Background
Leukocyte Function Antigen-1 (LFA-1), an αβ heterodimer integrin, is necessary for leukocyte adhesion and migration and is important in the formation of the immunological synapses [
1‐
3]. LFA-1 also has prominent roles in T cell costimulation [
4,
5] and transendothelial migration [
6]. The LFA-1 ligands, intracellular adhesion molecules (ICAMs) -1, -2, and -3, differentially bind to LFA-1 and regulate its adhesion [
7‐
9]. LFA-1 is a mediator of T cell driven inflammatory diseases such as psoriasis[
10], rheumatoid arthritis[
11], and multiple sclerosis [
12], and is a pharmaceutical target for the prevention of the rejection of organ transplantation [
13].
Peripheral blood lymphocytes (PBLs) primarily express LFA-1 as opposed to the other beta-2 integrins, such as MAC-1 (CD11b/CD18) and p150,95 (CD11c/CD18), and serves as a primary adhesive molecule for T cells. However, natural killer (NK) cells, eosinophils, neutrophils, monocytes, and dendritic cells also express LFA-1 and other integrins. Impairment of LFA-1/ICAM interactions, using monoclonal antibodies or in LFA-1 knockout mice, disrupts adhesion and migration of neutrophils, monocytes, eosinophils, and NK cells in several disease models of inflammation. Pathologies in these models include thioglycollate-induced peritonitis, delayed delayed-type hypersensitivity, asthma, and susceptibility to bacterial and viral infection [
14‐
16].
Significant advances have been made in understanding the structural properties of LFA-1 and the conformational states that LFA-1 adopts upon ligand binding [
17‐
19]. However, it is not clear how LFA-1 governs interactions in the context of multiple ligands or if LFA-1 signaling mechanisms are similar in different cell types. Previously, we showed that LFA-1 transmitted distinct intracellular signaling events that, in the presence of CD3 and CD28 stimulation, enhanced T cell activation thresholds and polarized T cells towards IFNγ-producing effector cells[
20]. Since human effector T cells differentiate into memory T cells, and LFA ligands are known to be important for enhancing the interaction at the immunological synapse, we investigated how the three endogenous LFA-1 ligands influenced commitment of human T cells to adopt a T
H1 memory phenotype. In particular, it is of interest as to whether the combinatorial integration of signaling induced by LFA-1's interactions with its ligands achieved an internal signaling threshold that committed T cells to end-fate decisions such as differentiation into a particular memory cell subset.
We find that the ICAM ligands differentially promote cell survival. In the presence of CD3 and CD28, stimulation with either ICAM-2 or ICAM-3 suppressed activation of caspase-3. Phospho-proteomic profiling showed that p38 and p44/42 phosphorylation was enhanced and rate of activation was increased. Production of TGFβ1 was decreased when cells were treated with CD3/CD28 plus ICAM-2 or ICAM-3, but not ICAM-1, indicating that different ICAMs can alter cytokine production that can influence TH1/TH2 T cell development. In long-term differentiation experiments, CD3/CD28 plus ICAM-3 or ICAM-2 generated a higher frequency of IFNγ producing CD4+CD45RO+CD62L-CD11aBright cells than stimulation with ICAM-1. Only ICAM-3/CD3/CD28 stimulation resulted in differentiation to the CD4+CD45RO+CD62L-CD11aBrightCD27- phenotype that transmigrated in response to chemokines. Upon stimulation with MIP3α, the highly differentiated CD4+ memory T cells that were generated by ICAM-3/CD3/CD28 produced a higher frequency of phospho-p44/42 and phospho-JNK positive cells than those stimulated with the other ICAMs, indicating a functionally responsive CD4+ memory cell subset.
These results suggest that signal activation thresholds for CD4+ memory T cell differentiation are integrated with, and bounded by, the combinatorial integration of input signals of CD3, CD28 and LFA-1. However, only ICAM-2 and ICAM-3 showed important responses to the production of highly differentiated functional memory CD4+ T cells in vitro. T cell commitment drives many adaptive immune system processes and thus these results are relevant to situations where multiple ligands interact with a single receptor to drive different underlying processes. The analysis shown here demonstrates that it is possible to distinguish these events at the single cell level.
Discussions and Conclusion
In this report we present evidence that the ICAM ligands of LFA-1 induce distinct signaling activation events that have functional consequences for human CD4+ memory T cell differentiation in vitro. We undertook a comprehensive approach to dissect the differences amongst cells treated with the ICAM ligands within the context of T cell activation and differentiation. Using multiparameter surface phenotyping in conjunction with intracellular phospho-protein profiling, extracellular cytokine profiling, and chemotactic functional assays, we found that distinct CD4+ memory T cell subsets were generated when cells were stimulated with ICAM-3/CD3/CD28 and ICAM-2/CD3/CD28, and contrasted those generated by ICAM-1/CD3/CD28, or CD3/CD28 stimulation.
Intracellular activation of p38 and Erk1/2 was sensitive to the ICAM ligand used in combination with CD3/CD28 stimulation. Both p38 and Erk1/2 are known to integrate signals to regulate both mitogenesis and differentiation in various cellular systems and these two MAPKS are integral for T cell development, activation, and differentiation [
32,
33]. Single cell quantitation of the intracellular phosphorylation of p38 and the ratiometric measurement of phosphorylated Erk1/2 revealed that the integration of CD3, CD28 and LFA-1 were dependent on the LFA-1 ligand used for stimulation. We surmise that sub-optimal levels of active Erk1/2 result in the first stages of T cell activation (initial cell division, activation marker expression, and cytokine production), but are not sufficient to dictate full conversion into highly differentiated T cells. This can be concluded since not all ICAM stimulations promoted the full conversion of CD45RA
+ to CD45RO
+ cells in T cell differentiation assays
in vitro. Intriguingly, when the ICAM ligands were tested in combination, ICAM-1 and ICAM-2 enabled cell progression to the CD45RA
+CD45RO
+ stage, whereas ICAM-1 stimulation alone did not allow cells to pass the CD45RA
+ stage. Also, the combination of ICAM-1 and ICAM-3 enabled some cells in the population to progress to the full CD45RO
+ stage, whereas neither ICAM-1 nor ICAM-3 alone allowed progression past the CD45RA
+ stage (Fig.
3). These observations suggest a complex regulation of signaling events by LFA-1 dependent on the combination of ligands present. Although ICAM-1 and ICAM-3 have been observed at the immunological synapse in separate studies [
34,
35], the dynamic interaction of among the three ICAM ligands at the synapse has not been reported. Thus, thee results presented here suggest that the combinatorial integration of ICAM ligand interactions with LFA-1 have important and unexpected functional consequences for T cell biology.
Both ICAM-2/CD3/CD28 and ICAM-3/CD3/CD28 stimulations had higher Erk1/2 signaling thresholds and correlated with less activation of caspase-3 and higher levels of intracellular BCL2 levels than other treatments. These stimulation regimes also generated highly differentiated CD4
+ memory T cells that were functionally responsive to chemotactic agents. In murine models, Erk1/2 has been implicated in regulation of both positive and negative selection of developing T cells [
33,
36‐
38]. This developmental process is dependent on activation-induced cell death mechanisms that remove autoreactive T cells from the periphery. Improper elimination of autoreactive T cells leads to several forms of autoimmunity[
39,
40]. Inhibition of programmed cell death is also a prominent feature of various forms of T and B cell lymphomas [
41,
42]. The relationship between intracellular signaling thresholds and cell death mechanisms in human naïve CD4+ T cells are currently not understood, however, the methodologies employed in this study can be used to resolve the interconnectedness of these two processes.
Activation of CD4 T lymphocytes in the presence of specific cytokines causes differentiation into distinct effector T
H subsets with different immunoregulatory properties. TGFβ1 is an immunosuppressive cytokine that has been observed to suppress the proliferative response of CD4
+CD45RO
+ lymphocytes [
43] and inhibit the production of IFNγ [
44]. In our experiments, it was noted that CD3/CD28 and CD3/CD28/ICAM-1 stimulation generated the highest levels of TGFβ1 and consequently, these two stimulation regiments also had the lowest frequency of IFNγ producing T cells (Fig.
4C) and suppressed production of CD4
+CD45RO
+ memory cells (Fig.
5B). Therefore, the enhancement in signaling imparted by LFA-1 upon binding ICAM-2 or ICAM-3 in human CD4+ T cells can attenuate T cell differentiation. Mechanistic understanding on how LFA-1 regulates phosphatase activities may shed some light as to how ligand binding correlates with differential intracellular activities and subsequent cellular outcomes.
A clear understanding of the distinctive tissue distributions of the ICAM ligands, and their roles in determining function of T cells, in the human system has not been completely resolved. ICAM-1 (CD54) has a wide tissue distribution on both hematopoietic and non-hematopoietic cells, can be up regulated upon cellular activation and is viewed as the prominent LFA-1 ligand at the immunological synapse in model systems of cell-to-cell contact and of leukocyte rolling [
3]. ICAM-1 is closely related to ICAM-3 (CD50), which is constitutively expressed at high levels on leukocytes and epidermal dendritic Langherans cells and can also be up regulated upon activation on endothelial cells. These two ICAMs contrast with ICAM-2 (CD102), which is broadly expressed on leukocytes and constitutively expressed at high levels on vascular endothelium and is not up regulated upon cellular activation [
45]. Interestingly soluble forms of the ICAMs exist in human blood and have been correlated with disease indications [
46,
47]. The physiological significance of soluble ICAMs in the blood is unknown; however, given the results presented here, one can surmise that the presence of extracellular LFA-1 ligands might potentiate the response of T cell activation and subsequent differentiation.
It is interesting to note that ICAM-1 is the only LFA-1 ligand that has been studied extensively for its adhesive contribution in the immune synapse formation and has been previously implicated to promote increased T
H1 differentiation [
43,
48]. From biophysical experiments, ICAM affinity interactions have been calculated using recombinant proteins to suggest ICAM-1 has the highest affinity for the LFA-1 receptor [
49]. However, more recent studies have suggested a dynamic conformational change of LFA-1 that has been attributed to the discrepancy of 2D and 3D off-rate measurements and requires adjustment of the ligand and receptor densities to accurately estimate an affinity constant [
50].
Because structurally, the ICAMs exhibit geometrical differences, with ICAM-1 being the only ICAM reported to require dimerization for activity [
51‐
54] our data suggest further experimentation is necessary to account for the biological differences observed in human memory T cell differentiation in vitro. At present only one published study using ICAM-1-/- splenic antigen presenting cells has demonstrated a delayed response in generating pathogenic CD4+ effector cells in a murine model of diabetes [
55], however a comparative study of the different ICAMs in disease models has not demonstrated. Future work will require follow-up in T cell stimulation by selective ICAM-deficient antigen presenting cells in both murine models and human systems.
When T cells interact with antigen presenting cells (APC), these cells display multiple ICAM ligands at their surface. It has been observed that one T cell can interact with several APCs [
56], thereby altering the potential density of locally present interacting LFA-1 ligands. Thus, it is plausible that LFA-1 on any given CD4+ T cell is presented with multiple opportunities to interact with one or more of its ligands, and that the density of ligand interaction governs the intracellular events regulated by LFA-1. This coupled with the dynamic range of
in vivo peptide-MHC interactions and the number of co-stimulatory molecular interactions warrants further studies into the combinatorial matrix of influential intracellular signaling thresholds that dictate T cell fates. The studies presented here show clearly that interaction simultaneously with different LFA-1 ligands gives rise to different outcomes during memory T cell generation
in vitro. Thus, cells are capable of interpreting the presence of distinct ICAM molecules presented simultaneously and collating a response that is distinct from their response to any individual ICAM. Whether this is due to action upon all receptors simultaneously or different ICAM ligands, when presented in combination, seek out distinct LFA-1 receptors, with at present unknown modifications or abilities, on the cell surface remains to be determined.
We previously demonstrated that LFA-1 lowers T cell activation thresholds[
20] and we recently showed that signaling through LFA-1 can activate human NK cells[
57]. The present work demonstrates that LFA-1 signaling mechanisms can potentiate T
H1 development and that the combinatorial integration of ligand dependent LFA-1 signaling regulates the development of memory T cell development. Our results also illustrate that quantifiable intracellular signaling thresholds, as imposed by LFA-1, can regulate T cell commitments. We expect that such thresholds, like Erk1/2 phosphorylation, serve as signaling checkpoints that regulate T
H1/T
H2 development. These may be important in T cell-dependent autoimmune diseases, including rheumatoid arthritis and multiple sclerosis, where aberrant cell activity leads to pathological outcomes. The identity of such checkpoints, and the regulatory kinases and adapter proteins that instruct these processes, may serve as novel areas for pharmaceutical intervention for controlling autoreactive T cells in autoimmune diseases and possibly in T-cell cancers
Methods
Immunological and Chemical Reagents
Anti-human CD3, CD4, CD45RA, CD45RO, CD11a, and CD27 direct conjugates (FITC/PE/PercP/PerCPCy5.5/APC/Pe-Cy7/APC-Cy7), cleaved caspase-3-PE, BCL2-FITC, and CD128b, CXCR3, CXCR5, CCR9, CCR5, CXCR4, CCR7, CCR6, CCR1, CCR2, CCR4, CDw128 (all on AX647) were obtained from PharMingen. ICAM-2 mAb and ICAM-2-FITC were from IC2/2 Research Diagnostics. Phospho-specific antibodies to p44/42 (T201/Y202), Gsk3β (Y279), Ikkα (S32/36), PLCγ1 (Y783), Lck (Y505), Zap70 (Y319), p38 (T180/Y182), Stat1 (Y701), Stat3 (Y705), Stat3 (S727), Stat5 (Y694), Stat6 (Y694), and PKA (S114) conjugated directly to Alexa dyes were from BD-Biosciences. TruCount beads were from BD-Immunocytometry systems. LFA-1 antibody clones TS1/22 and TS1/18 were obtained from the Developmental Hybridoma Studies Bank. Protein and chemical reagents used (and vendors) included fluorescein isothiocyanate (FITC, Pierce), Alexa Fluor dye series 488, 546, 568, 647, 680, 700 and CFMDA (Molecular Probes). PMA, ionomycin, and propidium iodide were purchased from Sigma. Recombinant human ICAM-1-FC, ICAM2-FC, ICAM3-FC were from R&D Systems. Recombinant cytokines IL-2, IL-4, IL-6, IL-10, IL-12, IFNγ, and TNFα were obtained from PharMingen. Recombinant chemokines IP-10, MIB1β, and MIP3α were from R&D Systems. Secondary antibodies to mouse and rabbit IgG were obtained from Santa Cruz Biotechnologies. Control treatments consisted of mouse IgG (for antibodies), 1% BSA (for proteins), or 0.01% DMSO vehicle (for chemicals). Secondary crosslinkers were evaluated and optimized for stimulations.
Cell Culture
Human peripheral blood lymphocytes were obtained by Ficoll-plaque density centrifugation (Amersham Pharmacia) of whole blood from healthy donors (Stanford Blood Bank) and depleted for adherent cells. Magnetically activated cell sorting (Dynal) was used to negatively isolate naïve CD4+ cells for studies as indicated. Human cells were maintained in RPMI, 5% human sera AB (Irvine Scientific), and 1% penicillin-streptomycin glutamate (PSQ). Blood from 42 donors were used for these studies. U-bottom Nunc-Immuno plates with MaxiSorp surface were used for immunology assays. Standard ELISA techniques were used to coat plates, briefly, 200 μL of ligands at concentrations of 0.01–1 μg in PBS, pH7.4 were added to wells and incubated at 4°C overnight. For comparative purposes, stoichiometric ratios of either two or three antibody or ligand were maintained constant by equal mixing prior to adsorption. Excess antigen coating solutions were removed and plates were blocked with complete media for 30 min before being incubated with cells. The MaxiSorp surface is a modified, highly charged polystyrene surface with high affinity to molecules with polar or hydrophilic groups and has a high binding capacity for proteins, including globular antibodies in proper orientation. Maximum binding capacity in a monoloayer is 650 ng/cm2.
Flow Cytometry
Intracellular and extracellular staining was performed as described[
58]. Intracellular probes for active kinases were made by conjugating phospho-specific antibodies to the Alexa Fluor dye series as described [58, 59]. Kinetic analyses were performed by direct application of fixation buffer in time-synchronized 96-well plates maintained at 37°C. 200 μL of 2% paraformaldehyde was added to 100 μL of 0.5 × 10
6 cells, stimulated as indicated, and the mixture was pipetted up and down three times to ensure even mixing. Plates were kept in a 37°C water bath during the process. Fixation was performed for 10 min at 37°C, and plates were then centrifuged (1500 RPM, 5 min, 4°C) and processed for flow cytometric staining. Flow cytometry data are representative of at least three independent experiments. Figure legends indicate specific replicate and donor repeatability. Intracellular cytokine staining was performed as suggested by manufacturer of intracellular IFNγ-APC stain (BD-Pharmingen). The proliferation assay was performed on sorted naïve CD4+ T cells labeled with 1 μM 5-carboxymethylfluorescein diacetate (CMFDA, Molecular Probes) in PBS for 5 min at 37°C. The mixture was added to coated plates (2 × 10
5 cells/well) in RPMI medium 1640 containing 5% human AB sera. After 96 hrs, the cells were stained for CD4 and analyzed. Absolute cell counts were done using TruCount beads. Data was collected on a FACSCalibur (four-color) using Cellquest software or an LSRII (12-color) machine with DiVA software (Becton-Dickinson) and analyzed using Flowjo software (Treestar). Clustering analysis, heatmap visualization, principle component analysis, and scatter plot analysis was performed in Spotfire software. Comparison algorithms used for data analysis were supported by Flowjo. Two algorithms (Overton and SED) were used to calculate the percentage of positive cells found in the sample and not in the control. Two algorithms (Kolmogorov-Smirnov (K-S) and Probability Binning (Chi(T) or PB) were used to determine the statistical difference between samples. The PB algorithm has been shown to detect small quantitative differences between two populations.
Multiplex bead assays
Cytokine detection was performed using either Cytometric Bead Arrays (CBA, PharMingen) to detect TH1/TH2 cytokines (IL-2, IL-4, IL-5, IL-6, IL-10 TNFα, and IFNγ) on a FACSCalibur machine or using Beadlyte multiplex kits (Upstate Biotechnologies) to detect (IL-2, IL-3, IL-4, IL-5, IL-6, IL-7, IL-9, IL-10, IL-12, IL-13, IL-15, TNFα, IFNγ, MIP1α, Eotaxin, MCP, and GMCSF) on a Luminex machine. The apoptosis CBA kit was from PharMingen. Protocols suggested by the manufacturers were used.
Transmigration assay
Cells were placed (0.25 × 106 cells in 100 μL) in the upper well of 24-well transmigration chambers (5 μm pore, Transwell, Costar Corp.) 100 ng SDF-1, Rantes, MIP3β, MIP1α, MIP1β, MCP-1, IP-10, or IL-8 (in 0.5 μL media) was then added to the lower well. Plates were incubated for 24 hrs at 37°C and cells that migrated to the lower chamber were counted using TruCount beads.
Acknowledgements
The authors acknowledge support from BD Biosciences-PharMingen, technical expertise from Laurie Gilmour and Jill Taylor (BD-Immunocytometry systems), reagents from BD Biosciences, Upstate Biotechnologies, Biosource International, Dynal Inc, Aventis Pharmaceuticals, Bio-Rad, and the Herzenberg laboratory (Stanford University) and support and advice from David Parks and Richard Stovel (Stanford FACS facility). We are grateful to Khoua Vang and Howard Gus for administrative support. ODP was supported as a Bristol-Meyer Squibb Irvington Fellow, by a Dana Foundation human immunology award, and from the NHLBI proteomics contract N01-HV-28183I. GPN was supported in this work by NIH grants P01-AI39646, AR44565, AI35304, N01-AR-6-2227, A1/GF41520-01, N01-HV-28183I and a grant from the Juvenile Diabetes Foundation.
Authors' contributions
OP conceived of the study and participated in the design and coordination of all the experiments, analyzed the data and drafted the manuscript. DM carried out the flow cytometry assays and participated in the data analysis. GPN helped revise the manuscript. All authors read and approved the manuscript.