The PH domain is a modular protein domain that affects membrane localization of proteins based on modulation of phosphoinositide lipids in the plasma membrane. Btk, the most extensively studied of the Tec family kinases, translocates to the plasma membrane via a PH domain–PI-3,4,5-P3
interaction, which is required for efficient tyrosine phosphorylation and activation, in response to antigen receptor stimulation (40
). Whether the T-cell-specific Tec family kinase Itk is regulated by a similar mechanism has only recently begun to be addressed (7
). Our previously published observation of constitutive membrane localization of Itk in unstimulated Jurkat T cells (68
), an observation that has subsequently been confirmed by two other labs (7
), is in disaccord with Itk following the model of Btk activation. The obvious moiety mediating Itk's aberrant membrane localization is its PH domain. Like Btk, the other Tec-family PH domains have the highest binding affinity for PI-3,4,5-P3
). Therefore, it seemed reasonable to consider the possibility that the metabolism of D3-phosphorylated phosphoinositides is inappropriately regulated in Jurkat T cells, leading to a basal accumulation of PI-3,4,5-P3
and inappropriate recruitment of Itk to the plasma membrane. The simplest scenario for the creation of such an imbalance in D3-phosphoinositide metabolism requires either that PI3K be hyperactive or that PTEN be hypoactive.
Given that PTEN is functionally absent in a number of human cell lines and that mutations within PTEN
have been noted previously in a Jurkat subline (64
), we reasoned that Jurkat T cells might lack a functional PTEN
allele. This could potentially explain both the transformed phenotype of these cells and the constitutive targeting of Itk to the membrane. Indeed, we found that Jurkat leukemic T-cell lines do not express detectable levels of PTEN protein, even though PTEN could be readily detected in normal human T cells isolated from peripheral blood. The fact that the antibody cocktail that we used to detect PTEN recognizes epitopes at both the N and C termini of PTEN suggested that either no PTEN mRNA was being made or the translated protein was unstable. A recent report showing that appropriately sized PCR products could be amplified from a Jurkat cDNA library with PTEN exonic primers argued for the latter possibility (75
). However, the demonstrated existence of an intronless PTEN
) on chromosome 9p21 that is highly homologous to PTEN
, with more than 98% identity (13
), made us concerned that this approach might be subject to false positives due to the almost unavoidable contamination of cDNA libraries with small amounts of genomic DNA. To avoid potential ψPTEN
-related ambiguities, we used Northern blot analysis to directly detect PTEN
transcripts in normal and JTAg T cells. This analysis revealed comparable levels of PTEN message in both JTAg and normal T cells, suggesting a posttranslational mechanism in the PTEN deficiency in Jurkat T cells.
We therefore considered the possibility that the JTAg PTEN
gene contains mutations that result in an unstable translation product. Previous work by Sakai et al. indicating, by PCR–single-strand conformation polymorphism analysis, the presence of mutations in exon 7 of PTEN
in a Jurkat subline caused us to focus on this region of the gene (64
). In our analysis of PTEN
exon 7 we found no wild-type sequences. All sequences contained one or the other of the described mutations (Fig. B) (64
), suggesting that both PTEN
alleles of JTAg are affected. Both mutations cause premature termination of PTEN translation. PTEN with a truncation within this exon is expected to have minimal phosphatase activity and to be highly unstable (33
). Interestingly, PTEN mutations in this region are associated with many cases of Cowden disease and Bannayan-Zonana syndrome—human disorders associated with loss of PTEN function (55
). Our results thus support the notion that mutations in both alleles of the PTEN
gene in Jurkat T-cell lines encode the production of truncated PTEN that has low phosphatase activity and is subject to rapid degradation. This leads to a functional deficiency of PTEN in these cells, resulting in basal accumulation of D3-phosphorylated phosphoinositides.
It is likely that defective PTEN expression, and the resultant accumulation of PI-3,4-P2
, contributes to the transformed phenotype of Jurkat T cells. This supposition is supported by the fact that restoration of expression of PTEN ultimately leads to loss of growth and increased cell death beginning approximately 24 h after transfection (reference 75
and our unpublished observations). This effect is likely to be mediated largely by activated Akt. PI-3,4-P2
recruits Akt, via its PH domain, to the plasma membrane, where Akt becomes phosphorylated on Thr-308 by PDK1, which is juxtaposed to Akt at the membrane by virtue of high-affinity binding of the PH domain of PDK1 with PI-3,4,5-P3
. Akt then autophosphorylates on Ser-473 and becomes fully active (71
). Active Akt provides a strong antiapoptotic signal (10
) and has been found to protect T cells from Fas-mediated and activation-induced cell death (21
). It is unclear why CD3 cross-linking, which does activate PI3K in Jurkat T cells (11
), did not cause increased Akt phosphorylation. Possibly, most of the available Akt was already phosphorylated due to high levels of PI-3,4-P2
. Alternatively, this may indicate TCR-stimulated activation of other phosphatases, such as SHIP, which may have held the levels of PI-3,4-P2
fairly constant (24
The degree of Itk partitioning to the membrane fraction was directly correlated with the amount of Ser-473 phospho-Akt recovered from the cells. Given the close correlation between the cellular levels of PI-3,4-P2
and the ability of Akt to become phosphorylated at Ser-473 (16
), this would suggest that the ability of Itk to bind to the plasma membrane is directly dependent upon the concentration of D3-phosphorylated PI in the plasma membrane. This hypothesis is supported by the observation that membrane targeting of Itk required an intact PH domain (7
). Both pharmacologic inhibition of PI3K and exogenous expression of PTEN could release Itk from the membrane and relocalize it to the cytosol, which is consistent with the hypothesis that the constitutive targeting of Itk to the plasma membrane was due the accumulation of PI-3,4,5-P3
in the plasma membrane as a result of the absence of PTEN expression in these cells.
The time course that we observe for the loss of Akt phosphorylation and loss of Itk from the cell membrane using inhibitors of PI3K predicts a gradual decline in PI-3,4-P2
levels due to inefficient catabolism of basally accumulated D3-phosphorylated phosphoinositides in the absence of PTEN. This would explain why Mills and colleagues were not able to block activation of Itk by 30 min of Jurkat cell treatment with pharmacologic inhibitors of PI3K (51
). The first time point at which Akt Ser-473 phosphorylation became completely undetectable was 3 h after the addition of wortmannin. By 8 h after wortmannin treatment, basal Akt phosphorylation was still undetectable; however, by this time, CD3 cross-linking could stimulate Akt phosphorylation. An explanation for these observations is perhaps provided by the fact that wortmannin is an irreversible inhibitor of PI3K that is unstable in aqueous solution (59
). This being the case, it is likely that by 8 h all of the wortmannin has degraded and sufficient translation of new PI3K has occurred to support Akt phosphorylation when the PI3K was activated by TCR stimulation. Sixteen hours following wortmannin addition, sufficient PI3K would be resynthesized to support Akt activation in the absence of TCR stimulation. The ectopic expression of wild-type PTEN in JTAg T cells was also able to reduce basal Akt phosphorylation. As in the cells incubated with wortmannin for 8 h, the Akt in the PTEN-expressing cells showed increased phosphorylation in response to TCR stimulation. This seems reasonable, since PTEN and Akt should have equal access to the D3-phosphorylated phosphoinositides, allowing for Akt phosphorylation prior to catabolism of the D3-phosphorylated phosphoinositides by PTEN. Together these results provide strong evidence for the basal accumulation of D3-phosphorylated phosphoinositides in JTAg T cells as a result of the absence of PTEN activity. In addition, these data support the idea that TCR engagement itself, without a contribution from CD28, can activate the antiapoptotic pathway mediated by activated Akt in Jurkat T cells.
It has remained an open question whether or not Itk is regulated by the same general mechanism as has been demonstrated for Btk. The Btk activation model holds that efficient activation requires the activation of PI3K to create high-affinity membrane binding sites for the PH domain of Btk. This then results in the recruitment of Btk to the plasma membrane, whereupon it becomes phosphorylated by activated membrane-resident Src family PTKs (61
). The fact that Itk requires an intact PH domain and the presence of PI-3,4,5-P3
in the plasma membrane in order to be activated in response to CD3 cross-linking strongly supports the notion that Itk must be able to translocate to the plasma membrane as a function of the interaction between its PH domain and PI-3,4,5-P3
in order to be activated in response to TCR engagement, and is consistent with other recently published results (7
). However, it should be noted that membrane recruitment is not sufficient for Itk activation; additional TCR-initiated events requiring the kinase activity of ZAP-70 and expression of the membrane linker protein LAT (68
) and the cytoplasmic adapter protein SLP-76 (R. L. Wange and X. Shan, unpublished data) also are required in order for Itk activation by TCR engagement to be observed.
One caveat against full conformity to the Btk model is the fact that we have been unable to detect translocation of Itk to the plasma membrane in JTAg T cells that express PTEN ectopically and are stimulated by CD3 cross-linking. This is unlikely to be a result of the overexpression of PTEN, since, in the same cells, OKT3-stimulated phosphorylation of Akt on Ser-473 could be observed. A more likely explanation is that the translocation is rapid and transient and occurs earlier than the 2-min-poststimulation time point that was analyzed in these studies. This supposition is based upon the fact that full activation of Itk kinase activity under our stimulation conditions occurs after only 45 s of stimulation (68
) and that most of the active kinase is recovered in the cytosolic fraction (R. L. Wange and X. Shan, unpublished data). Nonetheless, our results are in good general agreement with the Btk-derived model of Tec family kinase activation in response to antigen receptor stimulation, and they clearly demonstrate the importance of PI3K and PTEN in regulating the subcellular redistribution and activation of Tec family kinases.
Over the past 5 years there has been substantial controversy in the literature regarding the importance of PI3K in the process of T-cell activation, especially as to whether PI3K is a necessary signaling component of the CD28 coreceptor (9
). An analysis of the literature reveals that most of the studies conducted with normal T cells have supported a role for PI3K in CD28 function (23
), while studies carried out with Jurkat T-cell lines have consistently found PI3K to be dispensable for CD28 function in these cells (38
). The finding that Jurkat T-cell lines lack PTEN expression, and therefore have high basal levels of D3-phosphorylated phosphoinositides, may provide an explanation for the discrepant results reported for these two model systems.
In addition to uncovering an explanation for the unexpected constitutive membrane localization of Itk to the plasma membrane, we also provide evidence that PTEN deficiency may contribute to maintaining an elevated basal signaling state in Jurkat cells. This is suggested by both the elimination of basal Akt Ser-473 phosphorylation and the 50% reduction in basal NF-AT reporter activity in cells expressing PTEN compared to those that do not. Akt and NF-AT activity are linked by GSK3, which phosphorylates NF-AT, causing nuclear expulsion of the transcription factor (4
). In its active state, Akt phosphorylates and inactivates GSK3, potentially promoting enhanced NF-AT activity due to nuclear accumulation. In addition, Itk has also been more directly implicated in the activation of NF-AT (28
Our results also indicate that PTEN may play a role in limiting TCR-initiated signals. This is suggested by the approximately 50% reduction in both Itk kinase activity and PLC-γ1 tyrosine phosphorylation in response to CD3 cross-linking, and by the more-transient Erk activation kinetics that were observed in the PTEN-expressing cells. An inhibitory effect of PTEN expression in Jurkat T cells upon Erk activation has also been reported recently (75
), confirming this observation. Since Tec family kinases are thought to play a role in PLC-γ1 tyrosine phosphorylation and activation, it is intriguing that the greatest effect of PTEN expression upon TCR-stimulated signaling events, other than reduced Akt phosphorylation and Itk kinase activity, was seen upon PLC-γ1 tyrosine phosphorylation. We are currently in the process of developing JTAg cell lines that will inducibly express PTEN, and we hope that this reagent will allow us to look at the effect of PTEN expression on more-distal TCR-initiated signaling events, such as transcriptional activation and interleukin 2 production.
There are a host of other signaling molecules, involved in TCR signaling, that either possesses PH domains themselves or are affected by PH domain-containing proteins. Further investigation will be required in order to determine which additional signaling pathways, other than those mediated by Itk and Akt, are affected by the PTEN-deficient status of Jurkat T cells. A number of important signaling molecules that could potentially be affected by the PTEN status are depicted in Fig. . One would predict that Tec, the other PH-domain-containing Tec-family kinase that is expressed in T cells, would behave similarly to Itk and would also show constitutive plasma membrane association in Jurkat cells. As for Itk, the degree of Tec activation that could be achieved following TCR stimulation would be dependent upon the PTEN status of the cell. Also PLC-γ1 and Vav, which are both key effectors of TCR signaling, both possess PH domains, and PLC-γ1 has been shown to be regulated by PI3K in a PH domain-dependent manner (27
). Ras pathways might be altered through the PH domain-containing upstream effectors Ras-GAP, which stimulates the GTPase activity of Ras and is thought to be a Ras effector, and SOS, which is a guanine nucleotide exchange factor for Ras. Both molecules possess PH domains with preferences for D3-phosphorylated phosphoinositides (63
). All PKC isozymes have been found to be substrates for PDK1, the same PH domain-containing kinase that phosphorylates Akt (15
). PDK1 phosphorylates a site within the activation loop of PKC; phosphorylation of this site is associated with increased kinase activity. Consequently, there is the potential that multiple TCR-initiated signaling pathways are affected by the PTEN status of the cell. However, given that TCR stimulation itself leads to PI3K activation and the accumulation of D3-phosphorylated phosphoinositides, in some ways the PTEN deficiency can be thought of as providing a partial TCR signal. In the absence of any compensatory down-modulatory effects (yet to be determined), one might expect that the biggest difference between PTEN-replete and -deficient cells would be related to the rate of onset and termination of signals. Also, with the exception of the activation of Akt, most signaling events that are regulated by PI3K also require a secondary signal of some sort, such as activation of ZAP-70 in the case of Itk and PLC-γ1 activation, or the production of diacylglycerol, in the case of PKC activation. Therefore, it seems likely that only a subset of PI3K-dependent pathways will be significantly perturbed in Jurkat T cells as a consequence of the absence of PTEN.
FIG. 9 Multiple TCR-stimulated signaling pathways are sensitive to the level of D3-phosphorylated phosphoinositides. Opposing activities of PI3K (heavy green arrow) and PTEN (red arrow) are shown regulating the pool of D3-phosphorylated phosphoinositides. The (more ...)
It is interesting that of the more than 100 different proteins that have been found to have PH domains, only the Tec family kinases possess tyrosine kinase activity, and as we and others have shown, an intact PH domain is required for antigen receptor-mediated activation of this class of kinases. To our knowledge, no other tyrosine kinase has yet been shown to be subject to negative regulation by PTEN. It is tempting to speculate, therefore, that the ability of PTEN expression in JTAg T cells to reduce or eliminate tyrosine phosphorylation of PLC-γ1, as well as the proteins at 21 (TCRζ), 36 and 38 (LAT), and 40 kDa in response to CD3 cross-linking, indicates that these proteins are substrates of either Itk or Tec. Such a finding would be significant, given how little is known about the substrate repertoire of Tec family kinases, and would be the first indication of a role for these kinases in the phosphorylation of LAT and TCR subunits. An alternative hypothesis would be that PTEN may directly dephosphorylate these proteins via its phosphotyrosine phosphatase activity. We are currently working to differentiate between these two hypotheses.
Much of the credit for the rapid advances achieved over the last decade in our understanding of the signaling events that are initiated in T cells upon TCR stimulation has to go to the Jurkat T-cell line, which has demonstrated tremendous predictive power for the signaling events that occur in normal T cells. The advantages of this system are many and include the substantial body of data that has already been collected with this system, the ready availability of standardized reagents, and the relative ease and cost-effectiveness of growing large numbers of cells for biochemical manipulation. But what has made the Jurkat system indispensable as a means for dissecting T-cell signaling pathways has been the amenability of these cells to molecular biological manipulation, ranging from transient expression of exogenous cDNAs to the relative ease with which somatic mutations can be induced in these cells. A great many mutant Jurkat T-cell lines that lack important signaling molecules have consequently been generated. These have proven to be invaluable tools for the performance of structure-function studies of these molecules and have aided tremendously in our understanding of the roles of these molecules in TCR signaling. However, the discovery that Jurkat T cells lack a key component responsible for regulating the level of D3-phosphorylated phosphoinositides points up the need for caution in using this model system to study signaling events that are under the control of these phospholipids.