Forden's Immunology Home page

Mainstream immunology now recognizes that danger-signals stimulate T cells both directly (via receptors and reverse-signalling ligands on T cells) and indirectly (e.g. via dendritic and other APCs), although most immunologists still refer to signalling molecules like MICA not as "danger signals," but with more specific nomenclature such as "infection" or "cellular transformation" signals. Regardless of the terminology preferred, the unproven part of my hypothesis is that reference cells affect lymphocyte cytotoxicity in a reverse manner, i.e. that low-levels of danger-signals on non-cognate targets increase lymphocyte cytotoxicity, and conversely, that high-levels of danger signals on non-cognate targets decrease cytotoxicity.

Sensitivity of these tests for disproving theories
The three-party cytotoxicity tests described herein will discover whether the level of danger-signals on non-cognate targets affects cytotoxicity against cognate targets. The known reverse-signalling effect of FasL suggests that adding Fas^hi bystanders would increase cytotoxicity. The previously proposed danger-theory would predict that adding NKG2D ligands to a cytotoxicity assay would always increase cytotoxicity. In contrast, the reference-cell theory predicts that such additional cells would decrease cytotoxicity. As for adding low-danger, non-cognate targets, one would expect they would consume time and metabolic resources from the effectors and would therefore reduce net cytotoxicity. In contrast, the reference-cell theory predicts that low-danger non-cognate targets should increase cytotoxicity against cognate targets. Therefore both the low-danger-reference and high-danger-reference assays described below form very sensitive ways of distinguishing between successful and unsuccessul theories.

Since I do not have access to an immunology lab (nor time to perform these experiments) I describe here what I believe to be optimal tests to confirm or disprove the reference-target hypothesis. Instead I have much experience and training in engineering signal-processing systems. That experience has lead me to propose the reference-cell hypothesis as a likely explanation for various poorly understood effects, including peripheral tolerance and possibly tumor tolerance. The reference-cell hypothesis is directly analogous to the common engineering design practice of differential signal sensing.

The experiments will resemble bystander toxicity assays where non-cognate targets are added to standard two-party (effector and target) toxicity assays. Heretofore the additional cells have been called "bystanders", and cytotoxicity against them has been what was measured. I propose measuring how the non-cognate targets affect cytotoxicity against cognate targets. If the presence of non-cognate targets bearing high levels of danger signals, reduces cytotoxicity against cognate targets, then we have confirmation that we can call the non-cognate targets "reference" cells instead of just "bystanders". In order to reduce the complexity of interpretation I would recommend that all three cell parties (cognate targets, effectors, and reference targets) be syngeneic with each other. As referenced at length in my paper on Ag correlation, many cytoxicity assays have shown that syngeny is mandatory for FasL probing of (non-cognate) "bystanders" after Tc engage cognate targets, so reference cells at least must be syngeneic with effectors.

CD8+ effectors are more likely than CD4+ Tc, to produce comprehensible results in these experiments. There are both specific and general reasons for the preference for killer rather than helper T cells. Suzuki and Fink's data show that CD8+ cells proliferate in response to ligating their FasL with cognate targets' Fas, whereas CD4+ cells do not. This is compatible with the theory that antibody-mediated responses are directed against necrotically lysing viruses, whereas cytotoxicity is directed against viruses that do not burst open their host cells in order to escape. CD4+ cells seem to have more known interactions with other immune cells, so their interactions with potential reference cells could be more difficult to interpret. When CD4+ cells are eventually tested for their response to reference cells, one should be careful to discriminate between CD25+ regulator cells and CD25- helper cells, since they would be likely to behave in opposition to each other. It would be interesting to test whether CD4+ cells react to opsonization as a danger signal. Of course it makes no sense to consider opsonization as a danger-signal from the point of view of the immune system as a whole, because the immune system produces the opsonization. However, individual cells need to process data semi-autonomously, so an immune system in which some of its components treat markers produced by other immune cells as danger-signals, could be efficient as a distributed processing system.

Target cells
The cognate (primary) and non-cognate (reference) targets should be as similar as possible to each other, except for cognate-antigen and danger-signal presentation. Similarity is strictly necessary because in-vivo Tc face many extraneous local cell types that should not be used to measure the reference levels of danger signals of the local tissue of interest. For example, if Tc must evaluate the danger-signals presented by muscle cells, nearby nerve fibers and blood vessel cells would be likely to display different levels of danger signals than healthy muscle cells. Therefore it is quite likely that Tc have evolved to be unresponsive to background levels of danger-signals as presented by nearby non-cognate cells of dramatically different tissue types.

Ideally none of the assay cells would be transformed cell lines, but if culture of transfected, non-transformed cells is too difficult or slow, then parental and transfected immortalized cell lines could be used. The parental line could be transfected to express Fas receptor, MICA, and OVA. OVA would serve as the cognate antigen; Fas and MICA would be danger signals. Eight cell lines could result, ranging from cells expressing no more than background levels of Fas and MICA, and no OVA (which I believe is restricted to hamsters), to cells expressing large quantities of all three molecules. More cell lines could be created if transfected cells express varying amounts of the transfected genes. The estimate of eight cell lines assumes the transfection success would be binary for each of the three transfected genes; i.e. that transfected cells either express uniform amount of the targeted protein, or they express or increase their expression not at all. Intermediate expression of transfected genes would be very useful for advanced experiments, but the simple assays listed below require cells with just high or low (or zero in the case of OVA) expression levels.

Apoptosis of primary targets (bearing e.g. OVA by transfection or pulse-loading) would be measured, for example by ⁵¹Cr release. Therefore the cognate targets should be pulsed with ⁵¹Cr prior to the assay. The reference targets should not be loaded with ⁵¹Cr; this is the primary difference between the assays I propose, and many published assays on cytotoxicity against bystanders.

Preparation of effector cells
Modified 2006 Dec. 21. Inject OVA (or other peptide you choose for the experiment) with adjuvant s.c. into mice. If this technique results in a Th2, rather than the desired Tc1 response, try injecting instead, peptide pulsed, irradiated target cells as vaccinators. Two months after the last vaccination, harvest the lymph nodes or spleen and culture the lymphocytes in a plastic vessel to remove adherent cells. I suggest two months (rather than two weeks) to allow Tc stimulated by DC to recover from that acute stimulation; When highly stimulated by dendritic cells, Tc need to temporarily ignore the lack of danger-correlation to early viral proteins because some viruses have evolved to temporally separate danger-signals from their epitopes. By allowing the responder Tc to shift into a memory-state instead of testing them during high-activation after priming, they are more likely to respond to relatively lower levels of danger-signals on cognate targets, by reducing cytotoxicity.

Culture danger-signal controlled stimulator cells with OVA. The stimulator cells for culture with lymphocytes from vaccinated mice should be chosen to match the danger-presentation characteristics of the primary targets those responder lymphocytes will later face. Verify OVA is loaded onto MHC class 1 molecules of the stimulators. Stimulators should be washed after irradiation and before adding them to the culture so that excess target peptides will not contaminate reference cells. Culture responder lymphocytes with IL-2 and mixtures of reference cells and irradiated stimulators. Reference cells are similar to stimulators but neither loaded with OVA nor irradiated. Reference cells cultured with responders in vitro should have the various combinations of danger levels as those to be used in the cytoxicity assays described in the table below. Responders will therefore be cultured with stimulators that match the primary targets they will later face in the assays. Reference cells added during effector expansion should also match those that that set of effectors will face during assay. After expansion of responders has been verified for most combinations of danger-levels in reference and stimulator cells, ⁵¹Cr assays can begin.

Published results should include the expansion rates of effectors cultured with the various combinations of stimulator and reference cells. All the assays should begin with identical numbers and concentrations of effectors, primary (cognate) target cells, and reference cells, so that expansion rate in stimulation culture will not directly control total cytotoxicity, although it could influence cytotoxicity per effector. The only difference between primary targets and stimulator cells, is that primary targets would have been pulsed with ⁵¹Cr. All the described cells (stimulator, reference, and effector) must be syngeneic.

Assay guidelines
As per standard practice, cytotoxicity should be measured with varying ratios of effectors to targets. My hope is that when low-danger reference cells are added, high cytotoxicity can be achieved even when targets greatly outnumber effectors. The fact that effectors usually need to outnumber targets in order to kill a substantial fraction of them in four hours, seems an indication that something has been missing from our assays. My hope is that the missing ingredient is reference targets. It is desirable to test high ratios of targets to effectors for another reason, too; then effectors will be less likely to use other effectors as references. Duke shows that effectors seem to prefer non-lymphocytes as reference targets (bystanders affected by FasL on effectors that kill cognate targets). However, restricting reference effects to cells that we specifically add as references, will make our data easier to interpret. I recommend that the initial tests use equal numbers of reference and cognate targets.

In addition to these described assays, it would be helpful to at least at first, simultaneously run assays of cytotoxicity against bystanders, just to verify that the experiment successfully duplicates conditions previously published in which syngeny was found necessary for bystander lysis. Descriptions of such assays can be found in:

• Smyth MJ, Krasovskis E, Johnstone R; "Fas Ligand-Mediated Lysis of Self Bystander Targets by Human Papillomavirus-Specific CD81 Cytotoxic T Lymphocytes" Virology, July 1998, p. 5948–5954)
• Smyth MJ, "Fas Ligand-Mediated Bystander Lysis of Syngeneic Cells in Response to an Allogeneic Stimulus" The Journal of Immunology, 1997,158: 5765-5772.

Assay cell sets
The following discussion uses the notation (low) for cells that display low levels of a given danger signal. Ideally such levels would be those typical of normal, mature cells in a given tissue type. However it may be difficult to culture healthy, mature cells, so blast cells might be a reasonable alternative. If necessary, transformed cell lines might be used, in which case (low) might mean absolutely zero expression of a given danger signal. I do not know whether complete extinction of danger signals will be interpreted by immune cells as a danger signal in itself. 

All primary targets will express the cognate Ag, e.g. OVA. No reference target will express the cognate Ag. 

Confounding effects of competition
NEW: 2006 Dec. 21. A positive result from the cytotoxicity inhibition assays, could be interpreted as due to competition for the attentions of effectors by high-danger bystander targets, rather than as due to effectors comparing the danger signals on cells serving as references. Although a 50% decrease in cytotoxicity could reasonably be blamed on competition, a decrease of 90% or more, would be more plausibly the result of comparision effects. But how to prove that? One could condition the effectors with both cognate targets and non-cognate reference cells for hours or days, then separate effectors out and drop them into assay wells with only labelled cognate targets. The addition of reference cells to a conditioning regimen would not influence later cytotoxicity via competition effects, as long as the reference cells had been effectively excluded from the ensuing assay.