ML-IAP [melanoma IAP (inhibitor of apoptosis)] is an anti-apoptotic protein that is expressed highly in melanomas where it contributes to resistance to apoptotic stimuli. The anti-apoptotic activity and elevated expression of IAP family proteins in many human cancers makes IAP proteins attractive targets for inhibition by cancer therapeutics. Small-molecule IAP antagonists that bind with high affinities to select BIR (baculovirus IAP repeat) domains have been shown to stimulate auto-ubiquitination and rapid proteasomal degradation of c-IAP1 (cellular IAP1) and c-IAP2 (cellular IAP2). In the present paper, we report ML-IAP proteasomal degradation in response to bivalent, but not monovalent, IAP antagonists. This degradation required ML-IAP ubiquitin ligase activity and was independent of c-IAP1 or c-IAP2. Although ML-IAP is best characterized in melanoma cells, we show that ML-IAP expression in normal mammalian tissues is restricted largely to the eye, being most abundant in ciliary body epithelium and retinal pigment epithelium. Surprisingly, given this pattern of expression, gene-targeted mice lacking ML-IAP exhibited normal intraocular pressure as well as normal retinal structure and function. The results of the present study indicate that ML-IAP is dispensable for both normal mouse development and ocular homoeostasis.
Apoptosis is a genetically regulated form of cell death that is mediated by death receptors belonging to the TNF (tumour necrosis factor) receptor family or through mitochondrial pathways . Both pathways activate caspases, the cysteine-dependent aspartyl-specific proteases that execute demolition of the cell . The IAP (inhibitor of apoptosis) proteins interact with a variety of inducers and effectors of apoptosis . The ability to block cell death induced by many different pro-apoptotic stimuli make IAP proteins central regulators of cell fate. ML-IAP (melanoma IAP; also known as BIRC7, livin or KIAP) is expressed highly in melanomas and several other tumour types, where it contributes to resistance to death stimuli [4–9]. The related XIAP (X-linked IAP) binds and inhibits caspases 3, 7 and 9 directly, whereas ML-IAP limits amplification of the apoptotic signal by Smac (second mitochondrial-derived activator of caspase) [2,10–12]. Located in the mitochondria of healthy cells, Smac is released into the cytosol during apoptosis, where it binds and inhibits XIAP [6,13,14]. ML-IAP binds Smac with high affinity and sequesters Smac away from XIAP [15,16]. Certain IAP proteins are also E3 ubiquitin ligases that use their C-terminal RING domains to ubiquitinate themselves and several of their binding partners [17,18].
Small-molecule IAP antagonists (Smac mimetics) bind select BIR (baculovirus IAP repeat) domains of IAP proteins, thereby blocking IAP protein interactions with caspases and Smac . When IAP antagonists bind the BIR3 domain of c-IAP1 (cellular IAP1) and c-IAP2 (cellular IAP2), they stimulate c-IAP auto-ubiquitination and proteasomal degradation [19–21]. IAP ubiquitin ligase activity is critical for IAP-mediated pro-survival signalling as well as IAP antagonist functions [17,18]. IAP RING domain dimerization is essential for IAP ubiquitin ligase activity, and antagonists appear to stimulate this E3 activity by promoting dimer formation [22,23]. Recent biochemical and structural studies have shown that, in the absence of IAP antagonists, BIR3 through RING domains of c-IAP1 form a compact monomeric structure that prevents RING dimerization [22,24]. Binding of IAP antagonists to the c-IAP1 BIR3 domain blocks critical BIR3–RING interactions, and conformational rearrangements allow RING dimerization and instigate E3 ligase activity .
In the present study we demonstrate that a bivalent IAP antagonist causes proteasomal degradation of ML-IAP, which is dependent on the BIR domain of ML-IAP but independent of c-IAP1, c-IAP2 or XIAP. Interestingly, monovalent IAP antagonists did not stimulate ML-IAP E3 ligase activity and promote its degradation. ML-IAP expression in normal mouse tissues appears restricted to the eye, but genetic ablation of ML-IAP did not disrupt general mouse development or any retinal structure and ocular function, suggesting that ML-IAP is not essential for eye homoeostasis in mammals.
Cell lines, antibodies, transfections and reagents
MeWo, Colo 829, Malme-3M and SK-MEL28 human melanoma cells were obtained from A.T.C.C. KMS18 multiple myeloma cells were from JCRB (Japanese Collection of Research Bioresources). c-IAP1- and c-IAP2-knockout, and matched wild-type MEFs (mouse embryonic fibroblasts) were provided by Dr John Silke (Walter and Eliza Hall Institute of Medical Research, Melbourne, Australia) [20,25]. All cells were grown on 50:50 DMEM (Dulbeco's modified Eagle's medium)/FK12 medium supplemented with 10% FBS (fetal bovine serum), 10000 units/ml penicillin and 10000 μg/ml streptomycin. KMS18 cells were transfected by electroporation using Amaxa Nucleofactor technology (Lonza). Plasmids expressing FLAG–c-IAP1, FLAG–c-IAP2, FLAG–c-IAP1 RING mutant (H588A), Myc–ML-IAP, Myc–ML-IAP ΔRING and Myc–ML-IAP BIR mutant (D138A) have been described previously [5,19,26]. ML-IAP RING mutant (H251A) was generated by site-directed mutagenesis. The primary antibodies against c-IAP1 were purchased from R&D Systems (affinity-purified goat antibody) or Alexis Biochemicals, anti-ubiquitin and anti-XIAP antibodies were from Cell Signaling Technology, anti-FLAG M2 antibody was from Sigma, anti-Myc antibody was from Roche, anti-tubulin antibody was from ICN Biomedicals, and monoclonal and polyclonal antibodies against ML-IAP were generated at Genentech . MG132 and Z-VAD-FMK (benzyloxycarbonyl-Val-Ala-DL-Asp-fluoromethylketone) were purchased from Calbiochem, and IAP antagonists MV1, BV6  and GDC-0152  were from Genentech. Western blot analyses were performed as described previously .
Recombinant protein production and ubiquitination assays
Full-length ML-IAP (residues 1–298) was subcloned into the pET-15b bacterial expression vector (Novagen) and the protein was expressed in Escherichia coli strain BL21-Gold (DE3) (Agilent Technologies). Overnight cultures were diluted 1:100 and grown at 37°C in Luria–Bertani medium with 50 mg/ml carbenicillin to a D600 of 0.8 with vigorous shaking (200 rev./min). IPTG (isopropyl β-D-thiogalactopyranoside) was added to a final concentration of 0.5 mM to induce protein expression. After IPTG induction, expression was allowed to proceed overnight at 16°C, with shaking at 200 rev./min. Cells were pelleted and resuspended in 50 ml/l buffer A [50 mM Tris (pH 7.5), 150 mM NaCl, 5 mM DTT (dithiothreitol) and Roche complete protease inhibitors]. Cells were homogenized, microfluidized and centrifuged (27000 g for 30 min at 4°C). Protein was purified by passing through a Ni-NTA (Ni2+-nitrilotriacetate) agarose column (Qiagen), followed by gel-filtration chromatography. Protein was then loaded on to a Superdex 75 gel-filtration column (Pharmacia) equilibrated with buffer A (without protease inhibitors). Eluted fractions were analysed by SDS/PAGE and ESI (electrospray ionization)–TOF (time-of-flight) LC (liquid chromatography)-MS and stored at 4°C. Reconstituted ubiquitination assays with recombinant ML-IAP and E2 enzymes were performed as described previously [19,28]. Recombinant UbcH5a/b/c, UbcH6, UbcH7, E1 enzyme and ubiquitin were purchased from Boston Biochem.
SD-OCT (spectral domain-optical coherence tomography) and FA (fluorescein angiography)
All procedures involving animal experiments adhered to the ARVO (Association for Research in Vision and Opthalmology) guidelines for the use of animals in ophthalmic and vision research, using protocols approved and monitored by Genentech's Institutional Animal Care and Use Committee. Longitudinal assessments of SD-OCT and FA (Spectralis HRA+OCT, Heidelberg Engineering) were obtained from 1- to 6-monthold wild-type and homozygous mutant mice. This cohort comprised wild-type males (n=10) and females (n=8), and ML-IAP−/− males (n=8) and females (n=10). An aged cohort, 20–25 months old, was also evaluated. For these experiments, mice were anaesthetized with an intraperitoneal injection of a solution of ketamine (16 mg/kg) and xylazine (5 mg/kg) and maintained on a heating pad at 37°C. The pupils were dilated with a topical drop of 1% tropicamide (Bausch & Lomb). Bilateral SD-OCT volume scans of 19 slices with a distance between slices of 200 μm were performed for at least three regions of each eye under study (superior nasal, superior temporal and central inferior areas). Retinal thickness was obtained in all cases.
Angiography assessment was performed in a single eye for each mouse as part of the SD-OCT session. Following an intraperitoneal injection of 0.1 ml of 2.5% fluorescein sodium (Alcon), images were obtained for early (30–180 s) and late (300–400 s) phases after injection.
IOP (intraocular pressure)
Non-invasive tonometry was measured in conscious 6-month-old mice restrained in a plastic sleeve in the early afternoon hours (13:30–15:00 h). Bilateral assessments were performed using the TonoLab rebound tonometer with disposable probes (Icare). An average of six tests were recorded per eye. Each group comprised 16–22 eyes, wild-type male (n=10 mice; 20 eyes) and female (n=8 mice; 16 eyes) mice, and ML-IAP−/− male (n=8 mice; 16 eyes) and female (n=10 mice; 20 eyes) mice.
Prior to ERG assessment, mice were adapted to darkness overnight and subsequently the whole manipulation was performed in dim red light. For these studies, the groups consisted of eyes from wild-type males (n=10) and females (n=8), and eyes from ML-IAP−/− male (n=8) and female (n=10) mice. Mice were anaesthetized with an intraperitoneal injection of a solution of ketamine (16 mg/kg) and xylazine (5 mg/kg) and maintained on a heating pad at 37°C. The pupils were dilated as described above. A topical drop of 2.5% hypromellose (Goniovisc, Sigma Pharmaceuticals) was instilled in each eye immediately before placing the contact lens electrodes (The Electrode Store). A reference electrode was placed on the forehead, with a ground electrode placed in the tail. Flash-induced ERG was recorded bilaterally using the Espion E2 system and Ganzfeld dome (Diagnosys). The electrical responses of the retina were recorded simultaneously from both eyes. Retinal responses were recorded under dark adaptation to white light flashes at five steps ranging from −4 to 1.5 log(cd·s·m−2), capturing rod-driven rod–cone mixed responses as well as cone-dominated responses at the highest intensities [28a]. For each light-intensity stimulus, ten 3-s consecutive stimuli were averaged, with an interval between light flashes in scotopic conditions of 10 s for dim flashes and up to 60 s for the highest intensity flashes. OPs (oscillatory potentials) were isolated using white flashes of 1.5 log(cd·s·m−2) in a recording frequency range of 100–10000 Hz. The amplitudes of the a-wave and b-wave, and peak-to-peak OPs were measured off-line.
The IAP antagonist BV6 promotes ML-IAP degradation
We, and others, have shown that IAP antagonists stimulate c-IAP1 and c-IAP2 auto-ubiquitination and proteasomal degradation [19,20]. We examined the fate of ML-IAP following IAP antagonist treatment in several melanoma cell lines. Addition of the bivalent IAP antagonist BV6  stimulated degradation of ML-IAP in MeWo and Colo 829 cells. Consistent with previous studies , c-IAP1 was also degraded (Figure 1A). The proteasome inhibitor MG132 inhibited degradation of ML-IAP, whereas the pan-caspase inhibitor Z-VAD-FMK had no effect, indicating ubiquitin–proteasome system involvement in ML-IAP degradation (Figure 1A). We found that c-IAP1 was degraded much faster than ML-IAP. For example, c-IAP1 was not detected in cellular lysates after 10 min of BV6 treatment, whereas complete ML-IAP degradation took more than 1 h (Figure 1B).
IAP antagonist BV6 promotes degradation of ML-IAP
Next, we mutated the ML-IAP BIR or RING domains (Supplementary Figure S1 at http://www.BiochemJ.org/bj/447/bj4470427add.htm) and determined the effect of these mutations on ML-IAP stability in KMS18 cells. Wild-type ML-IAP was degraded fully after BV6 treatment, but mutation of the Zn2+-co-ordinating residue His251 to alanine or deletion of the entire RING domain prevented ML-IAP degradation in response to BV6 (Figure 1C). Similarly, mutation of an IAP antagonist-binding residue in the BIR domain (Asp138 to alanine) of ML-IAP also prohibited ML-IAP degradation (Figure 1C). Therefore the IAP antagonist BV6 stimulates ML-IAP degradation, and requires the RING domain and functional antagonist-binding site on the BIR domain of ML-IAP.
Degradation of ML-IAP occurs independently of other IAP proteins
Next we investigated whether other IAP proteins are important for IAP antagonist-stimulated ML-IAP degradation using several different cellular models wherein IAP protein expression was down-regulated, ablated genetically or accomplished through ectopic transfection. Simultaneous knockdown of c-IAP1 and c-IAP2, or knockdown of XIAP, did not affect BV6-induced ML-IAP degradation (Figures 2A and 2B). Consistent with these data, BV6 caused degradation of ML-IAP expressed ectopically in wild-type, c-IAP1-deficient or c-IAP2-deficient MEFs (Figure 2C). Ectopically expressed ML-IAP was also degraded after BV6 treatment of KMS18 multiple myeloma cells, which lack Birc2 (c-IAP1 gene) and Birc3 (c-IAP2 gene) [29,30] (Figure 2D). Finally, co-expression of ML-IAP with RING mutant c-IAP1 H558A, which interferes with BV6-induced c-IAP degradation, did not affect ML-IAP degradation (Figure 2D). Collectively, these experiments suggest that BV6 promotes proteasomal degradation of ML-IAP independently of c-IAP1, c-IAP2 or XIAP.
c-IAP1, c-IAP2 or XIAP are not required for degradation of ML-IAP
Bivalent, but not monovalent, IAP antagonists stimulate ML-IAP degradation
Next, we compared bivalent BV6 with monovalent compounds MV1 and GDC-0152 [19,27] in their ability to induce ML-IAP degradation in MeWo and Colo 829 melanoma cells. MV1 and GDC-0152 both caused efficient degradation of c-IAP1, but they did not decrease ML-IAP stability (Figure 3A). Consistent with results described above, BV6 promoted degradation of both c-IAP1 and ML-IAP (Figure 3A). This result suggests that bivalent IAP antagonists may stimulate the ubiquitin ligase activity of ML-IAP, whereas monovalent agents cannot. We explored this possibility using purified recombinant full-length ML-IAP (Supplementary Figure S2A at http://www.BiochemJ.org/bj/447/bj4470427add.htm). We have shown previously that the ML-IAP RING domain can bind ubiquitin conjugating (E2) enzyme UbcH6 and members of the UbcH5a/b/c family, but not other E2 enzymes . In agreement with those results, ML-IAP incubated with UbcH5a, UbcH5b or UbcH6 in a reconstituted ubiquitination assay yielded a prominent ubiquitin smear (Supplementary Figure S2B). MV1 or GDC-0152 added into this assay did not significantly alter ML-IAP ubiquitin ligase activity (Figure 3B). In contrast, BV6 enhanced the ubiquitination signal and allowed its detection earlier (Figure 3B). Collectively, these data suggest that monovalent and bivalent IAP antagonists differentially regulate the stability of ML-IAP: bivalent antagonists can stimulate ML-IAP ubiquitin ligase activity and cellular degradation, whereas monovalent antagonists cannot.
Bivalent IAP antagonist stimulates ML-IAP degradation
Expression of ML-IAP in murine tissues
We next investigated the physiological role of ML-IAP by determining where ML-IAP is expressed in mice, and by generating a mouse strain with genetic ablation of the Birc7 gene encoding ML-IAP. Human ML-IAP is expressed in several human cancers and, to a lesser degree, in a few normal tissues, such as adult kidney [6,32]. Unexpectedly, mouse Birc7 mRNA was detected exclusively in eye samples (Figure 4A). When we revisited ML-IAP expression in normal human tissues, BIRC7 expression was higher in eye tissue than in the kidney (Figure 4B). In both rat and mouse eyes Birc7 showed the highest expression in the complex of RPE (retinal pigment epithelium) and choroid capillaries (Figure 4C). BIRC2, BIRC3 and XIAP showed a much broader pattern of expression in the same samples and were not more abundant in mouse or human eyes (Supplementary Figures S3–S5 at http://www.BiochemJ.org/bj/447/bj4470427add.htm).
Analysis of BRIC7 mRNA expression in mouse, rat and human tissues
ML-IAP-deficient mice were generated in which the first four coding exons were replaced with a LacZ/Neo cassette (Supplementary Figure S3). Quantitative real-time PCR analysis and Western blotting with an anti-mouse ML-IAP antibody confirmed ML-IAP protein and mRNA expression in wild-type, but not ML-IAP−/−, RPE (Supplementary Figure S3). Expression of Birc2, Birc3 or Xiap mRNA was not affected by ML-IAP deficiency (Supplementary Figures S4A–S4C). Mice lacking ML-IAP did not exhibit any developmental abnormalities or immune-system-related disorders, and they aged comparably with their wild-type littermates (results not shown), suggesting that ML-IAP is not critical for overall homoeostasis.
Normal retinal and vessel architecture in ML-IAP-deficient mice
SD-OCT, an increasingly important diagnostic tool in ophthalmology, provides ultra-high-resolution analysis of retinal architecture in the live animal [33–35]. To assess whether the deletion of ML-IAP alters the general morphological structure of the retina or contributes to any retinal degeneration over time, we performed a series of SD-OCT and cSLO (confocal scanning laser ophthalmoscopy) assessments on mutant and wild-type animals. The overall morphological structure in the ML-IAP−/− retina was unaltered (Figure 5A) at first assessment relative to wild-type littermates, and no further changes were noted over the first 6 months of life (Figure 5B). At 6 months of age the average retinal thickness (mean±S.D.) was 262±6 μm for wild-type males, 258±7 μm for ML-IAP−/− males, 258±7 μm for wild-type females and 257±14 μm for ML-IAP−/− females. Assessments in a much older cohort (20–25 months) revealed some alterations (Figure 5C), including geographic atrophy, that were not unique to the mutant eyes, but rather appeared to be age-related and similarly manifested in an age-matched wild-type group of mice.
Normal retinal and retinal vessel architecture in ML-IAP−/− eyes
Vascular structure and integrity were also examined over the same time frame by using FA in addition to the acquired fundus images. For these studies, we restricted our imaging to the right eye. Early- (results not shown) and late-phase recordings revealed a normal vessel morphology and pattern (Figure 5D). As there was no evidence of leakage from the retinal vasculature, we conclude that loss of ML-IAP had no effect on vessel stabilization of the inner blood–retinal barrier. The outer barrier, comprised of the single layer of RPE, also appeared to be intact as there was no fluorescein signal derived from the underlying choroid capillaries.
Normal IOP and retinal function in ML-IAP-deficient mice
ML-IAP is expressed in the CBE (ciliary body epithelium), a circumferential tissue in the front of the eye that performs several functions including aqueous humour production. Eyes of examined animals appeared normal by gross inspection, indicating that IOP was not extremely abnormally low or high. We determined the IOP in wild-type and ML-IAP−/− male and female cohorts. Pressures were monitored through 20–24 weeks of age. IOP at weeks 22–24 for wild-type males was 15.3±4 mmHg (1 mmHg=0.133 kPa) (mean±S.D.), 16±3 mmHg for ML-IAP−/− males, 20±6 mmHg for wild-type females and 18±5 mmHg for ML-IAP−/− females. These pressure measurements were not significantly different between male or female ML-IAP−/− animals in relation to their respective wild-type littermates, nor were differences between genders significant. This is consistent with the normal outward appearance of the eyes and the health of the anterior and posterior segment.
To determine possible effects of ML-IAP deficiency on retinal function, a series of full-field scotopic ERG experiments was performed on ML-IAP−/− mice, and on their corresponding ML-IAP+/+ littermates in a longitudinal manner from 1 to 6 months of development. At 1 month, the amplitude and timing of the scotopic a- or b-waves did not show significant differences between wild-type and mutant mice (Figure 6A). To determine the contribution of third-order neurons to light-induced ERG, OPs were isolated from the electrophysiological recordings. The OP recorded in response to high-intensity light stimuli under scotopic conditions showed no significant differences between genotypes (results not shown). At 6 months (Figure 6B) there was again no indication of reduced retinal function. Throughout the assessment period, all electrophysiological responses indicated that the retinal function in the ML-IAP−/− mutants was not altered.
Normal retinal function in ML-IAP-deficient eyes
The ability of IAP proteins to block cell death mediated through death receptor and mitochondrial pathways and to promote survival signalling pathways, as well as their elevated expression in many human cancers, makes these proteins attractive targets for anti-cancer therapies . One of the major strategies for targeting IAP proteins involves Smac-mimicking IAP antagonists, whose potential to fight human tumours is currently under investigation in several clinical trials . Both monovalent and bivalent IAP antagonists induce rapid auto-ubiquitination and proteasomal degradation of c-IAP1 and c-IAP2 [19,20]. IAP antagonists trigger this proteolytic outcome by causing conformational changes in c-IAP proteins that allow RING dimerization and unlocking of their ubiquitin ligase activity . However, it was not known whether IAP antagonists affect the stability of ML-IAP. In the present study we have shown that, similarly to cIAPs, IAP antagonists can promote proteasomal degradation of ML-IAP. Although c-IAP1 was reported to influence the stability of other IAP proteins, especially c-IAP2 and XIAP [37,38], down-regulation of any of these IAPs did not affect steady-state levels of ML-IAP or IAP antagonist-stimulated degradation of ML-IAP. Interestingly, only the bivalent antagonist BV6 induced ML-IAP degradation, whereas monovalent agents did not affect its stability. The explanation for this stark difference with cIAP degradation was provided by a reconstituted ubiquitination reaction, where the bivalent antagonist, but not monovalent compounds, stimulated ML-IAP E3 ligase activity. ML-IAP is a prototype of a simple IAP protein with a single BIR and a RING domain . Lacking the UBA domain and CARD (caspase-recruitment domain) present in c-IAP proteins , ML-IAP is not expected to engage conformational states that could potentially regulate its E3 ligase activity. Thus bivalent antagonist may bridge two ML-IAP molecules to boost its ubiquitin ligase activity.
Surprisingly, in normal tissues, ML-IAP is almost exclusively expressed in eyes. For 12 years we have referred to this protein as ML-IAP and discussed its expression in a variety of human cancers . High expression of ML-IAP in the eyes has eluded the scientific community, possibly because eye tissues are not included in most commercially available tissue collections. Nevertheless, it is now clear that in humans and rodents ML-IAP is principally expressed in ocular structures. The restricted expression of ML-IAP in the RPE and CBE, both of neuroectoderm origin, focused our phenotype characterization to the structures and functions of the eye.
The RPE, a single polarized layer of cuboidal-columnar pigmented cells between the retina and the choroid, provides the outer blood–retinal barrier and functions at a very high metabolic rate. The apical RPE surface has an intimate interaction with the light transducing PRs (photoreceptors) of the retina. With microvilli and cylindrical cytoplasmic sheaths that enclose the ends of the PRs, the RPE provides nourishment and pigments of the visual cycle to the rods and cones and also removes the continuously shed outer segments of these cells (reviewed in ). Imaging studies including SD-OCT, cSLO and FA, along with functional ERG indicated that this RPE/PR functional unit was intact in the mutant eye.
The ciliary body is a double layer of cuboidal epithelium composed of an inner transparent layer and outer pigmented layer . Like the RPE, the pigmented CBE also performs a barrier function referred to as the blood aqueous barrier. The primary function of this triangular tissue located near the front of the eye are 3-fold: (i) to secrete the aqueous humour that nourishes the avascular lens and cornea, (ii) to maintain IOP for the appropriate shape and optical properties of the globe and (iii) for accommodation/ to change the lens shape for near or distant vision (reviewed in ). The normal outward appearance, health of the anterior and posterior segment and normal IOP of the ML-IAP−/− eye indicated overall normal development and function of the ciliary body.
Genetic ablation of ML-IAP did not reveal any apparent defect during development or at later stages of life. It is possible that ML-IAP is more relevant in cancer cells that are constantly on the verge of death, which requires that they maintain expression of all anti-apoptotic weapons in their arsenal. In addition to ML-IAP, eyes of humans and rodents express other IAP proteins, making ML-IAP a potentially redundant factor for regular ocular functions. Future studies involving crosses of ML-IAP-deficient mice with mice lacking other apoptotic regulators might yet reveal a potential role for ML-IAP in eye homoeostasis.
baculovirus inhibitor of apoptosis repeat
ciliary body epithelium
cellular inhibitor of apoptosis
confocal scanning laser ophthalmoscopy
inhibitor of apoptosis
mouse embryonic fibroblast
retinal pigment epithelium
spectral domain-optical coherence tomography
second mitochondrial-derived activator of caspase
Eugene Varfolomeev performed ML-IAP expression studies; Jasmin Dynek conducted ML-IAP protein stability studies; Kim Newton managed the ML-IAP-knockout colony; Anna Fedorova and Kurt Deshayes produced recombinant ML-IAP protein; Elham Moradi and Jennifer Le Couter performed ocular studies; Jiping Zha performed pathological examination; and Domagoj Vucic conceived and supervised the study, and wrote the paper with input from Eugene Varfolomeev, Wayne Fairbrother, Kim Newton, Jasmin Dynek, and Jennifer Le Couter. All authors provided intellectual input in study design and data analyses.
We thank Erin Dueber, Tanya Goncharov, Dorothy French and members of the Early Discovery Biochemistry and Physiological Chemistry Departments for reagents, suggestions and comments.
All authors are employees of Genentech, Inc.
Present address: Health Interactions, San Francisco, CA 94111, U.S.A.
Present address: Crown Bioscience Inc., 6 Beijing West Road, Jiangsu Province, 215400, People's Republic of China
3To whom correspondence should be addressed (email firstname.lastname@example.org).