Never Say Die: Survival Signaling in Large Granular Lymphocyte Leukemia

Abstract

Large granular lymphocyte (LGL) leukemia is a rare disorder of either mature cytotoxic T- or NK- cells. LGL leukemia is characterized by accumulation of cytotoxic cells in blood and infiltration in bone marrow, liver and spleen. Here we review clinical features of LGL leukemia. We focus our discussion on known survival signals believed to play role in pathogenesis of LGL leukemia and their potential therapeutic implications.

Introduction

Large granular lymphocyte (LGL) leukemia is a rare disorder of cytotoxic lymphocytes. It was first described as clonal proliferation of LGL involving blood, marrow, and spleen.¹ There are two varieties of LGL leukemia depending upon the cell of origin – leukemia of cytotoxic T-lymphocytes (T-cell LGL leukemia or T-LGL leukemia) and that of natural killer (NK) cells (NK-cell LGL leukemia or NK-LGL leukemia).² T-LGL leukemia is characterized by expansion of clonal CD3+ CD8+ T-cells (cytotoxic T-lymphocytes, CTL), while NK-LGL is characterized by expansion of CD3− LGL cells.

Despite important differences in origin and functions, CTL share several characteristics with NK-cells.³ Together, they are commonly referred to as cytotoxic cells. In peripheral smear examination, LGL are larger than most lymphocytes (almost double the size of a red blood cell) and contain azurophilic granules in cytoplasm. Hence, they are also referred to as large granular lymphocytes.

LGL are crucial effector arm of the immune system. They actively survey for virus-infected or transformed cells. Upon recognition, LGL make brief contact with target cells and induce apoptosis in the latter.^3,4 LGL use various strategies to induce apoptosis in the target cells. One way to induce apoptosis in target cells is by delivering an extracellular signal through death receptors such as Fas (CD95). Interaction of Fas with its ligand known as Fas ligand (CD95L, FasL) induces apoptosis in Fas-bearing cells. While expression of Fas is widely distributed, expression of FasL is relatively restricted to activated LGL.⁵ This distribution pattern allows for surveying of variety of cells by LGL and taking remedial actions if needed.

Another mechanism of apoptosis induction is through actions of cytotoxins. LGL contain various cytotoxins such as perforin (pore forming protein, pfp) and granzyme B (GrB) in azurophilic granules that can be localized to the synapse formed between effector and target cells. Perforin punches pores in target cells and facilitates delivery of other cytotoxins into the target cells. Among known cytotoxins, GrB is a powerful inducer of apoptosis. Once in target cells, GrB cleaves and activates various proteins such as caspases, filamin, nuclear poly(ADP-ribose) polymerase (PARP) and Bid leading to cell death in both caspase dependent and independent manners.^3,5

Clinical features of LGL leukemia

Signs and Symptoms

Clinical features of LGL leukemia have been described in greater detail elsewhere.^2,6,7 Clinically, LGL leukemia is a disorder of middle-aged individuals with median age being around 50 years. Both T- and NK-LGL leukemia may manifest as an indolent disorder or an aggressive leukemia. Aggressive NK-LGL leukemia is one of the most aggressive tumors known to humans with median survival being a few months following diagnosis. Fortunately, more than 85% LGL leukemia manifest as indolent T-cell disease with median survival of more than 10 years.^2,7,5

About a third of patients with T-LGL are asymptomatic and diagnosed coincidentally. The rest present with an array of symptoms depending upon underlying pathology. Recurrent infections due to coexistent neutropenia is a common feature of LGL leukemia Other hematologic conditions may be associated with LGL leukemia, including hemolytic anemia, pure red cell aplasia, cyclic neutropenia and aplastic anemia. Some patients present with B-symptoms such as fever, unexplained weight loss and night sweats. While rheumatoid arthritis is the most common autoimmune condition seen in LGL leukemia, LGL leukemia may be associated with a wide spectrum of autoimmune conditions.^7,8

Immunophenotype and Diagnosis

T-cell LGL leukemia is characterized by expansion of CD3+ CD8+ T-cell receptor (TCR)-αβ T-cells though rarely CD3+ CD4+ CD8+ TCRαβ or CD4− CD8− TCRγδ T-cells are involved.² It was recently shown that leukemic T-LGL have CD3+ CD8+ CD45RA+ CD62L− phenotype consistent with effector-memory RA T-cells (T_EMRA).^2,9 Leukemic T-LGL can be considered as malignant expansion of T_EMRA cells. Leukemic T-LGL often express CD57+ which is a marker for mature T-cells. NK-LGL leukemia is characterized by CD3− CD56+ and/or CD16+ cells.²

LGL leukemia is diagnosed using hematological workup and examination of peripheral smear. Persistent elevated LGL count, as demonstrated by a typical LGL morphology on peripheral smear, points towards LGL leukemia. Flow cytometry and molecular clonality studies (such as TCR rearrangement assays) can be useful in suggesting or establishing clonal origin of cells in T-LGL leukemia. Absence of TCR-restriction can make diagnosis difficult for NK-LGL, but the study of NK-receptor repertoire may suggest clonality in NK-LGL leukemia patients.^2,7,10

Treatment

The mainstay of treatment in LGL leukemia is immunosuppressive therapy rather than chemotherapy. LGL leukemia patients with indolent course usually do not require any treatment. In such cases, regular follow-up is important to establish course of the disease and monitor associated symptoms.^7,10 Indications for therapy include symptomatic anemia/neutropenia or severe anemia (transfusion dependent) or neutropenia (absolute neutrophil count <500/μl). While there is no established standard of care, choices of immunosuppressive therapy include methotrexate, cyclophosphamide, cyclosporine A. Corticosteroids such as prednisone may be used to hasten the clinical response. Treatment options such as fludarabine, anti-CD52 monoclonal antibody (alemtuzumab), antithymocyte globulin (ATG) or splenectomy are considered second-line options in case of failure with the first-line treatment.⁷

Pathophysiology of CTL homeostasis

Activation-induced cell death in T-cell homeostasis

In periphery, antigen encounter by an antigen-specific naïve T-cell leads to proliferation of that T-cell. CD8+ T-cells require three stimuli to be fully activated: i) antigen-receptor signaling that is mediated through TCR-CD3 complex, ii) co-stimulation provided by CD28 with its ligands CD80 or CD86, and iii) a ‘third signal’ which is provided by cytokines such as IL-12, type I interferons (IFN)-α and β, or type II interferon such as IFN-γ.^11,12

Within days, vigorous proliferation leads to increase in antigen-specific T-cells by about 50,000-fold.¹³ This is accompanied by acquisition of effector functions.¹⁴ Unchecked proliferation and cytotoxicity of CTL is not desirable due to the risk of developing autoimmunity or malignancy. Thus, most of these cells are selectively eliminated following antigen clearance.¹⁴ This process is called activation-induced cell death (AICD), in that apoptosis of CTL follows and is induced by TCR-mediated activation. AICD is important in maintaining T-cell homeostasis as well as tolerance to self-antigen in periphery.¹⁵ One mechanism of AICD is through interaction of Fas with FasL.^5,15 Activation of CTL upregulates the surface expression of both Fas and FasL. This ensures that activated CTL can be effectively eliminated via FasL-mediated apoptosis either through autocrine or paracrine fashion after infection is cleared.

Uncoupling of AICD in LGL leukemia

Gene expression profiling carried out using microarray technique showed a unique gene expression signature in LGL leukemia PBMC.¹⁶ Expression Analysis Systematic Explorer (EASE)¹⁷ suggested that while leukemic LGL show expression pattern in agreement with acquisition of effector functions, they showed severe dysregulation in apoptotic machinery. Various genes known to have pro-apoptotic function were downregulated while those with known anti-apoptotic functions were upregulated. Thus, in leukemic LGL the processes of activation and apoptosis that are normally tightly coupled, appeared to have been uncoupled, leading to inhibition of AICD.¹⁶

Biology of Fas-mediated apoptosis

Fas is a member of tumor necrosis factor receptor (TNFR) family of proteins that plays role in CTL-mediated apoptosis of target cells including other activated CTL.^14,18 Interaction of FasL with its receptor leads to trimerization of Fas. The cytosolic portion of the trimerized receptor complex binds to an adaptor protein known as Fas-associated death domain (FADD). Collectively, this complex is called death-inducing signaling complex (DISC).¹⁸ Formation of DISC allows for binding and activation of a zymogen known as procaspase-8. Recruitment of procaspase-8 to DISC leads to its cleavage and formation of heterotetramers of two p10 and p18 subunits each.¹⁹ Activated caspase-8 is the key-initiator of death receptor-mediated apoptosis activating other caspases (known as effector caspases). Activation of caspase cascade eventually culminates in cell death. DISC formation and activation of caspase-8 are, in part, negatively regulated by cellular FADD-like IL1-converting enzyme (FLICE)-inhibitory protein (c-FLIP, see below).¹⁴

Dysregulation of Fas-mediated apoptosis in leukemic LGL

Like activated normal CTL, leukemic LGL express abundant Fas and FasL on their surfaces. However, while normal activated CTL readily undergo Fas-FasL mediated apoptosis, leukemic LGL are resistant to FasL-mediated apoptosis.²⁰ This raises two possibilities: i) Fas-FasL apoptotic machinery is deficient in leukemic LGL or ii) some constitutively present survival signals keep leukemic LGL alive even in face of Fas-mediated death signals. Both these possibilities have been investigated thoroughly.

Mouse models harboring defective Fas-FasL apoptotic machinery show severe defects in immune system homeostasis. gld and lpr mice harbor dysfunctional FasL and Fas mutations respectively. Both mice show similar phenotype with accumulation of CD4+ CD8+ double positive lymphocytes, lymphadenopathy, splenomegaly and autoimmune diseases. Similar mutations in humans result in autoimmune lymphoproliferative syndrome (ALPS).^20,21

LGL leukemia patients show persistent lymphocytosis, hypergammaglobulinemia, splenomegaly and significant association with autoimmune diseases. However, they do not have lymphadenopathy and lymphocytosis is typically that of CD4− CD8+ cells. Moreover, in contrast with lpr and gld mice as well as ALPS patients, LGL leukemia patients do not carry any known mutation in Fas or FasL.^2,22 In vitro treatment with interleukin-2 (IL2), phytohemagglutinin and IL2, or ceramide sensitize leukemic LGL to Fas-mediated apoptosis.^9,23 Also, inhibitors of various survival signaling pathways (such as AG490 and FTY720, see below) restore Fas-sensitivity in leukemic LGL suggesting intact Fas-FasL apoptotic machinery.^16,23 Together, these data suggest that although leukemic LGL are capable of undergoing Fas-mediated apoptosis, various survival signals keep them from doing so (Figure 1).

Figure 1. MODEL OF APOPTOSIS-RESISTANCE IN LEUKEMIC LGL.

Cytotoxic cells kill the target cells by one of the two mechanisms. Activated cytotoxic cells upregulate FasL on their surfaces, while target cells express Fas. This facilitates Fas-FasL mediated apoptosis in the target cells (upper panel). Another mechanism is through exocytosis of cytolytic granules from CTL into the target cells. Similar mechanism operates to maintain T-cell homeostasis. CTL use Fas-FasL mediated apoptosis to induce apoptosis either in self or another antigen-specific T-cells (middle panel). Leukemic LGL, despite expressing abundant Fas and FasL, are resistant to Fas-mediated apoptosis. It is believed that survival signaling overcomes Fas-mediated apoptotic signals in leukemic LGL leading to their accumulation (lower panel). CTL - cytotoxic T-lymphocyte; Mt. pathway – mitochondrial pathway.

Role of soluble Fas and FasL in LGL leukemia

Known mutations in Fas or FasL genes were not found in LGL leukemia. However, various isoforms of both Fas and FasL were found in sera of LGL leukemia patients. Sera from LGL leukemia patients contained elevated levels of soluble form of Fas receptor (sFas). sFas blocked Fas-mediated apoptosis in both activated normal PBMC and IL2-treated leukemic LGL. It was proposed that sFas may work as a decoy for FasL, resulting in Fas-resistance phenotype of leukemic LGL.²² Indeed, the source of these sFas variants in serum was traced to LGL leukemia PBMC. LGL leukemia PBMC was found to express alternative spliced Fas variants not seen in normal naïve or activated PBMC. When overexpressed, these variants were secreted in supernatant. The supernatant containing these sFas variants blocked Fas-mediated apoptosis of leukemic LGL.²²

Presence of soluble form of FasL (sFasL) is also known. sFasL can be produced by alternative splicing or proteolytic cleavage of membrane bound FasL by matrix metalloproteinase (MMP) family of enzymes.^24,25 LGL leukemia patients have high amounts of sFasL in serum whereas normal serum does not contain any detectable levels of sFasL. It was suggested that neutralizing antibody to sFasL or MMP inhibitors may be used in treating LGL leukemia with autoimmune diseases.^22,26

Abnormal DISC formation

DISC formation is the immediate downstream event of Fas-FasL ligation in Fas-mediated apoptosis. Both short and long forms of FLIP (known as c-FLIP_S and c-FLIP_L respectively) contain caspase-homologous regions, enabling them to be recruited to DISC. However, FLIP lack proteolytic capabilities of caspase-8. Thus, recruitment of FLIP to DISC inhibits the execution of Fas-induced apoptosis signals mediated by caspases. FLIP not only inhibit Fas-mediated apoptosis, but are also known to induce NF-κB- and Erk-mediated proliferation in T-cells.¹⁴ Downregulation of FLIP is seen towards the end of CTL response and correlates with their increased sensitivity to Fas-mediated apoptosis. In contrast, leukemic LGL express higher basal levels of both isoforms of FLIP that may contribute to Fas-resistant phenotype of leukemic LGL.⁹

Role of Ras-Mek-Erk Signaling Pathway

Ras has a well-established role in tumor biology. Mutations of Ras occur in about 30% of all human cancers.²⁷ Ras family of proteins belongs to guanidine triphosphatase (GTPase). In its active (GTP-bound) form, Ras engages various downstream effector pathways that play essential role in cellular responses such as survival, proliferation and differentiation.²⁷

Ras cascade is the major pro-survival regulator following T-cell activation.²⁸ Ras is activated following activation signal delivered to receptor tyrosine kinases (RTK). For cytokines and growth factors, this means binding of ligand to its receptor. For T-cells, immunoreceptor tyrosine-based activation motifs (ITAM) of TCR translate the signal of antigen engagement to Ras. Ras signaling can be activated by Src-family kinases (SFK), platelet-derived growth factor (PDGF), or sphingosine-1-phosphate (S1P) through Gα₁₂.^29–32 This allows Ras to bind to GTP and get activated.

Ras activity is also regulated by post-transcriptional regulations. Prenyl transferases are a class of enzymes that include farnesyl transferases (FT) and geranylgeranyl transferases (GGT). Prenyl-transferases modify Ras activity by adding either one or two hydrophobic moieties on C-terminus of RAS. This modification is essential to anchor Ras on cytosolic leaflet of cellular membranes which is a pre-requisite for activation.^27,28

Once activated, Ras phosphorylates and activates Raf-1. Raf-1 phosphorylates MAPK-extracellular regulated kinase (ERK) kinase (MEK) resulting in its activation. Activated MEK in turn phosphorylates ERK resulting in its translocation to nucleus. In the nucleus, ERK activates various transcription factors, including Fos and Jun of Ets family, which contribute to resulting in proliferation, differentiation or survival of cells.^27,28,33 Ras signaling also promotes survival by directly promoting the transcription of FLIP and MCL1.^34,35 In Jurkat T-cells MAPK/ERK activity is inversely proportional to Fas-sensitivity and anti-apoptotic activity of MAPK/ERK signaling overrides Fas-mediated apoptotic signals.³⁶

Dysregulation of Ras-Mek-Erk signaling in pathogenesis of LGL leukemia

LGL leukemia patients harbor constitutively active form of Ras (H-Ras-GTP) in their PBMC.³⁷ Ras-Mek-Erk signaling was found to be constitutively activated in leukemic LGL. Inhibition of RAS, MEK or ERK induced apoptosis in as well as restored Fas-sensitivity in leukemic LGL.³⁷ Inhibition of Ras either using chemical inhibitor FTI2153 or by overexpressing dominant negative form of Ras, induced apoptosis in leukemic LGL by inhibiting ERK activity. Similar results were obtained using MAPK inhibitors (PD98059 or U0216) that induced apoptosis in leukemic LGL and restored Fas-sensitivity in Erk-dependent manner. These results suggest that overactive RAS and MEK lies upstream to ERK in mediating survival signals in leukemic LGL.³⁷

Therapeutic implications

R1150777 (Zarnestra, Tipifarnib) is a farnesyl transferase inhibitor (FTI) designed to inhibit Ras pathway. Zarnestra is being investigated as a potential treatment in various tumors including leukemia.³⁸ Since Ras isoforms such as H-Ras requires farnesylation for malignant transformation activity,³⁹ it was hypothesized that by inhibiting farnesylation of Ras, Zarnestra would inhibit Ras-mediated signaling in LGL leukemia. A clinical trial was conducted on eight LGL leukemia patients using this drug. While none of the patients achieved clinical response, interesting biological responses were observed in most of these patients. One patient with NK-LGL leukemia had improvement in symptoms and signs of pulmonary hypertension while receiving Zarnestra (Epling-Burnette et al, in press).

Role of PI3k-Akt Signaling Pathway

Among downstream effector pathways of Ras cascade, phosphoinositide-3-kinase (PI3k) -v-akt murine thymoma viral oncogene homolog (Akt) mediated signaling has well established role in metabolism, survival and proliferation. With such cell stimulating profile, it is hardly any surprise that dysregulation of PI3k-Akt signaling is seen in various human tumors. This dysregulation is commonly caused by the deletion or mutation of a negative regulator of PI3k-Akt signaling, phosphatase and tensin homolog (PTEN).⁴⁰

In T-cells, PI3k-Akt signaling plays role in T-cell receptor mediated activation and proliferation. PI3K is activated upon membrane relocation. This relocation is mediated by the interaction of PI3K with RAS or SFK, or through the direct interaction of PI3K with cytokine receptors at the SH2-domain binding site.^33,40,41 The most studied component downstream of PI3K is a serine/threonine kinase - AKT (also known as protein kinase B, PKB). Activation of AKT accounts for many of the biological functions of PI3K. To promote proliferation, AKT acts through inducing cyclin D1, and mammalian target of rapamycin (mTOR) pathway, as well as downregulating forkhead box class O (FOXO) transcription factors and cyclin-dependent kinase inhibitors (CKI) such as p21^WAF and p27^KIP1.

One of the main targets downstream to Akt in pro-survival signaling is NF-κB signaling (see below). Further, AKT phosphorylates Bcl-2 antagonist of cell death (BAD), preventing its interaction with anti-apoptotic factor Bcl-x_L. This leaves Bcl-x_L to exert its anti-apoptotic functions. AKT phosphorylates and inhibits caspase-9, thus inhibiting apoptosis. Another anti-apoptotic phosphorylation target of AKT is MDM2 – a negative regulator of tumor suppressor p53. Phosphorylated MDM2 binds to p53, expediting its degradation and interfering with tumor suppression effects of the later.^42,43 Following activation, PI3k-Akt signaling can enhance survival of T-cells through inhibition of Fas clustering and DISC formation. This mechanism may contribute to abnormal DISC formation and Fas-resistance phenotype observed in leukemic LGL.^9,20,44

Dysregulation of PI3k-Akt pathway in LGL leukemia

In LGL leukemia PBMC, SFK maintain PI3K in its constitutively activated form as assessed by phosphorylation of AKT and glycogen synthase kinase-3 (GSK3). Inhibition of SFK or PI3K induced apoptosis in leukemic LGL is accompanied by inhibition of ERK1/2 activity. It was proposed that SFK-mediated activation of PI3k-Akt pathway results in constitutive ERK activity in leukemic LGL. This placed Akt upstream to Erk in LGL survival signaling. Inhibition of this pathway at any level – namely inhibition of SFK, AKT or ERK –induced apoptosis in leukemic LGL.^37,41

Therapeutic Implications

Due to their role in promoting proliferation and survival, PI3k and Akt are good candidates for anti-tumor therapy. Inhibitor of PI3K such as PI-103 and ZSTK474 have shown promise as anti-tumor agents.⁴² It would be interesting to see if these novel agents possess therapeutic potential in treatment of LGL leukemia.

Role of Nuclear factor kappa-B (NF-κB) Signaling Pathway

NF-κB was first identified as an enhancer of immunoglobulin κ-light chain in activated B-cells.⁴⁵ Subsequently, it was realized that NF-κB plays an essential role in hematopoiesis, inflammation, as well as survival and proliferation of adaptive immune system cells. Thus, NF-κB is now recognized as a critical player in almost all the aspects of immune responses.^12,45 Normally, NF-κB is found in cytoplasm as a complex with the inhibitor of NF-κB (IκB). This complex keeps NF-κB from both entering to nucleus and binding to DNA, thus depriving its transcriptional functions. This inhibition is void once IκB is phosphorylated by (IκB)-kinase complex (IKK). Phsophorylation of IκB, leads to its ubiquitination and proteosomal degradation. IKK is a known AKT substrate, rendering NF-κB downstream to PI3K-AKT pathway. In summary, phosphorylation of IKK by AKT leads to activation of the IKK, which in turn leads to phosphorylation and degradation of IκB, leading to NF-κB activation.⁴²

Activation of NF-κB is one of the most characterized pathways in antigen-receptor signaling in both B- and T-cells. Activation of NF-κB downstream of TCR ligation facilitates antigen specific proliferation and maturation of lymphocytes into effector cells. NF-κB orchestrates T-cell activation by providing a milieu to proliferate (such as inducing IL2 production) and acquire effector functions (such as promoting transcription of RANTES, Fas and FasL).⁴⁶

The most important of NF-κB functions is protection of T-cells against AICD.⁴⁷ This function is executed primarily through promoting the expression of pro-survival Bcl-2 family members and inhibitors of apoptosis (IAPs).⁴⁶ Deficiency or inhibition of NF-κB activity results in failure of activation, either due to lack of proliferation or premature onset of apoptosis suggesting crucial role of NF-κB signaling in mounting effective T-cell response.^12,48–50

Abnormal NF-κB signaling in survival of leukemic LGL

The pivotal role that NF-κB plays in CD8+ T-cell activation and survival along with an established role in various aspects of tumorigenesis as well as inflammation makes NF-κB an interesting candidate to study in LGL leukemia. Gene expression signature of leukemic LGL showed that c-Rel was overexpressed in leukemic LGL.¹⁶ We recently found that leukemic LGL show constitutively active NF-κB. We further found that the inhibition of NF-κB resulted in apoptosis of leukemic LGL. Inhibition of Akt led to inhibition of NF-κB activity whereas, inhibition of NF-κB activity did not affect Akt phosphorylation. This suggests that NF-κB acts downstream of PI3k-Akt pathway in leukemic LGL. We also found that NF-κB maintains expression of Mcl-1 independent of STAT3 activity.⁵¹

Role of Jak-Stat Signaling Pathway

Janus kinase-signal transducers and activators of transcription (Jak-Stat) signaling cascade plays role in conferring survival in various tumors. It is also known to be activated following T-cell activation. JAK proteins are kinases that phosphorylate STAT proteins. STAT proteins are latent transcription factors that, upon activation, transcribe various known anti-apoptotic genes. Jak-Stat pathway plays an essential role following cytokine signaling. Four members of Jak family and seven members of Stat family are known in humans. Stat family members may form homo- or heterodimers resulting in various combinations that may have overlapping but distinct transcription profiles.^52,53

Upon activation of cytokine or growth factor receptor, aggregation of receptors leads to transphosphorylation of JAK leading to its activation. Activated JAK can now bind to STAT. STAT proteins contain an N-terminus dimerization domain, a central DNA-binding domain and a C-terminus transactivation domain. Phosphorylation of tyrosine and threonine (believed to be mediated by ERK) of the dimerization domain allows STAT proteins to form homo- or heterodimers. Dimerized STAT proteins then translocate to nucleus where they exert their transcriptional activities by binding to enhancer regions of the target genes including Bcl-xL, myeloid cell leukemia sequence 1 (Mcl-1), IAP-family of protein survivin (BIRC5), cell-cycle regulator cyclin D1, c-Myc and vascular endothelial growth factor (VEGF).⁵⁴ While physiological activation of STATs last for a few minutes to a few hours, constitutively activated STATs are frequently found in a wide variety of human tumors.³³ Transformations mediated by oncogenes such as v-src, v-abl, v-fps, v-fes, v-eyk and Gα₁₂ have been shown to be mediated by STAT especially by STAT3.^23,55,56

Role of Jak-Stat pathway in pathogenesis of LGL leukemia

Leukemic LGL harbor constitutively activated STAT1 and/or STAT3 but not STAT5. STAT3 and/or STAT1 dimers from LGL leukemia patients’ PBMC showed DNA binding activity equivalent to that of in vitro activated normal PBMC, suggesting that leukemic LGL are activated in vivo. Inhibition of JAK2/3 using small molecular tyrosine kinase inhibitor AG490 induced apoptosis in leukemic LGL as well as restored Fas-sensitivity. Specific inhibition of STAT3 using antisense to STAT3 induced significant apoptosis as well as restored Fas-sensitivity in leukemic LGL.

The promoter region of human MCL1 - a BCL2 family member important for maintaining mitochondria integrity - contains a STAT3 binding site. In leukemic LGL, STAT3 binds to this site and induces expression of MCL1. The induction of apoptosis through STAT3 inhibition correlates with decreased MCL1 expression indicating a role of anti-apoptotic protein MCL1 in survival of leukemic LGL.²³

Therapeutic implications

Recently, there is a great interest in finding specific methods to inhibit Stat3 signaling. Small molecules inhibitors (JSI124 and platinum (IV) compounds such as CPA-7) or peptide-based inhibitors specific to STAT3,^57–60 have shown promising results in specifically inhibiting Stat3-mediated signaling. Given the role of Jak-Stat pathway in survival of leukemic LGL, it will be interesting to see if these novel agents can be used as therapeutic strategies in LGL leukemia patients.

Role of Sphingolipid Rheostat

Sphingolipids are biologically active lysophospholipids that act either directly or as second messengers to regulate diverse biological functions such as survival, proliferation, calcium homeostasis and migration. One important sphingolipid, ceramide (N-acyl sphingosine), can be synthesized either de novo (by condensation of serine and palmytoyl-CoA) or by catabolic pathway (by breaking down sphingomyelin). Ceramide is a pro-apoptotic molecule that is synthesized in cells following wide variety of stress or death signals, including Fas-FasL interaction. Ceramide can be deacytylated into sphingosine by ceramidases such as acid ceramidase (ASAH1) or phosphorylated to ceramide-1-phosphate (C1P) by ceramide kinase. Sphingosine, then, can be phosphorylated by one of the two sphingosine kinases (SPHK) into sphingosine-1-phosphate (S1P). Though structurally closely related to ceramide, S1P is a pro-survival molecule. Given that pro-apoptotic (such as ceramide and sphingosine) and anti-apoptotic (such as S1P and C1P) sphingolipids exist in a fast exchanging equilibrium, it has been proposed that the relative amount (termed sphingolipid rheostat), rather than their absolute quantity, determines cell fate.^42,61,62

Role of sphingolipids - especially that of S1P - in oncogenesis, metastasis and angiogenesis is well established.⁶³ S1P, presumably through S1P-receptor (S1PR), plays protective role in T-cell survival by protecting cells against ceramide and Fas-FasL mediated apoptosis.^42,64,65 The components that can lead to excessive S1P production or constitutive S1P-mediated signaling may act as an oncogene. The roles of three components of sphingolipid rheostat are well studied - acid ceramidase, SPHK and Gα₁₂.

Acid ceramidase functions upstream to SPHK in sphingolipid metabolism (Figure 2). Upregulation of ceramidase is a survival mechanism used by variety of human tumors to combat ceramide production that normally follows various apoptosis-inducing insults.^66–68 A chemical inhibitor of acid ceramidase, N-oleoylethanolamine (NOE), selectively induces apoptosis in various tumor cells.⁶⁹

Figure 2. SPHINGOLIPID RHEOSTAT AND ITS ROLE IN SURVIVAL OF LEUKEMIC LGL.

Sphingolipid signaling is determined by balance between pro- and anti-apoptotic sphingolipid molecules. It is proposed that deregulation in sphingolipid signaling shifts the balance towards survival in leukemic LGL. NOE is an inhibitor of acid ceramidase (ASAH1), while SKI-I and II selectively inhibit sphingosine kinase (SPHK). FTY720 is an immunomodulator that acts as an functional antagonist of sphingosine-1-phosphate (S1P)-mediated signaling. Treatment with any of these molecules selectively induced apoptosis in leukemic LGL.

SPHK is overexpressed in variety of tumors and is considered an oncogene.^67,70 Activation of various survival signaling pathways implicated in tumorigenesis result in activation of SPHK. SPHK in turn activates many of these pathways constituting a positive feedback loop. For example, PDGF and vascular endothelial growth factor (VEGF), are known to activate SPHK. Persistent elevation of phosphatidylinositol (3,4,5) trisphosphate results in activation of SPHK linking PI3k-Akt cascade to sphingolipid signaling. Similarly, ERK activation leads to activation of SPHK linking Ras-Mek-Erk signaling to sphingolipid rheostat. Activated SPHK feeds into several proliferative and pro-survival pathways by activating ERK, PI3K and NF-κB.^42,71 SPHK promotes cell survival by increasing cellular concentration of S1P while reducing concentrations of ceramide and sphingosine. This puts SPHK in very crucial position where various survival signaling pathways converge, rendering it an interesting candidate for anti-tumor therapy.

Gα₁₂ or GNA12 is a G-protein that is coupled with various G-protein coupled receptors (GPCR). In particular, Gα₁₂ is coupled with S1PRs S1P₁ and S1P₅. Constitutive active form of Gα₁₂ is sufficient to induce transformation that is at least partially mediated by activating STAT3. The components believed to play a role in Gα₁₂-mediated transformation include JAK3, PDGFα, and PI3K.⁵⁶ All these components are known to be deregulated in LGL leukemia suggesting involvement of Gα₁₂ in survival of leukemic LGL.^23,41,51

S1P-receptors in normal CD8+ T-cells and leukemic LGL

In immune system, S1P acts in autocrine or paracrine manner, either intracellularly or through one of the S1PR on the cell surface. There are at least five highly-specific S1PR known in humans – S1P₁ through S1P₅.⁶² In T-cells, S1PR-mediated signaling plays an important role in egress from lymphoid organs following activation.⁷² S1P also has a protective role against Fas-mediated apoptosis in various T-cell lines as well as in human PBMC.^16,64,65

S1P₁ is the most predominant S1PR in human naïve CD8+ T-cells, while other receptors are expressed at very low levels.¹⁶ In contrast to normal CD8+ T-cells, S1P₅ was found to be the predominant S1PR on leukemic LGL. Following activation, S1P₁ and S1P₅ are further downregulated in normal CD8+ cells. S1P₁ was downregulated in LGL leukemia PBMC suggesting that leukemic LGL are activated in vivo.^16,73 In contrast, S1P₅ was overexpressed by 4.5-fold compared to naive normal CD8+ cells and by 235-fold compared to activated normal CD8+ cells. Whether S1P₅ plays any role in survival of leukemic LGL remains to be established.

Dysregulation of sphingolipid rheostat in LGL leukemia

The sphingolipid rheostat described above, is dysregulated in LGL leukemia. The first clue about involvement of sphingolipid signaling in survival of leukemic LGL came when human S1P₅ was originally identified as an overexpressed expressed sequence tag (EST) using DNA library from LGL leukemia PBMC.⁷⁴ Later, analysis of gene signature established that sphingolipid metabolism as well as Gα₁₂-mediated signaling was enriched in LGL leukemia PBMC. Acid ceramidase was identified as a core enriched component in leukemic LGL. In normal PBMC, acid ceramidase is downregulated to undetectable levels following activation. This is in agreement with Fas-sensitive phenotype of activated normal PBMC. In contrast, leukemic LGL express abundant acid ceramidase presumably facilitating breakdown of ceramide, perhaps explaining Fas-resistant phenotype observed in leukemic LGL.^16,20

It was proposed that disruption of sphingolipid rheostat in a way that tilts the balance in favor of ceramide (and other pro-apoptotic molecules) and away from S1P (and other anti-apoptotic molecules) should lead to induction of apoptosis in in leukemic LGL (Figure 2). Indeed, inhibition of acid ceramidase using NOE induced significant apoptosis in leukemic LGL.¹⁶ Similarly, inhibition of SPHK using SPHK inhibitors SKI-I and SKI-II leads to induction of apoptosis in leukemic LGL.⁵¹ Inhibition of S1P-mediated signaling using FTY720 led to induction of apoptosis in leukemic LGL but not in normal LGL. In contrast, disruption of sphingolipid metabolism outside the rheostat did not induce apoptosis in leukemic LGL.¹⁶ FTY720 treatment restored sensitivity to Fas-mediated apoptosis suggesting a role of S1P-mediated signaling in protection against AICD.¹⁶

Therapeutic opportunities

FTY720 or Fingolimod is a novel immunomodulatory compound in clinical trials for post-renal transplant and in autoimmune conditions such as multiple sclerosis (MS). FTY720 is an analogue of S1P that binds to four out of five S1PR (except S1P₂) and acts as a functional antagonist.⁷³ Given that FTY720 selectively induces apoptosis of leukemic LGL and that LGL leukemia patients often have coexisting autoimmune diseases, it is possible that FTY720 may have a therapeutic role in these patients.

Sphingosine kinase inhibitors-I and -II (SKI-I and II) are known to have anti-tumor activity and have proven their anti-tumor activity in mouse models.⁷⁵ Development of newer, more selective inhibitors targeting SPHK offer exciting therapeutic opportunities.⁷⁶

Neutralizing antibody to S1P has been proposed as a candidate for blocking survival and angiogenic signaling in various tumors. It works as a ‘sponge’ to absorb extracellular S1P - leading to abolition of S1P-mediated protection against apoptosis.^63,70 However, there are no reports of neutralizing antibody to S1P as a potential in therapeutic in leukemia. It remains to be seen if it might be a potential therapy for LGL leukemia.

Serine Proteinase Inhibitor 9 (SERPINB9)

During down-phase of T-cell activation, activated CTL induce apoptosis in other activated CTL through granule exocytosis and death receptor pathways. Leukemic LGL contain perforin and GrB as well as other proteases in their granules. Leukemic LGL also express abundant Fas and FasL on their surface and possess intact Fas-machinery. Given these findings, it is puzzling that leukemic LGL are resistant to apoptosis mediated by either of these pathways. It is believed that there should be survival signals operating in leukemic LGL that would inhibit both granule exocytosis and Fas-FasL mediated apoptosis. One such candidate is a homologue of cowpox virus protein cytokine response modifier A (CrmA) known as SERPINB9 (PI6). SERPINB9 binds to and inactivates serine proteinases including GrB and caspase-8.^3,77,78

Upregulation of SERPINB9 is sufficient to inhibit both GrB and Fas-FasL mediated apoptosis in various in vivo and in vitro models.⁷⁹ It is hypothesized that upregulation of SERPINB9 may protect leukemic LGL from killing each other thus helping to escape AICD. Microarray analysis suggested that while activation led to downregulation of SERPINB9 in normal CD8+ cells, it was significantly upregulated in leukemic LGL. We confirmed these findings using quantitative real-time PCR (qRT-PCR). qRT-PCR analyses show that leukemic LGL express greater than 30-fold higher SERPINB9 compared to activated CD8+ cells from healthy donors (Figure 3).

Figure 3. SERPINB9 IS OVEREXPRESSED IN LEUKEMIC LGL COMPARED TO ACTIVATED NORMAL ENRICHED CD8+ CELLS.

Quantitative real-time PCR was performed and analyzed as described.16 Primers used for SERPINB9 are : Forward 5’-AGATGGCCCAGGCACTGTC-3’ and Reverse: 5’- AGAGCCTGTTGGCCGTTCTC-3’. Figure shows that SERPINB9 is overexpressed in T-LGL leukemia PBMC (●, n=9) by more than 30-fold compared to activated enriched normal CD8+ T-cells (○, n=3; P<0.05). Each dot represents expression in an individual PBMC sample, while bar (−) represents mean of observations in a group.

Cross-talk of survival pathways and network modeling approach

Signaling cascades serve triple purposes – they carry message from cell surface into nucleus, amplify the signal, and regulate the cell in order to adapt to the environment. In doing so, survival pathways do not work in isolation. Instead, they are involved in dynamic and complex interaction (Figure 4). Since experimental results are usually focused on limited interactions of components in one pathway, a systematic analysis of pathway crosstalk is desirable to fully understand the cause of such abnormal survival of leukemic LGL.

Figure 4. CROSS-TALK AMONG VARIOUS SIGNALING PATHWAYS IN LEUKEMIC LGL.

Various survival signaling pathways interact and cross-regulate each other at various levels. It is this dynamic interaction that leads to survival in leukemic LGL. Figure shows known survival signaling pathways and their interactions. Legends used: red - upregulation and/or constitutive activation, green - deregulation, blue – downregulation/or constitutive inhibition; white - no direct information available in LGL leukemia; golden – state of a cell or its component such as apoptosis and mitochondrial integrity loss, rectangle – intracellular component, circle - extracellular component; blue edges – activation/enhancement, red edges – inhibition or downregulation.

Network modeling has been increasingly utilized to better understand complex and interactive biological systems.^80,81 Experimentally obtained signaling pathway information can be translated into a network by representing proteins, transcripts and small molecules as network nodes, and denoting the interactions between nodes as edges.⁸⁰ The simplest discrete models, called Boolean models, assume two possible states for each node in the network: ON (meaning above threshold) and OFF (below threshold). With Boolean operators AND, OR and NOT, the biological functions by which upstream regulators act on a downstream node can be readily translated into logical statements.

We constructed a T-LGL leukemia survival signaling network through integrating signaling pathways involved in normal CTL activation and known deregulations of survival signaling in leukemic T-LGL. This network was subsequently translated into a predictive discrete dynamic Boolean model. By simulating node deregulations corresponding to known signaling abnormalities, we concluded that the minimum condition to reproduce all known deregulations in leukemic T-LGL is constitutive presence of IL15, presence of PDGF, and initial T-cell activation signal. The predictions made by network analysis were further validated experimentally.⁵¹

Concluding Remarks

LGL leukemia is a disorder resulting from dysregulation of apoptosis. The uncoupling of activation and AICD provide a unique opportunity to reveal the survival signaling pathways responsible for persistence of leukemic LGL. In this review, we summarized the up-to-date knowledge of survival deregulations in leukemic LGL. In addition, we proposed a systematic approach to examine network interactions utilizing LGL leukemia as a model disease.

Maintenance of long-lived and functional cytotoxic cells is essential for rapid and effective response against chronic viral infections and tumors. Mere presence of activated CTL against a viral or tumor antigen is not sufficient for generating long-lasting effector response in vivo as highlighted by recent failure of CTL-based HIV vaccine.⁸² Leukemic LGL are long-term surviving, antigen-specific functional cytotoxic cells. Thus, understanding of survival mechanisms in leukemic LGL may provide insight in vaccine development against chronic viral infections (such as HIV) and tumors.

On the other hand following antigen clearance, tight control of number and function of cytotoxic cells is desirable since deregulation of cytotoxic cell homeostasis may lead to autoimmune disorders. Clinically, LGL leukemia is associated with a wide array of autoimmune disorders. Thus survival signaling in leukemic LGL may also help decipher pathogenesis of autoimmune disorders.