The term “autophagy” means “self-eating” derived from Greek. It was first mentioned by Christian De Duve in 1963 1 , and has been used since to describe a bulk degradation process by lysosome-dependent mechanism. Autophagy functions to degrade protein aggregates, maintain the homeostasis of organelles, such as mitochondria, peroxisomes and ribosomes, and destroy intracellular pathogens 2 . The selectivity of autophagic degradation is thought to be achieved by recognizing post-modification such as ubiquitination 3 or acetylation on proteins 4 5 . Several autophagy receptors or adaptors, including SQSTM1/p62, NBR1 and HDAC6, have been identified, and they are considered to function by recognizing and recruiting ubiquitinated protein aggregates to be degraded through the autophagy pathway 6 . Until now, several types of autophagy-mediated degradation have been described. These include: 1) macroautophagy that is used to sequester cytoplasmic materials such as organelles and intracellular pathogens by de novo formation of double-layer membranes 7 : 2) microautophagy that is used to engulf a part of the cytoplasm by the invagination of lysosomal membrane into lysosome lumen 8 ; 3) chaperone-mediated autophagy that is used to transport specific cytosolic proteins by chaperones to lysosomal degradation 9 . Macroautophagy will be discussed in this review and is herein referred to as autophagy.

The autophagy machinery contains more than 30 autophagy-related (Atg) genes; most of which are highly conserved from yeast to mammal. When autophagy is induced by stressed conditions such as starvation, the first step is the nucleation of phagophore membranes (Figure 1), also called pre-autophagosomal structures 10 or isolation membrane, which likely originates from the endoplasmic reticulum, Golgi complex, mitochondria, endosomes and/or the plasma membrane 11 . In nutrient rich condition, the mammalian target of rapamycin (mTORC1) kinase is activated by class I PI3K and amino acids to inhibit the autophagy pathway by associating with and inactivating the ULK1/2 (Atg1 in yeast) complex (including ULK1/2, Atg13 and FIP200), which is essential for the induction of autophagy 12 13 14 (Figure 1). Under growth factor deprivation or nutrient starvation, the activity of mTORC1 is inhibited by energy sensor AMP activated protein kinase (AMPK). The ULK1/2 complex is also directly activated by AMPK-mediated phosphorylation, resulting in the translocation of ULK1/2 complex to the membrane of endoplasmic reticulum 14 15 16 17 . The ULK1/2 complex works mechanistically, likely through the recruitment of the Vps34 (class III PI3K)-Beclin-1 complex to the site of autophagosome generation to produce phosphatidylinositol-3-phosphate (PI3P), which is enriched on the inner surfaces of the phagophores, and to recruit PI3P binding proteins including WIPI-1 (Atg18 in yeast), small GTPase Rab5 and Atg14 for autophagy initiation 13 18 19 20 21 22 (Figure 1).

After phagophore membrane formation, the phagophores are elongated by two ubiquitin-like proteins, Atg12 and microtubule-associated protein 1 light chain 3 (LC3, Atg8 in yeast), to form enclosed double-membrane vesicles as known as autophagosomes in order to sequester a part of the cytoplasm (Figure 1). In this process, Atg12 is first activated by E1-like ubiquitin activating enzyme Atg7, transferred to E2-like ubiquitin conjugating enzyme Atg10, and then conjugated to a lysine residue (K130) of Atg5 23 . The covalently linked Atg12-Atg5 and another membrane-bound factor, Atg16L1, further form a complex, which functions to expand the phagophore membrane; this complex dissociates from the membrane when autophagosomes are formed 24 25 26 . In a second step, full-length LC3 precursor is translated and immediately cleaved by the protease Atg4B to produce LC3-I (cytosolic form) with a free glycine residue 27 28 . Upon autophagy induction, LC3-I is conjugated to phosphatidylethanolamine 10 by the functions of E1-like ubiquitin activating enzyme Atg7 and another E2-like ubiquitin conjugating enzyme Atg3 to produce LC3-II 29 30 31 . The Atg12-Atg5-Atg16L1 complex has been reported to guide LC3-I to the phagophore membrane, and to function as E3-like ubiquitin ligase to promote the lipidation of LC3-I by PE 32 33 . LC3-II is specifically located on autophagosome structures, making it a commonly used specific marker for identifying autophagosomes 10 . LC3-II, which is located on inner membrane of autophagosomes, is eventually degraded after the fusion of the autophagosome with lysosome; however, the LC3-II protein on the outer membrane can be recycled and reused after delipidation by Atg4 28 .

A maturation step of autophagy is the sequential fusion of autophagosomes with endosomes and lysosomes to form autolysosomes; this fusion leads to the eventual degradation of the content of autophagosomes 34 35 (Figure 1). Recent reports suggested that SNARE 36 , ESCRT-III 37 , small GTPase Rab7 38 39 , and HSP70 40 41 are involved in autophagosome maturation. Other relevant findings include that the UVRAG protein is able to interact with Vps34-Beclin-1 complex to activate GTPase Rab7 and autophagosome-lysosome fusion 42 and that the Rubicon protein suppresses the maturation of autophagosomes by interacting VPS34-Beclin-1 complex 43 . The latter observation indicates that the VPS34-Beclin-1 complex can also regulate autophagosome maturation depending on selective protein association. Additionally, the TECPR1 protein is thought to form a complex with Atg12-Atg5 and PI3P to enhance the fusion of autophagosomes with lysosomes 44 . At the same time, it should be noted that functional lysosomes are also needed for autophagosome maturation. Thus, a deficiency of Lamp2, which is an essential constituent of the lysosomal membrane, causes autophagosome accumulation and disrupts proper autophagy-mediated degradation 45 . Moreover, the disruption of lysosomal acidification by bafilomycin A (BFA, an inhibitor of the lysosomal vacuolar-ATPase) or chloroquine (a lysosomotrophic agent to increase pH in lysosomes) strongly impairs autophagosome-lysosome fusion 46 47 . The mechanism of how lysosomal acidification influences autophagosome-lysosome fusion, however, needs further exploration.

Genetic mutations of several autophagy-related genes have been linked to human diseases. For example, Beclin-1 has been suggested to suppress tumorigenesis and progression of breast cancer 48 . The monoallelic deletion of Beclin-1 has been observed in 40-75% of human breast, ovarian, and prostate malignancies 49 . UVRAG is found to be monoallelically mutated in human colon cancer, and UVRAG has been suggested to act by inhibiting the proliferation and tumorigenic activity of human colon cancer cells 50 51 . By Genome-Wide Association Study (GWAS), IRGM1 (autophagy-stimulatory immunity-related GTPase) and Atg16L1 have been identified to be associated with the pathogenesis of chronic inflammatory bowel diseases, such as Crohn’s disease 48 52 . The somatic mutation of LAMP-2 has been linked to Danon disease, which exhibits cardiac hypertrophy and the accumulation of autophagosomes and lysosomal glycogen in cardiac muscle cells causing clinical symptoms of cardiomyopathy, myopathy and mental retardation 53 54 . The deficiency of LAMP-2 in mice also results in similar vacuolar cardioskeletal myopathy 53 . SQSTM1/p62 is an autophagy receptor, which recognizes and sends ubiquitinated substrates to be degraded by autophagy. Mutations in the ubiquitin associated (UBA) domain of SQSTM1/p62 have been reported to be associated with about 30% of Paget’s bone disease, which has disordered NF-κB-dependent osteoclast function and is characterized by focally increased and disorganized bone remodeling 55 . These collective examples raise the notion that perturbed activity of the autophagy pathway influences genomic instability and normal cellular metabolism 52 .

The link between autophagy and cancer development has been broadly established. Autophagy can clear toxic aggregates and damaged mitochondria which may produce reactive oxygen species (ROS) that cause DNA damage 56 57 , and autophagy has attributed roles in chromosome instability, including aneuploidy and gene amplification 58 59 60 . Moreover, a deficiency of autophagy results in failed degradation of SQSTM/p62, which plays a role in activation of NF-κB and inflammation-mediated tumorigenesis 56 61 62 . Thus, conceptually, autophagy serves to reduce environmental insults and neutralize events that favor cellular transformation. Indeed, in cellular transformation, it has been commonly regarded that apoptosis provides a protective mechanism in inciting the death of aberrantly transformed cells. In that context, it is increasingly recognized that apoptotic cell death of abnormal cells can be complemented by apoptosis-independent autophagy-dependent cell death 61 63 , especially in the elimination of transformed cells.

In a related aspect, the function of autophagy as a provider of nutrient and energy also contributes to tumor survival, especially under metabolically stressful condition such as nutrient starvation and hypoxia 64 . This concept is supported by the clinical observation that biallelic loss of Beclin-1 has not been seen in cancer patients 49 65 , and by in vitro experiments showing that autophagy deficiency achieved by small interfering RNA targeting Beclin-1 or Atg5 reduces cellular proliferation and increases the death of cancer cells 61 66 67 . Additionally, activation of autophagy is observed within cancer cells treated with chemotherapy or radiotherapy. Thus although these cancer therapies are designed to kill most cancer cells, it is a concern that by triggering increased autophagy they incite a reactive response that helps the residual cancer cells survive and resist extreme stress 64 . It is thus reasonable to consider a cancer treatment approach that combines traditional anti-cancer chemotherapy with autophagy inhibitors such as hydroxychloroquine. Several clinical trials are underway examining the effect of autophagy inhibitors on increasing the sensitivity of cancers to chemotherapy 68 69 .

Autophagy also functions as a cellular defense to remove invading pathogens, in a process termed xenophagy; and autophagy can serve to deliver antigen fragments of pathogens for MHC class II presentation to activate the adaptive immune system 70 . Many types of bacteria have been reported to be targeted by autophagic degradation. For example, Mycobacterium tuberculosis can be targeted by autophagy, and its clearance is enhanced by cellular starvation and exposure to lipopolysaccharides. The clearance of Toxoplasma gondii can be decreased by treatment with Bafilomycin A (an autophagy inhibitor) or Beclin-1 siRNA 71 . Ubiquitination of proteins is likely a crucial step for the clearance of invading bacteria. NDP52 (nuclear dot protein 52kDa) functions as an autophagy receptor that recognizes ubiquitinated Salmonella enterica and captures it into autophagosomes by interacting with LC3 72 . Recently, Watson et al. have observed that the cellular STING-dependent pathway recognizes extracellular bacterial DNA, triggering the intracellular ubiquitination of bacterial proteins, and that SQSTM/p62, NDP52 and the DNA-responsive kinase TBK1 are used for autophagic degradation of bacteria 73 . However, it should be noted that many bacteria have evolved countermeasures to combat the cell’s autophagic defense. For example, Legionella pneumophila and Brucella abortus do induce cellular autophagy, but can thwart the maturation step of autophagy in order to facilitate pathogenic replication 70 . Similarly, Mycobacterium tuberculosis can interfere with autophagosome-lysosome fusion through its ESAT-6 Secretion System-1 (ESX-1) 74 .

Interestingly, most viruses, with a few exception such as vesicular stomatitis virus (VSV), appear to have evolved mechanisms to evade cellular clearance by autophagy 75 . Many viruses have developed counteracting mechanisms to escape autophagic degradation 66 76 . For instance, several herpesviruses, including herpes simplex virus type 1 (HSV-1), bovine herpesvirus type 1 (BHV-1), human cytomegalovirus (HCMV), Kaposi’s sarcoma-associated herpesvirus (KSHV), herpesvirus saimiri (HVS) and molluscum contagiosum virus (MCV), can capably suppress autophagy. For HSV-1, the viral protein ICP34.5 interacts with Beclin-1 to inhibit autophagy induction 77 78 .

Some DNA viruses including Epstein-Barr virus (EBV), varicella-zoster virus (VZV), adenovirus, human papillomavirus 16 (HPV16), simian virus 40 (SV40), human parvovirus B19 (HPV-B19) and hepatitis B virus (HBV), activate portions of the autophagy pathway and employ this process to enhance viral replication. Thus, autophagy-induced cell death assists the final step of the adenovirus life cycle to release virus particles 79 . Many RNA viruses, including VSV, coxsackievirus B4 (CVB4), coxsackievirus B3 (CVB3), poliovirus, dengue virus-2 (DENV2), dengue virus-3 (DENV3), rotavirus, hepatitis C virus (HCV), influenza virus A, have been observed to induce autophagy, but inhibit autophagosome-lysosome fusion. 80 . For poliovirus and HCV, autophagy induction seems to provide cell membranes for RNA replication 81 82 83 84 . For influenza A virus, the viral M2 protein inhibits autophagosome-lysosome fusion, possibly inhibiting MHC antigen presentation of influenza A virus proteins to reduce host immune response 85 . Below, we will discuss in greater depth lessons on autophagy learned from hepatitis B virus and human retroviruses.

The human hepatitis B virus (HBV) is the prototype member of a family of small, enveloped DNA viruses called Hepadnaviridae that infect a restricted number of mammals and birds. Despite the existence of effective vaccines, HBV remains one of the most significant human pathogens with an estimated 2 billion people infected worldwide, of whom 350 million are chronic HBV carriers. Chronic hepatitis B is a major risk factor for severe liver diseases including liver cirrhosis and hepatocellular carcinoma (HCC). HCC is the fifth most common cancer, and the third leading cause of cancer death in the world 86 .

Although chronic HBV infection has been epidemiologically linked to the development of HCC for more than 40 years, the mechanisms by which HBV infection results in HCC are still unclear. The hepatitis B virus X protein (HBx) has generally been viewed as an oncoprotein in viral carcinogenesis. In order to favor virus replication, HBx subverts cellular activities such as signal transduction, transcription, autophagy, and proliferation. In doing so, HBx apparently induces the accumulation of cellular dysfunctions and damage, ultimately leading, in the case of viral persistence, to cancer development.

HBV enhances and uses autophagy for its replication. However, the mechanisms responsible for autophagy induction as well as the step of HBV replication affected by autophagy are still controversial 87 88 . Upon viral infection, autophagy can be triggered by direct mechanisms like the recognition of viral element that promote autophagy protein expression, or by indirect mechanisms like virus triggered cellular stress. For example, during infection a large amount of viral proteins are synthesized and unfolded, and the misfolded proteins can activate ER stress response. HBV can use direct and indirect mechanisms to induce autophagy (Figure 2).

Different groups have shown that HBV expression is correlated with autophagy induction 87 88 89 . Two of these publications observed that while HBV expression induces the formation of early phagosomes, the rate of autophagic protein degradation is seemingly not increased 87 88 90 . These results argue that HBV acts on the early step of phagosome formation. As noted above, some viruses induce early steps in the autophagy process, but delay phagosome maturation, in order to promote viral replication 91 92 . Further work will be needed to determine carefully whether and how HBV blocks the formation of late phagosomes. Interestingly, Tang and collaborators have observed autophagy induction by HBV only under starvation condition. The discrepancy between this work and other studies could stem from the different cell lines used in these studies. Alternatively the models used: transfected HBV genome versus integrated HBV genome could lead to different levels of viral protein expression, and increasing the level of HBV protein can induce ER stress 93 thereby activating autophagy 94 . For future clarity, it will be necessary to assess autophagy in the context of authentic HBV infection.

The exact steps of HBV replication that are regulated by autophagy remain to be identified. In extant publications the role of autophagy on the early steps of HBV infection is not addressed; however, current data do support that autophagy impacts late steps of HBV replication, increasing HBV production. Indeed, using either an inhibitor of PI3KC3 or via the silencing of enzymes essential for the formation of autophagosomes, Sir et al. showed that inhibition of autophagy had a marginal effect on HBV transcription and HBV RNA packaging, but suppressed HBV DNA synthesis, suggesting an enhancement of HBV DNA replication by autophagy 88 . They further confirmed the role of autophagy in the production of HBV virions in vivo using HBV transgenic mice with liver-specific knockout of Atg5. They demonstrated in this model that the formation of autophagosome is essential for HBV DNA synthesis in the cytoplasm 90 . Again, in this study the HBV DNA is integrated into the mouse genome, preventing direct extrapolation of the findings to in vivo HBV infection. Moreover, these studies are somewhat in contradiction with the study of Li and collaborators who reported that the autophagy machinery is needed for efficient envelopment of the nucleocapsids at the ER membrane and has only a slight effect on HBV DNA replication 87 . It is unclear the reasons for the differences; in both studies, the HBV genome is transfected into hepatoma cells, albeit using different techniques. However, it may be that slight differences in the level of viral protein expression and the cell lines employed could account for the discrepancies between the two studies.

If the impact of autophagy on HBV replication remains a matter of discussion, the mechanisms leading to the induction of autophagy by HBV remain also unclear (Figure 2). It was first suggested that the viral regulatory protein HBx was directly involved in starvation-induced autophagy via the up–regulation of Beclin-1 expression 89 . In that report, the authors showed that HBx, a known weak transcriptional activator, transactivates the Beclin-1 promoter in hepatic and hepatoma cell lines. They next demonstrated that silencing of Beclin-1 expression by siRNA blocked the induction of autophagy by HBx, suggesting that HBx acts via the transcriptional activation of Beclin-1. These authors, however, did not include control experiments with a transactivation-deficient HBx mutant and their siRNA knock down results do not formally address increased Beclin-1 transcription by HBx, rather the findings solely indicate that Beclin-1 is essential for induction of autophagy by HBx. This is an important point because in another publication, Sir et al. did not observe induction of Beclin-1 expression by either HBV or HBx. Rather, those investigators observed that HBx interacted with PI3KC3 and enhanced the latter’s activity 88 .

HBV can also induce autophagy indirectly via the induction of cellular stress 94 95 96 (Figure 2). In searching for the mechanism leading to autophagy upon HBV expression, Li and collaborators found that the expression of HBV small surface protein (SHBs) induced ER stress and subsequently the activation of three signaling pathways PERK, ATF6 and IRE1. They further demonstrated that the blockade of any of these three UPR (unfolded protein response) signaling pathways blocked autophagy induction. Their study supports the idea that induction of ER stress by SHBs is the inducer of autophagy 87 . Moreover, the authors observed an interaction between SHBs proteins and the autophagosome marker LC3, suggesting that this interaction could be involved in the enveloping process of HBV virions (Figure 2). How autophagy enhances viral envelope acquisition needs further investigation. One should note that another group has reported findings in contradiction with the notion that SHBs proteins increase autophagy without enhancing the rate of protein degradation or that autophagy favors virus replication. Indeed, Lazar and collaborators showed that HBV activates UPR through the increase of EDEM1 expression, which negatively controls viral particle production 97 . They demonstrated that EDEM1 expression leads to the degradation of HBV envelope proteins L, M and S by autophagy. However, Lazar and coworkers studied the role of EDEM1 on viral surface protein stability in HEK293T cells that over-expressed viral envelope proteins. Whether envelope proteins are degraded in the setting of authentic HBV infection and replication was not addressed.

Human immunodeficiency virus-1 (HIV-1) is the causative agent for acquired immunodeficiency syndrome (AIDS) 98 99 ; the virus infects over 30 million individuals worldwide and causes approximately 3 million deaths each year. HIV-1 infects and replicates in CD4⁺ T cells and macrophages 100 101 . After entry into cells, HIV-1 replication is challenged by cellular autophagic degradation 102 and/or by host cell restriction factors 103 104 105 , such as APOBEC3G 106 107 , BST-2/Tetherin 108 109 110 , TRIM5α 111 112 , SAMHD1 113 114 115 116 , and microRNAs 117 118 . However, HIV-1 has evolved means to counter these defense mechanisms to overcome these cellular restrictions. For example, HIV-1 uses viral accessory protein Vif to promote the degradation or exclude the virion incorporation of APOBEC3G 106 119 ; the Vpu protein to counter the effect of BST-2/Tetherin 109 120 121 , and the Tat protein to modulate cellular miRNA activity 122 123 .

Regarding autophagy, HIV-1 apparently subverts this cellular defense process in a manner to benefit viral replication. In macrophages, the viral accessory protein Nef 124 125 , by interacting with Beclin-1, has been found to block the maturation step of autophagy and thus acts to prevent the destruction of HIV-1 126 (Figure 3A). Treatment of BFA, an inhibitor of autophagosome-lysosome fusion, enhances accordingly HIV-1 production 127 . Additionally, immunity-associated GTPase family M (IRGM), which interacts with Atg5 and Atg10, has been reported to be another target of Nef for the accumulation of autophagosomes and HIV-1 production 80 . It has been observed that a Nef-deficient HIV-1 cannot overcome autophagic degradation and replicates less efficiently 126 . Overall, the current findings are that the early steps of autophagy contributes to HIV-1 replication (Figure 3A), and consistent with this notion, HIV-1 Gag is seen colocalized with LC3-enriched autophagosomes; and treatment of cells with the autophagy inhibitor 3-methyladenine (3-MA) and siRNA-mediated knockdown of Beclin-1 or Atg7 significantly reduces the yield of HIV-1, while the autophagy inducer rapamycin enhances virus production 126 . A recent study also found that vitamin D treatment can inhibit HIV-1 replication through initiating and promoting the maturation of autophagy, and that treatment of cells with BFA and knockdown of Beclin-1 and Atg5 counter the inhibitory effect of vitamin D 128 . These results suggest approaches that increase autophagosome-lysosome fusion could potentially be useful anti-HIV-1 therapeutic strategies.

HIV-1 infection of CD4⁺ T cells is not identical to its infection of macrophages. Unlike reports from macrophages, HIV-1 infection in MOLT-4 T lymphoblast cell line and CD4⁺ T cells has been suggested by one research group to inhibit autophagy as measured by reduced LC3-II or Beclin-1 levels 127 129 (Figure 3B); this indicates that in T-cells the process of autophagy may be a net negative for HIV-1 replication 130 . However, this notion is somewhat unsettled because another group has reported that autophagy is induced by HIV-1 infection of CD4⁺ T cells, as shown by increased levels of Beclin-1 and LC3 131 (Figure 3B). Moreover, they found that the levels of ULK1, Atg4D, Atg5 and Atg12 conjugates were also increased by HIV-1 or HIV-2 infection, and that autophagy inhibitor 3-MA and Beclin-1 siRNA were able to inhibit HIV-1 replication in Jurkat T cells 131 . Elsewhere, the stable knockdown of autophagy-related genes, such as Atg5 and Atg16 also was seen to inhibit HIV-1 production in SupT1 cells 132 . These results suggest that activation of autophagy is a net positive for HIV-1 in T cells; indeed, the relationship between HIV-1 and autophagy in T-cells remains incompletely understood and requires further investigation.

Further complicating the picture is a postulated role of HIV-1 on the autophagy status of uninfected bystander CD4⁺ T cells. Through interacting with CXCR4 or CCR5, soluble circulating HIV-1 env protein induces autophagy to trigger apoptosis in uninfected CD4⁺ T cells (Figure 3B), accounting in part for the clinical depletion of CD4⁺ T cells 127 133 134 . The apoptotic cell death induced by env protein can be fully inhibited by treating cells with 3-MA and BFA, or using siRNAs to knock down Beclin-1 and Atg7, indicating a link between autophagy and apoptosis through autophagy-related proteins 130 133 .

Another human retrovirus is the Human T-cell Leukemia Virus type 1 (HTLV-1), which was identified a few years prior to HIV-1, and is the etiological agent for a human lymphoproliferative malignancy, adult T-cell leukemia (ATL), and chronic inflammatory diseases, including HTLV-1-associated myelopathy (HAM)/tropical spastic paraparesis (TSP) 135 136 137 138 . HTLV-1 infects approximately 10 to 20 million individuals worldwide 139 . The virus infects CD4⁺ T cells, CD8⁺ T cells, B cells, macrophages and fibroblasts; this diversity of infection occurs possibly because of the ubiquitous distribution of its hypothesized receptors (glucose transporter 1, heparan sulfate and proteoglycans and neuropilin-1) 140 141 . Empirically, HTLV-1 primarily targets CD4⁺ T cells, resulting in persistent NF-κB activation by the viral regulatory protein Tax, leading to the clonal expansion of CD4⁺ T cells 142 143 144 145 146 .

A recent report showed that HTLV-1-infected T cells exhibit increased autophagy and that the genetic disruption of Beclin-1 decreased the viability of HTLV-1-transformed T cells 147 (Figure 3C). HTLV-1 Tax was found to interact with the Vps34-Beclin1 complex in IKKβ –dependent fashion 147 . Additionally, HTLV-1 infection and Tax expression have been found to induce autophagy; and in this setting, blocking the autophagosome-lysosome fusion was shown to benefit virus replication (Tang et al., submitted) (Figure 3C). Mechanistically, the ability of HTLV-1 Tax to activate NF-κB pathway correlates with its induction of autophagosome accumulation (Tang et al., submitted).

Separately, it has been reported that CD4⁺ T cells immortalized by HTLV-2 Tax protein have increased LC3-II compared to Jurkat T cells, and that autophagy inhibitors (3-MA and chloroquine) inhibit the proliferation and induce the apoptosis of HTLV-2 Tax-immortalized T cells 148 (Figure 3C). HTLV-2 Tax was shown to interact with Vps34, IKKβ and Beclin-1, and shRNA-mediated knockdown of IKKβ or Beclin-1 expression reduced HTLV-2 Tax-induced accumulation of LC3-II, providing a possible mechanism for how HTLV-2 Tax activates autophagy 148 . Going forward, a comparison of similar/different mechanism(s) shared by HTLV-1 and −2 Tax proteins in autophagy induction would be informative.

A recent study suggested that the degradation of IKK (inhibitor of kappa B kinase) induced by geldanamycin inhibition of Hsp90 (heat shock protein 90) is through the autophagy, not the proteasome, pathway. In Atg5-deficienct cells with impaired autophagy, IKK degradation induced by geldanamycin treatment is attenuated, indicating that in this setting, autophagy plays a key role 149 . Additionally, treatment with autophagy inhibitors increased the survival of ATL cells when their Hsp90 protein is inhibited by geldanamycin treatment 150 151 . These results implicate autophagy as playing a physiological role in the death of ATL cells.