Genetic evidence for a predominant role of PI3Kβ catalytic activity in ITAM- and integrin-mediated signaling in platelets

The phosphatidylinositol 3-kinase (PI3K) is implicated in platelet activation downstream of G protein–coupled receptors (GPCRs), immunoreceptor tyrosine-based activation motif (ITAM)–bearing receptors, as well as integrins.¹  The different PI3K isoforms are grouped in 3 classes, I, II, and III. Class I PI3K includes the p85-associated members PI3Kα, PI3Kβ, and PI3Kδ (class IA), as well as the PI3Kγ isoform (class IB). Although all class I members are expressed in platelets, PI3Kγ and PI3Kβ are considered to play a major role in platelet physiology,^1,2  but their relative contribution is still controversial.

PI3Kγ is activated by G protein βγ dimers, and, in platelets, it has been implicated in Rap1b activation and integrin α_IIbβ₃–supported aggregation downstream of the G_i-coupled P2Y12 adenosine 5′-diphosphate (ADP) receptor.^3,4  This model has however been challenged by other studies, proposing a predominant role of PI3Kβ rather than PI3Kγ in ADP-induced platelet activation.^5,6  In addition, the finding that PI3Kγ also functions as a scaffold protein⁷  supported a model for a kinase-independent contribution of PI3Kγ in ADP-mediated platelet activation.⁸ 

In platelets, PI3K is also activated upon stimulation of the ITAM-bearing collagen receptor glycoprotein VI (GPVI)/FcR γ-chain.⁹  Although there is evidence that a class IA member is involved,¹⁰  its identity is still unknown. Similarly, integrin α_IIbβ₃ recruitment activates PI3Ks. The involvement of a class II member has been hypothesized,¹¹  but a possible role for class I isoforms clearly emerges from a more recent study.¹² 

The specific contribution of PI3Kγ and PI3Kβ in platelet activation has remained largely elusive, mainly because of the early embryonic lethality caused by genetic ablation of PI3Kβ in mice,¹³  and because the interpretation of the analysis of PI3Kγ knockout mice turned out to be complicated by the kinase-independent action of this isoform.⁷ 

In this work we report the first comparative analysis of platelet activation in mice homozygous for the kinase dead forms of PI3Kβ (PI3Kβ^KD) or PI3Kγ (PI3Kγ^KD), in which lysine at positions 805 or 833, respectively, had been substituted with an arginine residue. Our results demonstrate that while both isoforms are implicated in signaling by GPCRs, PI3Kβ plays a unique role in platelet activation downstream of GPVI and integrin α_IIbβ₃.

Generation and characterization of PI3Kβ^KD and PI3Kγ^KD mice has been previously reported.^7,14  Blood was withdrawn from the inferior vena cava of anesthetized animals, and washed platelets were prepared as described.¹⁵  Platelet aggregation was measured in a lumiaggregometer and followed for up to 5 minutes upon addition of ADP (in the presence of 200 μg/mL fibrinogen), U46619, or convulxin. Phosphorylation of Akt was evaluated on whole platelet lysates by immunoblotting with anti–phospho-Akt (Ser473) antibody, followed by reprobing of the membrane with anti-Akt antibody (Cell Signaling Technology). Rap1b activation was measured by a pulldown assay using glutathione S-transferase (GST)–tagged RalGDS-RBD. Analysis of platelet adhesion and spreading on immobilized fibrinogen was performed on coated glass coverslips, and results were quantified by fluorescence microscopy analysis upon staining with tetramethylrhodamine isothiocyanate (TRITC)–phalloidin (Sigma-Aldrich). Further details on the source of the reagents and on the experimental procedures adopted are reported in previous works.^5,15,16  The use of mice for our experimental work was approved by the Ethics Committee of the University of Pavia.

Tail bleeding time was not altered in both PI3Kγ^KD and PI3Kβ^KD mice (data not shown), indicating that activity of neither PI3K isoforms is essential for global hemostasis. Expression of p110γ and p85 was normal in PI3Kβ^KD platelets, as was expression of p110β and p85 in PI3Kγ^KD platelets (supplemental Figure 1, available on the Blood website; see the Supplemental Materials link at the top of the online article). However, we observed that expression of p110γ in PI3Kγ^KD platelets and expression of p110β in PI3Kβ^KD platelets, albeit clearly evident, was reduced by approximately 50% to 60%. Although the reason for this enzyme instability is unclear, this finding is consistent with previous observations in other tissues.¹⁴ 

ADP-induced platelet aggregation was clearly impaired, but never suppressed, in both PI3Kγ^KD and PI3Kβ^KD mice (Figure 1Ai). The PI3Kβ inhibitor TGX221 further reduced, but did not abolish, ADP-induced aggregation of PI3Kγ^KD platelets. Similarly, comparable defective ADP-induced Rap1b activation was observed in both genotypes, although Rap1b activation largely occurred independently of both PI3Kγ and PI3Kβ activity (Figure 1Aii). TGX221 did not further significantly reduce Rap1b activation in either ADP-stimulated PI3Kγ^KD platelets or PI3Kβ^KD platelets. Figure 1Aiii also shows that ADP-induced Akt phosphorylation was completely suppressed in both genotypes. This surprising and unexplained observation is actually in line with previous independent genetic and pharmacologic data,^2,5,16,17  and it clearly indicates a complex interplay between the 2 isoforms in ADP signaling. These observations demonstrate that ADP-induced Rap1b activation and aggregation are only partially regulated by PI3Kγ and PI3Kβ kinase activity, and they question the role of Akt in these processes.

The limited and redundant contribution of PI3Kγ and PI3Kβ to platelet activation by GPCR stimulation was even more evident when U46619 was used as an agonist. In this case, inhibition of platelet aggregation was reduced only by approximately 30% in both genotypes, and there was no additional effect of TGX221 on either PI3Kγ^KD platelets (Figure 1Bi) or PI3Kβ^KD platelets (data not shown). The latter finding confirms the specificity of this inhibitor. U46619-induced activation of Rap1b was dependent on both PI3Kγ and PI3Kβ (Figure 1Bii). In fact, although at 1 μM U46619 activation of Rap1b was only minimally affected by inactivation of PI3Kγ, but was almost ablated in PI3Kβ^KD platelets, upon stimulation with 5 μM approximately 50% of Rap1b activation was independent of either enzyme. In contrast, U46619-induced Akt phosphorylation was exclusively promoted by PI3Kβ, as it was undetectable in PI3Kβ^KD platelets, but occurred normally in PI3Kγ^KD platelets. Accordingly, TGX221 completely inhibited Akt phosphorylation in both wild-type and PI3Kγ^KD platelets. Once again, these results indicate that Akt is not a necessary mediator of PI3K contribution to platelet activation.

Important differences between PI3Kγ^KD and PI3Kβ^KD mice were found upon stimulation of the ITAM-bearing receptor GPVI by convulxin. Aggregation of PI3Kγ^KD platelets was only slightly reduced compared with wild type, but it was almost completely suppressed in PI3Kβ^KD platelets (Figure 1Ci). Moreover, addition of TGX221 to PI3Kγ^KD platelets completely abolished the residual convulxin-induced aggregation. The predominant contribution of PI3Kβ in convulxin-promoted platelet activation was confirmed by the analysis of Akt phosphorylation and Rap1b activation (Figure 1Cii-iii). Both of these events were completely undetectable in stimulated PI3Kβ^KD platelets but occurred normally in PI3Kγ^KD platelets. Moreover, convulxin-induced Akt and Rap1b activation in PI3Kγ^KD platelets was completely suppressed by TGX221. GPVI-promoted Rap1b activation is partially dependent on secreted ADP,¹⁸  and actually it was decreased in the presence of apyrase. However, even in the presence of apyrase, convulxin-induced activation of Rap1b was still dependent exclusively on PI3Kβ but not on PI3Kγ enzymatic activity (Figure 1Cii). These results indicate that the catalytic activity of PI3Kβ, but not of PI3Kγ, is involved in GPVI signaling and is required for Akt phosphorylation, Rap1b activation, and platelet aggregation.

Finally, we analyzed the involvement of PI3Kβ and PI3Kγ in integrin α_IIbβ₃-mediated outside-in signaling. An evident and significant reduction of adhesion and spreading on immobilized fibrinogen was observed in the PI3Kβ^KD but not in the PI3Kγ^KD platelets (Figure 2A). Quantitative analysis confirmed that apparent reduction of PI3Kγ^KD adhesion was not statistically significant, and that, in contrast, the defect of PI3Kβ^KD platelet adhesion was more pronounced and statistically significant both at an early time point and late time points where a reduction of approximately 50% was measured (Figure 2B). According to previous findings,¹²  we found that catalytic inactivation of PI3Kγ slowed the initial rate of platelet spreading on fibrinogen, but this initial small defect, which was not statistically significant, was completely recovered within 60 minutes (Figure 1C). In contrast, PI3Kβ^KD platelets were almost completely unable to spread on fibrinogen at every time point analyzed (Figure 1C).

In conclusion, our comparative analysis of murine platelets expressing catalytically inactive forms of PI3Kβ and PI3Kγ demonstrated that both of these isoforms are involved, at different extents and with a complex interplay, in GPCR-mediated platelet activation, suggesting the potential usefulness of double selective inhibitors. In contrast, we have demonstrated that PI3Kβ is essential for the tyrosine phosphorylation-based signaling pathways triggered by GPVI or α_IIbβ₃. PI3Kβ has been recently suggested as being mainly activated by GPCRs.¹⁹  Our data demonstrate that PI3Kβ is a common signaling platform shared by both ITAM signaling and GPCRs, but they also indicate that it plays a predominant role in ITAM- and integrin-mediated platelet activation.

This work was supported by grants from the Ministero dell'Istruzione, Università e Ricerca Scientifica (PRIN 2006; C.B.) and from the Regione Lombardia and the University of Pavia (project REGLOM16; I.C.).

Contribution: I.C. and L.S. designed and performed experiments and analyzed data; L.C. and C.G. performed experiments; E.C. provided vital reagents and performed experiments; C.B. analyzed data and edited the manuscript; E.H. contributed vital new reagents, analyzed data, and edited the manuscript; and M.T. designed research, analyzed data, wrote the manuscript, and provided overall direction.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

The current address for Dr Stefanini is Department of Medicine, Thomas Jefferson University, Philadelphia, PA 19107.

Correspondence: Mauro Torti, Department of Biochemistry, University of Pavia, via Bassi 21, 27100 Pavia, Italy; e-mail: mtorti@unipv.it.