Does CXCR3 chemokine receptor expression by CD8⁺ T cells affect their moving towards or only their binding to virus-infected monocytes?

Hickman et al.¹ established a vaccinia virus (VV) mouse ear-infection model, and investigated the importance of CXCR3 expression by CD8⁺ T cells for their ability to locate and bind to infected cells, and to help eradicate the virus. The receptor CXCR3 can bind the chemokines CXCL9 and CXCL10, of which genes in this mouse ear model were found to be abundantly expressed by virus-infected tissue¹. Using a number of techniques, Hickman et al. assayed the migration of CD8⁺ T cells into the infected ears, including into dense areas of infected monocytes, and to the draining lymph nodes. Among the techniques used were intravital multiphoton microscopy (MPM) experiments, which allowed them beautifully to follow the migration of individual CD8⁺ T cells. Recipient mice included wildtype (wt; albino C57BL/6), CXCR3^-/- and CD8^-/- genotypes, while CD8⁺ lymphocytes used for grafting were isolated from CFP OT-I and CXCR3^-/- DsRed OT-I transgenic mice. The recombinant vaccinia viruses used were VV-NP-eGFP and VV-NP-S-eGFP, of which only the latter contains the CD8⁺ T cell determinant peptide sequence SIINFEKL which can be recognized in the MHC context of this system by the T cell receptor transgenically expressed in OT-I mice. Hickman et al. investigated a number but not all possible combinations of these mice, cells and viruses, and their different effects on virus replication and CD8⁺ T cell migration.

Findings by Hickman et al. were that (1) viral clearance by Cxcr3^-/- CD8+ T cells was somewhat impaired at intermediate times of infection, (2) migration of Cxcr3^+/+ and Cxcr3^-/- CD8⁺ T cells into infected ears and also to draining lymph nodes was similarly efficient, (3) both Cxcr^+/+ and Cxcr3^-/- CD8⁺ T cells could be found in dense areas of VV-infected monocytes, although the relative abundance was about 1.4-fold higher for Cxcr3^+/+ CD8⁺ T cells (see their Figure 5D), (4) Cxcr3^-/- CD8⁺ T cells were more motile than Cxcr3^+/+ CD8⁺ T cells in dense areas of infected monocytes (see their Figure 5F), (5) Cxcr3^+/+ CD8⁺ T cells probably could bind better to virus-infected cells than found for Cxcr3^-/- CD8⁺ T cells (see their Figures 5H and 5I), and (6) could also kill these cells more efficiently (see their Figure 6).

Together with the high-quality pictures and videos, the above listed data, most of which appear solid, contained enough substance for publication. For reasons that are not supported by the data, however, in their narrative Hickman et al. exaggerated the effect of the Cxcr3 knockout on viral clearance, and speculated as a major model of explanation that Cxcr3 knockout reduced penetration of CD8⁺ cells into infected cell areas. Here I expand on the two main issues, I and II, followed by a direct listing of the eight questions/issues as I placed them on the Immunity site (the first two therein overlap with the below items I and II). I moved the discussion to F1000Research to improve the visibility of what I hope should be a valuable public discussion.

(I) Upon comparison of the development of VV-infection in ears of wt and Cxcr3^-/- mice, Hickman et al. obtained results that they presented in their Figure 3F and summarized as “Likewise, we saw no overt changes in morbidity after epicutaneous VV infection; however, Cxcr3 deficiency interfered with viral clearance at intermediate times of infection (days 5–7)”. This statement is correct in stressing the modesty of the effect, but neglects that at day 5 post infection (p.i.) the virus titer is more than four-fold higher in wt than in Cxcr3^-/- mice. This fact should have been either noted and discussed or the data re-evaluated in case of an error. This issue would probably not have remained unnoticed if the main narrative had properly stressed that the Cxcr3^-/- knockout effect on virus clearance was rather modest and restricted to days 5–7 p.i. Instead, however, the Summary section of the article says “Cxcr3^-/- mice exhibited dramatically impaired CD8⁺-T-cell-dependent virus clearance”, and this interpretation may have prevented co-authors, editors and reviewers from taking a thorough look at the day 5 p.i. result.

(II) Whether the Figure 3F issue will seriously affect our understanding of CXCR3 function is questionable, but that might well be the case with another issue. By proper quantification, Hickman et al. show that the Cxcr3^-/- CD8⁺ T cells are more motile and less abundant than Cxcr3^+/+ CD8⁺ T cells in dense areas of infected monocytes (see their Figures 5D and 5F), which may be explained by differences in binding to the virus-infected cells (see their Figures 5H and 5I). However, without support by proper evidence and at least without proper quantification, the authors portray as their major model that the Cxcr3^-/- CD8⁺ T cells have relative difficulties to enter dense areas of infected monocytes. The evidence should come from data presented in Figure 5C and movie S6, but the data is finally drawn together in a rather subjective and non-quantitative interpretation by the authors “Overall, however, whereas high numbers of wt T cells entered fields of virus-infected cells, Cxcr3^-/- T cells hesitated at the perimeter of heavily infected areas of the tissue (Figure 5C; Movie S6).” Their Figure 5D shows that the percentages of Cxcr3^+/+ CD8⁺ T and Cxcr3^-/- CD8⁺ T cells are on average around 44% and 32%, respectively, and I wonder, even if the relatively small differences in presence levels could in part be explained by differences in penetration efficiencies, how these penetration efficiencies could reliably be judged without proper quantification. The proper assay for quantifying the penetration efficiencies would probably be the MPM technique shown in Figure 5E, but if I interpret the Figure 5E results correctly, the Cxcr3^-/- CD8⁺ T cells have no apparent deficit in penetrating the virus-infected region, they only more readily leave. The lower presence of the Cxcr3^-/- CD8⁺ T cells compared to the Cxcr3^+/+ CD8⁺ T cells might be entirely explained by the lesser binding to the virus-infected cells, causing faster migration within this region and easier leaving of the region. It seems to me that, because CXCR3 is a chemokine receptor, Hickman et al. forcefully concentrate on a model different from binding. In my opinion, they should have focused on trying to find reasons for the differences in binding efficiencies, rather than assume without proper evidence that Cxcr3^-/- CD8⁺ T cells have difficulties infiltrating regions of infected cells. Hickman et al. expressed their belief in the reduced penetration model in their title “CXCR3 chemokine receptor enables local CD8(+) T cell migration for the destruction of virus-infected cells”, in their description of the Figure 5D result “Although wt and Cxcr3^-/- T cells were equally distributed outside of infected areas of the dermis, a higher percentage of wt T cells than of Cxcr3^-/- T cells penetrated virus-infected areas (Figure 5D)”, and at several other sites in the article.

In summary, Hickman et al. produced good and interesting data, but they harmed their own article by an exaggerated and misleading narrative. I would like to invite Hickman et al. to address the above concerns, as well as a few other issues as they were already placed on the Immunity site:

1. The major conclusion of the authors regarding the importance of CXCR3 for clearance of vaccinia virus from infected mouse ears appears to be summarized in their statement: “Likewise, we saw no overt changes in morbidity after epicutaneous VV infection; however, Cxcr3 deficiency interfered with viral clearance at intermediate times of infection (days 5–7)”. I wonder how this conclusion agrees with the day 5 result depicted in Figure 3F, which says that the virus-titer in wt mice is about four-fold higher. This observation is unlikely to be within chance variation, given the small SEM values and the fact that “All experiments were repeated at least three times with n = 2–5 mice/group”. Probably the p<0.0001 value depicted in the figure refers to this measurement.

2. The authors conclude that Cxcr3^-/- CD8⁺ cells show reduced penetration of dense areas of infected monocytes compared to wt CD8⁺ cells. However, the only parameter that appears to be measured reliably is the number of cells present, and if I interpret Figure 5E correctly, the Cxcr3^-/- cells are not impaired in their ability to penetrate the infected area but only leave more easily (and then often come back again). From Figure 5C and movie S6, I can’t follow the interpretation by the authors that “Overall, however, whereas high numbers of wt T cells entered fields of virus-infected cells, Cxcr3^-/- T cells hesitated at the perimeter of heavily infected areas of the tissue (Figure 5C; Movie S6).”

3. The authors conclude that “Cxcr3^-/- CD8⁺ T Cells activate and home normally”. Whereas most of their experimental data were obtained at day 5 p.i., for the “normal homing” conclusion in regard to draining lymph nodes they looked at day 2 p.i. (Figure 4). As far as I understand, for “homing” some degree of specificity should be shown. For the presence of transferred cells in infected ears the authors provide evidence of active homing in Figure S1, because, as they show, the transferred cells do not traffic to uninfected ears. I can see little evidence that the transferred cells detectable in the draining lymph node already at day 2 after infection are there because of active homing? Also in regard to their Figure 1B findings, day 2 p.i. may not be an appropriate time point to examine specific homing.

4. The authors write “Together, these data show that CXCR3 is not required for SIINFEKL-expressing VV-induced activation of OT-I CD8+ T cells in vivo”, which most readers will interpret as that the SIINFEKL peptide was relevant. This is potentially misleading as the authors do appear not to have analyzed the effect of the SIINFEKL-epitope on the activation of Cxcr3^-/- OT-1 CD8⁺ cells.

5. It might be easier to interpret the findings if data measuring the outcome of the infection in CD8^-/- hosts in the context of the different cell transfers were included. The report seems to concentrate on measurements at day 5 after infection. Did, for example, the mice survive?

6. This is a study about clearance of virus from the mouse ear. Figure S3B shows that about 50% of the infected cells are CD45-negative, and consist mostly of keratinocytes as the authors explain. The virus infections of monocytes and keratinocytes should probably be understood as communicating barrels. Then why is it that for most experiments the authors chose not to show any results for the CD45-negative cells, although they systematically show virus titers and numbers of infected monocytes per ear? I feel this is an exaggeration of the principle to avoid unnecessary complexity in scientific stories.

7. In Figure 4B the authors show that two days after infection the endogenous CD8⁺ cells in the draining lymph nodes do not express detectable IFNγ levels. But, apparently, at this day the inflammation had hardly started (see Figure 2C). Nevertheless, in other experiments measuring at day 5 p.i., at which the inflammation is pronounced, the authors seem to use the Figure 4B result to define all IFNγ⁺ CD45⁺ cells as transferred CD8+ T cells (see Figures 6B-D and the main text discussion of these results). With regard to IFNγ expression by endogenous cells, it would be interesting to have further justification why the data of a non-inflamed situation (day 2 p.i.) were extrapolated to an inflamed situation (day 5 p.i.).

8. As a small point, it is unclear what is shown by the green color in Figure 4D, where green stains CD69 as well as virus infected cells.