PI3K are a family of lipid kinases known to phosphorylate the inositol ring of the 3-OH group in inositol phospholipids. They are further classified into three classes: Class I, II, and III based on primary structure and regulation[1]. However, till date only Class I, assisting in tight regulation of this pathway, has been shown to be associated with cancer. Class I PI3K is a heterodimeric enzyme, with a catalytic and a regulatory subunit. The catalytic subunits for class I PI3Ks are p110α, p110β, p110γ, and p110δ. It is further subdivided into class1A, encompassing p110α, p110β, and p110δ with a p85α, p85β, and p55γ regulatory subunit, and class1B consisting of only p110γ with p101, p84, and p87PIKAP regulatory subunits. A typical regulatory subunit has several protein-protein interacting domains, one of them, the inter-SH2 domain (iSH2) interacts with the p110 catalytic subunit, stabilizing p110 and its activities[2]. Reports have suggested the p110α subunit encoded by the PIK3CA gene as being the only catalytic subunit associated with several cancers[3]. p110α typically links with p85α, which is the most highly expressed regulatory subunit. The substrate for class1 PI3K, phosphatidylinositol-4,5-biphosphate (PI-4,5-P₂), generates the second messenger phosphatidylinositol-3,4,5-triphosphate (PI-3,4,5-P₃).

Serine/threonine protein kinase Akt belongs to the AGC [named after the protein kinase A, G, and C families (PKA, PKC, PKG)] family of kinases. Three highly homologous isoforms (Akt1, Akt2, and Akt2) of Akt have been identified so far. Structurally, Akt is mainly comprised of three domains: an N-terminal pleckstrin homology (PH) domain, a central kinase catalytic domain (CAT), and a C-terminal extension (EXT) containing a regulatory hydrophobic motif (HM). Phosphorylation of residues in both the catalytic and C-terminal extension domain is essential for complete activation of Akt downstream of PI3K signaling. PDK1 selectively phosphorylates Thr (308) on the CAT domain of Akt, while the kinases responsible for phosphorylation of Ser (473) on the EXT domain of Akt are still unknown[4]. Thus, phosphorylation of both Ser (473) and Thr (308) residues on Akt is required for its complete activation[5,6]. Fully activated Akt further regulates several processes downstream, along with positive regulation of mTOR and thereby mediating mTOR activation.

The mTOR protein, a 289-kDa serine/threonine kinase, is a master regulator of cell growth. It can be distinguished into two distinct multi-protein complexes; mTOR complex 1 (mTORC1) and mTOR complex 2 (mTORC2). The mTORC1 complex is composed of mTOR, regulatory-associated protein of mTOR (Raptor), mammalian lethal with SEC13 protein8 (MLST8), PRAS40, and DEPTOR. It functions downstream of Akt, integrating the Akt and mTOR pathway with 4E-BP1 and S6K, which are immediate downstream targets of the mTORC1complex. The mTORC2 complex, on the other hand, is composed of rapamycin-insensitive companion of mTOR, MLST8 and mammalian stress-activated protein kinase interacting protein 1. This complex functions upstream of Akt/PKB and plays a role in complete activation of Akt by phosphorylating Akt at the Ser473 residue. Hence, downstream substrates of the mTORC2 complex include Akt[7] as well as PKC[8,9]. Hence, both complexes are important for effective regulation of the Akt/mTOR dual pathway, with the mTORC1 complex responsive towards growth factors, nutrients, energy, or oxidative stress indirectly while the mTORC2 complex plays an important role towards Akt activation to assist in complete activation of the PI3K/Akt/mTOR pathway.

Akt acts as a central regulator of cell survival by interacting with anti-apoptotic signals both transcriptionally and post translationally. Akt phosphorylates Bad, a Bcl-2 family of anti-apoptotic proteins at Ser-136 and Caspase-9, a protease at Ser-196, thereby partially blocking cell death and supporting cell survival signals. Akt also regulates anti-apoptotic functions transcriptionally by translocating into the nucleus and regulating the transcription of the forkhead box O (FoxO) family of transcription factors. The FoxO family of transcription factors regulate cell death signals via expression of various members of both intrinsic and extrinsic modes of apoptosis as well as cyclin-dependent kinase inhibitors. Upon nuclear translocation, Akt represses the transcription of FoxO1, FoxO3, and FoxO4, thereby enhancing cell survival signals[11].

Akt also plays an important role in regulating cell cycle progression in normal cells. It either directly phosphorylates or indirectly regulates the protein expression levels of several molecules of cell cycle progression at the G1/S and G2/M phase of the cell cycle. These substrates are mentioned in Table 1.

Cellular metabolism of carbohydrates into proteins, nucleotides, and lipids is a fundamental aspect of cell growth and proliferation, with nutrients acting as a fuel for cell growth. mTOR plays a crucial role in regulating this metabolism in response to nutrient availability. Of the two mTOR complexes, the mTORC1 complex plays a key role in regulating cellular metabolism[12]. It receives signals of activation from nutrients and growth factors. For example, during metabolism of carbohydrates, there is a spike in insulin levels, which activates the mTORC1 complex of the Akt/mTOR pathway via inhibition of the TSC1/2 complex by phosphorylation of TSC2 at multiple sites to inhibit TSC1[13]. In this process, eventually Ras homolog enriched in brain (Rheb), a small GTPase belonging to the Ras family of guanine-nucleotide binding proteins that enhances apoptotic signalling at cellular levels[14], is inhibited upon TSC1 complex inactivation. The mTORC1 complex is also stimulated in the presence of amino acids by promoting the conversion of Ras-related GTP-binding protein (RAG) heterodimers (RAGA or RAGB, and RAGC or RAGD) into their active conformation, which further assists in mTORC1 complex cellular localization from the cytoplasm to the surface of the lysosome where it binds to inactivated RHEB[15-17]. The activated mTORC1 complex also tightly regulates pathways such as the AMP-activated protein kinase (AMPK) pathway by preventing its activation in the presence of a high ATP/AMP ratio. However, in the absence of energy in cells, AMPK gets activated by phosphorylating TSC2 at Ser1387 and Raptor from the mTORC1 complex at Ser-792, resulting in mTORC1 inactivation[18,19]. After mTORC1 activation and subsequent complete activation of the Akt/mTOR pathway, immediate downstream substrates of mTORC1 complex such as S6K (ribosomal S6 kinase), 4E-BP1, and ULK1 (UNC-51 like kinase) are phosphorylated at different residues. Interestingly, activated S6K further phosphorylates Insulin receptor substrate-1 (IRS-1), upstream of mTORC1. Phosphorylation of IRS-1 at serine residues by S6 kinases prevents IRS-1 functions and thereby PI3K activation[20]. This negative feedback loop of the PI3K/Akt/mTOR pathway is an important aspect of maintaining homeostasis in cellular metabolism, protein synthesis, and cell growth.

PI3K inhibitors are classified into three classes based on their selectivity for the ATP-binding cleft on PI3K and pharmacokinetic properties: pan-class I, isoform-selective, and dual PI3K/mTOR inhibitors[49,50].

Pan-class I inhibitors have inhibitory effects against each isoform of p110 (PIK3CA). Several pan-class I inhibitors are under investigation for GC targeted therapy, since PIK3CA gene mutations comprise 25% of gastric tumors, resulting in PI3K dysregulation in GC[51]. The first report of a molecular agent inhibiting PI3K was quecertin, which was, however, a non-specific kinase inhibitor[52]. Eventually, more specific pan-class I inhibitors were identified, such as Wortmannin and a quecertin analogue, LY294002. Although both LY294002 and Wortmannin exhibited potent PI3K-inhibitory properties, there were considerable limitations for them to proceed towards clinical trials[52-54]. LY294002 showed non-specific targeting, a short half-life, and toxicity in vivo[55,56], while Wortmannin had limitations involving biological stability and short half-life[57].

To further improve on the pharmacological availabilities, a structural analogue of Wortmannin, PX-866 was developed.

PX-866: PX-866 is a semisynthetic pan-class-1, Wortmannin analogue with inhibitory concentrations in nanomolar ranges and better efficacy and a safer pharmacokinetic profile than Wortmannin. Pre-clinical in vivo studies have shown its anti-cancer effect against several xenograft models of various cancers[57,58]. It is currently in Phase II clinical trials for patients with glioblastoma and head and neck cancer[59,60].

PX-866 recently also came under limelight for a multicenter trial for advanced solid tumors including gastric tumors. Data from the trial show that PX-866 can be administered with endurable toxicity for patients with advanced solid tumors[61].

NVP-BKM120 (Buparlisib): NVP-BKM120 is a potent pan-class I PI3K inhibitor with its activity in nanomolar ranges for all the isoforms of Class I PI3K. Preclinical investigations have revealed its effectiveness in a diverse range of cancer cell lines, with increased sensitivity in tumors harboring PIK3CA mutations[62]. Similar results were also observed in a panel of GC cell lines[63]. Additionally, combination therapy using PI3K and STAT3 inhibitors showed better efficacy and a synergistic effect in GC cell lines harboring KRAS mutations. The STAT3 pathway is also known to be constitutively activated in GC[64]. Although preclinical studies on BKM120 in GC are still ongoing, it has reached Phase II clinical trials for other cancers such as brain, breast, colorectal, endometrial, NSCLC, and renal cell carcinoma[49,65-67]. Thus, BKM120 has a potential to progress into clinical trials for GC treatment using targeted therapy.

ZSTK474: ZSTK474, a pan-class 1 PI3K inhibitor inhibits all the four isoforms of PI3K and exhibits anti-tumor activity in vivo against human tumor xenograft models[68-70]. In vitro studies in GC cell lines suggest combination therapy of ZST474 and IGFR inhibitors for treating IGFR-positive cancers to overcome any intrinsic resistance to inhibition of PI3K/Akt/mTOR signaling, since over-expression of IGFR correlated with increased tyrosine phosphorylation on Insulin Receptor substrate, leading to increased PI3K activation. Hence, combination therapy with both ZST474 and IGFR inhibitors on GC cells with high IGFR expression exerted a superior therapeutic response[71].

BAY80-6946: BAY80-6946 is synthesized by Bayer healthcare and is a highly potent, selective, and reversible pan-class I inhibitor working in nanomolar concentrations against all the isoforms of p110. However, it shows preferential activity against p110α and β than p110 γ and δ in tumor cell lines and xenograft models[72]. BAY80-6946 demonstrated acceptable safety profiles in phase I clinical trials for advanced solid tumors, and therefore, exhibiting a potential to be progressed to phase II clinical trials for patients with advanced solid tumors.

PI3K isoform specific inhibitors were designed with an aim to provide comparable or superior efficacy than pan-class I inhibitors. Some of them under investigation for GC treatment will now be discussed.

BYL719: BYL719 is an α-isoform specific PI3K inhibitor working at nanomolar concentrations with minimal activity against other PI3K isoforms[73]. With the PI3K/Akt/mTOR pathway being frequently dysregulated in GC, BYL719 exhibited its inhibitory effects in synergy with another inhibitor LJM716, a ligand dependent as well as independent HER3 inhibitor, in GC xenograft models[74]. Interestingly, the combination study was done in HER2 positive GC cell lines, suggesting the sensitivity of this drug towards HER amplifications. BYL719 also recently completed Phase1b clinical trial for advanced stage GC in a combinational study with the HSP90 inhibitor AUY922 in patients whose tumors either harbour molecular alterations of PIK3CA or HER2 amplification[75].

INK1117: INK117 is another novel, selective p110α inhibitor. It is particularly more effective and sensitive to tumors bearing PIK3CA mutations. With good oral bioavailability in preclinical xenograft studies, it has entered a phase-I study for advanced solid tumors including GC, to evaluate its safety, tolerability, pharmacokinetic and pharmacodynamic properties[76].

PI3K/mTOR dual inhibitors inhibit PI3K and the downstream mTOR kinase activity by binding to the ATP-binding cleft of these enzymes. Relative to the single inhibitors, these drugs have the benefit of inhibiting mTORC1 and mTOCR2, as well as all the isoforms of PI3K. Evidence has suggested that the mTORC1/S6K axis has a “two-edge sword”-like function in activation of the PI3K/mTOR pathway by promoting growth signals downstream of Akt, as well as mediating a potent negative feedback loop that restrains signaling via the insulin/IGFR and other RTKs. Dysregulation of this negative feedback loop has been reported to contribute towards resistance in cancers subjected to single inhibitors[77]. Hence, the need of dual PI3K/mTOR inhibitors arises with an aim to discover drugs with low toxicity and good pharmacokinetic profile.

NVP-BEZ235: NVP-BEZ235 is a novel dual ATP-competitive PI3K and mTOR inhibitor for p110α/β/γ/δ and mTOR kinase, with inhibitory doses at nanomolar ranges. It first entered phase trials for breast cancer[78]. The effectiveness of BEZ235 has been investigated in both PIK3CA mutated and wild type cell lines in vitro and in xenograft models in vivo. The first group reporting an effect of BEZ235 on gastric xenografts showed reduced tumor growth for NCI-N87 but not MKN-45 or MKN-28 xenografts. Interestingly, the reduction in tumor growth correlated with thymidine kinase expression and not PI3K/mTOR pathway inhibition[79]. Another group demonstrated in vitro increased sensitivity of AGS, PIK3CA mutated cells than for NCI-N87 and MKN-45, wild type PIK3CA GC cells[63]. Clinically, the response rate for BEZ235 was highest for patients with PIK3CA mutations than those without this mutation[80]. However, another recent study showed an increased anti-tumor response with BEZ235 alone or in combination with nab-paclitaxel in NCI-N87, AGS, and SNU-16 GC cells, independent of PIK3CA mutation status in vitro and in a SNU16 xenograft model in vivo[81]. Hence, with increasing preclinical studies focusing on using NVP-BEZ235 for GC targeted therapy, and its gaining popularity in clinical trials for other cancers, NVP-BEZ235 might be a good potential candidate drug to be considered for clinical trials for solid tumors such as GC.

VS-5884: VS-5884 is a dual PI3K and mTOR inhibitor, inhibiting all the isoforms of PI3K and both mTOR complexes (mTORC1 and mTORC2), with nanomolar inhibitory concentrations for a panel of cancer cell lines in vitro and increased sensitivity towards cell lines harboring PIK3CA mutations. It also exhibits a favourable pharmacokinetic profile in vivo. VS-5884 shows a statistically significant inhibition of tumor growth in HER-over-expressing GC xenograft models (NCI-N87). This drug also exhibits a synergistic response in these xenograft models with geftinib, an EGFR inhibitor (EGFRi), currently in phase II trials for GC treatment. Since this drug has proven its efficacy for monotherapy and combination therapy in GC xenograft models[82], these data provide a rationale for testing VS-5884 in early phase clinical trials for GC patients.

PI-103: PI-103 is an ATP-competitive PI3K and mTOR inhibitor with variable sensitivities towards different isoforms of p110 and mTOR at nanomolar concentrations. It was recently assessed for its synergistic effect to enhance the anti-tumor response for GC both in vitro and in vivo with 5-FU treatment. This study suggested PI-103 usage for enhancing 5-FU chemotherapy for GC; with 5-FU currently being used to treat GC patients but demonstrating limitations due to inter-variability in response rate of these patients. This synergistic effect of PI-103 with 5-FU was also associated with PIK3CA mutations and reduction of downstream effectors of PI3K/Akt/mTOR pathway and thymidylate synthase, an enzyme that generates thymidylate precursors for DNA synthesis in vitro[83].

The Akt signaling cascade controls a spectrum of tumorigenesis events such as cell growth, proliferation, survival, angiogenesis, invasion, and metastasis, as well as activation of the mTOR signaling cascade. Several mutations or amplifications in the Akt/mTOR signaling cascade contribute towards constitutive activation of Akt, which includes PTEN mutations and PIK3CA mutations, as well as over-expression/amplification of Akt itself[84-87]. With activated Akt playing a crucial role in tumorigenesis, several Akt inhibitors have been designed that have entered preclinical as well as clinical trials. Akt inhibitors that are currently being investigated for GC treatment include:

AZD5363: AZD5363 binds to and inhibits all the isoforms of Akt (Akt1, Akt2, and Akt3) with potency in the nanomolar range. Reports have suggested increased sensitivity towards AZD5363 in cancer cells harboring PIK3CA mutations, PTEN mutations, or HER-2 amplifications both in vitro and in vivo[88]. This sensitivity of AZD5363 towards the activating mutations was also tested in GC xenograft models, one with a PIK3CA mutation and another with PTEN loss. AZD5363 exhibited a relatively more significant anti-tumor response towards PIK3CA mutant GC xenografts than those with PTEN loss alone. Interestingly, for GC xenografts with inactivated PTEN treated with a combination of AZ5363 and taxotrene, a synergistic and potent anti-tumor response was observed rather than monotherapy with AZD5363[89]. Hence, selection of patients based on their mutational status would be beneficial for targeted therapy, which can eventually lead to more effective and tailored therapy either with single agents or in combination.

MK-2206: MK-2206 is a highly selective, allosteric Akt inhibitor, with higher potency for Akt1 and Akt2 isoforms than Akt3. Its efficacy was investigated both in vitro and in vivo as a single agent as well as in combination with several chemotherapeutic drugs or molecular targeted drugs (EGFRi) to overcome any potential resistance. This drug enhances the anti-tumor response in combination therapy, making it a suitable and promising agent for the second line of therapy in cancer patients receiving chemotherapy or targeted therapy[90]. This drug is currently used in phase II trials as a second-line therapy for gastric and gastroesophageal cancer[91].

Perifosine: An oral anti-cancer agent and an Akt inhibitor, Perifosine has entered clinical trials for major human cancers. eIF4E is a downstream effector of the Akt/mTOR pathway, and increased levels of phosphorylated eIF4E and total-eIF4E correlate with increased GC in tumor tissues. A recent study showed that Perifosine treatment of GC cells with increased eIF4E expression (p-eIF4E) down-regulated eIF4E expression, and thereby exerting an inhibitory effect on the Akt/mTOR pathway. Also, the combination of eIF4E inhibitor with Perifosine in these GC cells further sensitized the cells towards more effective treatment[92]. Another study of combination therapy in GC revealed the effectiveness of Perifosine in combination with a miR-27a inhibitor, an oncogene that contributes to drug resistance in GC cells[93]. With more studies identifying the molecular mechanisms of Perifosine inhibition in GC, a recent study shows Perifosine inhibiting tumor growth in GC cells via inhibition of the Akt/GSK3β/C-MYC signaling pathway, with significant down-regulation of AEG-1 (Astrocyte elevated gene), a gene reported to play an important role in cellular processes such as proliferation, apoptosis, and invasion[94]. Hence, Perifosine is a good potential therapeutic Akt inhibitor when used in combination therapies to overcome drug resistance, with scope for further progression into clinical trials.

TCN-PM: Triciribine Phosphate Monohydrate (TCN-PM) is a potent Akt inhibitor inhibiting all the three isoforms of Akt. In Phase I studies of patients with solid tumors, where TCN-PM was administered to patients with increased p-Akt levels (as assessed by immunohistochemical staining), a moderate reduction in p-Akt was observed after single TCN-PM therapy, which may have been possibly due to a small sample size. Further studies to confirm its availability as a single agent as well as its efficacy in combination treatments would help to promote its importance for phase II clinical trials for GC[95].

mTOR is often dysregulated in GC, with several preclinical studies suggesting mTOR as a potential therapeutic target. mTOR forms two types of complexes to perform its cellular function based on its interacting partner and substrate specificity, these being the mTORC1 and mTOR2 complex. The mTORC1 complex is rapamycin sensitive[96]; rapamycin being the first mTOR inhibitor developed, while the mTORC2 complex is rapamycin insensitive. mTOR inhibitors are classified into two types based on their specificity for mTOR complexes: Rapalogs and mTORC1/2 inhibitors.

Rapalogs: Rapamycin and its analogs (referred as rapalogs), first form a complex with the intracellular receptor FK506 binding protein 12 (FKBP12) and then bind to a domain separate from the catalytic site of mTOR, preventing mTOR function. Rapalogs are effective against the mTOCR1 complex[97]. Some of the rapalogs under preclinical and clinical studies for GC treatment are as follows:

Temsirolimus: Temsirolimus binds to FKBP12, and the resultant protein-drug complex prevents mTORC1 activity. A Phase I clinical study determined a favorable toxicity profile, maximum tolerated dose, pharmacokinetics, and anti-tumor efficacy in patients with advanced cancer including GC, making it a favorable drug to proceed towards phase II trials[98].

Everolimus - Everolimus is another oral mTORC1 complex inhibitor that has demonstrated good safety and clinical tolerability profile in Phase I trials for several cancers including GC. Phase II trials for patients with advanced GC treated with Everolimus exhibited a good median progression free survival (PFS). Phase III trials for previously treated GC did not show a significant overall survival benefit vs best supportive care patients. However, the PFS for six-months and the safety profile was significant, which highlights the need for predictive biomarkers for Everolimus treatment response in order to obtain better efficacy for this drug[99].

Ridaforolimus: Ridaforolimus, a rapamycin analog, is under clinical investigation with its well defined anti-tumor response in preclinical studies. It showed a synergistic anti-tumor response in patients with solid tumors including GC, in a Phase Ib trial, where ridaforolimus was given in combination with capecitabine, a prodrug that converts into FU[100].

Although mTORC1 inhibition by rapamycin analogs results in substantial tumor growth inhibition, drug resistance has been reported due to a negative feedback loop in the PI3K/Akt/mTOR pathway either via RTKs upregulation with increased Akt activation or crosstalk of PI3K with Ras signaling, leading to MAPK pathway activation. Hence, mTORC1/2 complex inhibitors have gained interest owing to their ability to act as ATP-competitive inhibitors of mTOR kinase activity.

PP242: PP242 significantly inhibits mTOR kinase activity, inhibiting both mTORC1 and mTORC2 complex activities. It shows superior anti-proliferative and anti-angiogenesis effects in GC cell lines in vitro relative to rapamycin, indicating its promising potential as a therapeutic drug in future for GC[101].

Other mTORC1/2 inhibitors that have been under investigation for solid tumor treatment include AZD2014, AZD8055, and OSI-027; however, there are currently no reports on their usage for GC treatment.