PI3Kδ hyper-activation promotes development of B cells that exacerbate Streptococcus pneumoniae infection in an antibody-independent manner

Streptococcus pneumoniae is a major cause of pneumonia and a leading cause of death world-wide. Antibody-mediated immune responses can confer protection against repeated exposure to S. pneumoniae, yet vaccines offer only partial protection. Patients with Activated PI3Kδ Syndrome (APDS) are highly susceptible to S. pneumoniae. We generated a conditional knock-in mouse model of this disease and identify a CD19+B220− B cell subset that is induced by PI3Kδ signaling, resides in the lungs, and is correlated with increased susceptibility to S. pneumoniae during early phases of infection via an antibody-independent mechanism. We show that an inhaled PI3Kδ inhibitor improves survival rates following S. pneumoniae infection in wild-type mice and in mice with activated PI3Kδ. These results suggest that a subset of B cells in the lung can promote the severity of S. pneumoniae infection, representing a potential therapeutic target.

A well done and interesting paper suggesting there is a novel type of B cell population with Breg characteristics in the lung that the the PI3Kδ hyper-activation mutation increases in number and this causes susceptibility to pneumococcal infection. Overall the data are convincing and experiments well done, and succinctly described in an easy to follow way. I do have some points / issues to raise: Major points: 1. I really think the data need an infection experiment in the knockin mice v wild-type experiment with IL10 depletion via antibody; this is necessary to directly link the new type of Breg population to the increased susceptibility to pneumococcus. An alternative would be to do this in B cell depleted mice using antiCD20 antibody or similar if this population expresses this.
2. most of the difference in survival occurs after 24 hours (I think, difficult to tell as the x axis goes out to 700 hours) and there is no clear mechanism in the paper as to why survival is different; except the interesting data for the Ighmtm1 mice. So I do think we need bacterial CFU and histology for (a) later time point(s) when the survival curve is clearly different, plus cytokine levels just to get a feel for what is happening. The histology is especially important given that the mechanism should be some sort of immuno-dampening effect. Although it will probably take more experiments than this to know for sure what the mechanism of improved survival is, I think that is beyond the scope of this paper; just some data to give a feel for what is happening would be important otherwise it is all a bit too mysterious Minor points: 3. a lot of supplementary fig 7 I think needs to be in the main paper: bacterial CFU at 24 hours, total IL10, cell specific IL10 especially the B cell data), effects of nemiralisib on IL10 levels and the volcano plot. It looks like the total IL10 levels are not much affected which weakens their case a bit (but timing / localisation etc may make it hard to find a clear difference in IL10 in lung homogenates) 4. the x axis for the survival curves should be shortened / made into a broken x axis as the tail of parallel survival does not help and we need the detail for the earlir timepoints. 5. Fig. 1 S pneumoniae strain type and inoculum size need stating as it does in other infection experiments 6. line 145 re. p110δD910A mice not having increased susceptibility, in these experiments the mice immune are naive to pneumococcus so we would not expect antibody to be protective, apart from natural IgM 7. the protective effect of T cell expression of p110δE1020K is very interesting any explanation for this effect? Also this protective effect could have cancelled out the susceptibility of B cell p110δE1020K mice; yet in fact the all cell p110δE1020K knock out has similar henotype to the B cell KOs alone. Any suggestions as to why the B cell defect is dominant for the phenotype of the total KO mouse? Some thoughts in the discussion line 306 would be useful 8. the exclusion of natural IgM as a mediator for the increased susceptibility would be strengthened if the authors measured IgM recognition of S. pneumoniae in sera taken from uninfected mice using flow cytometry 9. the Ighmtm1 phenotype is very interesting; again further development of this observations is probably beyond the scope of this paper Reviewer #2 (Remarks to the Author): The severity of pneumococcal infections depends largely on host´s susceptibility. Here, the authors indicated in a comprehensive study the a subset of B cells (CD19+B220-) induced by hyperreactivity of PI3Kδ. Upon this induction the host -here mouse in experimental designs and in a translational study patients suffering on APDS -are more susceptible against pneumococcal infection. Importantly, the translational and therapeutic aspect could be considered here, as a PI3K inhibitor -nemiralisib -was used and the results indicated that the inhibitor protects against fulminant pneumococcal infections. Noteworthy, the population of CD19+B220-cells secrets IL-10 in an expanded manner, which is consequently reduced by nemiralisib. However, the increased susceptibility due to the expansion of the B cell population occurs only in the first stages of the infection. In conclusion, this is an excellent study, perfectly designed and carried out and of broad interest as well.
specific comments: • the experiments are described in an excellent way and all information needed are provided. Here, it is also important to mention that the authors have not only provided the names of antibodies used but also all the other information needed to repeat the experiments. • the study defined the cell populations and IL-10 producing cells. Because not all readers are immunologists, the author can add some more information about the gating strategy engaged by their FACS analysis and in addition, explain the values (mostly as %) given as insets in the histograms or dot plots • Figure 2C and D: the Western blots are shown below the panel for stimulated and treated cells. However, one can only speculate that the blots are shown for stimulated cells and not for nemiralisib treated cells. Please clarify.
• The description of figure 3B is, at least for the reviewer, confusing. Please rephrase and improve the description of the data shown in Fig. 3B. Does n=5 mean 5 mice? • the labeling of some figures (e.g. as in Fig. 2, 7, 8 and supplementary figures) have to be explained in a better way and without repetitions. The meaning of the -and + is not always easy to follow and the symbols can be mentioned simply in the legend. • the WHO reference (ref 1) has to be updated The introduction is clear and focused. However, part of the introduction summarizes again the results of the study (starting in line 54). This is nice, but unnecessary. This part can simply be shorten and if desired, in parts shifted to the discussion.

Reviewer #3 (Remarks to the Author):
Overall this is a very interesting manuscript. Knockin mice with the p110 delta E1020K mutation equivalent to that seen in patients with CVID present with an expanded B220-CD19+ population which expresses IL-10 and also infiltrates the lungs. These cells are assumed to be transitional B cell-like and B-reg like cells and the authors implicate them in the susceptibility to more severe S. pneumoniae infection seen in these mice. There is considerable enthusiasm for these extensive studies but there are some relatively easily filled lacunae that should be addressed. The evidence that the B220-CD19+ population is pathogenic is indirect but the phenotype of these cells could be clarified in a couple of ways a) Do the CD19+B220-B cells express the AA4.1/CD93 marker at levels comparable to wild type murine transitional B cells? b) To partially rule out the possibility that these cells are generated primarily in a hyperactive Tcell dependent fashion, the frequency of these B220-CD19+ B cells should have been examined in the Mb-1 Cre version of the E1020K p100delta knockin mouse that the authors describe. It is formally possible that these B cells are not transitional but more plasmablast-like or activated B cells.
c) The discussion should make it clear that it has not been firmly established that the B200-CD19+ cells are pathogenic, though they well might be Shiv Pillai

Reviewers' comments:
Reviewer #1 (Remarks to the Author): A well done and interesting paper suggesting there is a novel type of B cell population with Breg characteristics in the lung that the the PI3Kδ hyper-activation mutation increases in number and this causes susceptibility to pneumococcal infection. Overall the data are convincing and experiments well done, and succinctly described in an easy to follow way. I do have some points / issues to raise: Major points: 1. I really think the data need an infection experiment in the knockin mice v wild-type experiment with IL10 depletion via antibody; this is necessary to directly link the new type of Breg population to the increased susceptibility to pneumococcus. An alternative would be to do this in B cell depleted mice using antiCD20 antibody or similar if this population expresses this.
Since IL-10 is a key regulatory cytokine and previous work showed that a complete lack of IL-10 leads to increased mortality following S. pneumoniae infection, we studied the effect of partial/local IL-10 depletion via the following two strategies.
1. Homozygous Il10 ITIB mice are IL-10 hypomorphic, therefore, we used heterozygous reporter mice for IL-10 detection. However, we exploited this property to achieve reproducible systemic downregulation of IL-10 levels. Infection of Il10 ITIB homozygous p110δ E1020K and WT mice as well as Il10 ITIB WT p110δ E1020K and WT mice revealed reduced serum IL-10 levels in Il10 ITIB homozygous mice at death (supplementary fig 7A). However, reduced systemic IL-10 levels did not alter the survival outcome of p110δ E1020K or WT mice compared to Il10 ITIB WT mice (supplementary fig 7B). 2. We considered that while systemic IL-10 may be protective, high levels of local IL-10 in the lung environment during early stages of infection could be detrimental. Therefore, we treated WT mice with daily intranasal anti-IL-10 JES5-2A5 (200µg). These studies showed a non-significant trend towards improved survival (supplementary fig 7C). However, when we depleted IL-10 from B cell conditional p110δ E1020K-B mice, they showed increased mortality compared to isotype control treated mice supplementary fig 7D.
These results lead us to conclude that, while IL-10 production is a key characteristic of the B220 -B cell population described here, it is not the main/only mechanism whereby these cells contribute to pathology during early S. pneumoniae infections. We chose the 24h time-point for CFU/cytokine analysis, because it is just before the mice start showing clinical signs. We believe the increased susceptibility of p110δ E1020K mice is due to a mechanism at work during the first hours after infection. For instance, the inhaled p110δ inhibitor nemiralisib had no effect when administered as soon as 8 hours after infection, but did improve disease outcome when administered from 24h prior to infection. Some animals, especially from the p110δ E1020K groups reach the humane endpoint from 48h onwards. This complicates the interpretation of results at later timepoints, because study groups contain individuals with a wide variation in health status which could affect experimental readouts independently of genotype. In addition, analysis at later timepoints will invariably select for individuals surviving to that timepoint. We have analysed lungs from mice by immunohistochemistry, but have not observed changes that shed further light on the mechanism of action. So far, using histology or flow cytometry, we have not observed significant differences in inflammatory infiltrates into the lung. Hence, we prefer to focus on the data that we have included in this manuscript which clearly highlights phenotypic and functional changes as consequences of altered PI3Kδ activity.

Minor points: 3. a lot of supplementary fig 7 I think needs to be in the main paper: bacterial CFU at 24 hours, total IL10, cell specific IL10 especially the B cell data), effects of nemiralisib on IL10 levels and the volcano plot. It looks like the total IL10 levels are not much affected which weakens their case a bit (but timing / localisation etc may make it hard to find a clear difference in IL10 in lung homogenates)
We have moved Supplementary figure 7 into the main text to become Figure 7. Given that the Nat Comm guidelines allow for up to 10 display items, we also moved supplementary figure 2 (B cell phenotype in p110δ E1020K-GL mice) to the main paper (now Figure 3).

the x axis for the survival curves should be shortened / made into a broken x axis as the tail of parallel survival does not help and we need the detail for the earlier time points.
We have looked into this but do not find that broken axis show the results more clearly. The majority of animals are culled between 48-72h and we can show a snapshot of survival proportions at this time (See figure 1 below). However, we believe that the survival curve is more informative if it shows disease progression throughout the duration of the study (10 days). The median survival values also show the kinetics of disease progression: Wild-type 71.1h; p110δ E1020K 58h, p110δ D910A 113.47h, indicating that p110δ E1020K mice have accelerated disease development.

Fig. 1 S pneumoniae strain type and inoculum size need stating as it does in other infection experiments
We used the strain TIGR4, serotype 4. We added this information to the figure legend of Fig 1 and  line 77 of the text. This is also indicated on page 5 line 144 and materials and methods (page 12, lines 412, 425).
6. line 145 re. p110δD910A mice not having increased susceptibility, in these experiments the mice immune are naive to pneumococcus so we would not expect antibody to be protective, apart from natural IgM The key point of interest here is that D910A mice are devoid of so-called natural IgM antibodies as illustrated in Fig 4B and 4C. We changed the order of the graphs in the figure and moved the paragraph describing natural antibody levels to line 147 to clarify the link between disease susceptibility and natural antibody levels.

the protective effect of T cell expression of p110δE1020K is very interesting any explanation for this effect? Also this protective effect could have cancelled out the susceptibility of B cell p110δE1020K mice; yet in fact the all cell p110δE1020K knock out has similar henotype to the B cell KOs alone. Any suggestions as to why the B cell defect is dominant for the phenotype of the total KO mouse? Some thoughts in the discussion line 306 would be useful.
This is part of an ongoing study. We raise a number of possible explanations in the discussion we prefer not to speculate further until we have additional data. Fig 3F (now fig 4 B and C) shows that anti-PC IgM antibodies are equivalent in WT and p110δ E1020K mice, but absent in p110δ D910A mice. Anti-PC antibodies are generally thought to provide protection against encapsulated bacteria, including S. pneumoniae and are present in naïve animals.

the Ighmtm1 phenotype is very interesting; again further development of this observations is probably beyond the scope of this paper
We were very surprised to find that Ighmtm1 (μMT) mice are resistant to S. pneumoniae mediated pathology. This is independent confirmation that B cells cause increased susceptibility to S. pneumoniae and that antibodies are not required for protection at early stages.

Reviewer #2 (Remarks to the Author):
The severity of pneumococcal infections depends largely on host´s susceptibility. Here, the authors indicated in a comprehensive study the a subset of B cells (CD19+B220-) induced by hyperreactivity of PI3Kδ. Upon this induction the host -here mouse in experimental designs and in a translational study patients suffering on APDS -are more susceptible against pneumococcal infection. Importantly, the translational and therapeutic aspect could be considered here, as a PI3K inhibitor -nemiralisib -was used and the results indicated that the inhibitor protects against fulminant pneumococcal infections.

Noteworthy, the population of CD19+B220-cells secrets IL-10 in an expanded manner, which is consequently reduced by nemiralisib. However, the increased susceptibility due to the expansion of the B cell population occurs only in the first stages of the infection. In conclusion, this is an excellent study, perfectly designed and carried out and of broad interest as well.
We are very grateful for these supportive comments.
specific comments: • the experiments are described in an excellent way and all information needed are provided. Here, it is also important to mention that the authors have not only provided the names of antibodies used but also all the other information needed to repeat the experiments.
Thank you for this comment. We very much hope the detailed methods will be useful to readers.
• the study defined the cell populations and IL-10 producing cells. Because not all readers are immunologists, the author can add some more information about the gating strategy engaged by their FACS analysis and in addition, explain the values (mostly as %) given as insets in the histograms or dot plots.
We have changed the figures to make the gating strategy more obvious. We added supplementary Figure 8 showing the full gating strategy for identifying B220 -B cells. We added information to clarify which populations were gated on in the figures and in the figure legends. We clarified the meaning of numbers given in pseudocolour plots in the figure legends.
• Figure 2C and D: the Western blots are shown below the panel for stimulated and treated cells. However, one can only speculate that the blots are shown for stimulated cells and not for nemiralisib treated cells. Please clarify.
The blots show phospho Akt in unstimulated cells, stimulated cells and stimulated cells that have been nemiralisib-treated. We agree that Fig 2A and 2B could be clearer and have changed these, but we believe that Fig 2C and 2D are clearly marked.
• The description of figure 3B is, at least for the reviewer, confusing. Please rephrase and improve the description of the data shown in Fig. 3B. Does n=5 mean 5 mice?
Fig 3B is a survival curve (n=22). Does the reviewer mean Fig 2B? In that figure and elsewhere in the manuscript, each data-point represents one biological replicate (e.g. one mouse). We updated the figure legends to reflect this.
• The labeling of some figures (e.g. as in Fig. 2, 7, 8  We find it is often useful to summarise the key findings at the end of the introductory paragraph, but will take the editor's advice on this.

Reviewer #3 (Remarks to the Author):
Overall this is a very interesting manuscript. Knockin mice with the p110 delta E1020K mutation equivalent to that seen in patients with CVID present with an expanded B220-CD19+ population which expresses IL-10 and also infiltrates the lungs. These cells are assumed to be transitional B celllike and B-reg like cells and the authors implicate them in the susceptibility to more severe S. pneumoniae infection seen in these mice. There is considerable enthusiasm for these extensive studies but there are some relatively easily filled lacunae that should be addressed. The evidence that the B220-CD19+ population is pathogenic is indirect but the phenotype of these cells could be clarified in a couple of ways a) Do the CD19+B220-B cells express the AA4.1/CD93 marker at levels comparable to wild type murine transitional B cells?
No, The CD19 + B220 -B cells in both WT and p110δ E1020K mice express intermediate levels of CD93