Biogenesis and cytotoxicity of APOL1 renal risk variant proteins in hepatocytes and hepatoma cells.

Two APOL1 gene variants, which likely evolved to protect individuals from African sleeping sickness, are strongly associated with nondiabetic kidney disease in individuals with recent African ancestry. Consistent with its role in trypanosome killing, the pro-death APOL1 protein is toxic to most cells, but its mechanism of cell death is poorly understood and little is known regarding its intracellular trafficking and secretion. Because the liver appears to be the main source of circulating APOL1, we examined its secretory behavior and mechanism of toxicity in hepatoma cells and primary human hepatocytes. APOL1 is poorly secreted in vitro, even in the presence of chemical chaper-ones; however, it is efficiently secreted in wild-type transgenic mice, suggesting that APOL1 secretion has specialized requirements that cultured cells fail to support. In hepatoma cells, inducible expression of APOL1 and its risk variants promoted cell death, with the G1 variant displaying the highest degree of toxicity. To explore the basis for APOL1-mediated cell toxicity, endoplasmic reticulum stress, pyroptosis, autophagy, and apoptosis were examined. Our results suggest that autophagy represents the predominant mechanism of APOL1-mediated cell death. Overall, these results increase our understanding of the basic biology and trafficking behavior of circulating APOL1 from the liver.

were plated and cultured according to the manufacturer's instructions (Triangle Research Labs, Research Triangle Park, NC). Briefl y, cells were plated onto collagen-coated dishes at a density of 1.5 × 10 5 cells per cm 2 of culture dish, and medium was replaced with maintenance medium 6 h after plating. Cells were cultured in maintenance medium, with medium replaced every 24 h. All cell culture solutions were supplemented with 100 U/ml penicillin and 100 g/ml streptomycin. Cells were grown at 37°C in an atmosphere containing 5% CO 2 .

Transfection and selection of stable clones
To obtain tetracycline (Tet)-inducible cell lines stably expressing APOL1, McA cells were transfected in 100 mm dishes at 30% confl uence with 5 g pTet-On (Clontech) using Fugene HD transfection reagent (Promega, Madison, WI). Twenty-four hours posttransfection, cells were subjected to selection with DMEM containing 10% FBS supplemented with 500 g/ml G418 (Cellgro, Manassas, VA). Selection medium was replaced every 48 h for 10 days. Individual clones were isolated, expanded, and maintained in 250 g/ml G418. To generate inducible cell clones, the cDNAs of APOL1 G0, G1, and G2 were inserted into the pTRE2hyg plasmid (Clontech), which was transfected into McA Tet-On cells. Twenty-four hours posttransfection, cells were subjected to selection with DMEM containing 10% FBS supplemented with 500 g/ml G418 and 50 g/ml hygromycin B (Invitrogen, Waltham, MA). Selection medium was replaced every 48 h for 14 days. Individual clones were selected and maintained in DMEM containing 10% FBS supplemented with 250 g/ml G418 and 25 g/ml hygromycin B. To induce APOL1 expression, cells were incubated with 1 g/ml doxycycline (Dox) (BD Biosciences, San Jose, CA) for 16 h, unless indicated otherwise. Individual clones were analyzed for inducible APOL1 protein expression by immunoblot analysis. Several clones were characterized for each variant, and one clone from each variant was used for subsequent experiments. In some experiments, Dox-induced APOL1 expressing McA cells were incubated with tunicamycin (Sigma-Aldrich; 2 g/ml) to induce endoplasmic reticulum (ER) stress, 3-methyladenine (3MA) (5 mM; Sigma-Aldrich) to inhibit autophagy, leupeptin (100 g/ml; Sigma-Aldrich), a caspase inhibitor that blocks all forms of lysosomal turnover, MG132 (20 M; Sigma-Aldrich), a proteasome inhibitor, or Z-YVAD-FMK, a caspase-1 inhibitor (20 M, C1I; Millipore, Billerica, MA).

Metabolic radiolabeling and analysis
Unless indicated, cells were grown in 60 mm dishes and labeled for the indicated times with 100 Ci/ml [ Met/Cys in maintenance medium. Following each labeling period, medium was collected and cell debris were cleared by centrifugation and adjusted to 1× lysis buffer [50 mM Tris (pH 7.5), 150 mM NaCl, 1 mM EDTA, 1% Triton X-100, 1 mM PMSF, 10 g/ml pepstatin, and 10 g/ml leupeptin]. Cells were placed on ice and lysed in lysis buffer containing protease inhibitors.
The samples described above were adjusted to 0.2% BSA using concentrated stocks and then subjected to immunoprecipitation by addition of either 2 l rabbit anti-APOL1 antibody

Plasmid construction
pCMV5-APOL1. cDNAs of wild-type (G0) APOL1 , as well as G1 and G2 renal risk variants, were released from the pIRES2-EGFP-APOL1 vector (a gracious gift from Dr. Peter Antinozzi, Department of Biochemistry, Wake Forest School of Medicine) with Bgl II and Sal I restriction enzymes (New England Biolabs, Ipswich, MA) and inserted into the Bgl II-and Sal I-digested pCMV5 vector.
pCMV5-HSA-APOL1. To generate pCMV5-HSA-APOL1, the human serum albumin (HSA) signal peptide sequence was incorporated into the sense primer containing a terminal Bgl II site. The antisense primer contained a terminal Sal I site. Primers used for PCR amplifi cation were: forward, 5 ′ -TAAAGATCTTTTGGCA CAATGAAGTGGGTAACCTTTATTTCCCTTCTTTTTCT CTTT-AGCTCGGCTTATTCCGAGGAAGCTGGAGCGAGG-3 ′ ; reverse, 5 ′ -TAAGTCGACTCACAGTTCTTGGTCCGCCTGCAG-3 ′ . The pCMV5-APOL1 plasmid served as a template for PCR. PCR products were purifi ed, digested with Bgl II and Sal I, and inserted into the Bgl II-and Sal I-digested pCMV5 vector.
pTRE2hgy-APOL1. cDNAs of G0 APOL1 , as well as the G1 and G2 renal risk variants, were released from the pIRES2-EGFP-APOL1 plasmid with Nhe I and Sal I and inserted into the Nhe Iand Sal I-digested pTRE2hgy vector (Clontech, Mountain View, CA).

Generation of APOL1 transgenic mice
APOL1 was cloned into the pLiv11 vector under control of the apoE promoter and liver-specifi c enhancer region ( 13 ), as described above. The entire insert and fl anking regions were verifi ed by DNA sequence analysis. The pLiv11-APOL1 transgene was separated from the plasmid backbone by digestion with Sal I (5 ′ ) and Spe I (3 ′ ). Linearized plasmids were microinjected into pronuclei of fertilized B6D2F1/J mice by the Transgenic Mouse Core Facility at the University of Florida. Presence of the APOL1 transgene, apoE promoter, and hepatic control region were verifi ed by PCR, and immunoblot analysis was used to confi rm the secretion of APOL1 in plasma (data not shown).

Generation of anti-human APOL1 antiserum
cDNA corresponding to G0 APOL1 (amino acid residues 200-398 of the APOL1 precursor protein) was generated by PCR and cloned into the pMAL-C5E vector (New England Biolabs) to generate a maltose binding protein-APOL1 fusion protein in Escherichia coli . Primer sequences used for PCR were: forward, 5 ′ -TAAGGTACCGACAGAGGGAGGCAGCC-3 ′ ; reverse, 5 ′ -ACCGTCGACTCAGAAGGGTGCCAGACCC-3 ′ . Following purification via amylose affi nity chromatography, the fusion protein was injected subcutaneously into rabbits with incomplete Freund's adjuvant (Lampire Biologicals, Pipersville, PA). Whole serum was used at 1:1,000 dilution for immunoblot analysis and 5 l whole serum was used for immunoprecipitation (fi nal volume, 1 ml).

Cell culture
McArdle rat hepatoma 7777 (McA) cells and COS7 cells were obtained from America Type Culture Collection (Manassas, VA) and grown in DMEM containing 4.5 g/l glucose and 10% FBS. Human hepatocytes were a gift from Dr. Colin Bishop (Wake Forest Institute for Regenerative Medicine). Human hepatocytes MA; 34078) followed by direct visualization using a Fujifi lm LAS-3000 imager. In some cases, band intensities were quantifi ed using MultiGauge software. APOL1 protein was detected using the rabbit anti-APOL1 antibody (Sigma-Aldrich; HPA01885). Other antibodies used included rabbit polyclonal to GRP78 (Sigma-Aldrich; HPA018885), 5 l rabbit anti-maltose binding protein-APOL1 fusion protein serum (APOL1-MBP antibody), or 5 l anti-mouse serum albumin antibody (Abcam, Cambridge, MA; AB34807). After 18 h incubation with rotation at 4°C, 20 l of protein G-Sepharose (50:50 slurry; Amersham Biosciences, Buckinghamshire, UK) was added to each sample followed by an additional 3 h incubation. Beads were collected by centrifugation at 10,000 rpm for 10 s and washed three times with lysis buffer. Proteins were eluted from the beads by heating in SDS-PAGE sample buffer at 100°C for 5 min and fractionated by 12% SDS-PAGE. Gels were then dried and visualized using a FujiFilm BAS-5000 phosphorimager. In the indicated experiments, band intensities were quantified using MultiGauge software.

Immunofl uorescence microscopy
APOL1 stably transfected McA cells were plated onto collagencoated 60 mm dishes. Twenty-four hours later, APOL1 expression was induced with 1 g/ml Dox for 10 h. Cells were then fi xed in 3.7% formaldehyde in PBS for 20 min, followed by incubation for 1 h in PBS containing 10 mM glycine, 0.1% saponin, and 1% BSA. Cells were then incubated for 1 h in primary antibodies at the indicated dilutions in PBS containing 0.1% saponin and 1% BSA. A 1:500 dilution was used for both rabbit anti-human APOL1 monoclonal antibody (Abcam; AB108315) and mouse anti-rat protein disulfi de isomerase (PDI) antibody (Pierce, Waltham, MA; MA3-019). Cells were then incubated with rhodamine-conjugated goat anti-rabbit IgG (1:1,000) and/ or FITC-conjugated goat anti-mouse IgG, (1:200) (Jackson Im-munoResearch; West Grove, PA) for 1 h in PBS containing 0.1% saponin and 1% BSA. Cells were postfi xed with 3.7% formaldehyde in PBS and stained with DAPI, where indicated. Cells were then mounted with ProLong® Gold Antifade reagent (Invitrogen) and viewed with an Olympus BX51 fl uorescence microscope using either a 10× objective or a 100× oil objective.

Immunoblot analysis
Cells were lysed on ice with lysis buffer, centrifuged at 10,000 g for 5 min and supernatants mixed with concentrated SDS-PAGE sample buffer. Proteins were separated by SDS-PAGE and transferred to a polyvinylidene difl uoride membrane. After incubation with primary and HRP-conjugated secondary antibodies, membranes were incubated with SuperSignal West Pico chemiluminescent substrate (Thermo Scientifi c, Waltham,  Cell lysates (C) and media (M) were then collected and APOL1 was immunoprecipitated with anti-APOL1 (APOL1-MBP) antibody and analyzed by SDS-PAGE and phosphorimaging. The experiment was performed in triplicate. mean ± SEM. Statistical analyses were performed using Graph-Pad Prism software.

Expression and secretion of APOL1 in COS7 cells
The consequences of APOL1 expression were initially examined by expressing the G0 isoform in COS7 cells. Twenty-four hours after transfection, cells were radiolabeled with [ 35 S]Met for 4 h, and cell and media samples were subjected to immunoprecipitation using rabbit anti-APOL1. As displayed in Fig. 1A , APOL1-transfected cell and media samples revealed three bands by SDS-PAGE, which were not detected in the mock-transfected control (compare lanes 1 and 2 to lanes 3-8). Based on molecular mass, the upper band represents the full-length 46 kDa form of APOL1; the lower bands likely correspond to degradation products . As APOL1 is believed to associate primarily with lipoprotein particles in human plasma and is easily degraded during storage (data not shown), it is possible that secreted APOL1 could be degraded in labeling medium lacking FBS; another possibility is that APOL1 may require additional time for secretion. To address these possibilities, G0 APOL1-transfected COS7 cells were (BiP; Abcam; ab21685) and anti-LC3 (NOVUS, St. Louis, MO; NB100-2200).

APOL1 toxicity assay
G0 APOL1, and G1 and G2 renal disease variants were stably transfected into McA cells and incubated under varying conditions, as described in the Results, for the indicated time periods. G0 and disease variants (G1, G2) APOL1-induced cell death was directly visualized by microscopy and quantifi ed by measuring cellular lactate dehydrogenase (LDH) release into the medium using a cytotoxicity assay kit (Pierce; 88954) according to the manufacturer's instructions.

Flow cytometry
Stably transfected McA cells (1 × 10 6 ) were plated in 6-well plates and grown to 80% confl uence. Following Dox induction for the indicated time points, cells were harvested for fl ow cytometric analysis using annexin (Anx) V-FITC apoptosis detection kit I (BD Pharmingen, San Jose, CA) according to the manufacturer's instructions. Data were acquired on a BD FACSCalibur™ fl ow cytometer and analyzed using FlowJo software.

Data analysis
Data were analyzed using one-way ANOVA followed by Tukey's multiple comparison test to determine signifi cant differences among groups. A Student's t -test was performed to determine statistical signifi cance between two groups. P < 0.05 was considered to indicate statistical signifi cance. Values are represented as the the folding and secretion of mutant or poorly folded and secreted proteins ( 16,17 ). However, neither chemical chaperone was capable of enhancing secretion of the G0, G1, or G2 APOL1 isoforms ( Fig. 2A, B ; lanes 7-12).
APOL1 secretion was further examined in primary human hepatocytes to determine whether the poor APOL1 secretion effi ciency is specifi c to immortalized cell lines. After plating, primary human hepatocytes were subjected to continuous radiolabeling with [ 35 S]Met and [ 35 S]Cys for 4 h. Cell lysates and medium were then subjected to immunoprecipitation with anti-APOL1 and anti-albumin antibodies. Results revealed that APOL1 remained almost exclusively in the cell lysate, whereas HSA was effi ciently secreted into media ( Fig. 3A ), demonstrating that APOL1 is also poorly secreted from primary cultured hepatocytes.
To determine whether the ineffi cient secretion of APOL1 is also observed in vivo, we examined the secretion of human APOL1 in liver-specifi c transgenic mice. APOL1 expression was confi rmed in transgenic mice relative to nontransgenic littermates following injection with 200 Ci [ 35 S]Met and immunoprecipitation of plasma with prebleed rabbit sera or anti-APOL1 and anti-albumin antibodies ( Fig. 3B ). Immunoprecipitation of APOL1 transgenic mouse plasma from the indicated postinjection time points revealed that the kinetics of APOL1's appearance in plasma were similar to those of albumin ( Fig.  3C ), suggesting that the ineffi cient secretion of APOL1 observed previously is likely caused by in vitro culture conditions .

APOL1 may be targeted for proteasomal degradation
Ineffi cient secretion is a characteristic of proteins that are also targeted for presecretory degradation, likely due radiolabeled with [ 35 S]Met and [ 35 S]Cys for 4 or 24 h in the presence or absence of 10% FBS. Cells and media were then subjected to immunoprecipitation . Neither an extended labeling period nor supplementation with 10% FBS in the labeling medium improved the apparent secretion effi ciency of APOL1 (data not shown). To explore the molecular basis for the low secretion effi ciency of APOL1, we fi rst addressed whether the APOL1 signal peptide is ineffi cient in its ability to target the precursor protein to the ER membrane. When the native APOL1 signal peptide was exchanged with the signal peptide from HSA ( Fig. 1B ), no improvement in secretion was observed (compare Fig. 1A ,  lanes 3-8 with lanes 9-14). Hence, we concluded that the ineffi cient APOL1 secretion likely relates to its ineffi cient folding and/or traffi cking within the ER and Golgi.

In vivo APOL1 secretion is more effi cient than in vitro secretion from cell lines or cultured primary hepatocytes
We next determined whether the cell type used for APOL1 expression impacted its secretory effi ciency. Because APOL1 is abundantly expressed in the liver, which is likely the primary source of circulating APOL1 ( 8 ), McA rat hepatoma cells were stably transfected with APOL1 G0, G1, and G2 variants under the control of a Tet-inducible promoter, using methods described previously ( 14,15 ). When cells were induced with Dox for 16 h and metabolically radiolabeled with [ 35 S]Met for 4 h, abundant protein was detected in cell pellets; however, the amount observed in media samples remained very low ( Fig. 2A, B ; lanes 1-6), demonstrating that cell lines of hepatic origin are also incapable of effi cient APOL1 secretion. Several chemical chaperones, including 4-phenyl butyric acid (Sigma-Aldrich; 10 mM) and trimethylamine N -oxide (Sigma-Aldrich; 150 mM), have been shown to enhance to the nonspliced form (479 bp) ( Fig. 7A ). In contrast, treatment with tunicamycin (2 g/ml), a robust inducer of ER stress, caused signifi cant induction of XBP-1 splicing ( Fig. 7A ). Consistent with these fi ndings, tunicamycin induced the accumulation of BiP, a known transcriptional target of activated XBP-1, whereas BiP was not induced by Dox-induced expression of the G0, G1, or G2 forms of APOL1. ( Fig. 7B, C ), suggesting that ER stress is not likely a cause of APOL1-induced cell death. Interestingly, treatment with tunicamycin signifi cantly reduced the cellular toxicity of all three APOL1 forms relative to untreated cells ( Fig. 7D , +Dox). This cell protection may result from reduced APOL1 mRNA and protein expression after tunicamycin treatment, which is a known adaptation caused by the ER stress response ( 21 ) ( Fig. 7E ).

APOL1 toxicity is regulated by autophagy
Wild-type (G0) APOL1 has been reported to trigger autophagy-mediated cell death ( 19,22 ). To determine to an inherent property of the protein, such as ineffi cient folding or lack of an essential binding partner or cofactor ( 18 ). We therefore explored whether APOL1 undergoes presecretory degradation and, if so, by which pathway. Stably transfected McA cells were induced with Dox (1 g/ml) for 4 h and subjected to continuous radiolabeling with [ 35 S]Met and [ 35 S]Cys for 3 h in the absence or presence of leupeptin, a caspase inhibitor that blocks all forms of lysosomal turnover including autophagocytic degradation, or MG132, a proteasome inhibitor. As shown in Fig.  4A, B , no signifi cant differences were observed in cellular or media APOL1 abundance upon incubation with leupeptin. In contrast, MG132 increased the abundance of cellular APOL1 protein ( Fig. 4C, D ). While G0 appeared to be the most highly stabilized, MG132 also facilitated the accumulation of both the G1 and G2 isoforms. While the cellular abundance of each form of APOL1 increased under conditions of proteasome inhibition, no increase in secretion was observed.
As the retrograde translocation of proteins from the ER to cytosol is a prerequisite for proteasome-mediated ERassociated degradation (ERAD), it was also important to determine whether APOL1 is localized to the ER. Fluorescence microscopy revealed that APOL1 colocalized with a known ER marker, PDI, in stably transfected McA cells following induction with Dox for 10 h ( Fig. 5A, B ). Hence, it appears that APOL1 is retained within the ER and does not traffi c effi ciently to the Golgi within this time period, consistent with the lack of APOL1 protein in media fractions from transfected cells.

Cellular toxicity of APOL1 constructs
G0 APOL1 induces cell death in several tissues ( 19 ). We found that Dox-induced expression of all forms of APOL1 in transfected McA cells caused cell death, as evidenced by the loss of cell adherence and cell number after 22 h of Dox induction ( Fig. 6A ). To compare the cell death-inducing capability among G0 and disease variant APOL1 isoforms at similar levels of protein expression ( Fig. 6B ), an LDH release assay was performed. McA cells expressing G0 and G2 showed comparable levels of LDH release, whereas induction of the G1 variant led to the highest LDH release ( Fig. 6C ). While both APOL1 variants associate with kidney disease, our results suggest that the G1 variant is more toxic than G2, which retains similar cellular toxicity as that of G0 APOL1.

APOL1 expression does not activate the ER stress pathways
ER stress, also referred to as the unfolded protein response, constitutes an important mediator of cell death. As APOL1 G0, G1, and G2 are poorly secreted in vitro ( Figs. 1-3 ) and are targeted for proteasomal degradation ( Fig. 4 ), we next explored whether APOL1 activates ER stress. As a measure of ER stress, cleavage of XBP-1 mRNA by RT-PCR was examined. Upon ER stress, XBP-1 pre-mRNA is cleaved by IRE1 and ligated to form activated (spliced) XBP-1 ( 20 ). Dox induction of APOL1 expression did not increase the spliced form of XBP-1 (453 bp) relative caspase-3 Western blot analysis were unsuccessful due to lack of a specifi c rat antibody. Interestingly, fl ow cytometry also revealed signifi cant alterations in cell size/morphology in all APOL1-overexpressing populations following Dox induction, evidenced by an increase in the percent of small cells and concomitant decrease in large cells, indicative of cells undergoing or likely to undergo autophagic cell death ( 19 ) ( Fig. 10 ).
Pyroptosis is another type of programmed cell death triggered by caspase-1 following its activation by various infl ammasomes, resulting in cellular lysis and release of cytosolic contents into the extracellular space ( 23,24 ). To test whether pyroptosis is involved in APOL1-induced cell death, cells were incubated with caspase-1 inhibitor, Z-YVAD-FMK, during APOL1 induction. Interestingly, caspase-1 inhibition signifi cantly decreased LDH release from APOL1-overexpressing cells relative to control cells ( Fig. 11 ), suggesting that caspase-1-induced pyroptosis may also contribute to APOL1-induced cytotoxicity.

DISCUSSION
Although the mechanism by which APOL1 risk variant proteins kill trypanosomes has been defi ned, the role of these variants in development of APOL1 -associated diseases remains poorly understood. Furthermore, expression of G0 (wild-type) APOL1 appears to induce programmed cell death, although the exact mechanism has yet to be determined. Prior reports on the main mechanism of APOL1induced cell death implicate the autophagic pathway ( 19,22 ), while others have shown that APOL1 induces podocyte necrosis by compromising lysosomal membrane permeability ( 25 ). Major goals of the present study were to elucidate the pathway of APOL1-induced cell death, as well as general outcomes of APOL1 expression in liver whether autophagy was involved in the cytotoxicity of APOL1 and its disease variants in our inducible cell lines, we fi rst performed Western blot analysis to examine LC3-II accumulation, a reliable marker of autophagy. As shown in Fig. 8A , Dox-induced stably-transfected McA cells showed an increased amount of LC3-II induction relative to untransfected cells, suggesting that autophagy contributes to APOL1 G0-and APOL1 nephropathy variantinduced cell death. Considering that the lower level of G1 mutant expression induced the highest amount of LC3-II, it is likely that the G1 variant is more toxic to cells than G0 or G2, consistent with the LDH release assay results reported above ( Fig. 6C ). To confi rm that autophagy is the cause of subsequent cell death, the autophagy inhibitor 3MA was employed. Addition of 5 mM 3MA to induced cells attenuated APOL1-induced cell death, as evidenced by decreased LDH release ( Fig. 8B ), further indicating that APOL1-induced cell death is, at least partially, caused by activation of autophagy.

Pyroptosis may also contribute to APOL1-induced cell death
APOL1 was identifi ed in 2009 as a novel BH3-only protein ( 19 ). BH3-only proteins represent the pro-apoptotic subclass within the BCL-2 family of apoptosis regulators. We therefore explored a potential role for apoptosis in mediating cell injury and death mediated by APOL1 and its G1 and G2 variants. Following 18 h of Dox induction, fl ow cytometric analysis of annexin V/propidium iodide (PI)-stained cells revealed a signifi cant increase in doublepositive (Anx + PI + )-stained stably transfected APOL1 G1expressing cells, relative to the G0 and G2 isoforms ( Fig. 9 ), indicating a role for late apoptosis in APOL1 G1-induced cell death. However, no difference was observed between the parental McA cells and G0 or G2 isoforms. Attempts to confi rm increased apoptosis in the G1 isoform by cleaved APOL1 is predominantly localized within the ER and excluded from the Golgi, thereby explaining why it is not secreted within the time periods examined. Intracellular abundance of APOL1 and its variants was increased by incubation of cells with the proteasome inhibitor, MG132. Hence, both transfected APOL1 and endogenous APOL1 expressed in human hepatocytes fails to achieve a transport-competent form, and is therefore retained in the ER and subjected to retrograde translocation and proteasome-mediated turnover. Interestingly, we demonstrated that APOL1 expressed by mouse hepatocytes, in vivo, is efficiently secreted, suggesting that APOL1-folding and/or assembly has specialized requirements that cultured cells fail to support. Although all forms of APOL1 were retained in the ER, APOL1 expression in stably transfected McA cells did not enhance XBP-1 splicing or BiP expression, suggesting absence of an ER stress response. Indeed, treatment with cells, because the liver is the primary source of APOL1 proteins and exhibits highly effi cient secretory activity.
The intracellular traffi cking of APOL1 was assessed to gain a better understanding of its potential roles in human disorders. APOL1 is ineffi ciently secreted in COS7 and McA cell lines, as well as cultured primary human hepatocytes. Furthermore, ineffi cient secretion is not likely caused by the signal peptide, because replacement with the HSA signal peptide failed to improve APOL1 secretion from COS7 cells. Poor secretion and stabilization of APOL1 by proteasome inhibitors led us to consider the possibility that, under the cell culture conditions employed, APOL1 undergoes incomplete folding and/or assembly, making it a substrate for ERAD and possible activation of ER stress pathways.
To determine whether APOL1 proteins are targeted for ERAD, we examined colocalization with the ER marker, PDI. Immunofl uorescence microscopy confi rmed that tunicamycin, a potent inducer of ER stress, protected cells from APOL1-induced cell death, likely by decreasing APOL1 mRNA and protein levels.
As APOL1 expression did not appear to induce ER stress, we aimed to clarify the mechanism of cell death associated with expression of APOL1 proteins. We concluded that all forms of APOL1 are toxic, although based on LDH release assays, the G1 variant appeared to be the most cytotoxic among the three variants examined. APOL1-expressing cells treated with 3MA exhibited a signifi cant decrease in toxicity, suggesting that autophagy is a potential mechanism for APOL1-induced toxicity, in agreement with prior reports ( 19,22 ). Moreover, stable expression of G0, G1, and G2 APOL1 clearly induced autophagy, as demonstrated by increased LC3-II production. Consistent with differential toxicity among APOL1 isoforms observed by LDH release assays, the G1 variant induced the highest accumulation of LC3-II.
While APOL1 was previously identifi ed as a BH3-only protein, no evidence has suggested that APOL1-induced toxicity is directly related to apoptosis. However, it is possible, as for other BH3-only family members, that APOL1 may regulate the apoptotic pathway. Although our data did not indicate that apoptosis was the major mechanism of cell death in APOL1-expressing McA cells, as the percentage of annexin V + /PI + cells increased from only ‫ف‬ 2% to 4%, a signifi cant increase in late-stage apoptosis associated with the G1 isoform relative to the other APOL1 variants was observed following Dox induction. Due to sensitivities to both time and dose of APOL1 regarding cellular toxicity, it will be important to examine the role of apoptosis in APOL1-expressing cells at additional time points.
The in vitro and in vivo studies on APOL1 secretion and toxicity presented herein are aimed to shed light on the consequences of APOL1 risk variant overexpression. Our results strongly suggest that cell lines, such as those used in this study, likely retain APOL1 proteins due to ineffi cient secretion, resulting in cellular toxicity. In contrast, APOL1 is effi ciently secreted by hepatocytes in vivo, as evidenced by our hepatocyte-specifi c APOL1 transgenic mouse model, which likely explains why hepatic toxicity is not observed in African Americans with APOL1 renal risk variants. Thus, cells with high secretory capacity may escape the toxic effects of APOL1 expression, whereas those with low secretory capacity do not. Secreted APOL1 proteins may play an important role in the reduced risk of calcifi ed atherosclerotic plaque (subclinical atherosclerosis) in African Americans with two APOL1 renal risk variants ( 3,4 ). Results may further improve our understanding of the potential mechanisms of APOL1 -associated pathology in this population. While signifi cant differences in toxicity were observed among the G0 and risk variant forms of APOL1 (notably G1), the mechanism of renal risk variant-induced kidney damage requires further study.
Taken together, the current results provide new insights into the intracellular traffi cking and secretion of APOL1 and the mechanism of APOL1 -induced cell death. While all APOL1 isoforms are ineffi ciently secreted in vitro, we showed that APOL1 is effi ciently secreted in G0 APOL1 transgenic mice, suggesting the presence of an in vivo binding partner/cofactor, which will be an important factor to consider in future studies on APOL1 secretion and function. Upon examining the major pathways of programmed cell death, including ER stress, pyroptosis, autophagy, and apoptosis, our results indicate that autophagy represents the main mechanism of APOL1-induced hepatic cell death. Apoptosis and/or   pyroptosis may also contribute, albeit to a far lesser extent. Additional studies are required to further elucidate the mechanism(s) of APOL1-induced cell death and identify potential differences among APOL1 variant proteins that ultimately contribute to progression of nondiabetic nephropathy in African Americans with two APOL1 renal risk variants.
The authors gratefully acknowledge Dr. Colin Bishop (Institute for Regenerative Medicine, Wake Forest School of Medicine) and Ivy Mead for providing human hepatocytes and suggestions for their in vitro culture.