Single-cell transcriptome profiling and the use of AID deficient mice reveal that B cell activation combined with antibody class switch recombination and somatic hypermutation do not benefit the control of experimental trypanosomosis

Salivarian trypanosomes are extracellular protozoan parasites causing infections in a wide range of mammalian hosts, with Trypanosoma evansi having the widest geographic distribution, reaching territories far outside Africa and occasionally even Europe. Besides causing the animal diseases, T. evansi can cause atypical Human Trypanosomosis. The success of this parasite is attributed to its capacity to evade and disable the mammalian defense response. To unravel the latter, we applied here for the first time a scRNA-seq analysis on splenocytes from trypanosome infected mice, at two time points during infection, i.e. just after control of the first parasitemia peak (day 14) and a late chronic time point during infection (day 42). This analysis was combined with flow cytometry and ELISA, revealing that T. evansi induces prompt activation of splenic IgM+CD1d+ Marginal Zone and IgMIntIgD+ Follicular B cells, coinciding with an increase in plasma IgG2c Ab levels. Despite the absence of follicles, a rapid accumulation of Aicda+ GC-like B cells followed first parasitemia peak clearance, accompanied by the occurrence of Xbp1+ expressing CD138+ plasma B cells and Tbx21+ atypical CD11c+ memory B cells. Ablation of immature CD93+ bone marrow and Vpreb3+Ly6d+Ighm+ expressing transitional spleen B cells prevented mature peripheral B cell replenishment. Interestingly, AID-/- mice that lack the capacity to mount anti-parasite IgG responses, exhibited a superior defense level against T. evansi infections. Here, elevated natural IgMs were able to exert in vivo and in vitro trypanocidal activity. Hence, we conclude that in immune competent mice, trypanosomosis associated B cell activation and switched IgG production is rapidly induced by T. evansi, facilitating an escape from the detrimental natural IgM killing activity, and resulting in increased host susceptibility. This unique role of IgM and its anti-trypanosome activity are discussed in the context of the dilemma this causes for the future development of anti-trypanosome vaccines.

zone and B1 populations are most strongly depleted, by conversion to short lived effector memory/plasma cells, (5) depletion of follicular B cells is not as complete, and previous studies may have over-estimated this due to CD23 down regulation, (6) IgG responses (particularly IgG2c) are less effective against the parasites than IgM, and may represent an evasion mechanism

Part II -Major Issues: Key Experiments Required for Acceptance
Reviewer #2: This is a very interesting study that clarifies several important aspects, however some conclusions are based on relatively little data and some improvements in data and writing are needed. As the data are mostly observational, the experiments demonstrating functional mechanisms should be strengthened.
1) The data supporting the functional importance of natural IgM in vitro is limited. To support the conclusion that IgM is more effective than IgG, isolated IgM and IgG would need to be compared. Figure 6C.

Just for information to the reviewer, the original published data of that experiment are included below.
Naïve/post-infection serum should be compared. Is AID plasma more effective due to different quantity or quality of IgM Ab or other reasons?

Comment: this new representation of Fig 6B has not been included in the new manuscript. This was only done to have a visual representation of the answer to this question. Both graph lines are however part of the overall Fig. 6B as was the case in the initial document (result description lines 260-263).
2) The data supporting the functional importance of natural IgM in vivo is limited. Plasma transfer experiments should compare wild type and AID KO naïve/post-infection serum.

All experiments have been repeated, now including controls and 7dpi data, confirming that plasma from AID -/mice does have an improved activity when it comes to controlling early stage T. evansi parasitemia.
3) Line 272/73 suggest that IgG2c is "detrimental". Implying that IgG2c is somehow detrimental to the immune response does not seem justified. The data address whether expanded natural IgM is beneficial, but there is not data addressing whether IgG2c is detrimental. To support line 289, where is the evidence that IgG2c is a "decoy mechanism"? This conclusion is repeated several times, and in lines 435 and 460 another mechanism "competition for antigen" is suggested (even though the authors find that the "vast majority of plasma cells are IgM+" on line 385). As there is no evidence in the manuscript for any of these negative effects of IgG2c, these should not be stated as conclusions but as possibilities for future investigation. For this reviewer, the important challenge to focus on is how to boost or maintain IgM responses prior to/during infection, not how to suppress so called detrimental IgG responses.

Part III -Minor Issues: Editorial and Data Presentation Modifications
Reviewer #2: 1) Line 123. Instead of arbitrarily mentioning day 42, describe the survival curve showing when the mice start to die

Day 42 was chosen for the main late stage flow cytometer and scRNAseq experiments, as we never had a mouse that had succumbed to infection prior to this date in any of the presented or preliminary executed experiments. The survival curve that was present in the first version of the paper has now been described in a different wording (lines 122-123). We have now also clearly indicated the rational for the 42dpi choice for scRNAseq, in lines 140-142.
2) Line 127/28 -5-10 fold increases in IgG isotypes after infection cannot be described as low, just less than IgG2c. To justify the focus of IgG2c, the data comparing anti-VSG titers for different isotypes should be shown. Figure S1,

shows the robust effect on induction of IgM and IgG2c during T. evansi infection. For IgG1, IgG2b and IgG3, the induction was only observed after 4 weeks post injection, and never reached the level of either IgM or IgG2c. The word 'low' has been removed from the text (Line 128).
3) Starting with Fig 1, complete flow cytometry gating strategies need to be illustrated for all cell populations reported. Figure S6 indicating all gating strategies that we used for our flow cytometry analysis. 4) Line 131 -"decline in cell numbers" Description of flow cytometry data needs to consistently refer to altered cell frequencies, unless referring to absolute cell number data. It's not acceptable to bury absolute cell numbers in supplementary data and not even mention it. I think this should be included in the primary figure and explained clearly as it's critical for the reader to understand what's going on. Figure S2C). We have also adapted our description in main text (lines 129-135).

In the initial manuscript, we had used cell percentage in Figure 1E to show consistency between all results, especially when comparing flow cytometry and of scRNAseq analysis (of which the latter always refer to portion of cells). However, we fully agree with reviewer that referring cell frequencies as "decline in cell numbers" can causes confusion. To address that issue, we have swapped the cell percentage graph from figure 1E to figure S2C and replaced it with graph showing absolute cell number during infection (previously shown in
5) The absolute number data include erythrocytes but method for spleen cell preparation, counting and flow cytometry gating for this analysis are not explained. Figure S6F.

In our analysis, we used a graph-based clustering algorithm which is developed by the Seurat team in Seurat v3 package (Stuart, Butler, et al., Cell 2019). That approach is heavily inspired by PhenoGraph's principle (PhenoGraph, Levine et al., Cell, 2015) in which a K-nearest neighbor graph based on the euclidean distance in principle component analysis space is constructed, and the edge weights between any two cells is refined based on the shared overlap in their local neighborhoods (Jaccard similarity).
7) Line 176 -this concluding statement doesn't make sense as flow cytometry data for several populations have not been shown at this point We agree with the reviewer, and have removed this sentence.

We agree with the reviewer that the description in the initial version of the paper could have been confusing and we have re-phrased the sentence. (Line 228)
10) Line 360 "short-lived atMBCs" -what is the evidence they are short-lived?

The term 'short-lived' was not meant here is a de facto reference to the lifespan of these cells (which we have not assessed), but refers to the standard nomenclature used for the population. (Pérez-Mazliah et al. eLife 2018; Knox et al. Immunol Rev. 2019).
11) Line 379 -I don't think histological evidence of absence of GC has been clearly shown. The immunofluorescence images shown do not demonstrate this. Staining such as GL7/IgD/AID would be needed, and comparing infected mice spleen with positive control samples containing GC structures would be needed.

We fully agree with the comment of this reviewer and have rephrased the data description referring to the observations that relate to the destruction of follicular structure, rather than GCs. (Lines 34, 221, 232, 389)
12) Line 405 -how can AID deficiency affect response to IgG-inducing signals? More accurate to say AIDdeficient cells can't perform class switch recombination

We agree with the comment of the reviewer and have rewritten the text as suggested (Line 426-427), removing the term 'IgG-inducing signals'.
13) Is the single cell RNAseq data from a single mouse per time point or pooled cells from multiple mice?

We have included in our Materials and Methods, under "Single cell RNA sequencing" section (line 733), that the data was obtained from a single non-pooled sample, in contrast to the flow cytometer were multiple biological (non-pooled) replicates were used to support all the conclusions. (See also comment number 17)
14) Fig 3D -are the gene functional groups shown the most statistically significant ones, or selected based on other criteria? Figure 3D Fig 4E and F aren't related to A-D parts, thus it's not clear why these should be combined. I felt unclear on the takeaway conclusions regarding B1 cells -unlike MZ cells it seems they are expanded and activated rather than reduced at day 14? This reviewer was not familiar with the markers used to identify the B1 cell population. It would be helpful to compare with flow cytometry analysis of B1 cells as done for all of the other populations. Figure 4 (B1 and atMBC) Figure S5A.

All data in bar or line graphs is represented as the mean + SD value of 3 or 5 mice in one of three independent experiment or as mean + SD of 9 values combined. The survival curves represent combined data of three independent experiments. All specific information is added to each figure legend. The scRNAseq experimental results were obtained on one experiment without combining samples, as we are dealing with a non-cloned non-synchronized infection. However, virtually all numerical data obtained in scRNAseq is backed-up by 9 individual flow cytometry observation, mirroring all conclusion and allowing for statistical validation of conclusion with respect to cell population sizes.
18) Numerous typos and wrong words were noted throughout and need to be corrected.

We have tried to correct all errors to the best of our capacity.
Reviewer #3: This is an important manuscript and deserves publication.
There is a long history of work on approaches to a vaccine for trypanosomiasis. Often the work on the pathogen has become divorced form the immunology of the host response. This type of manuscript is well grounded in looking at a realistic animal model (no such models are perfect and some have been shown to be misleading, but this is reasonable) and in using a non-clonal population of a recent field isolate of the parasite.
The approaches used are complementary and produce an interesting result. Importantly, this result may well have general applicability in other pathogen infections such as malaria. The result -that trypanosomes drive an early-stage IgMs/IgG2c switch and terminal PC differentiation followed by reduction of B-cell compartment (little replenishment) -provides a now way of thinking about this infection.
This concept trypanosomes drive immune maturation to evade high avidity IgM interactions and hence subvert the immune response is novel and raises an important concept. Moreover, it is important that the parasite vaccine community is aware of this since focussing on only certain parameters such as general antibody levels etc may well not be good indicators of vaccine effectiveness in the field. Also, it provides a working hypothesis of why bacterial vaccines might fail in tryp infected animals.
This data along with the hypotheses outlined are important and will greatly assist progress of the field. They provide a new set of ideas.

Part II -Major Issues: Key Experiments Required for Acceptance
Reviewer #3: Overall I have no major points here. The authors use a number of different approaches ...cell analysis, population analysis and markers, plasma transfer, knockout mice to reach a holistic conclusion. I am not an immunologist but in this type of work I think one needs to focus on the holistic rather than always asking for more detail.
We once again thank that the reviewer highlights some important issues here, also relating to the apparent of differences between cloned T. brucei and T. congolense lab strains, and field-related isolates. As mentioned above, these reviewing comments will encourage us to further pursue this line of research.

Part III -Minor Issues: Editorial and Data Presentation Modifications
Reviewer #3: POINTS: The paper will be of interest to immunologists, parasite immunologists and those interested in the parasite. For the latter the complexities of immunology language use could be clarified in places. For instance in places IgM subtypes are defined but in other places the term "natural IgM" is used. I understand the general meaning of natural IgM and maybe it is just me -but I had a little difficulty understanding which population of IgM was being referred to …for instance on page 10…the terms Natural IgM, VSG binding IgM, IgMs are used. 43, 58, 302, 414, 416, 449,463,477. There are a number of English usage / typos that would be usefully amended before publication: for instance particularly in the discussion -Line 332 treat?

We thank the reviewer for this comment and have tried to use this to improve the readability of the manuscript. In the new document we have tried to be as consistent as possible. Textual alterations include in line
Line 360. lallter?
Line 400 requires result?
Line 417 require to speculate?
We have tried to correct all errors to the best of our capacity.