive" of disease stages with strongly indicate the potential as a biomarker to a more modest "associated with" disease stages.  Fig. 2).

Answer:
To gain insight into the potential functional alterations occurring in HTNV-infected CD8 + T cells, we performed scRNA-seq using CD8 + T cells sorted by positive selection with magnetic beads from PBMCs. We then infected the cells with HTNV (MOI=0.1) for 48 h. Not all of the CD8 + T cells were infected, but a part of them did.

Answer:
We apologize for having not described this clearly before. In the revised paper, we have rephrased to accurately describe the study design.

Response:
We thank the reviewer for these thoughtful suggestions, and agree that Figure 4 as it currently stands is not conclusive. We therefore take the suggestion to remove this figure from the revised paper. Follow up studies will be performed to examine this question more thoroughly.
As a matter of scientific interests, we would like to elaborate a bit on the rationale behind making such a decision. There are notable technical difficulties that led us decide not to pursuit this question immediately. Figure 1 shows the expression level of HTNV-NP protein as detected by flow cytometry in the form of high and low fluorescence intensity. Whereas the single cell sequencing technique quantifies high and low mRNA levels. To detect "low level" expression by either technique is difficult, draw comparison between the two low levels is not likely to yield convincing results.
Even though we tried a new protocol (Singleron GEXSCOPE® protocol) to capture infected cells, only a small number of viral infected cells were captured, making it very hard to directly analyze the transcriptome differences between infected and uninfected cells.
As the reviewer correctly pointed out, even large single-cell sequencing companies such as 10x and Singleron are unable to solve this problem satisfactorily. .
Based on above, we think not to include Figure 4 is a sensible thing to do at this moment.

"Related to my initial concerns on the robustness of data demonstrating HTNV infects CD8 T cells, the authors have done a nice job of validating the results with inclusion of staining of healthy controls and qPCR of sorted HTNV infected cells. However, this qPCR data (included in the rebuttal) should be included in the manuscript as it
validates their approach."

Response:
We thank the reviewer for this suggestion. We have added the referred results as Fig.   S3 in the revised manuscript. These results clearly show that viral load of NP-positive CD8 T cells was much higher than that of the NP-negative cells. A more detailed description of this was added to the revised results section (Page 6, line 164-169), as the follows: "We then sorted NP + and NP -CD8 + T cells prepared from freshly isolated PBMC from patient H20-2, and measured viral mRNA levels in these two cell populations using qPCR.Results showed that the viral load of NP-positive CD8 T cells was much higher than that of the NP-negative cells (Supplementary Figure 3). "