Do innate NK cells truly show memory-like response to antigens? Monobenzone [hapten] study

This week Immunity published study claiming NK cell response to melanocyte-targeting hapten, monobenzone, display adaptive-like behavior with memory component.

The concept itself is quite new and not fully developed  [first report of this kind was published in 2006]. I personally do not see that data presented so far support proposal that NK cells have canonical "adaptive memory". For example, one of the hallmarks of adaptive memory response is that secondary response is stronger and more durable compared to primary response. However, in this new paper both primary and secondary hapten-specific response by NK cells show equal magnitude.

To confirm memory response by NK cells, in my view, the authors should show that (a) NK cell response is antigen-[hapten] specific, (b) it is long-lived and (c) secondary response is stronger. 

There is no doubt that in this paper NK cells show antigen-specific response (condition A) and such response could persist for relatively long term (ranging from few weeks to few months, condition B). However, condition C has not been met. In crucial experiment, the authors transferred NK cells recovered after primary response into naive host and then challenged the host 1 week later. However, for some reason, the authors also applied hapten 1 day after NK cells transfer [and before secondary challenge]. This basically prevented testing of recall response.

Only experiment that the authors performed to show long-lived NK cell memory response was the experiment wherein hapten-primed mice were left to sit around for 4 months and then challenged for the second time. However, in this scenario it is hard to tell whether it is true memory response or hapten simply persisted in mice and kept NK cells in active "effector" phase. In addition, in this experiment, both primary and secondary responses were of equal magnitude.

In summary, so far I have not seen definite results in support of "canonical" NK cell memory.

David Usharauli

Cish is a NK cell checkpoint inhibitor

This week several science news outlets spotlighted new study from Nature Immunology showing anti-tumor effect of Cish deficient NK cells. This study showed that Cish deficient NK cells are hyper-responsive to its canonical cytokine IL-15 and show improved control of experimental tumors.

I analysed this study to determine if the buzz was deserving. In my view this study is in fact two independent [and not connected] studies put together artificially. The finding that Cish deficient NK cells have superior anti-tumor behavior is based on non-physiological experimental model.

First part of this study deals with cytokine sensitivity of Cish deficient NK cells. This is exclusively in vitro experiments. It does show that Cish deficient NK cells are hyper-responsive to NK cell canonical cytokines such as IL-15, or IL-12/IL-18 combo [and other activatory receptors].

Second part of this study, however, is completely dissociated with IL-15 story and jumps directly to tumor protection experiments with WT or Cish deficient NK cells. These experiments showed that Cish deficient mice are resistant to exogenously injected tumor cells and this protection depended on asiolo-GM1+ cells and IFN-γ.

Finally, the authors showed that adoptive transfer of Cish deficient NK cells into NK-deficient hosts (Ncr1^Mcl1Δ/Δ mice) also provided protection against i.v. injected melanoma cell metastasis.

In summary, this study proposed that Cish functions as a checkpoint inhibitor for NK cells.

My view:

(A) It is not clear whether Cish deficient NK cells alone is sufficient for anti-tumor effect [it is independent of CD8 T cells, but CD4 T cells involvement was not tested].

(B) It is not clear what role cytokines such as IL-15 or IL-12/18 play in vivo in Cish deficient mice.  

(C) One of the major differences between NK and T cells is that NK cell effector functions are not regulated in an antigen-specific manner but rather by sensing signaling balance between activatory and inhibitory membrane receptors [functions like a rheostat].

During ontogeny individual NK cell undergoes its own "adjustment" to its environment and can even "tolerate" naturally arising tumor cells. This is why experimental tumor models when tumors are injected exogenously, and appear in the body out of the blue so to speak, do not recapitulate natural interaction with NK cells and easily could produce biased, non-physiological response from first-time encountering NK cells. The more physiological tumor models will be to use spontaneously arising mouse tumor models.

David Usharauli

    

Antigen-specific NK cell memory

Do NK cells have memory? New study in Nature Immunology suggests they do. 

This study was done on rhesus macaques. Purified NK cells were sorted from spleen, liver or blood. Autologous, virus infected or antigen-pulsed DCs were used as targets. Control DCs were either uninfected or mismatched antigen-pulsed.

Initially, the authors observed that purified spleen (c) or hepatic (d) NK cells from SHIV-infected hosts showed cytotoxic response to Gag and Env, but not OVA pulsed autologous DCs.

Similarly, purified splenic NK cells derived from SIV-infected hosts showed antigen-specific cytotoxicity (a, b). Interestingly, blood derived NK cells showed no cytotoxic activity against antigen-pulsed DCs (c).   

Finally, the authors showed that splenic (c) and hepatic (d) NK cells derived from hosts vaccinated 5 years earlier displayed vaccine-matching antigen-specific DCs lysis. Here too, blood derived NK cells showed minimal antigen-pulsed DCs lysis (e).

Additional results indicated that NK-mediated lysis of antigen-pulsed DCs were mediated via NKG2C or NKG2A pathways. These NK receptors recognize MHC class I molecules loaded with leader peptides derived from other endogenous MHC molecules. However, it is not clear how this mechanism confers exogenous antigen-specific memory to NK cells.

In summary, these results suggest that NK cells could mount antigen-specific memory response. 

David Usharauli 

NK cell activation initiates type II diabetes

Type II diabetes is a complex syndrome involving endocrine, immune and metabolic abnormalities. It is well known that obesity can lead to type II diabetes. But how?

New study in Nature Immunology suggests that obesity-induced adipocyte stress activates NK cells  via NCR1 receptor driving IFN-γ mediated insulin insensitivity characteristic to type II diabetes.

The authors observed that NK cell depletion in mice fed high fat diet (HFD) ameliorated insulin insensitivity and glucose intolerance.

Ex vivo examination of visceral adipose tissue (VAT), a target tissue of type II diabetes, revealed that HFD induced expression of NK cell ligand in VAT detected by NCR1 (NKp46 in humans).

Interestingly, VAT but not subcutaneous (Sc fat) adipose tissue from HFD fed mice could stimulate NK cells.

In vivo experiments confirmed that NCR1 deficiency improved insulin sensitivity.

Finally, the authors showed that IFN-γ derived from NCR1 activated NK cells promotes inflammatory macrophages in VAT leading to glucose intolerance, that can be ameliorated with NCR1 blockade.

In summary, these results showed that HFD induces visceral adipose tissue stress that activates local NK cells via NCR1 ligand leading to inflammatory macrophage polarization and reduced insulin sensitivity. Targeting NK cell activation may interrupt this disease cycle and improve type II diabetes management.

I was always wondered why diet-induced obesity leading to VAT stress should activate inflammatory, M1 type macrophages? What is an evolutionary advantage for such response, in general? No idea.  

David Usharauli

Cancer-bacterial relationship: how wrong focus can spoil everything

New concepts are only useful if they can explain something or predict something. Sometimes scientific idea is good but using wrong model to test it could produce negative impression.

Here is the one example for a such scientific approach. This paper was published in journal Immunity few days ago. This study identified bacterial component, Fap2, that can protect cancer cells from NK cell cytotocixity through TIGIT interaction. This concept of bacteria protecting tumors against immune system is very interesting, so I decided to analyse it.

Initially the authors showed FITC labeled Fusobacterium nucleatum (F. nucleatum) strain 23726 could bind both hematopoietic and epithelial cancer cell line. Surprisingly, these cell lines pre-incubation of F. nucleatum strain 23726 showed resistance to NK cell natural cytotoxicity (but there was no effect on IFN-γ or TNF-α production). This inhibition of NK natural cytotoxicity was specific for F. nucleatum strains since Escherichia coli lacked this effect.

The authors speculated and then showed that inhibition of NK natural cytotoxicity by F. nucleatum strains could be due to its interaction with inhibitory receptor TIGIT.

Indeed, transfection of TIGIT-negative NK cell line with human TIGIT could reduce its natural cytotoxicity against EBV-transformed B cell line.

Similar reduction of NK cell natural cytotoxicity were observed when primary human NK cells were used, though effect was minimal when using human colorectal carcinoma cell line, RKO.

Next, the authors were able to show the actual clinical isolates of F. nucleatum strain recovered from human colon adenocarcinoma samples could inhibit NK cell cytotoxicity.

Using F. nucleatum transposon-based insertion-inactivation mutant library the authors showed that inhibition of NK cell natural cytotoxicity was reversed with Fap2 mutation.

Additional the authors showed that Fap2 mutant F. nucleatum strains failed to induce IL-2 production from cell line transfected with human TIGIT.

Fap2 mutant F. nucleatum strains also failed to interact with purified human TIGIT.

Finally, the authors showed F. nucleatum strains could (1) inhibit TIL cytoxocity against autologous melanoma cells in a Fap2-dependent manner and (2) inhibit peptide-specific T cell IFN- response.

In summary, these results show that F. nucleatum strains could interact with human TIGIT and this interaction reduces NK cell cytotoxicity in a Fap2-dependent manner but has no effect on NK derived IFN-γ or TNF-α production. Strangely, it appears that Fap2 inhibited both T cell cytoxicity and IFN-γ production.

In my opinion since the authors argue in the beginning that F. nucleatum strains are mostly found in human colon adenocarcinoma tissue, but showed that Fap2 effect on NK cytotoxicity against colorectal carcimona cell line, RKO, is negligent, it puts the significance of this finding into question.

David Usharauli