Kinesin-1 regulates antigen cross-presentation through the scission of tubulations from early endosomes in dendritic cells

Dendritic cells (DCs) constitute a specialized population of immune cells that present exogenous antigen (Ag) on major histocompatibility complex (MHC) class I molecules to initiate CD8 + T cell responses against pathogens and tumours. Although cross-presentation depends critically on the trafficking of Ag-containing intracellular vesicular compartments, the molecular machinery that regulates vesicular transport is incompletely understood. Here, we demonstrate that mice lacking Kif5b (the heavy chain of kinesin-1) in their DCs exhibit a major impairment in cross-presentation and thus a poor in vivo anti-tumour response. We find that kinesin-1 critically regulates antigen cross-presentation in DCs, by controlling Ag degradation, the endosomal pH, and MHC-I recycling. Mechanistically, kinesin-1 appears to regulate early endosome maturation by allowing the scission of endosomal tubulations. Our results highlight kinesin-1’s role as a molecular checkpoint that modulates the balance between antigen degradation and cross-presentation.

The report by Belabed et al describes a new role for the motor protein kinesin-1 by regulating endosomal maturation. Kinesin-1 is a motor protein, which moves along the microtubules using ATP hydrolysis and transport cargo in the cells. The authors propose that kinesine-1 (Kif5b) is responsible for the scission of endosomal tubulation allowing the maturation of early endosomes (EEA1+) into recycling (Rab11+) or late endosomes (LAMP1+), a process required for antigen cross presentation in dendritic cells ( Figure 7F). This work is interesting and new but some mechanistic is lacking. For example Kinesine 1C has been involved in the transport of Golgi apparatus to ER (Dorners et al, JBC 1998 ) and in Golgi tethering (Lee at al, eLife 2015). The authors should quantify the number of EEA1, Rab11 and LAMP1 positives vesicles at the steady state as well as monitor the morphology of ER and Golgi apparatus in wt and Kif5b deficient DCs. Mechanistically does Kif5b binds EEA1? What is the percentage of Kif5b-EEA1 positive vesicles (not very high Figure 6B)? If Kif5b recruits EEA1+ endosomes and promotes the scission of EEA1+ endosomes to Rab11+ and LAMP1+ compartments, why is MHCII antigen presentation not affected and why do we see more colocalization of OVAs in LAMP1+ lysosomes in Kif5b deficient DCs at 60 min in comparison to control cells ( Figure 4)? The rate of MHCI recycling molecules is very high especially in CD8+ DCs. Usually, MHC recycling is very efficient and difficult to assess. The use of primaquine, which slow the return of receptors on the cell surface by increasing their intracellular pool makes possible to measure recycling (Reid PA et al, Nature 1990). Can the authors explain why they have such high rate of MHCI recycling without using primaquine? How many cells were used for this experiment? Did the authors look at the recruitment of the V-ATPase and NADPH oxidase in Kif5b deficient endosomes? Is the recruitment of LRO to early endosomes also impaired in Kif5b deficient DCs? Did they monitor proteases activities in Kif5b deficient endosomes?
Minor points: Figure 1B: gating strategy should be described Figure 1D: It is surprising that BMDCs are better cells to cross present OVAs than CD11b and even CD8+ DCs. The uptake of OVAs seems to be the same between BMDCs and CD8+ DCs ( Figure 2A). What is the number of DCs used per well to perform this assay? The ratio DC: T cells? It is not describe in the Figure legend or in the Mat and Methods section. Figure 3C: statistics should be done Overall the IF figures seem a bit overexposed (especially Figure 4). Will be nice to see the edge of the cells.
Reviewer #2, expert in cross-presentation (Remarks to the Author): In their manuscript, the authors investigate the influence of Kif5b on cross-presentation and endosome trafficking. They use conditional Kif5b KO mice to demonstrate that Kif5b deficiency inhibits antigen degradation, endosome acidification and maturation and cross-presentation. Additionally, they demonstrate that early endosomes from Kif5b deficient cells are less motile and show impaired scission of tubules.
The manuscript is very clear and describes a novel role of Kif5b in cross-presentation, which is an interesting subject to the field. The experiments are very convincing and depicted in clear figures. The manuscript is already of high quality.
Nevertheless, I have the following remarks: 1) on page 6, the authors show that endosomes from Kif5b-deficient DCs display impaired acidification. This is misleading. Of course, impaired endosome acidification can be the cause of changes in pH. However, it could also be that the dextrane used does not reach acidic compartments because of impaired lysosome trafficking. This should be clarified in the text.
2) Top of page 7: here, the authors claim that they investigate the effect of Kif5b on the vacuolar pathway of cross-presentation. This is not true and misleading. In their experiments, the authors did not ditinguish between the vacuolar pathway and the endosome-to-cytosol pathway at all. Given the effect of Kif5b deficiency on endosome trafficking, it is very likely to have an effect on crosspresentation, regardless whether the antigens are cross-presented via the vacuolar or via the endosome-to-cytosol pathway. This also should be clarified in the text.
3) Middle of page 7: here, the authors describe a lower percentage of colocalization between OVA and LAMP1. However, Fig 4 clearly demonstrates that the colocalization is not lower but rather delayed. This should be reworded. 4) middle of page 9: here they state as a sort of conclusion that "Kif5b binds to EEA1-positive endosomes..... and allows subsequent sorting of Ag/MHC-I complexes". This should be reworded since they didn´t prove any of these statements and this reflects overinterpretation of the results. 5) Top of page 11, they state that "the cross-presentation defect in the absence of Kif5b resulted from an impairment in early endosome dynamics and membrane fission." Also this is not proved in the manuscript and reflects overinterpretation. Should be reworded.
6) The authors showed several effects of Kif5b deficiency: antigen degradation, endosome acidification, endosome maturation, reduction in MHC-I recycling, antigen trafficking towards lysosomes and tubulation scission. But which of these effects now are responsible for impaired antigen presentation? As the authors point out correctly in their discussion, it is generally accepted that rapid antigen degradation prevents cross-prestation, meaning that impaired antigen degradation and trafficking towards lysosomes might not be critically involved in inhibition of cross-presentation. Alternatively, if peptides derived from soluble antigens are loaded onto MHC I in endosomes (regardsless of vacuolar or endosome-to-cytosol pathway), they need to be transported towards the cell membrane, which presumably occurs via Rab11+ endosomes. Since the authors show that Kif5b deficiency prevents such protein transport, this might be the main reason for impaired crosspresentation. The discussion on how Kif5b might influence cross-presentation should be extended in the manuscript in general.

Sven Burgdorf
Reviewer #3, expert in innate instruction of adaptive immunity (Remarks to the Author): This manuscript investigates kinesin-1 in cross-presentation. This is not the first time that kinesin-1 has been reported to play a role in DC function and cross-presentation (this should be cited in the Introduction) and/or receptor recycling. Here, the authors investigate the response in more detail. Overall the manuscript lacks rigour and coherence.
A major shortfall of the manuscript is that there is an overinterpretation of a specific role for kinesin-1 in cross-presentation. The manuscript lacks in depth analysis of other forms of antigen presentation. The role of kinesin-1 in MHC II antigen presentation and direct MHC I antigen presentation needs to clarified in a more rigorous manner. Indeed it is surprising that MHC II antigen presentation is not affected given the significant impact of kinesin-1 KO on endosomal biology and antigen proteolysis. Likewise, direct MHCI presentation may be altered due to the role of kinesin-1 in MHCI recycling. The supplementary data in SFig4 suggests there is a defect in MHCII presentation of soluble OVA by CD8 DC. Moreover, the manuscript requires clarification regarding the different forms of antigen used.
Bead-OVA antigen presentation data should be included. The data for cross-presentation contradicts the accepted view that cDC1 are the more efficient cross-presenting cell type. For example, published data suggests there is little cross-presentation of soluble OVA by CD11b+ DC eg. Schnorrer. 2006. How do the authors explain this discrepancy between their cross-presentation outcomes and those previously reported? The link between cross-presentation of soluble antigen in vitro (Fig 1) and crosspresentation of tumour associated antigen in vivo (Fig 2) is not clear, given these are two very different types of antigen.
Tumour data is used as evidence for a role of kinesin-1 in cross-presentation in vivo. While there is clearly an impact of kinesin-1 on the growth kinetics of the tumour, this also occurs in the absence of OT-I transfer suggesting mechanisms other than OVA cross-presentation could be playing a role. Are other immune populations eg CD8 and CD4 T cells, B cells and NK cells functional and present in normal numbers in the kinesin-1 cKO? Concluding the changes in tumour growth and/or mouse survival are solely due to cross-presentation is an overintepretation of the data. To measure crosspresentation in vivo, the authors need to perform in vivo transfer of OTI and OTII cells and measure proliferation in response to different forms of (cross-presented) antigen.
In Figure 5, analysis of OVA trafficking shows higher colocalization with EEA-1 and less with LAMP and Rab11 early in the cKO but at later time points this is enhanced compared to WT. How do the authors explain this? Do these experiments take into account that in the WT there is significantly more antigen proteolysis compared to cKO?
It is unclear how the data in Figure 7 directly relates to kinesin-1. Given that tubule formation requires actin polymerisation then it is not surprising endosomal tubulations would not be present in cells without tubules. It is unclear how this data fits into the theme of the manuscript.
The Discussion is poorly written and needs to be more concise.
Throughout the manuscript there are many statements that over-interpret the data presented: "Our results provide the first genetically based evidence in support of this hypothesis, and highlight kinesin-1 as a critical regulator of Ag degradation and Ag sorting from the early endosomes." "Our results highlight kinesin-1's newly recognized role as a molecular checkpoint that modulates the balance between antigen degradation and cross-presentation." "Our results show that kinesin-1 (i) has an essential role in the Ag and MHC-I endocytic trafficking upstream of cross-presentation" "Taken as a whole, these results suggest that the defect in T cell priming observed in the absence of kinesin-1 was specific to cross-presentation." "In contrast, the presentation of endogenous Ag to MHC-I molecules and the presentation of exogenous Ag to MHC-II molecules in cKOKif5b mice was unaffected". "Taken as a whole, these data indicate that Kif5b binds to EEA1-positive sorting endosomes, drives the scission of tubular structures, enables vesicles to mature into recycling or late endosomes, and thus allows the subsequent sorting of Ag/MHC-I complexes" The majority of the statistical analysis of data is performed using an unpaired t test which is not the correct analysis for the type of data presented.

Minor Comments
More detail is required to explain Fig 1A in   The authors show here that in absence of Kif5b expression, cross presentation in DCs is impaired and anti tumor immune responses are reduced. They also show that endocytic degradation, acidification, recycling and cargo transfer to late compartments are all impaired.
The results are clearcut and the experience are well designed. The authors conclude that these defects in intracellular trafic cause reduced antigen cross presentation and impair anti tumor immune responses. They propose that Kif5b plays a critical role in the control of the balance between antigen degradation and cross presentation.
The main problem with this conclusion is that, based on previous studies (cited by the authors), the defects in endocytic functions and intracellular trafic in Kif5b KO dendritic cells would suggest increased and not impaired, antigen cross presentation. Several groups have shown that reduced degradation, acidification and lysosomal transport are ll associated with better cross presentation. The main difference with the Kif5b KO is that recycling is also impaired. It is therefore most likely that impaired cross presentation in Kif5b is in fact due, mainly, to defective MHC I re-cycling. Supporting this possibility with experimental data would strengthen the paper. What are the effects of nocodazole and latruculin B on MHC class I re-cycling? And in all cases this issue should be better discussed.
Other remarks: -In the tumor growth experiment it is important to monitor the T cell responses (eventually using a model tumor antigen). -In all figures showing different types of DCs, please keep the same order for the different DC-types.
-The summary scheme in Figure 7F is not very helpful... If kinesin 1 (does this means Kif5b?) is involved in both recycling and degradation (as the results show), it is not possible to know if the effect on cross presentation is due to either or both...