Extracellular bacterial lymphatic metastasis drives Streptococcus pyogenes systemic infection

Unassisted metastasis through the lymphatic system is a mechanism of dissemination thus far ascribed only to cancer cells. Here, we report that Streptococcus pyogenes also hijack lymphatic vessels to escape a local infection site, transiting through sequential lymph nodes and efferent lymphatic vessels to enter the bloodstream. Contrasting with previously reported mechanisms of intracellular pathogen carriage by phagocytes, we show S. pyogenes remain extracellular during transit, first in afferent and then efferent lymphatics that carry the bacteria through successive draining lymph nodes. We identify streptococcal virulence mechanisms important for bacterial lymphatic dissemination and show that metastatic streptococci within infected lymph nodes resist and subvert clearance by phagocytes, enabling replication that can seed intense bloodstream infection. The findings establish the lymphatic system as both a survival niche and conduit to the bloodstream for S. pyogenes, explaining the phenomenon of occult bacteraemia. This work provides new perspectives in streptococcal pathogenesis with implications for immunity.

Page 22, Line 18: Check sentence. Video 6,7: "podoplanin (magenta) -positive subcapsular and medullary sinuses": I would clarify that podoplanin is also staining fibroblastic reticular cells in the interior of the node. Not all podoplanin positive cells are lymphatic endothelium.
Reviewer #2 (Remarks to the Author): An excellent work describing the initial phases of within host spread of S. pyogenes, with detailed investigation of early time points in multiple organs; importantly testing more than one bacterial strain. The surgical skill and the quality of microscopy, in particular intra-vital imaging is exceptional. The definition of the central role of the GAS hyaluronic acid capsule and in particular the bacterial chemokine protease in this phenotype is a great plus to this research work.

Specific comments:
Page 6 line 19 if you have no experimental data relative to CD44 mediated adhesion, please remove reference to Cd44 here and maybe refer to it with a reference in the discussion. Page 11 line 20 the observation of about a quarter of lymph node apparently occupied by GAS possibly lysing surrounding cells (Fig 5e), would probably warrant a stronger term then "prominent foci". Is this an abscess like situation? Quite acute for an abscess, but neutrophil influx and cell lysis would point to an abscess or extended necrotising event.
Page 13 line 13 In the legend of video 9 it is stated that some bacteria appear to be localised within CD169+ cells and the video really appears to show this. It might be appropriate at least to mention it in the main text, as not everybody will read the legend to video 9.
Page 13 line 14 the lack of any effect of clodronate on the infection shows clearly lack of macrophage involvement. I may have missed it, but do the authors show that clodronate actually depleted the macrophages form the lymph nodes. Maybe a macrophage staining of a lymph node or spleen after clodronate treatment would clarify this.
Page 13 line 20. The conclusion of the sentence indicating that hypervirulent capsule/protease related phenotypes are responsible for macrophage inefficiency is not clear. In vivo normal GAS is already evading macrophage control of lymph and you do not show any localised hyperproduction of capsule or protease by normal GASs. I therefore do not see the causative link to macrophage inefficacy in normal GAS infection.
Page 15 line 13 at the beginning of the paper you build all your story of lymphatic spread of GAS on specific interaction of the hyaluronic acid structure with LYVE-1. Now you are stating that any capsule, irrespective of its molecular structure, does the same job. This needs some better or additional explanation, as the relative paragraph in the discussion does not really clarify.
Reviewer #3 (Remarks to the Author): In their manuscript, "Lymphatic metastasis of virulent extracellular bacteria drives systemic infection" (NComm19-38830), Siggins and colleagues report that S. pyogenes (SP) disseminate from a local site of infection via lymphatic conduits to a sequential series of draining lymph nodes (LN) and ultimately to the bloodstream. This transiting occurs via extracellular mechanisms (i.e. not shielded within phagocytes) and resists clearance by phagocytes. They identify streptococcal virulence mechanisms that facilitate this spread, particularly the hyaluronan capsule and a chemokine-cleaving protease, SpyCEP, which can inhibit PMN recruitment to a site of infection. They conclude by claiming this lymphatic spread mechanism is not unique to SP, but also is true for other "pathogenic extracellular bacteria" thereby invoking a "new perspective" in bacterial pathogenesis.

Major comments:
These investigators present convincing data that a local SP infection (mouse hindlimb) travels to an ipsilateral draining LN through a sequential series of LN to spleen and liver. Once the bacteria appear in the bloodstream at 24 hrs, the bacteria are detected in the contralateral LN. These data are supportive of their hypothesis that SP can disseminate beyond a local site of infection via lymphatics into the blood stream These data were acquired by technically demanding lymphadenectomy of inguinal, iliac, axillary, and brachial LN on the ipsilateral and contralateral sides of 3-6 mice per group. They also used fluorescent SP and confocal intravital microscopy to demonstrate the SP were not within phagocytes as they traversed the lymphatics, and provided video support. They show that most of the leukocytes within the lymphatics were not CD11b phagocytes but rather CD3+ T cells and rarely entered the LN parenchyma. SP also were able to evade killing by CD169+ macrophages within LN.
My major concern is when these investigators expand their findings with SP to other "extracellular" bacteria and claim that this phenomenon is widespread and not a "quirk" of streptococcal pathogenesis. Here their data are far less robust. First, before they can make that statement, I would want to see many additional non-SP strains showing this behavior rather than one strain of each species. Their studies with SP appear to have a particular relevance in that with SP soft tissue infection there is clinically evident lymphangitic spread, suggestive of lymphatic involvement. This is not the usual case with the Gram-negative bacteria (GNB) and perhaps for S. aureus as well. With regard to S. aureus, they are unable to show any spread at 6 hr (Fig 7) beyond the local site; however, USA 300 is very poorly virulent in mice and therefore would not be a good candidate for study. Thus, they cannot make any generalization about the role of capsule based on the confinement of USA300 (which has no capsule) to the injection site With regard to the GNB (Klebsiella, Pseudomonas and E. coli) they state that these strains were clinical isolates obtained from the bloodstream of patients with soft tissue infections (see Methods-Bacterial Strains), but there is little additional information, including clinical description of the soft tissue infection. The strain designation in the absence of additional information on the capsule or O serotype provides no help. (Extended Data Table 1 lists two strains from "BioAID collection"-what is that??) They also do not offer any evidence of the role of virulence factors for these strains as they did for SP. For example, do we know anything about the capsules of either EC or KP (Pseudomonas does not have a capsular polysaccharide)? Does the anti-lymphatic endothelial receptor -(LYVE-1) also play a role for these other bacteria?
Having shown their technical expertise in delineating the lymphatics and isolating cellular elements within the lymphatics, it would have been quite illuminating for them to examine the lymphatic fluid. For example, is there immunoglobulin within the lymphatics?? In their discussion they state (p. 16, l.25) that the lymphatics do not "routinely allow every trivial bacterial infection" to develop into a serious systemic disease. Might they speculate on what types of protective mechanisms may come to bear? If SP secrete superantigens, why is there not a greater immune response within the LN given the large number of T cells and antigen-presenting cells?? Certainly if the authors suggest that the persistent stimulus of superantigen-secreting SP might induce autoimmune disease, there should be some evidence of immune response (e.g. germinal center formation).

Specific Comments:
Given my comments above, the title should be changed to substitute "S. pyogenes" in place of "extracellular bacteria". They should also limit the conclusion in the abstract to SP and not "pathogenic extracellular bacteria", and new perspectives on SP pathogenesis rather than "bacterial" pathogenesis. They should explain why they want to identify podoplanin-are they citing it as a specific lymphatic vessel marker or as a link between cardiovascular and lymphatic systems? p. 9. Interestingly, the hyaluronan capsule's role in virulence appears to be not so much in evading phagocytes but in retaining SP within the lymphatics based on their interaction with LYVE-1-is that correct? If so, this would be different from the role of capsule in Klebsiella and E. coli. Is there any evidence of germinal center formation within the lymph nodes transited by the SP? How did the SP leave the lymphatics at 24 hr and enter the LN parenchyma?? Author response: Thank you for this positive summary of our work and for your helpful comments.
We have made the changes suggested and agree that they further strengthen the manuscript. Point by point responses and details are given below. values are commonly observed using this model and we have no reason to suspect they are outliers.
We have performed similar experiments before and obtained comparable results such as shown in Figure R1. Furthermore, the choice of statistical test employed (Mann-Whitney rank sum U test) minimises the impact of these '0' values, as it only compares the mean of ranks, rather than the actual values. Hence, we are confident that the data are robust and reproducible. Author Response: In addition to previously performing a similar experiment with LYVE-1 blocking antibody (Fig. R1), we have also used LYVE-1 knockout mice for the same experiment (Fig. R2). Both methods gave similar results -albeit with a different bacterial strain -that are fully in line with the data obtained for this manuscript. Therefore, we are confident that the antibody delivery is sufficient and the results definitive. The original experiments shown in Fig. R1 and R2 were performed prior to our appreciation of the mechanism of S. pyogenes lymphatic metastasis, and so at the time we did not assess the distant draining axillary lymph node, which necessitated repetition with the inclusion of the axillary lymph node in this work. The findings of these experiments were reported in Lynskey et al 1

Fig. R1. LYVE-1 functional blockade reduces GAS dissemination to draining lymph nodes. Dissemination of M18 GAS in murine soft-tissue infection following LYVE-1 mAb blockade (n = 22/group). Quantitative culture of GAS at site of infection (A), ipsilateral draining LN (B), spleen (C) and blood (D) 3 h post-infection. Lines depict median values in each case
(Mann Whitney; * = p<0.05, ** = p<0.01). (Control antibody group: circles indicate isotype control antibody and triangles indicate polyclonal control IgG). Author Response: Thank you for pointing out this potential ambiguity, this has now been clarified. Author Response: Thank you for this suggestion. We now include images from mice infected with hypervirulent or SpyCEP-deficient S. pyogenes in Extended Data Fig. 5. The low magnification images (panels d and g) demonstrate the greater bacterial: neutrophil ratio in mice infected with the hypervirulent SpyCEP+ strain compared with those infected with the SpyCEP-deficient strain. In contrast to mice infected with the hypervirulent parent strain, comparatively strong recruitment of neutrophils was observed in mice infected with SpyCEP-deficient S. pyogenes. Nevertheless, the bacteria were not completely controlled within the lymph node. As discussed above, the capability of SpyCEP to drive bacterial systematic spread was apparent when expressed in L. lactis (an avirulent, readily phagocytosed bacterium) (Extended Data Fig. 5i). Hence, SpyCEP does indeed affect both lymphatic dissemination and systemic bacterial burden, but this effect will vary depending on bacterial strain. Author Response: Thank you, this sentence has been corrected. Author Response: Thank you for this suggestion. This is now explained with the sentence:

Fig R2. Genetic deletion of LYVE-1 impairs GAS dissemination to draining lymph nodes. Dissemination of M18 GAS in murine soft-tissue infection in constitutive LYVE-1-/-mice (n = 7/group). Numbers of GAS recovered at site of infection (A), ipsilateral draining LN (B), spleen (C) and blood (D) 3 h post-infection. Lines depict median values in each case
"Podoplanin staining in the interior of the node is due to the presence of fibroblastic reticular cells" in the legends for Videos 6 and 7.

Reviewer #2 (Remarks to the Author):
An excellent work describing the initial phases of within host spread of S. pyogenes, with detailed investigation of early time points in multiple organs; importantly testing more than one bacterial strain.
The surgical skill and the quality of microscopy, in particular intra-vital imaging is exceptional. The definition of the central role of the GAS hyaluronic acid capsule and in particular the bacterial chemokine protease in this phenotype is a great plus to this research work.
Author Response: Thank you for this extremely positive summary of our work and for your encouraging compliments regarding its quality.  Author Response: Thank you for this helpful suggestion. We fully agree that there appears to be extended necrosis and that our previous description of the infected lymph node understated the degree of streptococcal invasion. Accordingly, we have now changed the text to: 'extensive prominent foci of invading bacteria were evident deeper in the lymph node parenchyma, often forming large necrotic regions'. (Page 12; lines 2-4). Page 13 line 13 In the legend of video 9 it is stated that some bacteria appear to be localised within CD169+ cells and the video really appears to show this. It might be appropriate at least to mention it in the main text, as not everybody will read the legend to video 9.
Author Response: Thank you, we agree. The main text now includes reference to this: 'The overwhelming majority of S. pyogenes were neither sequestered nor internalised by CD169+ cells, with only extremely rare instances of colocalization'. Our analysis of many sections from numerous lymph nodes indicates that the bacteria are overwhelmingly extracellular. However, the high magnification required to discern sufficient detail of internalization, means that it is not possible to give a truly representative image or video. My major concern is when these investigators expand their findings with SP to other "extracellular" bacteria and claim that this phenomenon is widespread and not a "quirk" of streptococcal pathogenesis.
Here their data are far less robust. First, before they can make that statement, I would want to see many additional non-SP strains showing this behavior rather than one strain of each species.
Author response: Our primary purpose was indeed to undertake a detailed investigation of lymphatic dissemination of S. pyogenes-precisely because of the clinical observation that bloodstream infections sometimes develop with no obvious portal of entry, and because of the manifest association between S. pyogenes and lymphadenitis and lymphangitis. Nonetheless, it would be disingenuous to ignore the possibility that other bacteria might utilise similar routes; the size of bacteria renders passive transit in lymphatic fluid likely yet it is a niche that has been underinvestigated. The bacteria used were mostly clinical strains that had caused bloodstream infections; we do not necessarily expect all strains of these other species to survive lymphatic spread.
Nevertheless, in response to the Reviewer's concerns over this issue, we have expanded our analysis of lymphatic spread of non-S. pyogenes bacteria (Fig. 7). The figure now includes data from 13 unique bacterial strains that represent 6 different species of non-S. pyogenes extracellular bacteria (3 Gram positive, 3 Gram negative). This is in addition to examination of 6 S. pyogenes clinical isolates and 5 isogenic mutant strains. Furthermore, to maximise relevance to human disease, we focussed on use of clinical, mostly bacteraemia isolates (CC15 and C22 strains of S. aureus; ST88 and ST131 strains of E. coli; and serotype Ia and V strains of S. agalactiae (GBS) (Fig. 7 and Extended Data   Fig. 5i). The data show that lymphatic-dominated spread can occur early during infection with a diverse range of bacteria; whether such dissemination has clinical consequences is unclear but could be of importance when considering development of cognate immunity or autoimmunity. We hope the Reviewer will understand that such investigations are beyond the scope of the current work.

Their studies with SP appear to have a particular relevance in that with SP soft tissue infection there is clinically evident lymphangitic spread, suggestive of lymphatic involvement. This is not the usual case with the Gram-negative bacteria (GNB) and perhaps for S. aureus as well.
Author response: S. pyogenes was chosen as the focus of the study precisely because of this clinically evident association. We agree that other extracellular bacteria do not exhibit this feature in humans, but, as outlined above, we felt that it was important to determine if other bacteria might demonstrate the ability to enter the lymphatic system and survive the early stages of an experimental infection. We have now highlighted the distinction between S. pyogenes (that does have a relationship with clinical lymphatic syndromes) and other bacteria (that do not) in the manuscript; the distinction may well related to the unusual interaction of S. pyogenes with LYVE-1 or a host-specific difference in ability to clear different types of bacteria.
Notwithstanding these caveats, viable Gram negative bacteria have been recovered from human lymph nodes in patients undergoing surgery for cancer 2-6 and were associated with increased risk of postoperative sepsis, though the routes of dissemination are poorly understood 4,6 . This is discussed briefly: (Page 21; lines [18][19][20][21]. As such, there is some clinical evidence that lymphatic spread of other bacteria could be of relevance to humans. It should be noted that the use of a soft tissue intramuscular hindlimb infection model across all bacteria is to allow analysis of a defined route of lymphatic drainage, as well as to give comparative consistency to experiments. We are not suggesting that all bacteria assessed would naturally first cause soft tissue infection and then routinely disseminate through the lymphatics to drive systemic spread. The focus of these experiments is to assess whether these bacteria, once given partial access to the draining afferent lymphatics (via i.m. injection), are able to pass through sequential draining lymph nodes in order to drive systemic infection. In natural infection, initial access to the lymphatics could involve many different routes. For example, during tonsillitis, S. pyogenes could transit in efferent lymph from the palatine tonsil to tonsillar lymph nodes and beyond.
With regard to S. aureus, they are unable to show any spread at 6 hr (Fig 7) beyond the local site; however, USA 300 is very poorly virulent in mice and therefore would not be a good candidate for study.
Author response: We thank the Reviewer for making this valid point and have now undertaken assessment of lymphatic dissemination using two additional S. aureus strains (Fig. 7f), both of which are clinical bloodstream isolates from STs that represent the top five causing bacteraemia in the UK.
These strains were able to reach local draining lymph nodes more efficiently than USA300, though still exhibited poor metastatic ability overall, compared to most other bacteria assessed. The text has been amended to describe these important new data and the data included in a fully revised Fig. 7

(Discussed Page 16; lines 3-7).
Thus, they cannot make any generalization about the role of capsule based on the confinement of USA300 (which has no capsule) to the injection site.
Author response: Thank you for highlighting this. On reflection, we agree that our comments regarding the contribution of capsule in the poor lymphatic spread of USA300 were overstated in the original manuscript. Accordingly, to address the Reviewer's concern, reference to the capsule of USA300 has been removed from the manuscript as a rationale for the findings.  7 . The cohort study protocol is now referenced in the manuscript. Extended Data Table 1 has been updated to include the additional contemporary clinical strains used in response to the Reviewer's comments and provides the maximum amount of clinical detail available. While contemporary clinical isolates lack the depth of phenotypic characterisation of well-studied laboratory strains, we believe their greater physiological relevance justifies this trade-off.

Strains), but there is little additional information, including clinical description of the soft tissue infection.
Additionally, we are cautious about the potential for misdirecting readers or prompting overinterpretation by giving undue prominence to phenotypic details of the bacteria used that may or may not be relevant: a point already conceded above regarding the acapsular nature of USA300.
They also do not offer any evidence of the role of virulence factors for these strains as they did for SP. For example, do we know anything about the capsules of either EC or KP?
Author Response: The non-S. pyogenes bacterial strains assessed in the paper were selected based on their clinical association with bacteraemia, rather than their production of capsule/extracellular polysaccharides or other virulence factors. The hypothesis we set out in the Discussion, that the physical properties bestowed by capsule/extracellular polysaccharides may be a prominent driver of lymphatic metastasis, is included as a basis for future exploration, rather than a mechanistic focus of the paper. Therefore, the virulence factors -which are not yet fully characterised in these clinical isolates -are not explored in the manuscript.
However, the identities of such virulence factors are currently under investigation. [Redacted] [Redacted] A comprehensive mechanistic study of the numerous bacterial species and strains that we have evaluated, conducted to the same high standards as undertaken for S. pyogenes, is a significant undertaking that itself would likely span several years work. We believe that attempting to amalgamate additional data concerning different virulence factors from other bacterial species would compromise understanding of the current manuscript, which is already data-rich and contains exciting and impactful findings. We hope that after reviewing our revisions that you will also share this opinion. We anticipate and hope that increased awareness of the extent of extracellular bacterial lymphatic metastasis in our model will lead other groups to consider and explore the phenomenon and consequences in their own bacterial infection and immunity models.

(Pseudomonas does not have a capsular polysaccharide)
Author Response: Thank you for highlighting this important distinction: that the alginate, Pel and Psl production by P. aeruginosa does not constitute a proper capsule. Accordingly, we have amended the text to include reference to 'extracellular polysaccharides' (EPS). Notably, hyaluronan is not covalently linked to the S. pyogenes cell surface either. Yet, as it is invariably referred to as 'hyaluronan/hyaluronic acid/hyaluronate capsule' in the literature, we will maintain convention to prevent confusion. Author Response: Thank you for this interesting suggestion. Though we share your opinion that analysis of the lymph could be valuable, we have not been able to achieve this experimentally for the flank lymphatics. During an immune response, immunoglobulin has been shown to be present in efferent lymph collected from the thoracic duct 9 and Ig in lymph could feasibly influence phagocytosis and infection of bacteria transiting within lymphatic vessels. However, in our model of acute bacterial infection, using specific pathogen free mice (which have not been prior exposed to S. pyogenes) and early ≤24-hour time points, there would not be sufficient time to mount an antibody response and hence immunoglobulins (other than natural antibodies) would not be expected to play a major role. It is an important question however, in humans, where prior immunity may be an important factor, and where intravenous immunoglobulin is sometimes administered to combat invasive S. pyogenes infection.
In their discussion they state (p. 16, l.25)  Author Response: Our work strongly suggests an important role for neutrophils. This is most clearly illustrated by the transformation of avirulent L. lactis into a bacterium capable of lymphatic and systemic spread upon heterologous expression of the chemokine-cleaving protein, SpyCEP, which impedes neutrophil recruitment (Extended Data Fig. 5i). Human neutrophils, unlike murine neutrophils, produce antimicrobial defensins 10,11 , and any effect of neutrophils in preventing lymphatic spread and development of systemic disease in mice may be even stronger in humans.
Speaking more broadly, analysis of invasive streptococcal disease indicates that those with chronic disease and skin lesions are more at risk 12,13 , similarly host genetic factors 14,15 and prior blunt force trauma 16,17 are implicated. We would speculate that even in health, a low level of sporadic bacterial transit to lymph nodes might occur but would quickly be controlled by the host. However, under certain conditions, this balance might swing towards bacterial replication: i.e. a particularly virulent strain 18 able to spread in greater numbers and more quickly due to increased lymphatic flow and access to the lymphatics following tissue injury, in a host predisposed to severe infection through immune status or genetic factors. Of these genetic factors, HLA polymorphism has been suggested to influence the susceptibility to severe S. pyogenes disease 14,15 . In other streptococcal bacterial invasive disease, genetic variation in NFKBIL2 which encodes the proinflammatory transcription factor NF-κB inhibitor IκB-R has been implemented 19 . Author Response: Streptococcal superantigens are highly specific to human MHC II (HLA) and so do not mediate notable T cell mitogenicity or other immune effects in non-humanised mice such as those used in the current paper 20 . As this study analyses early timepoints in acute bacterial disease, we do not believe this limitation is of significance. It may represent an important barrier to subsequent investigations into later time points (days to weeks) and longer study of the effects of lymphatic spread may necessitate the use of HLA class II transgenic mice. The authors have previously shown a marked expansion of draining lymph nodes in HLA II transgenic mice that was specific to the superantigen SpeA; at the time, bacterial transit to the draining lymph node had not been expected 21 .
Specific Comments: Given my comments above, the title should be changed to substitute "S. pyogenes" in place of "extracellular bacteria". They should also limit the conclusion in the abstract to SP and not "pathogenic extracellular bacteria", and new perspectives on SP pathogenesis rather than "bacterial" pathogenesis.
Author Response: In response to Reviewer 3's earlier suggestion we have undertaken a significant number of additional in vivo experiments with multiple clinical strains representing six non-S.
pyogenes species, consisting of both Gram positive and negative bacteria. The majority of these bacteria exhibited the behaviour of lymphatic-dominated spread and the data point to a generalised phenomenon, rather than something limited to S. pyogenes. Therefore, we feel the use of 'extracellular bacteria' is a fairer representation of the data and it might seem misleading or disingenuous to suggest the phenomenon was restricted to S. pyogenes. Furthermore, the terminology of the title is consistent with Journal convention. and will help the data reach scientists who study non-S. pyogenes bacteria, who we believe would be interested in the results of our work and its application to other bacteria. They should explain why they want to identify podoplanin-are they citing it as a specific lymphatic vessel marker or as a link between cardiovascular and lymphatic systems?
Author Response: Podoplanin was employed to stain lymphatic endothelium to highlight lymph node sinuses. An explanation has been added to all relevant figure legends and in the text where relevant: 'Podoplanin staining highlights lymphatic endothelial cells in the subcapsular and medullary sinuses, as well as fibroblastic reticular cells in the cortex of the lymph node'. p. 9. Interestingly, the hyaluronan capsule's role in virulence appears to be not so much in evading phagocytes but in retaining SP within the lymphatics based on their interaction with LYVE-1-is that correct? If so, this would be different from the role of capsule in Klebsiella and E. coli.
Author Response: Our data indicates that hyaluronan capsule plays three roles that contribute to lymphatic dissemination. Firstly, evasion of uptake by phagocytes (Fig. 5, Fig. 6, Extended Data Fig.   5a, h, Extended Fig. 6c,d); Secondly, promotion of retention within the lymphatics through specific interaction with LYVE-1 1 (Extended Data Fig. 4d, e); and thirdly, potential exploitation of its physical properties of hydrophobicity and polyanionic nature to enhance lymphatic uptake (Extended Data  Fig. 5b, c, e, f). Death of host cells and disruption of endothelial barriers may allow bacterial spread deeper into the lymph node. Damage to cells is likely mediated by bacterial cytotoxicity 23 , though host-mediated damage from cellular responses and inflammation could also contribute 24 . Furthermore, there are many bacteria in the blood at 24 hours post infection (Fig. 1c) and these bacteria could be visible in the blood vessels running throughout the lymph node parenchyma, possibly also contributing to invasion.  [17][18][19][20][21]. This data is consistent with previous reports and supports an action of neutrophils in clearing L. lactis from both the original site of infection, and particularly metastatic infection sites. p. 20., Methods. It appears that the bacteria were grown overnight before use and hence in stationary phase. Was there any comparison of bacteria grown to log versus stationary phase??
Author Response: We thank the Reviewer for raising this potentially important issue. In response to their concern, we have now undertaken a comparison of bacteria grown to log versus stationary phase (Extended Data Fig. 2b). Lymphatic spread was shown to occur with both stationary and log phase bacteria.
Mice: In their methods they claim to use FVB/n, BALB/c and C57BL/6 mice, yet nowhere in their figure legends do they indicate the strain of mice. And why the three different strains??
Author Response: We apologise for this careless omission. The strain of mice used is now stated in each relevant figure legend. FVB/n mice were used unless otherwise stated, consistent with our previous work 1 . BALB/c mice were used for data presented in Extended Data Fig 2a, to confirm that metastasis occurred in another common strain of mice. Finally, C57BL/6 mice were used in some pilot studies and Extended Data Fig. 5i, based on availability. Ultimately, S. pyogenes exhibited lymphatic spread in these three strains of mice.
We agree that the differences observed between strains are likely due to the relative ability to clear each strain (i.e. strain virulence), as development of bacteraemia is a good indicator of relative virulence in mice. Assessment of LD50 is not within the scope of our licenses, and we believe that a more quantitative metric such as bacterial clearance would reproduce the variation already observed between strains. Nonetheless, the other bacteria did appear to use the lymphatic system-which was the point of extending the experiments to other species, rather than to understand the wider question of differences in virulence or the mechanisms that might allow other bacteria to reach lymph nodes.
We indicated to the Editor that we were willing to adjust the content of the paper and title as requested and have now done so. The manuscript has been retitled 'Extracellular bacterial lymphatic metastasis drives Streptococcus pyogenes systemic infection' and references to data from other bacterial species have been removed, bar a brief section in the Discussion. We plan to explore the lymphatic metastasis of other bacterial species in further depth in future work.
In my comments about examining lymphatic fluid for immune constituents, I was not so much interested in looking for evidence of an adaptive immune response, particularly since they were looking very early in the immune response, but wondering if the lymphatic fluid had natural antibody, complement etc.
Author response: Thank you for further clarifying your query. We are unable to sample murine flank lymphatics to provide a direct experimental answer to the question. There is limited literature detailing lymph concentrations of natural antibodies and complement. However, complement has been reported to be present in low-levels in lymph 1 . Natural antibodies are produced by spleen resident B1a lymphocytes and by B1 lymphocytes of bone marrow. As B-1 cells are present at relatively low levels in lymph nodes, we predict that the concentration of natural antibodies in lymph would be even lower than the low levels measured in plasma 2 . Despite this, it is entirely plausible that natural antibodies and complement within lymph could contribute to antibacterial defence, and this is an interesting area for further study.