View metadata, citation and similar papers at core.ac.uk brought to you by CORE

provided by Elsevier - Publisher Connector Cell Metabolism Short Article

The -Risk SORT1 Facilitates PCSK9 Secretion

Camilla Gustafsen,1 Mads Kjolby,1,2,5 Mette Nyegaard,3 Manuel Mattheisen,4 Jesper Lundhede,1 Henriette Buttenschøn,3,6 Ole Mors,6,7 Jacob F. Bentzon,5 Peder Madsen,1 Anders Nykjaer,1,2 and Simon Glerup1,2,* 1The Lundbeck Foundation Research Center, MIND 2Danish Research Institute of Translational Neuroscience DANDRITE, Nordic EMBL Partnership 3The Lundbeck Foundation Initiative for Integrative Psychiatric Research, iPSYCH 4Center for Integrative Sequencing, iSEQ Department of Biomedicine, Aarhus University, Vennelyst Boulevard 4, DK-8000 C Aarhus, Denmark 5Department of Clinical Medicine, Aarhus University, and Department of Cardiology, Aarhus University Hospital, DK-8200 N Aarhus, Denmark 6Translational Neuropsychiatry Unit, Department of Clinical Medicine, Aarhus University, DK-8240 Risskov, Denmark 7Research Department P, Aarhus University Hospital, DK-8240 Risskov, Denmark *Correspondence: [email protected] http://dx.doi.org/10.1016/j.cmet.2013.12.006

SUMMARY tein convertase /kexin type 9 (PCSK9) (Awan et al., 2012), a that targets LDLR for degradation in lysosomes Circulating PCSK9 destines low-density either following their encounter in the biosynthetic pathway or receptor for degradation in lysosomes, resulting in by an extracellular mechanism, where secreted PCSK9 affects increased LDL . Accordingly, it is an hepatic LDLR levels (Costet et al., 2008; Lambert et al., 2012), attractive drug target for hypercholesterolemia, and thereby counteracting the beneficial effects of . PCSK9 results from clinical trials are promising. While the belongs to the superfamily of -like proprotein convertases physiological role of PCSK9 in cholesterol meta- (Seidah et al., 2003), but the effect on LDLR degradation seems to be independent of its proteolytic activity (McNutt et al., 2007). bolism is well described, its complex mechanism of However, the role of circulating PCSK9 in promoting hypercho- action remains poorly understood, although it is lesterolemia is beyond dispute and strongly supported by pre- known to depend on intracellular trafficking. We clinical experiments and clinical trials where directed here identify sortilin, encoded by the hypercholester- against the LDLR in PCSK9 effectively reduce LDL-C olemia-risk gene SORT1, as a high-affinity sorting re- levels (Roth et al., 2012; Stein et al., 2012). ceptor for PCSK9. Sortilin colocalizes with PCSK9 in are the main source of circulating PCSK9 the trans-Golgi network and facilitates its secretion (Lagace et al., 2006). The same cell type also expresses high from primary hepatocytes. Accordingly, sortilin-defi- levels of LDLR (Tavori et al., 2013), but the underlying mecha- cient mice display decreased levels of circulating nism for how PCSK9 is escorted through the biosynthetic PCSK9, while sortilin overexpression in the con- pathway and what decides whether PCSK9 here destines fers increased plasma PCSK9. Furthermore, circu- LDLR for degradation directly or continues to secretory vesicles is at present enigmatic. PCSK9 is a soluble protein unable lating PCSK9 and sortilin were positively correlated to interact on its own with adaptor coating the outer in a human cohort of healthy individuals, suggesting surface of subcellular compartments. Furthermore, the effect that sortilin is involved in PCSK9 secretion in hu- of PCSK9 on LDLR is independent of the LDLR intracellular mans. Taken together, our findings establish sortilin domain (Holla et al., 2010). Thus, the existence of PCSK9 sorting as a critical regulator of PCSK9 activity. receptors can be predicted. In support of this, the COPII adaptor protein Sec24A is part of an unknown sorting receptor complex, involved in the transfer of PCSK9 from the endoplasmic reticu- INTRODUCTION lum (ER) to the cis-Golgi (Chen et al., 2013). The sorting receptor sortilin belongs to the family of Vps10p- High levels of circulating low-density lipoprotein cholesterol domain receptors and has previously been linked to systemic (LDL-C) accelerate the progression of atherosclerosis, and the lipoprotein metabolism and risk of lowering of LDL-C is currently considered the most effective (Aulchenko et al., 2009; Kathiresan et al., 2008; Willnow et al., strategy in preventing coronary artery disease. Statins function 2011). The receptor family is characterized by the Vps10p by inhibiting HMG-CoA reductase, the rate-limiting in domain, consisting of a ten-bladed b-propeller that creates a cholesterol synthesis, thereby elevating LDL receptor (LDLR) tunnel cavity for the binding of soluble ligands. In addition, the re- expression to increase the uptake of LDL particles from the cir- ceptors contain a short cytoplasmic tail harboring recognition culation (Goldstein and Brown, 2009). However, some patients motifs for subcellular sorting adaptor proteins. The functional are insensitive to treatment or experience serious adverse role of sortilin in lipoprotein metabolism is unclear, and several effects. Furthermore, statins increase the expression of propro- mechanisms have been proposed (Kjolby et al., 2010; Musunuru

310 Cell Metabolism 19, 310–318, February 4, 2014 ª2014 Elsevier Inc. Cell Metabolism Sortilin Is a PCSK9 Sorting Receptor

et al., 2010; Willnow et al., 2011). Here, we identify sortilin as proPCSK9 IF was observed (Figure 1H). This was reflected by a high-affinity receptor for PCSK9 and find that it interacts calculation of Pearson’s correlation coefficient (PCC), indicating with PCSK9 in the trans-Golgi network (TGN) of HepG2 cells. an overall correlation between sortilin and PCSK9 IF of 0.16 ± PCSK9 displays altered subcellular localization, and its secretion 0.02—significantly higher than for sortilin/proPCSK9 IF (PCC = is hampered in sortilin knockout (KO) hepatocytes. Furthermore, 0.05 ± 0.02, p < 0.01). Mander’s colocalization coefficient overexpression or KO of sortilin increases or decreases plasma (MCC) of 63% ± 5.7% indicated that the majority of sortilin levels of PCSK9 in mice, respectively, suggesting that sortilin IF overlapped with that of PCSK9. Significantly less overlap facilitates the passage of PCSK9 through the late secretory between sortilin and mutant proPCSK9 IF was observed and pathway. evident from a lower MCC of 34% ± 2.5% (p < 0.01). We therefore compared the localization of sortilin with that of RESULTS subcellular markers. No colocalization was observed between sortilin and the ER marker calnexin (PCC = À0.03 ± 0.05) or Sortilin Is an Alternative PCSK9 Receptor the COPII vesicle marker Sec24A (PCC = À0.09 ± 0.03) (Fig- Similar to LDLR-related proteins, sortilin is a transmembrane ure 2A), rendering it unlikely that sortilin/PCSK9 complex forma- type I trafficking receptor with a putative role in lipoprotein meta- tion occurs in the early secretory pathway. Sortilin displayed bolism. Hence, we speculated if it might also be a PCSK9 target. some colocalization with the cis-Golgi marker GM130 (Fig- Using surface plasmon resonance (SPR), we found that PCSK9 ure 2A), as calculation of MCC revealed that 40% ± 6.8% bound to the immobilized extracellular domain of sortilin in a con- of GM130-positive structures contained sortilin, although their centration-dependent manner with a dissociation constant (Kd) overall IF did not correlate (PCC = À0.04 ± 0.03). Instead, we of 6 nM at pH 7.4 (Figure 1A). For comparison, the reported Kd observed clear colocalization with the trans-Golgi marker of PCSK9 for LDLR is in the range of 120–620 nM at this pH (Cun- TGN46 (Figure 2B), and the vast majority of TGN46-positive ningham et al., 2007; Fisher et al., 2007), indicating that PCSK9 structures were positive for sortilin (MCC = 82% ± 3.3%; shows considerably higher affinity for sortilin than for LDLR. PCC = 0.22 ± 0.03). Partial colocalization between sortilin and We further tested the pH dependency of the interaction and EEA1 was also observed (PCC = 0.14 ± 0.04), suggesting that found it to be highly pH sensitive, as the binding was of high sortilin could play a role in PCSK9 endocytosis, but triple staining affinity at physiological pH 7.4 and at slightly acidic pH 6.5, indicated that several EEA1-positive structures containing whereas no interaction was observed at acidic pH 5.5 (Fig- PCSK9 were devoid of sortilin (Figure 2C). These data suggest ure 1B). This stands in marked contrast to the interaction be- that the sortilin and PCSK9 interaction is mainly established tween PCSK9 and LDLR, which is strongly enhanced at acidic late in the secretory pathway, such as the TGN, but could also pH (Cunningham et al., 2007; Fisher et al., 2007). occur following endocytosis. PCSK9 directs the LDLR and related receptors ApoER2, To confirm this hypothesis, we transfected HEK293 cells with VLDLR (Poirier et al., 2008), and LRP1 (Canuel et al., 2013) PCSK9 cDNA together with sortilin or mock cDNA and metabol- for degradation in lysosomes. To assess if PCSK9 has this effect ically labeled newly synthesized proteins with S35 cysteine and on sortilin in -derived cells, we overexpressed methionine in the presence of brefeldin A, which inhibits protein PCSK9 in HepG2 cells, which led to a marked reduction of transport from the ER to the Golgi by preventing the formation of LDLR, while it had no effect on sortilin (Figures 1C and 1D). We COPII vesicles (Klausner et al., 1992). Following labeling, brefel- further incubated HepG2 cells with increasing concentrations din A was washed out, and PCSK9 was chased for the indicated of PCSK9. While PCSK9 potently reduced the level of LDLR, time periods (Figure 2D). In this experiment, the conversion of no change in sortilin levels was observed (Figures 1E and proPCSK9 did not depend on sortilin (Figure 2E). Importantly, 1F). The combined results show that sortilin is an alternative coIP of labeled sortilin was not observed until 4 hr chase (Fig- PCSK9 receptor, and that it, unlike LDLR and related receptors, ure 2F), further substantiating that the complex does not form is insensitive to PCSK9-induced degradation. in the ER or COPII vesicles. To test if there is indeed an interaction between PCSK9 and Sortilin is also synthesized as a proprotein encompassing an sortilin in cells, we performed coimmunoprecipitation (coIP) inhibitory propeptide that shields the Vps10p domain from ligand experiments. Mature PCSK9 was preferentially pulled down binding until its removal in the TGN (Munck Petersen et al., 1999). together with endogenous sortilin from cells transfected with a Importantly, PCSK9 binding was specifically inhibited by purified construct encoding wild-type (WT) PCSK9 (Figure 1G). However, sortilin propeptide (Figure 2G) but not by the peptide ligand the receptor could also pull down the proform of PCSK9 from neurotensin (Figure 2H), demonstrating that the sortilin/PCSK9 HepG2 cells overexpressing a propeptide cleavage-resistant interaction is not established prior to the TGN. mutant (proPCSK9), suggesting that sortilin has the potential to interact with both PCSK9 isoforms (Figure 1G). Sortilin Facilitates PCSK9 Secretion We speculated that sortilin might regulate the extracellular levels Sortilin and PCSK9 Interact Late in the Secretory of PCSK9, either by sorting in the TGN or by endocytosis, and we Pathway therefore generated primary hepatocytes from WT and sortilin To study where in cells sortilin may encounter PCSK9, we KO (Sort1À/À) mice and measured the accumulation of PCSK9 analyzed PCSK9-transfected HepG2 cells using confocal micro- in the media over a period of 18 hr (Figure 3A). Importantly, scopy. Interestingly, the bulk sortilin immunofluorescence (IF) PCSK9 secretion from KO hepatocytes was markedly attenu- appeared to colocalize with PCSK9 in the paranuclear region, ated compared to WT cells, showing that sortilin predominantly whereas significantly less overlap between sortilin and mutant exerts its effects on PCSK9 in the secretory pathway. This

Cell Metabolism 19, 310–318, February 4, 2014 ª2014 Elsevier Inc. 311 Cell Metabolism Sortilin Is a PCSK9 Sorting Receptor

ABC sortilin, Kd=6nM sortilin, pH +---mock PCSK9 -++-PCSK9 pH 6.5 300 1600 ---+proPCSK9 PCSK9 (nM) pH 7.4 proPCSK9 20 1200 200 PCSK9 800 LDLR 100 response units

response units 5 400 pH 5.5 sortilin 2 0 1 0 0.5 GAPDH 0 400 800 1200 0 400 800 1200 time/sec time/sec

LDLR LDLR D F sortilin sortilin 120

100 100 E 80 * PCSK9 (nM) 80 * 60 01020100 60 ** LDLR 40 40 ***

% of mock cells sortilin 20 *** 20 % of untreated cells 0 GAPDH 0 0 10 20 100 mock PCSK9 (nM) PCSK9 proPCSK9

H HepG2/PCSK9

G

lysate IP +++------mock ---+- ++- - PCSK9 sortilin PCSK9 Merged ----+ --++proPCSK9 HepG2/proPCSK9 sortilin

proPCSK9 PCSK9 IgG +- + - + - anti-sortilin -+ - + - + preimmune IgG sortilin PCSK9 Merged

Figure 1. Sortilin Is a High-Affinity PCSK9 Receptor (A) PCSK9 (0.5–20 nM) bound to the extracellular domain of sortilin with high affinity (Kd = 6 nM), as determined using SPR. (B) PCSK9 (20 nM) binding to sortilin is pH dependent, as binding was observed at pH 6.5–7.4 but lost at pH 5.5. (C–F) PCSK9 does not induce degradation of sortilin. LDLR and sortilin in HepG2 cells were analyzed upon transfection with WT PCSK9 or mutant proPCSK9 encoding plasmids (C and D) or by incubation for 24 hr with purified PCSK9, as indicated (E and F). Western blots (WB) for GAPDH are included as loading controls. Error bars indicate SEM; n = 3. Mutant proPCSK9 lead to a small reduction in LDLR, most likely caused by alternative processing. ***p < 0.001; **p < 0.01; *p < 0.05. (G) WB showing coIP of PCSK9 with endogenous sortilin from HepG2 cells transfected with PCSK9, mutant proPCSK9, or mock plasmid, as indicated. (H) HepG2 cells transfected with PCSK9 or proPCSK9 were stained for sortilin and PCSK9. The MCC of 63% ± 5.7% indicated that the majority of sortilin IF overlapped with PCSK9. The correlation between sortilin and PCSK9 IF was 0.16 ± 0.02, as estimated by PCC. Significantly less overlap between sortilin and mutant proPCSK9 IF was observed (MCC = 34% ± 2.5%, p < 0.01 and PCC = 0.05 ± 0.02, p < 0.01. Nuclei (blue) are stained using Hoechst. Scale bars are 10 mm.

312 Cell Metabolism 19, 310–318, February 4, 2014 ª2014 Elsevier Inc. Cell Metabolism Sortilin Is a PCSK9 Sorting Receptor

A

Calnexin sortilin Sec24A sortilin GM130 sortilin

B

sortilin TGN46 Merged

C

EEA1 sortilin PCSK9 EEA1 PCSK9 sortilin

D E 024818hours 100 100 PCSK9 PCSK9/sortilin proPCSK9 80 80 PCSK9 60 60 PCSK9 40 40 proPCSK9 PCSK9 (%) 20 PCSK9 20 proPCSK9 (%) 0 0 PCSK9/sortilin 051015 051015 hours hours

F GH sortilin, propeptide inhibition sortilin, neurotensin inhibition

propep + propep 6000 1200 PCSK9 024818hours 4000 propep + PCSK9 800 PCSK9 sortilin + neurotensin PCSK9 2000 buffer + PCSK9 400 sortilin neurotensin response units response units 0 0 PCSK9/sortilin 0 400 800 1200 1600 0 400 800 1200 time/sec time/sec

(legend on next page)

Cell Metabolism 19, 310–318, February 4, 2014 ª2014 Elsevier Inc. 313 Cell Metabolism Sortilin Is a PCSK9 Sorting Receptor

observation suggested altered subcellular localization of PCSK9 mixed age and gender, and compared it to their serum levels within hepatocytes of sortilin-deficient mice. Indeed, fraction- of PCSK9. Indeed, we found a significant positive correlation be- ation of intracellular structures according to their density showed tween sortilin and PCSK9 (r = 0.22; p < 0.05; n = 79) (Figure 4D), a significant change in the subcellular localization of both PCSK9 suggesting that sortilin expression affects circulating PCSK9 isoforms in KO compared to WT liver tissue (Figure 3C). This was levels in humans. For comparison, the correlation factor for evident from an increased concentration of proPCSK9 in frac- PCSK9 and LDL-C is r = 0.24 (Lakoski et al., 2009). tions of lower density (p < 0.01) and a reduction of mature PCSK9 in sortilin KO in higher-density fractions (p < 0.05) (Fig- DISCUSSION ure 3D). We observed no change in the localization of albumin, which is also secreted in a constitutive manner (Figure 3C). PCSK9 was discovered 10 years ago, and soon after, its role in We next measured PCSK9 in plasma from sortilin KO and WT LDL metabolism became clear. It is now well established that mice using ELISA and found that KOs had significantly lower PCSK9 destines LDLR for lysosomal degradation following their PCSK9 levels compared to WT (173 ± 15 ng/ml versus encounter in the secretory pathway (Poirier et al., 2009) or at the 242 ± 21 ng/ml; p < 0.01) (Figure 3B). This was accompanied cell surface (Lagace et al., 2006). It is less clear how this function by a 2-fold increase in LDLR protein levels in KO compared to is executed, as PCSK9 is a soluble protein without apparent pro- WT liver samples (Figures 3F and 3G). Strikingly, adenoviral- teolytic activity. Strikingly, the pH optimum for LDLR/PCSK9 mediated overexpression of human sortilin (Avsort) in the liver binding is pH 5.5, approximately the pH of secretory vesicles increased PCSK9 plasma concentration compared to control and late endosomes. For comparison, the pH in ER, Golgi, and virus (AvLacZ) (406 ± 41 ng/ml versus 254 ± 28 ng/ml; p < early endosomes is around 7.2, 6.5, and 6.3, respectively (Casey 0.01) (Figure 3E). The increase in circulating PCSK9 was accom- et al., 2010). In this pH range, PCSK9 shows only modest affinity panied by a 27% ± 8.4% (p < 0.05) reduction in hepatic LDLR for LDLR (Cunningham et al., 2007; Fisher et al., 2007). It is there- (Figures 3H and 3I). Together, these data suggest that sortilin fore tempting to speculate that the route of PCSK9 through the plays a critical role in escorting PCSK9 to constitutive secretory biosynthetic pathway or endosomal system depends on its inter- vesicles of hepatocytes with concomitant effect on LDLR in vivo. action with transmembrane sorting receptors other than LDLR. Many of the loss-of-function mutations in PCSK9 are in fact Circulating PCSK9 Correlates with Sortilin in Humans related to its defective passage through the secretory pathway We compared the expression of sortilin and LDLR in hepatic tis- (Nassoury et al., 2007). As such, the exit of PCSK9 from ER de- sues from human and mouse. Both species were found to have pends on the intracellular adaptor Sec24A (Chen et al., 2013), levels of sortilin corresponding to 25% of LDLR, as determined suggesting a regulatory point in the early secretory pathway. Sol- by qPCR analysis (Figure 4A). Sortilin is constitutively released uble proteins can undergo concentrative transport in the early from the cell surface following shedding by metalloproteinases biosynthetic pathway mediated by specialized groups of recep- (Hermey et al., 2006) and can therefore be detected in human tors releasing the cargo into the cis-Golgi compartment. Specific serum, serving as a proxy for receptor expression in tissues (Har- ER-residing sorting receptors enable soluble cargo to move ris et al., 2012). Once released from the cell surface, sortilin does faster through ER and Golgi, in contrast to the slower receptor- not influence PCSK9 activity, as we observed no changes in independent bulk flow pathway (Dancourt and Barlowe, 2010). extracellular PCSK9-induced downregulation of LDLR regard- At present, the identity of the PCSK9-specific ER sorting less of the presence of excess sortilin extracellular domain (Fig- receptor is unknown. However, our results suggest that sortilin ures 4B and 4C). Based on our mouse experiments, it is tempting provides an additional regulatory step of PCSK9 secretion in to speculate that altered sortilin levels might correlate with a the late secretory pathway, likely by facilitating the transport of change in circulating PCSK9 in human subjects. To this end, PCSK9 from the TGN to the plasma membrane. The TGN is we developed a sandwich ELISA for soluble sortilin using sorti- the main intracellular sorting hub, and soluble proteins may lin-specific antibodies for capture and detection (Figure 4C), here interact with transmembrane receptors, directing them to quantified sortilin in serum obtained from healthy individuals of their cellular destination (De Matteis and Luini, 2008). We find

Figure 2. PCSK9 and Sortilin Colocalize in the TGN (A) HepG2 cells stained for sortilin and subcellular markers. The PCCs were calculated to À0.03 ± 0.05, À0.09 ± 0.03, and À0.04 ± 0.03 for sortilin IF with calnexin, Sec24A, and GM130, respectively. A minor fraction of sortilin colocalized with GM130 (MCC = 3.6% ± 0.9%), while 40% ± 6.8% of GM130-positive compartments contained sortilin IF. Nuclei (blue) are stained using Hoechst. Scale bars are 10 mm. (B) HepG2 cells stained for sortilin and TGN46, showing clear overlap of sortilin IF with TGN46 (MCC = 28% ± 5.5%). Furthermore, 82% ± 3.3% of TGN46 IF was also positive for sortilin. The PCC for sortilin/TGN46 was 0.224 ± 0.028, significantly higher than that of sortilin/GM130 (p = 2.4 3 10À6). (C) Triple staining of HepG2/PCSK9 cells for EEA1, sortilin, and PCSK9; arrowheads indicate structures positive for EEA1 and PCSK9 staining, but with no signal for sortilin. The PCC for sortilin/EEA1 IF was 0.14 ± 0.04. (D–F) Metabolic labeling and IP of PCSK9 from transfected HEK293. This cell line expresses very low levels of endogenous sortilin. ProPCSK9 and PCSK9 in cell lysate at indicated time points were visualized by phosphorimaging (D), quantified by densitometry, and normalized with respect to the protein levels at time point 0 hr (errors bars indicate SEM) (E). CoIP of labeled sortilin with PCSK9 is observed in sortilin-transfected cells from time point 4 hr (F). (G) PCSK9 (20 nM) binding to sortilin is inhibited by the sortilin propeptide (propep) (10 mM). The SPR chip was first saturated by injecting two flow channels with propep (200–700 s) where after a new injection of propep was performed in the absence (flow channel 1, propep + propep) or presence of PCSK9 (flow channel 2, propep + PCSK9). Buffer alone was injected in the third flow channel at 200 s and PCSK9 alone at 700 s. The degree of inhibition was estimated by comparing the response difference between flow channels 1 and 2 to that of PCSK9 in flow channel 3. (H) PCSK9 (20 nM) binding to sortilin is not inhibited by coinjection with the neuropeptide ligand neurotensin (10 mM).

314 Cell Metabolism 19, 310–318, February 4, 2014 ª2014 Elsevier Inc. Cell Metabolism Sortilin Is a PCSK9 Sorting Receptor

A Sort1-/- WT 60 B

WT Sort1-/- 40 sortilin ***

20 PCSK9 PCSK9 in medium ** ** GAPDH 0 t=1 t=4 t=8 t=18 hours CD 1 2 3 4 5 6 7 fraction Sort1-/- proPCSK9 WT 1e+6 4e+5 PCSK9 p < 0.01 p < 0.05 8e+5 3e+5 albumin 6e+5 WT 2e+5 4e+5 proPCSK9 1e+5 2e+5 PCSK9 intensity proPCSK9 intensity PCSK9 0 0 123456 7 1234567 albumin fraction fraction

Sort1-/-

** E G 250 ** 400 200 F 300 150 WT Sort1-/- 100 200 ** LDLR liver LDLR (%) 100 50 ng PCSK9/ml plasma sortilin 0 0 -/- -/- WT WT Sort1 AvsortAvLacZ I Sort1

H 100

Avsort AvLacZ 80 *

LDLR 60 GAPDH 40 liver LDLR (%) LacZ 20 sortilin 0

Avsort AvLacZ

(legend on next page)

Cell Metabolism 19, 310–318, February 4, 2014 ª2014 Elsevier Inc. 315 Cell Metabolism Sortilin Is a PCSK9 Sorting Receptor

A other lipoprotein particles, while sortilin overexpression specif- ically increases LDL-C. Using a different sortilin KO model, 20 C another group reported severe hypercholesterolemia (Musunuru 15 B 100 et al., 2010), but this was later questioned in Strong et al. (2012), 80 0100 100 100 0 PCSK9 (nM) who stated that Sort1-deficient mice are not hypercholesterol- 10 00 0.5 1.5 1.5 sortilin (µM) 60 emic. Musunuru et al. (2010) also reported that massive sortilin 40 5 LDLR overexpression in the liver reduced LDL-C. Our present data

Receptor expression (%) 20 show that a more modest sortilin overexpression results in % of untreated cells 0 0 increased plasma PCSK9, in line with the findings in Kjolby Ldlr Sort1 LDLR SORT1 sortilin et al. (2010). However, it cannot be excluded that, at very high no addPCSK9PCSK9/ mouse human sortilin levels, sortilin may start to play a dominating role in the clearance D of LDL and even of PCSK9. 500 r = 0.22 The strongest interaction between sortilin and PCSK9 is at pH p < 0.05 400 6.5, while binding was completely lost at pH 5.5. These data fit with a role for sortilin in binding PCSK9 in the TGN (pH 6.5) 300 and releasing it in secretory vesicles (pH 5.5), thereby serving

200 to concentrate the PCSK9 flow into constitutive secretory vesi- PCSK9 ng/ml cles. As a consequence, the subcellular localization of PCSK9 is 100 altered in sortilin KOs, and its secretion is hampered in KO hepa-

0 tocytes. In a similar manner, sortilin has also been shown to assist 01020304050 sortilin ng/ml the secretion of interferon-g from T cells (Herda et al., 2012) and ApoB100 from hepatocytes (Kjolby et al., 2010). In this regard, it Figure 4. PCSK9 and Sortilin Levels Correlate in Humans is curious that PCSK9 and ApoB100 are found circulating together (A) Quantitative RT-PCR analysis of SORT1 and LDLR mRNA in human and of in a complex (Sun et al., 2012) and that a gain-of-function PCSK9 Sort1 and Ldlr mRNA in mouse liver tissue. Expression in percentage of the mutant dramatically increases ApoB100 secretion (Sun et al., b-actin gene (ACTB or Actb), with error bars indicating SD; n = 4. 2005). Thus, PCSK9 may be a contributing factor to the physical (B and C) Soluble sortilin does not influence PCSK9 activity. WB showing the link between sortilin and ApoB100. LDLR level in HepG2 cells incubated with 100 nM PCSK9 ± sortilin extracellular domain (B). Bar graph shows mean LDLR levels relative to untreated cells There is a strong link between PCSK9 and LDL metabolism (error bars indicate SEM; n = 3) (C). in humans, as exemplified in large population-based studies (D) Positive correlation between PCSK9 and sortilin in human serum, as showing a positive correlation between PCSK9, LDL-C, and measured by ELISA. Spearman correlation coefficient r = 0.22; p < 0.05; n = 79. increased risk of coronary artery disease (Awan et al., 2012; Mean levels were 229 ± 63 ng/ml and 23 ± 6.1 ng/ml for PCSK9 and sortilin, Cohen et al., 2006; Lakoski et al., 2009). Interestingly, we here respectively. also find a positive correlation between circulating PCSK9 and sortilin in a human cohort, indicating a link between sortilin levels that sortilin and PCSK9 colocalize in the TGN of HepG2 cells and and PCSK9 secretion in the general population. Hence, our that the presence of sortilin is important for secretion of PCSK9 findings show that sortilin is a regulator of PCSK9 secretion from primary hepatocytes. Strikingly, lack of sortilin or viral over- and could suggest that therapeutic targeting of this interaction expression has marked impact on circulating PCSK9 levels with may modulate circulating PCSK9 levels. concomitant effect on the levels of hepatic LDLR. This effect is most likely solely posttranslational, as altered sortilin expression EXPERIMENTAL PROCEDURES in mice has no effect on the transcription of the Pcsk9 or Ldlr Immunofluorescence (Kjolby et al., 2010). In analogy, the cardiovascular pheno- Cells were fixed in 4% PFA and permeabilized using 0.1% Triton X-100 before type of sortilin KO mice reported by Kjolby et al. (2010) is charac- incubation with primary antibodies and fluorescent secondary antibodies terized by marked reduction specifically in LDL-C and not in (Invitrogen). Nuclei were visualized with Hoechst dye (Sigma Aldrich). Images

Figure 3. Sortilin Facilitates PCSK9 Secretion (A) ELISA measurements of PCSK9 secreted into the media of primary hepatocytes derived from sortilin KO or WT mice. After 4 hr in culture, the medium was changed (t = 0 hr), and the fresh conditioned medium was subsequently harvested at the time points indicated. PCSK9 levels were normalized with respect to the PCSK9 level at t = 1 hr for each genotype (n = 4). (B) WB showing the expression of PCSK9, sortilin, and GAPDH in lysates of WT and KO hepatocytes. (C and D) Intracellular compartments of WT and KO liver homogenates (each a pool of three animals) were separated according to their density by sucrose gradient centrifugation, and fractions were analyzed by WB. Fraction numbers increase with increasing density. The fractionation was repeated twice with similar results. Bar graphs show densitometry quantification of proPCSK9 and PCSK9 in each fraction. Results were evaluated by two-way ANOVA. (E) PCSK9 levels measured by ELISA in plasma obtained from WT and Sort1À/À mice (n = 12 of each genotype) and from WT mice 14 days following the injection with adenovirus encoding sortilin (Avsort) (n = 4) or LacZ (AvLacZ) (n = 5). (F and G) WB of LDLR in liver homogenates from WT and Sort1À/À mice (F). Control WBs of sortilin and b-actin are shown. Bars (G) show quantification of LDLR band intensities normalized with respect to the LDLR level in WT mice (n = 12/genotype). (H and I) WB of LDLR in liver homogenates from WT mice with adenovirus-mediated overexpression of sortilin (H). Expression of sortilin, LacZ, and GAPDHin samples is shown as control. The bar graph (I) shows quantification of band intensities, normalized with respect to the LDLR levels in AvLacZ-injected animals. Error bars indicate SEM. ***p < 0.001, **p < 0.01, and *p < 0.05.

316 Cell Metabolism 19, 310–318, February 4, 2014 ª2014 Elsevier Inc. Cell Metabolism Sortilin Is a PCSK9 Sorting Receptor

were acquired on a Zeiss LSM780. Colocalization analysis was carried out Chen, X.W., Wang, H., Bajaj, K., Zhang, P., Meng, Z.X., Ma, D., Bai, Y., Liu, using Zen 2011, and at least 300 cells were examined for each com- H.H., Adams, E., Baines, A., et al. (2013). SEC24A deficiency lowers plasma bination in three independent experiments. cholesterol through reduced PCSK9 secretion. Elife 2, e00444. Cohen, J.C., Boerwinkle, E., Mosley, T.H., Jr., and Hobbs, H.H. (2006). Metabolic Labeling of Cellular Proteins Sequence variations in PCSK9, low LDL, and protection against coronary Newly synthesized proteins of human embryonic kidney (HEK293) cells trans- heart disease. N. Engl. J. Med. 354, 1264–1272. fected with PCSK9 cDNA together with sortilin or mock cDNA were metabol- Costet, P., Krempf, M., and Cariou, B. (2008). PCSK9 and LDL cholesterol: ically labeled with 200 mCi/ml S35-Cys/S35-Met (Pro-mix; GE Healthcare) in unravelling the target to design the bullet. Trends Biochem. Sci. 33, 426–434. Dulbecco’s modified Eagle’s medium (DMEM) without Cys/Met supplemented with 2% dialyzed FBS and brefeldin A (10 mg/ml; Pierce) for 4 hr at 37C. Cunningham, D., Danley, D.E., Geoghegan, K.F., Griffor, M.C., Hawkins, J.L., Following labeling, brefeldin A was washed out, and cells were incubated in Subashi, T.A., Varghese, A.H., Ammirati, M.J., Culp, J.S., Hoth, L.R., et al. DMEM with 2% FBS. At defined time points, cell lysates were harvested and (2007). Structural and biophysical studies of PCSK9 and its mutants linked subjected to IP using 0.5 mg anti-PSCK9. Precipitated proteins were visualized to familial hypercholesterolemia. Nat. Struct. Mol. Biol. 14, 413–419. by phosphorimaging after SDS-PAGE. Dancourt, J., and Barlowe, C. (2010). Protein sorting receptors in the early secretory pathway. Annu. Rev. Biochem. 79, 777–802. Sortilin and PCSK9 in Human Serum De Matteis, M.A., and Luini, A. (2008). Exiting the Golgi complex. Nat. Rev. Mol. Human serum samples were part of the Danish PRISME study and obtained Cell Biol. 9, 273–284. from employees within public service workplaces (Kaerlev et al., 2011). Sam- Fisher, T.S., Lo Surdo, P., Pandit, S., Mattu, M., Santoro, J.C., Wisniewski, D., ples from 79 subjects of mixed gender and age were analyzed. None of the Cummings, R.T., Calzetta, A., Cubbon, R.M., Fischer, P.A., et al. (2007). subjects had any reported use of cholesterol-lowering agents (e.g., statins). Effects of pH and low density lipoprotein (LDL) on PCSK9-dependent LDL PCSK9 and sortilin were determined using ELISA. GraphPad Prism version receptor regulation. J. Biol. Chem. 282, 20502–20512. 6.0 was used for nonparametric test for significant correlation. All experiments using human samples were performed according to national and institutional Goldstein, J.L., and Brown, M.S. (2009). The LDL receptor. Arterioscler. guidelines. Thromb. Vasc. Biol. 29, 431–438. Harris, L.W., Pietsch, S., Cheng, T.M., Schwarz, E., Guest, P.C., and Bahn, S. Statistics (2012). Comparison of peripheral and central schizophrenia biomarker Unless otherwise indicated, results were evaluated using Student’s t test, profiles. PLoS ONE 7, e46368. and error bars indicate SEM. Statistical significant p values are indicated by Herda, S., Raczkowski, F., Mittru¨ cker, H.W., Willimsky, G., Gerlach, K., Ku¨ hl, asterisks: *p < 0.05, **p < 0.01, and ***p < 0.001. A.A., Breiderhoff, T., Willnow, T.E., Do¨ rken, B., Ho¨ pken, U.E., and Rehm, A. (2012). The sorting receptor Sortilin exhibits a dual function in exocytic SUPPLEMENTAL INFORMATION trafficking of interferon-g and A in T cells. Immunity 37, 854–866. Hermey, G., Sjøgaard, S.S., Petersen, C.M., Nykjaer, A., and Gliemann, J. Supplemental Information includes Supplemental Experimental Procedures (2006). Tumour necrosis factor alpha-converting enzyme mediates ectodo- and can be found with this article online at http://dx.doi.org/10.1016/j.cmet. main shedding of Vps10p-domain receptor family members. Biochem. J. 2013.12.006. 395, 285–293. Holla, O.L., Strøm, T.B., Cameron, J., Berge, K.E., and Leren, T.P. (2010). A ACKNOWLEDGMENTS chimeric LDL receptor containing the cytoplasmic domain of the transferrin receptor is degraded by PCSK9. Mol. Genet. Metab. 99, 149–156. This study was funded by the Lundbeck Foundation (C.G., M.K., and A.N.), the Riisfort Foundation (C.G.), the Korning Foundation (C.G.), the Danish Heart As- Kaerlev, L., Kolstad, H.A., Hansen, A.M., Thomsen, J.F., Kærgaard, A., sociation (M.K.), the Hartmann Foundation (M.K.), the Danielsens Foundation Rugulies, R., Mikkelsen, S., Andersen, J.H., Mors, O., Grynderup, M.B., and (M.K.), the Kirsten Anthonius’ Mindelegat (S.G.), and the Danish Medical Bonde, J.P. (2011). Are risk estimates biased in follow-up studies of psychoso- Research Council (S.G.). PRISME-study group, Denmark is acknowledged cial factors with low base-line participation? BMC Public Health 11, 539. for serum samples. Anja Aagaard Pedersen is thanked for excellent technical Kathiresan, S., Melander, O., Guiducci, C., Surti, A., Burtt, N.P., Rieder, M.J., assistance. Cooper, G.M., Roos, C., Voight, B.F., Havulinna, A.S., et al. (2008). Six new loci associated with blood low-density lipoprotein cholesterol, high-density Received: August 9, 2013 lipoprotein cholesterol or triglycerides in humans. Nat. Genet. 40, 189–197. Revised: November 2, 2013 Kjolby, M., Andersen, O.M., Breiderhoff, T., Fjorback, A.W., Pedersen, K.M., Accepted: December 12, 2013 Madsen, P., Jansen, P., Heeren, J., Willnow, T.E., and Nykjaer, A. (2010). Published: January 28, 2014 Sort1, encoded by the cardiovascular risk 1p13.3, is a regulator of hepatic lipoprotein export. Cell Metab. 12, 213–223. REFERENCES Klausner, R.D., Donaldson, J.G., and Lippincott-Schwartz, J. (1992). Brefeldin A: insights into the control of membrane traffic and organelle structure. J. Cell Aulchenko, Y.S., Ripatti, S., Lindqvist, I., Boomsma, D., Heid, I.M., Pramstaller, Biol. 116, 1071–1080. P.P., Penninx, B.W., Janssens, A.C., Wilson, J.F., Spector, T., et al.; ENGAGE Consortium (2009). Loci influencing lipid levels and coronary heart disease risk Lagace, T.A., Curtis, D.E., Garuti, R., McNutt, M.C., Park, S.W., Prather, H.B., in 16 european population cohorts. Nat. Genet. 41, 47–55. Anderson, N.N., Ho, Y.K., Hammer, R.E., and Horton, J.D. (2006). Secreted PCSK9 decreases the number of LDL receptors in hepatocytes and in Awan, Z., Seidah, N.G., MacFadyen, J.G., Benjannet, S., Chasman, D.I., of parabiotic mice. J. Clin. Invest. 116, 2995–3005. Ridker, P.M., and Genest, J. (2012). , sub- tilisin/kexin type 9 concentrations, and LDL cholesterol response: the JUPITER Lakoski, S.G., Lagace, T.A., Cohen, J.C., Horton, J.D., and Hobbs, H.H. trial. Clin. Chem. 58, 183–189. (2009). Genetic and metabolic determinants of plasma PCSK9 levels. J. Clin. Canuel, M., Sun, X., Asselin, M.C., Paramithiotis, E., Prat, A., and Seidah, N.G. Endocrinol. Metab. 94, 2537–2543. (2013). Proprotein convertase subtilisin/kexin type 9 (PCSK9) can mediate Lambert, G., Sjouke, B., Choque, B., Kastelein, J.J., and Hovingh, G.K. (2012). degradation of the low density lipoprotein receptor-related protein 1 (LRP-1). The PCSK9 decade. J. Lipid Res. 53, 2515–2524. PLoS ONE 8, e64145. McNutt, M.C., Lagace, T.A., and Horton, J.D. (2007). Catalytic activity is not Casey, J.R., Grinstein, S., and Orlowski, J. (2010). Sensors and regulators of required for secreted PCSK9 to reduce low density lipoprotein receptors in intracellular pH. Nat. Rev. Mol. Cell Biol. 11, 50–61. HepG2 cells. J. Biol. Chem. 282, 20799–20803.

Cell Metabolism 19, 310–318, February 4, 2014 ª2014 Elsevier Inc. 317 Cell Metabolism Sortilin Is a PCSK9 Sorting Receptor

Munck Petersen, C., Nielsen, M.S., Jacobsen, C., Tauris, J., Jacobsen, L., Seidah, N.G., Benjannet, S., Wickham, L., Marcinkiewicz, J., Jasmin, S.B., Gliemann, J., Moestrup, S.K., and Madsen, P. (1999). Propeptide cleavage Stifani, S., Basak, A., Prat, A., and Chretien, M. (2003). The secretory propro- conditions sortilin/neurotensin receptor-3 for ligand binding. EMBO J. 18, tein convertase neural -regulated convertase 1 (NARC-1): liver 595–604. regeneration and neuronal differentiation. Proc. Natl. Acad. Sci. USA 100, 928–933. Musunuru, K., Strong, A., Frank-Kamenetsky, M., Lee, N.E., Ahfeldt, T., Sachs, K.V., Li, X., Li, H., Kuperwasser, N., Ruda, V.M., et al. (2010). From noncoding Stein, E.A., Mellis, S., Yancopoulos, G.D., Stahl, N., Logan, D., Smith, W.B., variant to phenotype via SORT1 at the 1p13 cholesterol locus. Nature 466, Lisbon, E., Gutierrez, M., Webb, C., Wu, R., et al. (2012). Effect of a monoclonal 714–719. antibody to PCSK9 on LDL cholesterol. N. Engl. J. Med. 366, 1108–1118. Strong, A., Ding, Q., Edmondson, A.C., Millar, J.S., Sachs, K.V., Li, X., Nassoury, N., Blasiole, D.A., Tebon Oler, A., Benjannet, S., Hamelin, J., Kumaravel, A., Wang, M.Y., Ai, D., Guo, L., et al. (2012). Hepatic sortilin regu- Poupon, V., McPherson, P.S., Attie, A.D., Prat, A., and Seidah, N.G. (2007). lates both secretion and LDL catabolism. J. Clin. Invest. 122, The cellular trafficking of the secretory proprotein convertase PCSK9 and its 2807–2816. dependence on the LDLR. Traffic 8, 718–732. Sun, X.M., Eden, E.R., Tosi, I., Neuwirth, C.K., Wile, D., Naoumova, R.P., and Poirier, S., Mayer, G., Benjannet, S., Bergeron, E., Marcinkiewicz, J., Soutar, A.K. (2005). Evidence for effect of mutant PCSK9 on apolipoprotein B Nassoury, N., Mayer, H., Nimpf, J., Prat, A., and Seidah, N.G. (2008). The secretion as the cause of unusually severe dominant hypercholesterolaemia. proprotein convertase PCSK9 induces the degradation of low density lipopro- Hum. Mol. Genet. 14, 1161–1169. tein receptor (LDLR) and its closest family members VLDLR and ApoER2. Sun, H., Samarghandi, A., Zhang, N., Yao, Z., Xiong, M., and Teng, B.B. (2012). J. Biol. Chem. 283, 2363–2372. Proprotein convertase subtilisin/kexin type 9 interacts with apolipoprotein B Poirier, S., Mayer, G., Poupon, V., McPherson, P.S., Desjardins, R., Ly, K., and prevents its intracellular degradation, irrespective of the low-density lipo- Asselin, M.C., Day, R., Duclos, F.J., Witmer, M., et al. (2009). Dissection of protein receptor. Arterioscler. Thromb. Vasc. Biol. 32, 1585–1595. the endogenous cellular pathways of PCSK9-induced low density lipoprotein Tavori, H., Fan, D., Blakemore, J.L., Yancey, P.G., Ding, L., Linton, M.F., and receptor degradation: evidence for an intracellular route. J. Biol. Chem. 284, Fazio, S. (2013). Serum proprotein convertase subtilisin/kexin type 9 and cell 28856–28864. surface low-density lipoprotein receptor: evidence for a reciprocal regulation. Roth, E.M., McKenney, J.M., Hanotin, C., Asset, G., and Stein, E.A. (2012). Circulation 127, 2403–2413. with or without an antibody to PCSK9 in primary hypercholester- Willnow, T.E., Kjølby, M., and Nykjaer, A. (2011). Sortilins: new players in lipo- olemia. N. Engl. J. Med. 367, 1891–1900. protein metabolism. Curr. Opin. Lipidol. 22, 79–85.

318 Cell Metabolism 19, 310–318, February 4, 2014 ª2014 Elsevier Inc.