Lymph node stromal cell populations inhibit naive T cell activation
In order to study the function of lymphoid stromal cells during T cell priming, we first established cell lines from the CD45 negative fraction of the LN, representing FRCs (“FRC.5”: gp38+, CD31−), LECs (“LEC.6”: gp38+CD31+) and BECs (“BEC.7”: gp38− CD31+) (). These lines, once established, were passaged up to 10 times from each freeze and re-sorted between major freezes. These lines displayed distinct morphological features, consistent with their functions in situ; FRCs featuring very extended lamellopodia on plastic or glass surfaces and both LECs and BECs having elongated phenotypes with long extension () and a high propensity to form side-by-side ‘bundles’ of cells in 2D cultures (not shown).
Inhibition of Initial T cell priming by Lymph node Stromal Populations.
We subsequently tested the effects of each line, as a 5% cellular proportion, when co-cultured with a standard stimulation culture of naïve CD8 T cells, using monoclonal T cells derived from lymph nodes of OTI TCR transgenic mice mixed with in-vitro generated bone-marrow derived dendritic cells (BMDCs). T cells were labeled with CFSE to track cell division and, as shown in , in the absence of their cognate antigen, SIINFEKL (SL8), they did not divide, regardless of the presence of lymph node stromal lines. When BMDCs were pulsed with SIINFEKL peptide, T cells typically underwent multiple rounds of cell division within 48 hours. However, addition of FRC, LEC or BEC cell lines each resulted in nearly complete inhibition of the induced proliferation.
As T cell activation by DCs typically occurs in the T cell zone of lymph nodes, scaffolded by FRCs, we focused predominantly on these stromal cells and lines for subsequent studies. We first confirmed that freshly isolated gp38+CD31− CD45− FRCs were similarly inhibitory by assaying them in our T+DC stimulation cultures. In these assays, we compared FRC isolates to a similar number of total lymph node cells and found the FRCs () again to be inhibitory for proliferation of CD8 T cells. To test for the specificity of inhibition, we also compared the FRC.5 line to mouse embryonic fibroblast lines (MEF) and fibroblasts (NIH-3T3) and found that whereas the latter lines also were modestly inhibitory for proliferation, the FRC.5 line consistently inhibited to a greater extent (). Given that the FRCs are present proximally in the LN during priming whereas BEC and LEC are not, we focused on the mechanism underlying their inhibition of the response.
Inhibition of T cell activation by FRCs depends upon interferon gamma production during priming
We first sought to determine the magnitude of T cell inhibition and if it involved interactions between FRCs and DCs or was a direct effect upon T cells, regardless of DC presence. In , we added FRC.5 cells at varying levels to CFSE-loaded T cells and primed with a selection of stimuli. A 10% fraction of FRCs provided nearly complete blockade of T cell proliferation in response to BMDCs loaded with SL8, plate-bound anti-CD3 +/− soluble anti-CD28, or PMA plus ionomycin. However, when a smaller FRC concentration was used, there was little or no effect in some conditions but partial inhibition in the presence of αCD3 and αCD28. This implied that FRCs were not simply globally repressing responses but, rather, that they did so specifically in response to cues derived from the strength or type of activation of the T cells. We quantified this finding by a different criterion in . Here, we scored for the percentage of cells in the culture that were in or beyond the first division, and term this “% blasts”. By this criterion, which measures the degree to which cells are allowed to begin cycling, a greater distinction in the threshold of FRC-based inhibition emerged. We interpret from this that FRCs can very effectively inhibit the first cell division in some conditions (i.e. at 2.5% or more) such that very few cells progress to becoming ‘blasts’. In contrast, at the lower fractions (i.e. 1%), they inhibit this to a lesser extent but, for some priming conditions (e.g. BMDC), can prevent the maximal extent of cell cycling.
Regulation of T cell activation by LN Stromal cells via an IFNγ mediated and proximity-dependent mechanism.
We used this same metric, while varying the number of BMDCs in the culture in order to test whether higher densities of DCs could overcome inhibition (). This revealed that the inhibition at the higher levels of FRCs could not be overcome with greater numbers of DCs, strongly suggesting that the inhibitory effect was not due to FRCs somehow confining DCs from T cells. The exclusion of such a mechanism was also supported by the ability of FRCs to inhibit T cell proliferation to the fully soluble PMA and Ionomycin stimulus ( and ).
Since it has been shown that distinct priming conditions result in differences in both the quantity and quality of cytokines produced by blasting T cells
, we sought to determine the effect of blocking select cytokines or adding them back to T+DC assays (). While we saw little effect of αIL4 or αIL-12 antibodies (data not shown), we found that addition of αIFNγ
almost fully restored proliferation of T cells cultured with DCs in the presence of FRCs. Interleukin 2 nor αIL-2 had a significant effect on FRC-mediated inhibition of CFSE dilution. This implied that, at a strength of stimulation resulting in strong IFNγ secretion, FRCs were licensed to inhibit the T cell response. That FRC might be responsive to IFNγ was also supported by their robust expression of the IFNγ receptor 1 ()
When FRCs were separated from T cells and DCs by a semi-permeable membrane, the inhibitory effect was lost (). Addition of IFNγ itself also could not restore the inhibition in the transwell context, suggesting that a lack of IFNγ diffusion across the transwell didn't explain this effect. Combined with , this implied that, while IFNγ was involved, the inhibitory mechanism required cell-cell or very close proximity between T Cells, DC, and the modulating FRC population.
Regulation of T cell priming by IFNγ-promoted Stromal iNOS
We therefore turned toward assessing mechanisms by which FRCs might inhibit T cell responses. Recent reports have shown that viral infection of FRCs results in their upregulation of PDL-1, an inhibitory molecule for T cells
. We assessed levels of PDL-1 as well as the costimulatory ligands CD86 () and CD80 (data not shown). Whereas CD80 and CD86 levels were low and remained unchanged, PDL-1 levels were indeed upregulated in FRCs treated with IFNγ. However, blocking these pathways did not appreciably alter the proliferation of T cells in our assays (), suggesting another mechanism. Since IFNγ appeared to be necessary for inhibition, we next examined the transcription of nos2
gene and protein expression and the production of nitric oxide (NO) which had previously been shown to be IFNγ-inducible in a selection of non-hematopoietic cells
as well as cells of hematopoietic origin
. The nos2
gene encodes the enzyme inducible nitric oxide synthase (iNOS), which is responsible for production of nitric oxide and can be specifically inhibited by the drug 1400W. As shown in , addition of IFNγ directly to FRC lines in the absence of T cells or DCs results in an approximate 10-fold increase in transcription of the nos2
gene, an effect that was not directly altered by blocking the enzyme itself with 1400W. Using antibodies to iNOS protein, we were able to detect significant signal by immunofluorescence in FRC lines and these levels were both quantitatively higher and more localized to the cytosol in the presence of T plus DC (, arrowhead highlights T-DC couple
). Flow cytometry confirmed an upregulation of total iNOS levels in response to T+DC augmented, compared to control (DC alone), and that this augmentation could be blocked with antibodies to IFNγ ().
IFNγ dependent Induction of NOS2/iNOS/Nitrate production in Lymphoid stromal cells.
We next measured accumulation of nitrite, a relatively-stable end-product of nitric oxide, in our cultures using a standard Griess assay
. Here, we found that addition of exogenous IFNγ to FRC cultures failed to increase production of nitrite to detectable levels (), an effect that may suggest that release or further activation of the enzyme requires additional signals delivered with T cell activation. However, nitrite was detected when T cells were stimulated in those cultures using SL8-loaded BMDCs, and this was considerably augmented by inclusion of FRC in a concentration-dependent manner. Nitrite could not be detected when unstimulated FRCs were plated. Finally, the nitrite that was detected when FRCs were mixed with T cells plus peptide-loaded DCs was blocked by the iNOS-specific inhibitor 1400W.
To test the emerging hypothesis that iNOS production was responsible for FRC inhibition of T cell responses, we mixed T cells plus SL8-loaded DCs together with FRCs in the presence or absence of iNOS inhibition. As shown in , we could completely restore proliferation of OTI CD8+ T cells by blocking iNOS function with 1400W. Similarly, we could restore proliferation using the drug L-NMMA, a broader-spectrum NOS inhibitor. When quantified as % blasts, this drug completely released cells to proceed into cycle (–c) with only very modest inhibition of complete cycling of cells as assessed by CFSE peaks.
A proximal role for iNOS during T cell priming by Lymphoid Stroma.
As iNOS has clearly been described as being produced by a variety of macrophages
, we wished to formally exclude that our BMDC were not in some way responsible for the production and inhibition. This was previously supported by our demonstration that FRC inhibited responses to anti-CD3 plus anti-CD28 or PMA/Ionomycin () and also the demonstration that 1400W augmented responses in the presence of FRC, but had no effect on basal proliferation in their absence (). Consistent with this, the stimulation of T cells using BMDC from mice genetically lacking the iNOS gene was nevertheless dramatically inhibited by stromal lines (). We noted that T cell inhibition by LEC or BEC stroma appears to operate via a similar iNOS/IFNγ dependent mechanism (data not shown).
Although it is currently not possible to generate a mouse in which deletion of iNOS is restricted to stroma and bone-marrow chimeras to approximate this fail to fully reconstitute the lymphoid compartment for unknown reasons (data not shown), we nevertheless sought to determine whether our findings were consistent with iNOS acting to attenuate T cell responses in lymph nodes. This was particularly important since previous reports had noted a requirement for iNOS in augmenting immune responses, likely as a result of its potent anti-microbial activity
. We therefore turned to the use of a germline nos2
(iNOS) knockout and immunized in ways that should engage the pathway we had studied in vitro and, conversely, in ways that might not engage this mechanism. We additionally chose to use tetramer staining to examine the response derived from endogenous repertoire rather than using a TCR-transgenic adoptive transfer system that may bias the outcome due to abnormally high numbers of antigen-responsive cells.
To recapitulate our in vitro studies in which BMDC prime IFNγ production and the ensuing iNOS production, we first immunized control or iNOS−/− mice in the flank using the same BMDCs pulsed with SL8 peptide. In this regimen, BMDC were always iNOS proficient and so any change due to the genetic background of the host would be due host cells beyond the APC. As shown in –g iNOS deficiency resulted in an overall trend towards an increased percentage of tetramer+ cells among total CD8 T cells and a statistically significant increase (1.8 fold) in the absolute number of tetramer+ CD8+ T cells in the lymph node at day 3.
As a comparison, we repeated this type of experiment but sought to direct antigens to a resident DC population in the absence of additional overt inflammatory cues. For this, we targeted the relatively non-inflammatory DEC-205+/CD8+ DC subset as antigen presenting cells
using anti-DEC205-OVA antibody-protein conjugates
. In contrast to what was observed using inflammatory DCs as the priming antigen presenting cells, iNOS deficiency consistently resulted in a mild decrease in T cell expansion as assessed using tetramer staining (–j
). Together this suggests that iNOS production during intra- lymph node priming in the first 3 days does represent a mechanism of regulating T cell responses, as predicted by our in vitro
experiments. Interestingly, and subject to further study, it also indicates that iNOS is differentially engaged and utilized, depending on the type of challenge and we would propose that this may depend upon the cell types that are stimulated to produce it and the local responses to this. The engagement of this pathway during the first days of lymph node priming will represent fruitful future study to understand the likely multiplicity of its action.
FRC inhibition depends on T cell production of IFNγ and is Th1 specific
A key prediction of these results is that T cells that do not make IFNγ will not engage this regulatory mechanism. To formally test this, we assessed FRC-mediated inhibition of OTI cells lacking the interferon gamma gene (IFNγ−/−). As shown in , OTI T cells lacking expression of IFNγ were completely refractory to FRC-mediated inhibition, even when they were present at fractions as high as 10% of T cells. Confirming that this was truly due to the loss only of IFNγ, addition of exogenous IFNγ to these cultures re-established inhibition of T cell cycling.
FRC-mediated inhibition of IFNγ producing CD4+ and CD8+ T cells but not Th2 cells.
This observation raised the possibility that FRCs might specifically block certain types of T cells that make IFNγ while sparing those that do not. A classic dichotomy exists in CD4 T cell subsets in which effector T helper 1 (Th1) cells produce IFNγ but no IL-4, whereas T helper 2 (Th2) cells produce IL-4 but no IFNγ. We therefore tested the prediction that FRCs would specifically block cycling of restimulated Th1 clones while having little or no effect on Th2 cells. As shown in , we found this to be the case: Th1 cells loaded with SE-efluor670 dye had a significantly higher MFI, indicative of less dilution and hence fewer rounds of cell division, when cultured with FRCs whereas Th2 cells showed no difference in MFI. Since blasts are heterogeneous in size and proliferate quickly from the start of this assay, we do not observe cells expressing undiluted levels of this dye. For this reason we did not quantify %Blasts for this assay. demonstrates that Th2 cells are insensitive to FRCs across a broad range of concentrations while Th1 cells proliferate less (higher SE-efluor670 MFI) as FRCs are added in increasing concentrations.
FRCs block cell-cycle progression but do not block initial TCR recognition
We next turned toward dissecting the dynamics of how iNOS functions during the priming response. We first assessed the ability of T cells to form stable conjugates with DCs over time and found no significant difference between T cells activated with BMDCs in the presence of upwards of 25% FRCs (). Similarly, analysis of calcium signaling profiles failed to detect significant differences in the kinetics of intracellular calcium rise over the first 12 minutes post BMDC contact, despite the readily available co-engagement of these cells (). When examining the dynamics of the priming reaction, the first sign of inhibition by FRCs occurred at approximately 24 hours when cultures were examined for homotypic T-T clusters, correlating with successful activation
. This timeframe has previously been shown to correspond to early IFNγ production in T cells
. OTI T cells cultured with SIINFEKL-pulsed DCs in the absence of FRCs displayed robust clusters averaging 1.2×104
in size while average cluster size in the presence of FRC was 0.1×104
(). This was, in part, mediated by the action of iNOS/NO production as homotypic clustering was partially restored by blockade with 1400W to an average size of 0.5×104
. The remaining inhibition may suggest a second mechanism that alters homotypic adhesiveness although this will be subject of future work.
Inhibition of T cell activation by LN stromal lines occurs largely at the level of cell-cycle progression.
Interestingly, when we assessed CD25 and CD69 expression, markers of early T cell activation, we found only a small difference in CD25 upregulation at 24h when FRCs were added and no difference in CD69 upregulation at this time, regardless of how many FRC were in the culture (). In comparison, however, when cells were assessed at 36 hours for cell-cycle progression and the expression of Ki67, a marker of cell cycling, the profiles of T cells activated in the presence of FRC were identical to those of naïve T cells, showing very significant inhibition of these measures. This effect was titratable: between 1 and 2.5% FRCs, both progression to G2/M and the expression of Ki67 were reduced (). From this we conclude that the interactions occurring after initial T-DC interaction, but prior to approximately 36 hours, lead to the predominant inhibition of T cell proliferation. From our other evidence, this is consistent with published reports that OTI T cells generate IFNγ within the first 24 hours
and (our data) that T cell IFNγ production leads to FRC expression of iNOS in a similar timeframe. Consistent with our previous data, inhibition of NO production reverses the block in cell-cycle and overall expansion of the T cells ()