Hsp40s play distinct roles during the initial stages of apolipoprotein B biogenesis

Abstract

Apolipoprotein B (ApoB) is the primary component of atherogenic lipoproteins, which transport serum fats and cholesterol. Therefore elevated levels of circulating ApoB are a primary risk factor for cardiovascular disease. During ApoB biosynthesis in the liver and small intestine under nutrient-rich conditions, ApoB cotranslationally translocates into the endoplasmic reticulum (ER) and is lipidated and ultimately secreted. Under lipid-poor conditions, ApoB is targeted for ER-associated degradation (ERAD). Although prior work identified select chaperones that regulate ApoB biogenesis, the contributions of cytoplasmic Hsp40s are undefined. To this end, we screened ApoB-expressing yeast and determined that a class A ER-associated Hsp40, Ydj1, associates with and facilitates the ERAD of ApoB. Consistent with these results, a homologous Hsp40, DNAJA1, functioned similarly in rat hepatoma cells. DNAJA1-deficient cells also secreted hyperlipidated lipoproteins in accordance with attenuated ERAD. In contrast to the role of DNAJA1 during ERAD, DNAJB1—a class B Hsp40—helped stabilize ApoB. Depletion of DNAJA1 and DNAJB1 also led to opposing effects on ApoB ubiquitination. These data represent the first example in which different Hsp40s exhibit disparate effects during regulated protein biogenesis in the ER and highlight distinct roles that chaperones can play on a single ERAD substrate.

INTRODUCTION

Each year, cardiovascular disease kills ∼18 million people, and one in four deaths in the United States is due to cardiovascular disease (Heron, 2017; Benjamin et al., 2019). Coronary artery disease (CAD) is the most common form of cardiovascular disease and arises when arteries narrow due to the deposition of cholesterol-filled macrophages that form plaques. This process, known as atherosclerosis, can lead to angina from lumenal narrowing or to a heart attack if a plaque ruptures and causes an acute thrombotic occlusion. Current treatments for CAD include drugs that target various aspects of lipid metabolism, including the synthesis, transport, and endocytosis of lipids and/or cholesterol (Goldstein and Brown, 2015; Adhyaru and Jacobson, 2018; Larsen et al., 2019; Libby and Everett, 2019). The most commonly used drugs are the statins. For the vast majority of individuals, statins represent an effective first-line treatment to lower the risk of CAD, but the response to statins varies widely. For example, the decrease in circulating levels of low-density lipoproteins (LDLs) can vary from 5–70% in an individual (Reiner et al., 2013; Reiner and Tedeschi-Reiner, 2013). This variation most often arises from polymorphisms in genes linked to statin metabolism (i.e., pharmacodynamics and pharmacokinetics) (Reiner, 2014; Sikka et al., 2011; Dadu and Ballantyne, 2014; Ward et al., 2019). Moreover, in some patients, statins elicit several side effects including muscle weakness, leg cramps, and myopathy, forcing select individuals to discontinue medication or to lower the dose. Therefore it is imperative to identify new therapeutic targets to combat CAD.

Triacylglycerol, phospholipids, cholesterol, and cholesterol esters are transported through the circulatory system in lipoprotein particles. The major structural component of these atherogenic very low-density lipoproteins (VLDLs), LDLs, and chylomicrons is apolipoprotein B (ApoB), a ∼550 kDa amphipathic protein (Bostrom et al., 1986; Olofsson et al., 1987; Boren et al., 1994). (In contrast, high-density lipoproteins [HDLs] are ApoAI and ApoAII associated.) ApoB aids in the assembly, secretion, transport, and cellular uptake of lipoproteins. In mammals, ApoB is expressed in the liver and small intestine in two different isoforms, ApoB100 and ApoB48, which represent the full-length and N-terminal 48% portion of the full-length protein, respectively (Chen et al., 1986, 1987, 1988; Powell et al., 1987; Hussain et al., 1996, 2005). Not surprisingly, there is a direct correlation between circulating ApoB and CAD: higher levels cause CAD, whereas lower levels are associated with normal homeostasis or, if severely limited, with hypobetalipoproteinemia, which presents as delayed development, hepatomegaly, steatorrhea, and cytolysis (Jang et al., 2020; Welty, 2020). Based on its central role in lipoprotein metabolism, ApoB has been directly targeted to treat CAD (Zimmermann et al., 2006; Boekholdt et al., 2014; Reyes-Soffer et al., 2016; Nandakumar et al., 2018). Unfortunately, a second generation ApoB-targeted antisense oligonucleotide-based drug (Mipomersen) is effective yet exhibits severe side effects (namely, liver damage) (Astaneh et al., 2021). Nevertheless, these data position ApoB as a bona fide therapeutic target.

As nascent ApoB is cotranslationally translocated into the endoplasmic reticulum (ER) through the Sec61 translocon, the apolipoprotein is lipidated by the ER lumenal microsomal triglyceride transfer protein (MTP) complex and a premature lipoprotein particle is formed (Wu et al., 1996; Fisher et al., 1997; Mitchell et al., 1998; Zhou et al., 1998). This pre-VLDL particle then exits the ER via a nonclassical pathway that requires enlarged COPII vesicles. The particle next enters the Golgi where it undergoes further lipidation and lipid remodeling before it is finally secreted (Gusarova et al., 2003; Siddiqi et al., 2003; Butkinaree et al., 2014; Sane et al., 2017). In contrast, if a cell lacks adequate lipids—or in the absence of MTP function—ER entry is slowed so continued translation exposes hydrophobic ApoB loops in the cytosol (Gordon et al., 1996). These potentially aggregation-prone loops are recognized by molecular chaperones, which in theory could help maintain solubility and facilitate ubiquitination by an ER resident E3 ubiquitin ligase, gp78 (Liang et al., 2003b; Fisher et al., 2011). Ultimately, ApoB is degraded by the cytosolic 26S proteasome with the aid of the AAA⁺-ATPase p97/VCP (Fisher et al., 1997, 2008; Rutledge et al., 2009). Over many years, our group and others have identified and characterized several classes of molecular chaperones, including BiP, heat shock protein (Hsp)70, Hsp90, Hsp110, Hsp104, protein disulfide isomerase (PDI), calnexin, and calreticulin, which play various roles during the folding and ER-associated degradation (ERAD) of ApoB (Zhou et al., 1995; Fisher et al., 1997; Ginsberg, 1997; Chen et al., 1998; Tatu and Helenius, 1999; Gusarova et al., 2001; Hrizo et al., 2007; Rutledge et al., 2013; Doonan et al., 2019).

Although Hsp70 is one of the many chaperones that regulate ApoB biogenesis, the function of this chaperone commonly requires an Hsp40 partner. Hsp40s, also known as J domain proteins (JDPs), are present in all cell types and assist Hsp70s by enhancing Hsp70 ATPase activity and facilitating substrate targeting and selectivity (Kampinga and Craig, 2010; Craig and Marszalek, 2017). Like Hsp70s, Hsp40s also play essential roles during protein folding, assembly, translocation, the prevention of aggregation, ubiquitination, and degradation (Kampinga et al., 2019). In addition, Hsp40s reside in one of three classes, which are based on sequence similarity and domain architecture. Recent work indicates that class A and B Hsp40 chaperones function in tandem with Hsp70 and an Hsp70 nucleotide-exchange factor, Hsp110, to solubilize small aggregates both in vitro and in vivo (Nillegoda et al., 2015, 2017; Kirstein et al., 2017). In addition, Hsp40s are one of the largest families of Hsps, and mammalian cells express >50 family members. This network of Hsp70 cofactors has expanded during evolution to accommodate the concomitant increase in the number of substrates in more complex organisms, and thus their roles are critical for the maintenance of protein homeostasis, or “proteostasis” (Rebeaud et al., 2021). To date, however, only one Hsp40 in the ER membrane, with an ER lumenal-facing J domain, has been shown to facilitate ApoB degradation (Oyadomari et al., 2006; Rutkowski et al., 2007). In contrast, the role of cytosolic Hsp40s in ApoB ERAD has never been explored.

Based on the importance of preventing ApoB aggregation before proteasome delivery, we set out to examine if cytosolic class A and B Hsp40s facilitate ApoB solubility and then target ApoB for ERAD. By using a yeast model we previously established (Hrizo et al., 2007; Grubb et al., 2012; Doonan et al., 2019), as well as a rodent cell line that endogenously synthesizes and secretes ApoB-containing lipoproteins, we observed—as anticipated—that class A (DNAJA1) and class B Hsp40 (DNAJB1) associate with ApoB. We then found that the class A Hsp40 facilitates ApoB degradation, but surprisingly the class B Hsp40 instead stabilizes the apolipoprotein. Yet, in line with these unexpected results, the ablation of the class A Hsp40 led to the accumulation of ubiquitinated ApoB, whereas class B knockdown decreased the levels of ubiquitinated ApoB. These data suggest that the chaperones act at different steps during ApoB maturation and more generally indicate for the first time that substrate-associated Hsp40s can play opposing roles during ERAD.

RESULTS

The ER-associated Hsp40, Ydj1, and the Ssa1 Hsp70 chaperone facilitate ApoB ERAD in yeast

Prior work established that cytoplasmic Hsp70 is required for ApoB degradation (Fisher et al., 1997; Gusarova et al., 2001). This discovery was made possible in part through the development of a system in which the contributions of reagents isolated from yeast cells and cell culture systems could be investigated in vitro (Gusarova et al., 2001). To test if Hsp40s also oversee ApoB stability—and to confirm that yeast Hsp70 is required for ApoB ERAD when expressed in this model organism—we utilized a Saccharomyces cerevisiae β-estradiol inducible expression system (Doonan et al., 2019). In brief, yeast strains were transformed with a linearized plasmid encoding a chimeric transcription factor, pACT1-GEV, and cells were selected in which the construct was stably integrated (McIsaac et al., 2011). This chimeric transcription factor is composed of a GAL4 DNA-binding domain, an estrogen-binding domain, and the VP16 transcription factor. The GEV chimeric transcription factor is constitutively expressed under the control of an ACT1 promoter but remains inactive in the cytosol and bound to Hsp90. Once β-estradiol is added, the hormone targets the estrogen receptor domain in GEV, thus releasing the Hsp90 repressor. The liberated GEV enters the nucleus, binds the upstream sequence of GAL promoters, and rapidly activates transcription of GAL-regulated genes.

To establish the anticipated role of Hsp70 in facilitating the ERAD of ApoB in yeast, we conducted cycloheximide chase assays in the ssa1-45 strain, which lacks the homologous SSA2, SSA3, and SSA4 genes and is viable at permissive temperatures due to the presence of a thermosensitive SSA1 allele (Becker et al., 1996). An isogenic strain containing a wild-type copy of SSA1 served as a control. The version of ApoB used for this analysis contains the N-terminal 29% of the protein, which is the smallest fragment capable of transiting through the secretory pathway and is still metabolically regulated in higher eukaryotes (Collins et al., 1988; Linton et al., 1993; McLeod et al., 1994). After ACT1-GEV versions of SSA1 and ssa1-45 yeast strains were generated, a cycloheximide chase assay was performed after cells had been shifted for 30 min to 37°C. Consistent with data from mammalian cells and in vitro (Fisher et al., 1997; Gusarova et al., 2001), the initial rate of ApoB29 degradation was significantly slowed in ssa1-45 yeast (Figure 1A).

FIGURE 1:

Ydj1 associates with and facilitates the degradation of ApoB29. (A) Cycloheximide chase assays were conducted as described in the Materials and Methods. The turnover of ApoB was examined in SSA1 (open circle) and ssa1-45 (closed circle) strains expressing ApoB29 after cells had been shifted to 37°C for 30 min and β-estradiol was present for 1 h. Equivalent ODs of cells were collected at the indicated time points and processed for quantitative Western blotting. Data from each time point were normalized to the G6PDH loading control. The amount of ApoB29 remaining over time was plotted relative to the amount of protein at time 0. (B) Cycloheximide chase assays were conducted in the HLJ1YDJ1 (open square), ∆hlj1 (closed square), ydj1-151 (open circle), and ∆hlj1ydj1-151 (closed circle) strains expressing ApoB29 as described above. (C) Cycloheximide chase assays were again conducted in each strain but in this case the yeast expressed a model ERAD substrate, Chimera A*, whose turnover is known to similarly require Ydj1 function. (D) Coimmunoprecipitation assays were performed as described in the Materials and Methods using an antibody against ApoB in wild-type cells expressing HA-tagged ApoB29, and precipitated proteins were subject to immunoblot analysis with the indicated antibodies. Note the presence of ApoB29 as well as the Ssa1, Ydj1, and PDI molecular chaperones in the “IP: ApoB” lane from the antibody-treated (+) sample. (E) Coimmunoprecipitation assays were performed as described in the Materials and Methods and in Figure 1D. In panels A–C, data were calculated from N = 5 independent experiments, ±SE, *p < 0.05. Statistical significance was calculated by Student’s unpaired t test.

We next explored the contributions of two cytosolic ER-associated Hsp40s with established roles in ERAD. Yeast express >20 Hsp40 isoforms (Kampinga and Craig, 2010; Kampinga et al., 2019), but of the dozen or so Hsp40s present in the cytosol, only Hlj1 and Ydj1 are associated with the ER membrane. Whereas Hlj1, a class B Hsp40, embeds into the ER membrane due to a C-terminal tail anchor, Ydj1, a class A Hsp40, is tethered to the lipid bilayer by a farnesyl moiety (Cyr et al., 1992; Becker et al., 1996; Hoe et al., 1998; Costanzo et al., 2001; Beilharz et al., 2003; Walsh et al., 2004). Therefore to specifically examine the contributions of Ydj1 and Hlj1 on ApoB stability, we used strains in which the gene encoding Hlj1 was deleted (∆hlj1) and/or that contained a temperature-sensitive YDJ1 allele (ydj1-151) since—in some backgrounds—the Ydj1 protein is essential (Caplan and Douglas, 1991; Cyr and Douglas, 1994). Next, GEV-expressing yeast were isolated and the temperature sensitivity of cells containing the ydj1-151 allele was confirmed (unpublished data). The ApoB29 plasmid was then introduced into the yeast and after β-estradiol was added for 1 h, the strains were shifted to 37°C. As shown in Figure 1B, ApoB29 was only stabilized in strains containing the loss-of-function mutation in Ydj1 (i.e., the ydi1-151 and ∆hlj1 ydj1-151 strains). We also ensured that a known ERAD substrate was stabilized in the GEV-integrated strains. To this end, each strain was transformed with a plasmid engineered to express Chimera A*, an ERAD substrate whose degradation requires Ydj1 (Preston et al., 2018). Consistent with previous findings, Chimera A* degradation was slowed only in the ydj1-151 and ∆hlj1 ydj1-151 strains (Figure 1C).

Besides stimulating and enhancing the ATPase activity of Hsp70s, Hsp40s aid in substrate recognition (Fan et al., 2003; Craig and Marszalek, 2017). To explore if Ydj1 identifies ApoB29 for ERAD, we conducted coimmunoprecipitation experiments. After cell lysis and antibody-associated precipitation, the presence of ApoB29 and distinct chaperones and other proteins in the precipitate was examined by Western blotting. ApoB29 coimmunoprecipitated with Ydj1 and to a lesser degree Ssa1, and as a positive control we found that PDI was ApoB associated (Figure 1, D and E), as previously reported (Grubb et al., 2012). In contrast, G6PDH, a cytosolic enzyme, was absent after precipitation. As a negative control for this experiment, we also examined if Sis1, which thus far has not been associated with the ERAD pathway, also precipitates with ApoB. As anticipated, Sis1 was absent from the ApoB-containing complex, in contrast to Ydj1 as well as PDI (Figure 1E). These data suggest that Ydj1 interacts with and regulates the ERAD of ApoB in yeast.

The DNAJA1 molecular chaperone associates with and facilitates the degradation of ApoB100 in mammalian cells

Among the model systems used to study ApoB biology, McArdle cells, a rat hepatoma cell line, are ideal because they endogenously express and secrete lipidated ApoB100 that resembles VLDL (Meex et al., 2011). The cells also faithfully recapitulate ApoB degradation under lipid-poor conditions. Moreover, specific candidate genes can be readily knocked down or overexpressed and their effects on ApoB100 stability and transport examined in pulse-chase experiments (Bostrom et al., 1986; Borchardt and Davis, 1987; Boren et al., 1992, 1994).

We began by examining the effects of knocking down DNAJA1 (also known as Hdj2) since this is one of the closest mammalian homologues of Ydj1 (∼46% sequence identity; see Supplemental Figure S1). Like Ydj1, DNAJA1 is also a class A Hsp40 (Stark et al., 2014; Craig and Marszalek, 2017). To optimize knockdown conditions, we tested five ORF-targeted siRNA oligonucleotides at several concentrations and at two time points post-transfection. In addition, since DNAJA1 modulates the biogenesis of select substrates (Okiyoneda et al., 2011; Abisambra et al., 2012; Rauch and Gestwicki, 2014; Dekker et al., 2015; De Mattos et al., 2020), we confirmed that knockdown failed to elicit a general stress response by measuring Hsp70 levels (see Materials and Methods and Supplemental Figure S2A). Under the final optimized conditions, DNAJA1 was reduced to ∼20% of the levels present in the scrambled siRNA control (Figure 2A). This extent of knockdown is in line with other studies in which the effect of DNAJA1 was measured after silencing (see Meshalkina et al., 2016; Parrales et al., 2016). We then conducted pulse-chase experiments in the presence of the scrambled siRNA or after knocking down DNAJA1 (Figure 2, B and C). In the control, ApoB100 was present in cell lysates after 15 min, and the protein was secreted after 60 min, consistent with published data (Bostrom et al., 1986; Boren et al., 1992, 1994; White et al., 1992). This delay reflects the fact that ApoB100 requires ∼30 min to be synthesized, folded, packaged into COPII vesicles for transport to the Golgi, and secreted (Benoist and Grand-Perret, 1997).

FIGURE 2:

The DNAJA1 molecular chaperone facilitates ApoB100 degradation. (A) McArdle cells were transfected with the most efficacious DNAJA1 siRNA (KD) or a scrambled siRNA (control) for 48 hr. A representative Western blot shows DNAJA1, DNAJB1, and actin levels after knockdown. (B) A pulse-chase experiment was conducted as described in the Materials and Methods after cells were starved for 1 h and then incubated with radioactive Met/Cys for 15 min (pulse). ApoB100 was immunoprecipitated using an anti-ApoB antibody and analyzed via SDS–PAGE and phosphorimaging analysis. A representative radiograph and graph of the intracellular ApoB100 levels over time are shown. (C) A representative radiograph and graph show the levels of secreted ApoB100 over time. (D) DNAJA1 was either knocked down and/or cells were treated with MG-132 for 1 h before a pulse-chase analysis was conducted. A representative radiograph shows ApoB100 levels under each condition. (E) Chemical cross-linking was performed as described in the Materials and Methods after cells were treated with 2 mM DSP for 30 min. Input (1%) and immunoprecipitated fractions were processed for SDS–PAGE and Western blotting. Actin was used as a negative control. The slower migration of the cross-linked immunoprecipitated ApoB100 indicates incomplete reduction of the cross-linked complex in the immunoprecipitated fraction (denoted with an asterisk). In panels A-D, data represent the means of N = 5 independent experiments, ±SE; *p < 0.05, ns indicates p > 0.05. Statistical significance was calculated by Student’s unpaired t test.

Next, we measured ApoB100 levels and stability over time. When DNAJA1 was silenced, both the intracellular and especially the secreted ApoB100 levels rose significantly compared to the scrambled siRNA control at both the 60- and 90-min time points (Figure 2, B and C; see closed arrowheads and graphs). The magnitude of this effect was mirrored by the increase in ApoB100 achieved when the proteasome is inhibited (Liang et al., 2000; Cardozo et al., 2002; Fisher et al., 2008) (also see gray bars, Figure 2D). This result suggests that DNAJA1 helps deliver ApoB100 to the ERAD pathway. Consistent with this view, combined knockdown of DNAJA1 and treatment with MG132 failed to further increase ApoB100 levels (Figure 2D).

To ensure that DNAJA1 depletion did not affect the biogenesis of all proteins that traverse the secretory pathway, we also measured albumin secretion from McArdle cells treated under control or DNAJA1 knockdown conditions. First, we confirmed that the DNAJA1 siRNA again led to profound depletion of the chaperone, yet neither actin nor Hsp70 levels were altered (Figure 2A and Supplemental Figure S2A). Next, albumin was immunoprecipitated from a lysate fraction that contained equal radioactive counts 15 and 60 min into chase. As shown in Supplemental Figure S2B, the levels of secreted albumin were identical, regardless of whether DNAJA1 was abundant.

Because Ydj1 coimmunoprecipitates with ApoB29 in yeast (Figure 1, D and E), we asked if DNAJA1 also binds ApoB100 in McArdle cells. To this end, cells were treated with the membrane-permeable chemical crosslinker, dithiobis(succinimidyl propionate) (DSP) (Mattson et al., 1993), and after solubilization under denaturing conditions, ApoB100 was immunoprecipitated with an antibody directed against the protein along with Protein G–Sepharose. Protein partners were identified by Western blotting. As displayed in Figure 2E (IP +ab), DNAJA1 was cross-linked to ApoB100. In contrast, an interaction was absent between the apolipoprotein and actin.

The solubility of ApoB is unaffected when DNAJA1 is depleted

Based on the action of other Hsp40s (Kampinga et al., 2019) and the fact that DNAJA1 binds ApoB100, we asked if DNAJA1 might act as a “holdase” to maintain ApoB100 solubility as it retrotranslocates from the ER en route to the proteasome. A cellular holdase might be critical given the highly hydrophobic nature of the protein (Segrest et al., 1992). Therefore a series of detergent solubility assays using a nonionic detergent, dodecyl maltoside (DDM), which solubilizes membrane integrated proteins but not aggregated proteins (Calcutta et al., 2012), was conducted. Lysates prepared from cells treated with the control or an siRNA directed against DNAJA1 were incubated with a range of DDM concentrations on ice, the samples were centrifuged at 100,000 × g, the soluble (supernatant) and insoluble (pellet) fractions were processed, and the presence of ApoB100 and ribophorin, an integral ER membrane protein control (Crimaudo et al., 1987), was examined by Western blot analysis. Contrary to our hypothesis, depletion of DNAJA1 had no effect on ApoB100 solubility (Supplemental Figure S3A). Similar results were obtained when a range of other detergents were examined in control or DNAJA1-depleted cells (i.e., NP-40, sodium deoxycholate, Triton X-100) and under a variety of conditions (unpublished data). However, when the solubility of the cystic fibrosis transmembrane conductance regulator, CFTR, an established aggregation-prone protein (Strickland et al., 1997; Estabrooks and Brodsky, 2020), was measured, the protein resided primarily in the insoluble pellet (P) in detergent-treated lysates (Supplemental Figure S3B). These data suggest that DNAJA1 facilitates ERAD independent of its potential solubilizing effects.

DNAJA1 deficiency increases ApoB100 lipidation

The secreted ApoB100 pool in serum normally contains fully lipidated lipoprotein particles (Boren et al., 1994), and in cell culture a range of particles can be detected, including VLDL, LDL, and HDL (Meex et al., 2011). Because DNAJA1 silencing enhanced ApoB100 secretion, we asked if ApoB100 was lipid-depleted under these conditions since the secreted pool might otherwise have been targeted for ERAD, or if the rescued ApoB100 remained accessible to MTP and the post-ER lipidation machinery (Gusarova et al., 2003). To measure the extent of secreted ApoB100 lipidation, we followed an established protocol (Boren et al., 1992, 1994; Sparks et al., 1992; Wang et al., 1993). First, DNAJA1 was knocked down or cells were treated with a scrambled siRNA control, as above. This step was followed by a pulse-chase analysis, and media samples were collected 90 min into the chase. Normalized radioactive counts from the media of the control and experimental cells were then diluted and overlaid on a 12.5–50% sucrose gradient, and after centrifugation to equilibrium (∼65 h), 12 fractions were collected and ApoB100 was immunoprecipitated from each fraction. We also measured the refractive index of each fraction along with a series of controls that together confirmed gradient integrity and identified the predicted positions at which VLDL- and LDL-like particles migrate (Figure 3A). Consistent with the second model above, DNAJA1 depletion resulted in the production of a slightly greater percentage of the most buoyant, lipid-rich particles that were VLDL-like (≤ 1.006 g/ml) (Smith et al., 1978; Tiwari and Siddiqi, 2012) (Figure 3B). Based on this result, we propose that DNAJA1 depletion slows ERAD, increases the efficiency of cotranslational translocation, and thus augments MTP association and lipid loading (i.e., VLDL assembly; also see Discussion). Of note, the relatively modest effect observed in the VLDL pool most likely reflects the fact that a significant amount of ApoB100 is still targeted for degradation even after DNAJA1 is depleted (Figure 2D).

FIGURE 3:

Depletion of DNAJA1 increases ApoB100 lipidation and ApoB100 puncta formation. McArdle cells were treated with an siRNA against DNAJA1 (DNAJA1 KD) or a scrambled control (Control KD), pulse labeled with radioactive Met/Cys for 15 min, and after the media were changed, an excess of unlabeled amino acids was provided. The media were then collected after 90 min into the chase and subjected to sucrose gradient ultracentrifugation before the indicated fractions were collected. (A) Refractive indices were measured for each fraction. (B) ApoB100 was precipitated from each fraction and processed by SDS–PAGE and phosphorimage analysis. The graph represents the relative amount of ApoB100 residing in VLDL (0.95–1.006 g/ml), LDL (1.006–1.063 g/ml), and HDL (1.063–1.210 g/ml) in the precipitated fractions. Data represent the means of N = 3 independent experiments, ±SE. (C) Confocal IF images from McArdle cells probed with antibodies against ApoB (green) and the smooth ER resident protein, HMG-CoA reductase (red). DAPI staining (nucleus) is also shown. Statistical significance was calculated by Student’s unpaired t test.

Under steady-state conditions and in the absence of excess exogenous lipids, ApoB most commonly forms puncta that reside in both the ER and in the post-ER compartments (Fisher et al., 1997; Mitchell et al., 1998; Butkinaree et al., 2014). Consistent with enhanced ApoB100 flux through the secretory pathway (Figure 2), DNAJA1 deficiency led to an apparent increase in the number and sizes of ApoB100 puncta in McArdle cells. As anticipated, ApoB100 residence was also distinct from HMG-CoA reductase, a component of the smooth ER (Figure 3C).

DNAJB1, a class B Hsp40, also associates with but instead stabilizes ApoB100

Hsp70s can have multiple Hsp40 partners, and heterologous pairs of Hsp40 isoforms are common (Liu et al., 2020). Moreover, class A and class B JDPs function with Hsp70 and Hsp110 family members to act as components of protein disaggregases (Nillegoda et al., 2017; Nillegoda et al., 2018; Wentink et al., 2020). In yeast, an alternate disaggregase is Hsp104 (Parsell et al., 1994). Hsp104 functions in tandem with Hsp70-Hsp40 (Glover and Lindquist, 1998), and prior work from our lab demonstrated that Hsp104 associates with and facilitates the degradation of ApoB29 in the yeast model (Doonan et al., 2019). Although DNAJA1 depletion did not alter ApoB100 solubility (Supplemental Figure S3) while supporting ERAD, we still anticipated that a class B Hsp40 that associates with DNAJA1 would similarly facilitate ERAD. To this end, we investigated the contribution of DNAJB1 (also known as Hdj1) on ApoB100 biogenesis. Notably, DNAJA1 and DNAJB1 interact as components of cellular protein disaggregases (Nillegoda et al., 2015, 2017, 2018; Kirstein et al., 2017; Wentink et al., 2020).

As above, we first optimized DNAJB1 knockdown conditions in McArdle cells (see Materials and Methods and unpublished data). Ultimately, a knockdown efficiency of ∼80% was achieved (Figure 4A), and as observed when DNAJA1 was silenced (Figure 2A), the depletion of DNAJB1 had no effect on Hsp70 steady-state levels (unpublished data). The amounts of DNAJA1 and actin were also unchanged (Figure 4). Next, the stability of ApoB100 was examined in pulse-chase assays using cells treated with the scrambled siRNA control or the siRNA directed specifically against DNAJB1. In this case—and unexpectedly in contrast to the effect of DNAJA1 depletion—DNAJB1 knockdown failed to slow ApoB100 turnover. Instead, there was ∼2-fold less ApoB100 in cells at the 60-min time point (Figure 4B; please note that the error bars are obscured by the symbol). There was also a pronounced ∼4-fold decrease in the amount of secreted ApoB100 compared with the control at both the 60- and 90-min time points (Figure 4C), suggesting that higher levels of secretion were maintained after the initial buildup of cellular ApoB100. This result also indicates that the lower level of ApoB100 in cells (Figure 4B) cannot be explained by increased secretion when DNAJB1 was silenced. A direct effect of DNAJB1 on ApoB100 biogenesis was further supported by the fact that DNAJB1 could be cross-linked to ApoB (Figure 4D), yet like DNAJA1, the knockdown of DNAJB1 also failed to alter ApoB100 solubility (Supplemental Figure S4). Thus, in contrast to the role that DNAJA1 plays in enhancing degradation, DNAJB1 appears to instead facilitate ApoB100 folding/maturation in the ER (but see below).

FIGURE 4:

The DNAJB1 molecular chaperone augments ApoB100 biogenesis. (A) McArdle cells were transfected with the most efficacious DNAJB1 siRNA (KD) or a scrambled siRNA (Control) for 48 hr. A representative Western blot shows DNAJB1, DNAJA1, and actin levels after knockdown. (B) A pulse-chase experiment was conducted as described in the Materials and Methods and in Figure 2. A representative radiograph and graph of intracellular ApoB100 levels over time are shown. (C) A representative radiograph and graph show the levels of secreted ApoB100 over time. (D) Chemical cross-linking was performed as described in the Materials and Methods and in Figure 2E. Actin was used as a negative control. Data represent the means of N = 5 independent experiments, ±SE; *p < 0.05. Statistical significance was calculated by Student’s unpaired t test. The mean of the data ± SD was also significant for time point 60 min in B.

Neither a class C Hsp40 nor another class B Hsp40 associate with ApoB100

To control for these knockdown experiments, the potential effect of a JDP that resides in another cellular compartment—yet is linked to proteostasis—was measured. We selected a class C Hsp40, DNAJC19, which resides in the inner mitochondrial membrane. DNAJC19 functions during protein translocation into the mitochondria (Heinemeyer et al., 2019), so defects might lead to cytoplasmic stress responses. In fact, mutations in the gene encoding this chaperone result in cardiomyopathy with ataxia (Janz et al., 2020). Therefore we followed the same experimental methods outlined above and again optimized DNAJC19 knockdown conditions. However, as shown in Supplemental Figure S5A, DNAJC19 knockdown was nearly complete but had no effect on the intracellular or secreted levels of ApoB100 (Supplemental Figure S5, B and C). Chemical cross-linking with DSP also indicated that DNAJC19 fails to associate with ApoB100 (Supplemental Figure S5D). As an additional negative control, we examined whether another class B cytoplasmic Hsp40, DNAJB6, cross-links to ApoB100. Although this chaperone is known to be involved in modulating the formation of protein aggregates and amyloids (Thiruvalluvan et al., 2020), DNAJB6 also failed to associate with ApoB100, in contrast to DNAJA1 and Hsp70 (Supplemental Figure S6).

DNAJA1 and DNAJB1 differentially alter ApoB100 ubiquitination

In some cases, Hsp40s select ERAD substrates prior to substrate ubiquitination, so Hsp40 depletion can correlate with reduced ubiquitination (Nakatsukasa et al., 2008; Preston and Brodsky, 2017). Because DNAJA1 depletion slowed ApoB100 degradation (Figure 2), we initially suspected that the chaperone would similarly act prior to ApoB100 ubiquitination and, therefore, that DNAJA1 silencing would decrease ubiquitin conjugation. To test this hypothesis, ApoB100 ubiquitination was examined in lysates from cells transfected with a ubiquitin overexpression plasmid and incubated in the presence or absence of MG132. After lysis, ApoB100 was immunoprecipitated, and ubiquitinated ApoB100 levels were analyzed by Western blotting. In contrast to our hypothesis, ubiquitinated ApoB100 accumulated when DNAJA1 was knocked down (see DMSO samples, left, Figure 5A). This result suggests that DNAJA1 acts after substrate ubiquitination in the ERAD pathway, perhaps by facilitating ApoB100 delivery to the proteasome (also see Discussion). As a control for this experiment, there was no effect on ApoB100 ubiquitination when DNAJC19 was depleted (Figure 5, A and D). Moreover, MG132 increased the amount of ubiquitinated ApoB100 in both knockdown and control conditions (Figure 5, A and B), consistent with the targeting of ubiquitin-modified ApoB100 to the proteasome, as well as with the data in Figure 2D.

FIGURE 5:

The depletion of DNAJA1 and DNAJB1 leads to opposite effects on the levels of ubiquitinated ApoB100. (A) ApoB100 ubiquitination was assayed in McArdle cells that also transiently expressed HA-tagged ubiquitin and were treated in the presence of a control siRNA or an siRNA directed to DNAJA1, DNAJB1, or DNAJC19 as described in the Materials and Methods. Cells were also pretreated with DMSO or MG132 for 1 h before they were lysed and ApoB100 was immunoprecipitated. The levels of ubiquitinated ApoB100 (Ub-ApoB) were detected via SDS–PAGE and Western blotting against ApoB100 and HA (to detect the ubiquitinated protein). A representative Western blot shows the levels of ubiquitinated Apo100. The arrowhead denotes the migration of unmodified ApoB100 at ∼550 kDa. (B–D) Graphs show the levels of ubiquitinated ApoB100 normalized to the corresponding levels of ubiquitinated ApoB100 in the presence of DMSO and in cells that had been treated with the scrambled (control KD) siRNA. Data represent the means of N = 5 independent experiments, ±SE; *p < 0.05, ns indicates p > 0.05. Statistical significance was calculated by Student’s unpaired t test.

In contrast to the effect of DNAJA1 ablation, DNAJB1 knockdown decreased ApoB100 ubiquitination in the absence of MG132 (DMSO, left, Figure 5, A and C). Nevertheless, the addition of MG132 magnified the level of ApoB100 ubiquitination to an equal degree, regardless of whether DNAJB1 was silenced (Figure 5C). This result suggests that DNAJA1 and DNAJB1 act at different stages during ApoB100 biogenesis, in accordance with their distinct effects on substrate stability. In contrast to these results, prior work indicated that DNAJB1 knockdown had no effect on the ubiquitination and degradation of mutant p53 (Parrales et al., 2016), but in another study there was an increase in mitogen inducible gene 6 (MIG6) ubiquitination when DNAJB1 was silenced (Park et al., 2015). Overall, these and other data highlight the varied roles that Hsp40s can play during ERAD and, more generally, on the maintenance of cellular proteostasis (see below).

DISCUSSION

ApoB100 is an atypical ERAD substrate. This large (∼550 kDa) apolipoprotein contains pause-transfer sequences, is ubiquitinated cotranslationally, and is regulated by presecretory degradation via the ERAD pathway (Boren et al., 1992, 1994; Yeung et al., 1996; Benoist and Grand-Perret, 1997; Fisher et al., 1997; Sakata et al., 1999; Brodsky and Fisher, 2008). Even though cytosolic Hsp70 has been known for many years to regulate ApoB100 turnover (Fisher et al., 1997; Gusarova et al., 2001), cytosolic Hsp40s that modify the stability of ApoB have not been reported.

In this study, we identified and characterized the roles of two ER-associated Hsp40s that regulate ApoB100 biogenesis in McArdle cells. We began by conducting a candidate-based screen in a yeast ApoB29 expression system. Yeast express approximately a dozen Hsp40 homologues, but by focusing on those linked to ERAD we discovered that Ydj1, a Class A Hsp40, associates with and facilitates ApoB29 degradation (Figure 1). While it is noteworthy that the absence of another JDP (i.e., Hlj1) had no effect on ApoB stability, and that no interaction with yet another homolog was apparent (i.e., Sis1), we acknowledge that additional Hsp40s might act with or in opposition to Ydj1. Indeed, in one study it was shown that only 4/13 of the yeast Hsp40s exhibited “specialty” functions (Sahi and Craig, 2007). Next, since prior efforts established that data from the yeast model can be translated into mammalian cell systems (Hrizo et al., 2007; Grubb et al., 2012), we investigated the effects of DNAJA1, a mammalian Ydj1 homolog, on ApoB100 biogenesis in a rodent hepatic cell line that synthesizes and secretes ApoB100. Consistent with data in yeast, DNAJA1 similarly associated with and facilitated the ERAD of ApoB100. Because Hsp40s regulate ERAD either before or after ubiquitination (Brodsky, 2007; Preston and Brodsky, 2017), we also asked if ApoB100 was ubiquitinated to a greater or lesser extent on DNAJA1 depletion. Ubiquitinated ApoB100 accumulated when DNAJA1 was silenced (Figure 5), suggesting that the chaperone acts after the gp78 ubiquitin ligase (Liang et al., 2003a) has modified ApoB100. Because chaperones, including Hsp40s, play critical roles in maintaining protein solubility, we also investigated if DNAJA1 depletion alters ApoB solubility, but no change in detergent solubility was noted after chaperone levels fell. These results suggest that DNAJA1 is dispensable for ApoB100 disaggregation and plays another role during ApoB turnover, perhaps facilitating the transfer of ApoB100 to the proteasome. Indeed, prior studies have suggested that the proteasome can serve as a platform for chaperones that participate in the transfer of ubiquitinated proteins (Verma et al., 2000). It thus remains likely that there are other compensatory disaggregases—or even other Hsp40s—that facilitate ApoB100 solubility (also see above). This hypothesis will be examined in future studies.

The decision to either degrade or secrete ApoB100 is made at the ER/cytosol interface and is tightly regulated. This mechanism represents a tug-of-war between ER lumenal factors that favor lipoprotein assembly (e.g., Hsp110 and the MTP complex) and cytosolic components that target ApoB100 for degradation (e.g., Hsp70, Hsp90, Hsp104, p58^IPK) (Boren et al., 1992, 1994; Benoist and Grand-Perret, 1997; Fisher et al., 1997; Gusarova et al., 2001; Hrizo et al., 2007; Grubb et al., 2012; Sirwi and Hussain, 2018; Doonan et al., 2019;). Because ApoB100 achieves a bi-topic orientation with respect to the ER membrane while undergoing cotranslational translocation, the apolipoprotein remains in close proximity to DNAJA1, which is anchored to the ER membrane (Cyr et al., 1992; Becker et al., 1996; Hoe et al., 1998; Terada and Mori, 2000; Costanzo et al., 2001; Beilharz et al., 2003; Walsh et al., 2004; Xu et al., 2019). p97/VCP has also been shown to coimmunoprecipitate with ApoB100 and facilitate ApoB100 degradation (Fisher et al., 2008; Rutledge et al., 2009), suggesting that DNAJA1 assists p97/VCP during retrotranslocation after the apolipoprotein is ubiquitinated. Therefore DNAJA1 depletion might increase the extent of cotranslational translocation, folding, and lipidation by the MTP complex, which favors transit into the ER lumen. This would explain the fact that DNAJA1 knockdown increased ApoB100 buoyancy/lipidation.

DNAJA1 is member of an interclass JDP complex (Nillegoda et al., 2015, 2017, 2018; Kirstein et al., 2017; Wentink et al., 2020), so we anticipated that DNAJB1, a class B Hsp40, would also play a role in ApoB100 degradation. Contrary to our hypothesis, DNAJB1 knockdown accelerated ApoB100 turnover, suggesting that the chaperone is an ApoB100 stabilizer. As observed with DNAJA1, DNAJB1 also cross-linked to ApoB100, but in contrast to DNAJA1, DNAJB1 depletion blunted ApoB100 ubiquitination. While initially unexpected, this result is consistent with the functional diversity noted among Hsp40s and even with their different effects on a range of substrates. For example, DNAJA1 and DNAJB1 aid in CFTR folding, but DNAJB12 and DNAJC5 facilitate its degradation (Meacham et al., 1999; Farinha et al., 2002; Zhang et al., 2002; Schmidt et al., 2009;). In turn, DNAJA1 knockdown decreases the maturation efficiency of the hERG channel (Walker et al., 2010) and triggers ubiquitination and degradation of a mutant form of p53 (Parrales et al., 2016). Thus, a given Hsp40 might either help protect or destroy different substrates—or have no effect—and different Hsp40s acting on a specific substrate might alternatively favor either ERAD or protein biogenesis (Meacham et al., 1999; Farinha et al., 2002; Zhang et al., 2002; Schmidt et al., 2009; Walker et al., 2010; Park et al., 2015; Parrales et al., 2016). Related Hsp40s have also been found to bind in different ways to a given substrate (Irwin et al., 2021). Similar phenomena have been observed for other chaperones (Brodsky, 2007; Preston and Brodsky, 2017). To date, there is no way to predict whether a chaperone will contribute to secretory protein folding or degradation.

The Hsp40 J domain stimulates the ATPase activity Hsp70 and uses a C-terminal region to bind substrates (Fan et al., 2003; Craig and Marszalek, 2017). DNAJA1 is a class A Hsp40 that contains an N-terminal J domain of ∼70 residues, a glycine phenylalanine-rich region (GF), and a zinc fingerlike region (ZFLR) that is followed by two structurally similar beta-barrel C-terminal domains (CTDs), CTD1 and CTD2, along with a dimerization domain (Qian et al., 1996; Kityk et al., 2018; Jiang et al., 2019). The J domain forms a secondary structure harboring four alpha helices, and the highly conserved HPD tripeptide motif, situated between the second and third helices. This domain targets the Hsp70 nucleotide-binding domain, thereby enhancing its ATPase activity (Qian et al., 1996; Kityk et al., 2018; Jiang et al., 2019). The ZFLR instead binds and aids in substrate folding (Lu and Cyr, 1998; Jiang et al., 2019). The CTDs are responsible for substrate binding, yet recent studies emphasized the roles of other domains in substrate recognition (Kampinga and Craig, 2010; Yu et al., 2015a,b; Schilke et al., 2017). In contrast, class B Hsp40s lack the ZFLR and may present fewer binding interfaces for substrate association compared with class A Hsp40s. Moreover, a recent study reported that class B Hsp40s exhibit a unique autoinhibitory mechanism in which the Hsp70 binding sites are initially blocked by a short regulatory domain, but inhibition is released upon interaction with the Hsp70 C-terminal tail (Faust et al., 2020). Together, differences in sequence and domain architecture between class A and class B Hsp40s expand substrate selectivity and specialization for Hsp70-dependent functions (Rebeaud et al., 2021).

Since both DNAJA1 (this study) and Hsp70 (Gusarova et al., 2001) associate with ApoB100, we hypothesize that DNAJA1 binds ApoB100 through the CTD(s). The Hsp40 might then recruit Hsp70 and stimulate Hsp70 activity via the J domain, thus coupling ApoB100 binding to ATP hydrolysis. In this scenario, DNAJA1 requires both the J domain and the CTD to regulate ApoB100 biogenesis. However, because Hsp40s act as homodimers and form intermolecular complexes with Hsp40s from another class via J domain-CTD1 salt bridges (Nillegoda et al., 2015, 2017, 2018; Kirstein et al., 2017), additional interaction sites might be available for substrate binding. As a consequence of ApoB100’s size and multiple hydrophobic/lipid binding sites, it is also likely that several DNAJA1 molecules simultaneously associate with ApoB100. Further studies with DNAJA1 constructs harboring mutations in specific domains, along with studies involving truncated forms of ApoB100, will help determine which domains are necessary for chaperone-mediated regulation of ApoB100 during biogenesis in the ER.

Like DNAJA1, DNAJB1 is also an Hsp70 cochaperone (Michels et al., 1999; Blard et al., 2007; Gao et al., 2015). In addition, DNAJB1 functions with mammalian Hsp110 and Hsp70 to disaggregate and refold luciferase aggregates and resolubilize alpha synuclein as well as tau aggregates (Hageman and Kampinga, 2009; Rampelt et al., 2012; Kuo et al., 2013; Mattoo et al., 2013; Nillegoda et al., 2015; Deane and Brown, 2017; Osaki et al., 2018; den Brave et al., 2020). Based on its role in stabilizing ApoB100, DNAJB1 might represent a therapeutic target since, in principle, its depletion would result in lower levels of circulating atherogenic lipoproteins. However, due to its participation in disaggregating toxic proteins, severe secondary effects would occur if DNAJB1 function was compromised. Nevertheless, it will be important to confirm the role of DNAJB1 in lipoprotein and cholesterol metabolism in vivo.

In sum, as overabundance of the atherogenic ApoB-containing particles is a major risk factor for CAD, a better understanding of the mechanism underlying the regulation of ApoB biogenesis may help offset the catastrophic effects of this disease. Drugs are in development that target many of the factors that impact ApoB100 biogenesis (Koishi et al., 1992; Yokota et al., 2000; Soti et al., 2005; Assimon et al., 2013; McConnell and McAlpine, 2013; Shevtsov et al., 2019; Lu et al., 2020; Park et al., 2020), some of which may also affect chaperone function. Thus, our work provides a necessary step toward the continued development of improved or complementary methods to treat CAD.

MATERIALS AND METHODS

Yeast strains, plasmids, and growth conditions

Yeast strains were grown at 30°C (unless otherwise indicated) using standard growth, media, and transformation conditions (Adams et al., 1997). All the strains used in this study are listed in Supplemental Table S1. To drive expression of ApoB29, yeast were transformed with pSLW1-B29-HA (Hrizo et al., 2007; Grubb et al., 2012; Doonan et al., 2019), which contains a URA3 selection marker and encodes ApoB29 with a triple HA tag at the C-terminus. The ApoB29 isoform causes hypobetalipoproteinemia in humans but is the shortest isoform that can successfully traffic through the secretory pathway and is regulated by lipids (Collins et al., 1988; Linton et al., 1993; McLeod et al., 1994). The expression of ApoB29 is under the control of a galactose inducible promoter, but because galactose is a nonoptimal carbon source and causes additional stress in yeast (Adams, 1972; Balch et al., 2008), we used a  β-estradiol inducible expression system driven by pACT1-GEV chimeric transcription factor.

To make yeast strains β-estradiol inducible, EcoRV (New England Biolabs) linearized pACT1-GEV plasmid was transformed into log phase yeast cells and integrated at the leu2 ∆0 site using a standard lithium acetate protocol (Adams, 1972; Veatch et al., 2009; McIsaac et al., 2011; Doonan et al., 2019). Transformants were selected on yeast peptone dextrose (YPD) plates containing 0.1 mg/ml nourseothricin (Werner Bioagents, Jena, Germany). To ensure selection of colonies with successful integration of the plasmid, positive colonies were restruck three times on YPD nourseothricin plates.

Yeast cycloheximide chase assays

Cycloheximide chase assays were conducted following published protocols (Hrizo et al., 2007; Grubb et al., 2012; Doonan et al., 2019). In brief, to examine the rate of ApoB29 degradation, yeast transformed with pSLW1-B29-HA were grown overnight in synthetic media lacking uracil (SC-ura) supplemented with 2% glucose at 30°C and were then diluted with the same media and grown to logarithmic phase (OD₆₀₀ = 0.4–0.6). Cells were harvested to obtain 1.0 OD₆₀₀ equivalent of cells per time point and resuspended again in the same media but supplemented with 300 nM β-estradiol to induce expression of ApoB. All temperature-sensitive strains were preincubated at 37°C for 30 min. To inhibit protein translation, cycloheximide at a final concentration of 50 μg/ml was then added and 1.0 OD₆₀₀ of cells was harvested as the 0-min time point. Equivalent amounts of cells were collected for the indicated subsequent time points. Total protein was precipitated using trichloroacetic acid (TCA) following published protocols and resolved by SDS–PAGE and quantitative Western blotting (Zhang et al., 2001). The triple HA-tagged ApoB29 was detected using an anti-HA-horseradish peroxidase conjugated antibody, and G6PD levels were examined as a loading control with an anti-G6PD antibody (Supplemental Table S2).

Coimmunoprecipitation assays in yeast

To probe for specific protein–protein interactions, published protocols for ApoB29 in yeast were utilized (Hrizo et al., 2007; Grubb et al., 2012; Doonan et al., 2019). In brief, yeast containing pSLW1-B29-HA were grown overnight at 30°C, the cells were diluted to 0.25 OD₆₀₀/ml, grown to an OD₆₀₀ of 0.5, and 300 nM β-estradiol was added. Cell growth was continued for 2 h at 30°C. Once the OD₆₀₀ reached 1.0, 100 OD₆₀₀ of cells were harvested, which were then resuspended in 1 ml Roche lysis buffer (150 mM NaCl, 50 mM Tris-HCl, pH 7.5, 0.1% NP-40, 20 mM sodium molybdate, 1 mM phenylmethylsulfonyl fluoride [PMSF], 1 µg/ml leupeptin, 0.5 µg/ml pepstatin, and cOmplete EDTA-free protease inhibitor cocktail tablet [Millipore Sigma], hereafter referred to PIs + PIC for protease inhibitors and protease inhibitor cocktail).

Next, the cells were lysed using acid-washed mini glass beads by vigorous agitation on a Vortex mixer 3× for 2 min each with 2-min intervals on ice after each agitation. The lysate was collected and centrifuged for 2 min in a 4°C microcentrifuge at 5000 rpm, and the supernatant was transferred to a new microfuge tube and centrifugation was repeated. The cleared lysates were transferred to new tubes, 30 µl of a 1:1 slurry of Protein A Fast Flow–Sepharose beads (GE Healthcare) were added to each lysate, and tubes were placed on an end-to-end rotator for 1 h at 4°C. The tubes were centrifuged at 5000 rpm for 5 min at 4°C, the lysate was transferred to fresh tubes, and 5% (50 µl) of the initial volume was saved as input and processed by TCA precipitation as described earlier. The remaining lysate was diluted to 1.5 ml with Roche lysis buffer + PIs + PIC, 50 µl of anti HA affinity matrix (Roche) or an equal volume of Sepharose 6B beads (negative control) were added to the lysate, and the samples were placed on an end-to-end rotator at 4°C overnight. To isolate the bead-bound proteins, the lysates were centrifuged at 5000 rpm for 2 min at 4°C in a refrigerated microcentrifuge, and the supernatant (unbound fraction) was saved and stored at –80°C. The beads were then washed with 1 ml of ice-cold Roche lysis buffer + PIs+ PIC and again spun at 5000 rpm for 2 min at 4°C. This step was repeated twice, and two additional washes were conducted with the Roche lysis buffer containing PIs+ PIC but supplemented with 300 mM NaCl. After the final wash, any residual supernatant was removed, the beads were resuspended in 50 µl of TCA of sample buffer, and the solution was incubated at 37°C for 20 min and finally processed by SDS–PAGE and Western blotting. Proteins were detected using anti-HA, anti-ApoB, anti-Ydj1, anti-Ssa1, anti-PDI, anti-rpL5, anti-Sis1, and anti-G6PD antibody (see Supplemental Table S2).

Cell culture

McArdle RH-7777 cells (McArdle cells; ATCC CRL-1601) were purchased from American Type Culture Collection (ATCC) and grown at 37°C in a 5% CO₂ incubator in DMEM high glucose pyruvate (Thermo Scientific, Life Technologies) supplemented with 10% fetal bovine serum (FBS) (VWR), 10% horse serum (HS) (Sigma-Aldrich), and 0.1 mg/m penicillin/streptomycin (Thermo Fisher Scientific) on collagen-coated 10-ml dishes (Cellcoat Collagen Type I coated Petri dishes, Greiner). Cells were detached with 3 ml TrypLE Express Enzyme (No phenol Red, Life Technologies) for 3 min at 37°C. To quantify cell numbers, 40 µl of resuspended cells were mixed with 40 µl of trypan blue stain (Life Technologies) and an aliquot was analyzed on a hemocytometer. Cell culture experiments were performed in a 1300 Series A2 biosafety hood (Thermo Fisher Scientific).

Plasmids used in this study included HA-tagged ubiquitin (Kamitani et al., 1997) and pcDNA3.1-CFTR (Han et al., 2018). For transfection, 2 × 10⁵ cells were seeded/well in 6-well collagen-coated plates (Greiner Bio-One), grown overnight, and treated with Lipofectamine 2000 following the manufacturer’s instructions using Opti-MEM I reduced serum medium (Thermo Scientific Life Technologies) that was treated with 10 µl Lipofectamine 2000 and Opti-MEM I reduced serum medium (Thermo Scientific Life Technologies) with 4 µg of plasmid DNA. The solutions were then mixed and incubated for 20 min, and 500 µl of the mix were added to each well, followed by the addition of 2 ml of complete media. For transfection using 10-cm dishes, reagents were scaled up sixfold. Cells were harvested or lysed 20 h after transfection.

For knockdown studies, the siRNAs in Supplemental Table S3 were purchased from Dharmacon/Horizon Discovery. Initially, a set of four individual siRNAs (ON-TARGET plus Set of 4 Upgrade siRNA) was used to optimize knockdown conditions, followed by experiments using the most efficient siRNA. Lyophilized siRNAs were resuspended in nuclease free water to a final concentration of 20 µM. Knockdown efficiency was measured after 48 and 72 h and 20 and 30 nM of each siRNA were used. In all experiments, equal concentrations of fluorescent scrambled siRNA (BlockIt Alexa Fluor Red Fluorescent Control Oligo Invitrogen) were used.

Pulse-chase assays and measurements of ApoB100 ubiquitination

Pulse-chase assays were performed after modifications of published protocols (Bostrom et al., 1986; Borchardt and Davis, 1987; Boren et al., 1992, 1994; Fisher et al., 1997; Gusarova et al., 2001; Grubb et al., 2012). McArdle cells (1 × 10⁵) were seeded in 6-well collagen coated plates (see above) in 2 ml complete media and grown overnight. The cells were starved for 1 h in media lacking Cys/Met (DMEM) but containing L-Gln for 1 h at 37°C, washed with DPBS (Dulbecco phosphate-buffered saline), and then incubated in starvation media containing 100 mCi of S-35 (EasyTag Express 35-S Protein labeling mix; PerkinElmer) supplemented with 2% L-Gln. After 15 min, the cells were washed twice with DPBS and 1 ml of chase media (starvation media/2% Gln/5% FBS/5% HS/5%, 100 µM nonradioactive Cys, and 40 µM Met) was added to each well. Cells and media were collected at 15, 60, and 90 min into the chase. The media were collected into tubes with 25 µl of 1 mM PMSF. The cells were washed once with DPBS and lysed with 700 µl RadioIP assay buffer (RIPA) buffer (50 mM Tris-Cl, pH 7.4, 150 mM NaCl, 1% NP-40, 0.5% deoxycholate, 0.1% SDS, and PIs + PC) for 20–30 min on ice. The lysates were then centrifuged at 1600 × g for 20 min in a microcentrifuge at 4°C, and the supernatant was collected. Radioactive counts were determined for each lysate and media sample in a scintillation counter, and immunoprecipitation reactions were set up (for each cellular lysate/well) with 680 µl DPBS, 120 µl 5× NET buffer (150 mM NaC1, 5 mM EDTA, and 50 mM Tris, pH 7.4, 0.5% Triton X-100, and 0.1% SDS), 5 µl of anti-ApoB antibody, 50 µl of a 1:1 Protein G bead slurry, 600 µl of lysate, and PIs + PIC. For the media samples, each immunoprecipitation contained 360 µl of double distilled water, 240 µl of 5× NET buffer, 5 µl of anti-ApoB antibody or anti-albumin antibody, 50 µl of a 1:1 Protein G bead slurry, 800 µl of sample, and PIs + PIC. After an overnight incubation at 4°C, the tubes were centrifuged at ∼1600 × g for 1 min at 4°C in a microcentrifuge, and the beads were washed three times with 1× NET buffer with PIs + PIC. Next, the residual IP wash buffer was aspirated, and the beads were suspended in 75 µl urea sample buffer (8 M urea, 1% SDS, 10 mM EDTA, 150 mM Tris-Cl, pH 6.8). Proteins were resolved on 5% polyacrylamide SDS gels, and after the gels were dried and exposed to a phosphorimager screen, the radiograph was developed using a Typhoon FLA 7000 or Amersham Typhoon phosphorimager (GE Healthcare). Phosphorimages were collected at 50 µm resolution using a PMT voltage of 1000 V, and signals were quantified using ImageJ.

To measure the extent of protein ubiquitination, McArdle cells were transfected with an siRNA targeting the gene of interest, as described earlier, and on the next day cells were transfected with the HA-tagged ubiquitin expression plasmid (see above). After an overnight incubation, the media were aspirated, and fresh complete media with either the vehicle control (DMSO) or 25 mM of MG-132 (Selleck Chemicals) were added and the cells were incubated at 37°C for 1 h. Next, the media were aspirated and 500 µl of TrypLE were added. Following a 5-min incubation, the cells were harvested, and the cell pellets were lysed with 1 ml RIPA buffer + PIC supplemented with 25 µM MG-132 and 10 mM NEM for 20–30 min on ice. After centrifugation at ∼1600 × g for 20 min at 4°, the supernatant was collected and precleared with 30 µl of a 1:1 slurry of Protein G–Sepharose (GE Healthcare 17061801) beads for 1 h at 4°C. The cleared lysate was then collected after a ∼1000 × g spin for 5 min at 4°C, and protein levels were normalized using a BCA assay (ThermoScientific) following the manufacturer’s instructions. A fraction (20%) of the total lysate was saved for each reaction as the input, and equal amounts of protein plus 120 µl 1× NET buffer and 100 µl DPBS (Life Technologies) including PIC, 25 µM MG-132, and 50 µl of a 1:1 slurry of Protein G–Sepharose (GE Healthcare) and 3 µl anti-ApoB antibody (Calbiochem) were added to each immunoprecipitation. The tubes were incubated at 4°C overnight, centrifuged at ∼1600 × g for 1 min at 4°C, and washed three times with 1× NET including PIs + PIC, and the beads were suspended in urea sample buffer. Proteins were resolved on 5% polyacrylamide SDS gels, and ubiquitinated ApoB was detected with HA-HRP conjugated antibody. Another set of blots was probed with anti-ApoB antibody.

Other biochemical methods

To perform chemical cross-linking in McArdle cells, 1 × 10⁶ cells were seeded in 10-cm dishes in 10 ml and grown overnight, as above. The cells were washed twice with DPBS, and 2 mM DSP (ThermoFisher) was added at room temperature for 30 min. Next, stop solution (20 mM Tris-Cl, pH 7.5) was added and the cells were incubated for another 15 min before the cells were trypsinized and isolated as above. The cells were lysed in RIPA buffer, and immunoprecipitation reactions were set up as described for the pulse-chase protocol. To cleave DSP, 75 µl of urea sample buffer supplemented with 50 mM of freshly prepared DTT was added to each sample, which was then incubated at 37°C for 30 min.

Sucrose gradient analysis of ApoB-containing lipoproteins was conducted following published protocols (Boren et al., 1992, 1994) using cells that were seeded at 6 × 10⁵ cells/well and grown overnight in complete medium. The next day, siRNA transfections were performed, as above, and after 48 h, a pulse-chase was conducted as described above. However, media were collected only at the 90-min time point. Next, a gradient was set up with 2 ml of 50 and 25% (wt/vol) sucrose onto which the isolated media containing identical amounts of radioactivity diluted to 12.5% (wt/vol) sucrose concentration in a total volume of 5 ml was loaded. Each solution also contained 50 mM Na₃PO₄, pH 7.4, 150 mM NaCl, 5 mM EDTA, and PIs + PIC. Finally, 3 ml of media were layered on top. After centrifugation in a SW Ti41 rotor in a Beckman LM centrifuge at 150,000 × g for 65 h at 4°C, 12 1-ml fractions were carefully collected from the top, and radioactivity and the refractive index of each fraction were calculated using a scintillation counter and an Abbe refractometer, respectively. Next, an overnight immunoprecipitation with anti-ApoB antibody and Protein G beads as performed, as outlined above. Samples were subjected to SDS–PAGE, and the gels were dried and exposed to a phosphorimager screen, which was developed using a Typhoon FLA 7000 or Amersham Typhoon phosphorimager (GE Healthcare). Phosphorimages were collected at 50 µm resolution using PMT voltage of 1000 V, and the signal was quantified using ImageJ.

To assay ApoB100 solubility in various detergents, cells were grown overnight and siRNA-mediated DNAJA1 knockdown was conducted as described above. Scrambled control and DNAJA1-depleted cells were incubated with the indicated concentrations of dodecyl D-maltoside (DDM) (Sigma Aldrich) in 0.3 M sucrose, 0.1 M KCl, 2.5 mM MgCl₂,1 mM sodium-free EDTA, 10 mM PIPES, pH 6.8, and PIs + PIC on ice for 20 min. After centrifugation at 100,000 × g at 4°C for 1 h, the soluble (supernatant) and insoluble (pellet) fractions were processed for SDS–PAGE and developed by Western blot analysis with anti-ApoB antibody. The levels of DNAJA1 were also examined with DNAJA1 antibody, and ribophorin was used as a control (Crimaudo et al., 1987).

Indirect immunofluorescence (IF)

McArdle cells were grown on fibronectin-coated coverslips and transfected with scrambled siRNA (control) or an siRNA targeting DNAJA1, as described above. After 48 h, cells were fixed with 4% paraformaldehyde solution and then incubated with primary antibodies against ApoB (1:100; Santa Cruz Biotechnology 393636) and HMGCR (1:100; Biorbyt orb340767) at 4°C overnight. Cells were then washed with 1% bovine serum albumin solution and incubated with Alexa dye conjugated secondary antibodies for 1 h in the dark. Mounting solution containing DAPI was used to label the nucleus. Images were taken on a Nikon A1R confocal microscope with 60× objective and analyzed with Fuji software (National Institutes of Health ([NIH]).

Materials, quantification, and statistical analysis

The commercial sources and catalogue numbers for other purchased reagents are listed in Supplemental Table S4.

For quantitative Western blots, proteins were visualized using SuperSignal West Pico or the Femto Chemiluminescent kit (ThermoFisher) along with the respective Pro-Signal Pico or Femto developer solutions (Prometheus) using a Bio-Rad universal Hood II Imager. Image J software (NIH) were used to obtain and quantify images. Sample size comprised the stated number of independent biological replicates that individually also included three to four technical replicates. Statistical significance was calculated by Student’s unpaired t tests using the GraphPad statistical software.

For pulse-chase experiments, samples were normalized based on radioactive counts, and the signal after phosphorimager analysis (see above) was quantified using ImageJ (NIH). Radiographs were saved as TIFF files (8/16 bit) to collect raw integrated densities. Student’s unpaired two-tailed t tests were performed on GraphPad, average and standard errors were calculated, and significance was determined. Independent experiments were biological replicates which included the average of two to three technical replicates. For ubiquitination assays, data were normalized to the corresponding ApoB100 levels. Statistical significance was again determined by conducting unpaired t tests via the GraphPad statistical software.

Abbreviations used:

ApoB

Apolipoprotein B

ApoB100

full-length apolipoprotein B

ApoB48

N-terminal 48% of the apolipoprotein B polypeptide

CAD

coronary artery disease

CFTR

cystic fibrosis transmembrane conductance regulator

CTD

C-terminal domain

DDM

dodecyl maltoside

DPBS

Dulbecco phosphate-buffered saline

DSP

dithiobis(succinimidyl propionate)

ER

endoplasmic reticulum

ERAD

endoplasmic reticulum-associated degradation

FBS

fetal bovine serum

HDL

high-density lipoprotein

Hsp

heat shock protein

IF

indirect immunofluorescence

JDP

J domain protein

LDL

low-density lipoprotein

MTP

microsomal triglyceride transfer protein

PDI

protein disulfide isomerase

PMSF

phenylmethylsulfonyl fluoride

TCA

trichloroacetic acid

VLDL

very low-density lipoprotein

YPD

yeast peptone dextrose

ZFLR

zinc fingerlike region.