Previous studies have shown that constitutive knockout of Bloc1s1 in mice produces embryonic lethality (Scott et al., 2014; Zhang et al., 2014). To study the function of BLOS1 in liver lipid metabolism, we generated a conditional knockout mouse mutant (cKO mice) by crossing Bloc1s1 floxed mice (loxp mice) with Albumin (Alb)-cre mice, while littermates lacking cre gene served as control group (Figure 1—figure supplement 1a). Deletion of Bloc1s1 was confirmed in hepatocytes isolated from cKO mouse liver by genomic PCR (Figure 1—figure supplement 1b). Because antibody to BLOS1 was not applicable, we used two other BLOC-1 subunits (Pallidin and Dysbindin) to monitor the loss of BLOS1 as the depletion of BLOS1 leads to the destabilization of other BLOC-1 subunits (Zhang et al., 2014). Indeed, the content of both subunits was significantly reduced in cKO mouse livers and hepatocytes, suggesting the depletion of BLOS1 (Figure 1a,b).

Figure 1 with 1 supplement see all

Download asset Open asset

Figure 1—source data 1

Full image of CBB staining showing the comparison of control and cKO mice plasma proteins.

Representative figure of CBB staining result of control and cKO mouse plasma proteins separated by SDS-PAGE. Arrow indicates the protein band which was further verified as apoE.

https://cdn.elifesciences.org/articles/58069/elife-58069-fig1-data1-v2.pptx

Download elife-58069-fig1-data1-v2.pptx

To investigate whether BLOS1 deficiency could affect liver lipid droplet content, we performed Oil Red O staining on frozen sections of mouse livers and found that cKO mice had fewer lipid droplets in the liver when fed on chow diet, and the accumulation of lipid droplets after starvation was largely inhibited in cKO mouse liver (Figure 1c–e). Considering that many important constituents of plasma are secreted by liver, we examined plasma samples of cKO mice and their control littermates by non-reduced SDS-PAGE analysis. Although there were no changes in several abundant plasma proteins, such as albumin and transferrin, a protein band at about 34 kD was consistently increased in plasma samples from cKO mice (Figure 1f). Mass spectrometry identified this protein as apolipoprotein E (ApoE) (Figure 1—figure supplement 1c), and immunoblotting confirmed the increase of ApoE in cKO mice plasma (Figure 1g).

ApoE is a core protein component of very-low-density lipoprotein (VLDL) and high-density lipoprotein (HDL). For this reason, lipoproteins in plasma samples from cKO and loxp mice were pre-stained with Sudan Black B and then separated by gradient native polyacrylamide gel electrophoresis. Using lipoproteins purified from pooled mouse plasma by sequential ultracentrifugation, we determined different lipoprotein bands in native gels (Figure 1—figure supplement 1d, left). We observed that in cKO mouse plasma, VLDL level was decreased, LDL was increased, and HDL had a tendency to increase (Figure 1h). Native gels subjected to Oil Red O staining also showed similar results (Figure 1—figure supplement 1d, right). The reduction of VLDL in cKO plasma apparently could not account for the increase of ApoE content. We wondered whether ApoE is increased in the HDL fraction of lipoproteins. To address this point, we separated unstained lipoproteins from cKO and loxp mice using native gels, and then performed a second dimensional SDS-PAGE and immunoblotting to the gel slices of target lanes containing separated lipoproteins. Indeed, we observed that there was more ApoE in the HDL slice of cKO mice (Figure 1i).

Collectively, these results suggest that cKO mice have abnormal lipid metabolism, mainly characterized by reduced lipid droplets in the liver, and altered plasma lipoprotein compositions with increase of LDL, reduction of VLDL, and increase of ApoE in HDL.

Despite the reduced VLDL content, a common constituent of LDL and VLDL, ApoB, was significantly increased in cKO mouse plasma (Figure 2—figure supplement 1a), suggesting that increase of LDL in cKO mouse plasma was likely the main effect of BLOS1 deficiency on lipoproteins. VLDL is largely converted to LDL in plasma. The reduced VLDL in cKO mouse plasma is unlikely the major cause of increased LDL. Liver is the primary site for plasma LDL clearance, the accumulation of LDL in cKO mouse plasma suggested that LDL clearance by cKO mouse liver could be impaired. To test this hypothesis, we examined the endocytosis of LDL in primary hepatocytes using purified mouse LDL labeled with DiI (LDL-DiI). We observed that LDL endocytosis in cKO hepatocytes was reduced at the initial stage (Figure 2a,b), suggesting that less LDL binds to the cell membrane of cKO hepatocytes. We then detected the LDLR protein level by immunoblotting. We found a significant reduction of LDLR in cKO mouse liver (Figure 2c,f and Figure 2—figure supplement 1b). Immunofluorescence staining of LDLR in primary hepatocytes isolated from both control and cKO mice also showed significant reduction of LDLR in cKO mice (Figure 2—figure supplement 1c).

Figure 2 with 1 supplement see all

Download asset Open asset

The reduction of LDLR in cKO mouse liver was not attributable to the dysfunction of either the BLOC-1 or BORC complexes, as both the pa mice (lack of Pallidin subunit in BLOC-1) and Kxd1-KO mice (lack of KXD1 subunit in BORC) showed no obvious changes in liver LDLR levels (Figure 2d–f). This suggests that the effect on LDLR is specific for BLOS1 alone. Considering that the uptake of LDL by other tissues (except the liver) is unaffected in cKO mice, the increase of plasma LDL is mainly attributable to impaired liver clearance of LDL.

The reduction of LDLR caused by BLOS1 deficiency was further confirmed in Hep G2 cells with stable knockdown of BLOC1S1 (BLOC1S1-KD cells) (Figure 2g). The potential down-regulation effect of BLOS1 deficiency on LDLR transcriptional levels was excluded by qRT-PCR in two different sets of primers (Figure 2h) (actually, the mRNA level of LDLR was slightly increased in BLOC1S1-KD cells), suggesting a post-transcriptional effect on LDLR. In addition, reduction of LDLR level (as well as TfR level) in BLOC1S1-KD cells was recovered after the inhibition of lysosomal degradation by leupeptin (Figure 2i), suggesting excessive degradation of LDLR or TfR due to the loss of BLOS1. We also noted that the content of PCSK9, a known negative regulator of LDLR degradation, was not affected in cKO mouse livers (Figure 2—figure supplement 1b).

We next investigated whether BLOS1 could physically interact with LDLR. LDLR co-immunoprecipitated with FLAG-tagged BLOS1 (Figure 2—figure supplement 1d) and GST-BLOS1 fusion protein could pull-down LDLR from liver tissue lysate (Figure 2—figure supplement 1e), indicating an interaction between BLOS1 and LDLR. We further constructed different truncations of BLOS1, and then performed the GST pull-down experiments using mouse liver lysates. The interacting region on BLOS1 was narrowed down to residues 76–100 (Figure 2—figure supplement 1f–h and j). Furthermore, the cytosolic domain of LDLR at its C-terminus could pull down BLOS1 (Figure 2—figure supplement 1i,j).

Taken together, these results showed that loss of BLOS1 in liver could lead to impaired LDL clearance from plasma which is caused by excessive lysosomal degradation of LDLR, and residues 76–100 on BLOS1 interacts with the cytosolic domain of LDLR at the C-terminus.

To further explore the underlying mechanism of how BLOS1 regulates LDLR level, we first performed immunocytochemistry (ICC) to observe the subcellular distribution of BLOS1. In Hep G2 cells, overexpressed BLOS1 with different tags all showed puncta distribution patterns in the cytosol (Figure 3a). When fused to a large tag such as GFP, BLOS1-GFP mostly formed aggregates (Figure 3a, left). However, when fused with small tags such as Myc, FLAG or HA, BLOS1 localized to smaller puncta structures distributed more evenly in the cytosol (Figure 3a, right). We used small tagged BLOS1 in the following immunostaining assays in Hep G2 cells. We co-stained BLOS1 overexpressing cells with different organelle markers and found that BLOS1 partially colocalized with lysosome/MVB marker CD63 (Figure 3c), and almost no colocalization of BLOS1 with mitochondria (labeled by Cytochrome C) was observed (Figure 3b). In addition, a small fraction of BLOS1 colocalized with the early endosome marker EEA1 (Figure 3d), and more BLOS1 puncta colocalized with TfR labeled recycling endosomes (REs) (Figure 3e) or LDLR-positive vesicles (Figure 3f, see also Figure 3g for quantification), indicating a potential role of BLOS1 in the endocytic and recycling pathway.

Figure 3

Download asset Open asset

Interestingly, when overexpressed in mouse primary hepatocytes, BLOS1 showed a tubular distribution in addition to the scattered puncta (Figure 3h). This observation promoted us to examine the relationship between these tubular structure and cytoskeletons. We found that these tubular structures colocalized well with α-tubulin labeled microtubules (Figure 3i, bottom) but not with actin filaments (labeled with phalloidin) (Figure 3i, top). The microtubule-localized pattern of BLOS1 in primary hepatocytes suggests that BLOS1 may be associated with microtubules.

Since BLOS1 has no reported or predicted microtubule-binding motif, we then explored whether other microtubule binding proteins were involved in the microtubular distribution of BLOS1 in primary hepatocytes. Three motor proteins (KIF3B, KIF13A, and KIF16B) from the kinesin superfamily, which have been reported to function in the anterograde transport of vesicles in endocytic system, were taken as candidates. Among these three KIF proteins, KIF3B belongs to the kinesin-2 family, while KIF13A and KIF16B are classified in the kinesin-3 family. The co-immunoprecipitation (co-IP) assays revealed that BLOS1 interacted with both KIF3B and KIF13A, but not KIF16B (Figure 3j). In agreement with these results, we found that the tubular structures that BLOS1 labeled were KIF13A-positive (Figure 3k). Together, these results suggest that BLOS1 may play a role in endosomal trafficking through interacting with KIF proteins on the microtubules.

We noticed that LDLR mainly localized peripherally underneath the cell membrane in primary hepatocytes (Figure 4a, left), while in Hep G2 cells, LDLR showed a scattered distribution (Figure 4a, right). We further explored the different distribution pattern of LDLR in Hep G2 cells and mouse primary hepatocytes. We found that the peripherally distributed LDLR in primary hepatocytes were mostly located on clathrin-coated vesicles (labeled by clathrin light chain A, Clta) (Figure 4—figure supplement 1a), while only a portion of LDLR were colocalized with CLTA in Hep G2 cells (Figure 4—figure supplement 1b), suggesting that LDLR in primary hepatocytes were recycled faster to the cell membrane to achieve their physiological role to recycle the majority of LDLR in plasma. In addition, LDLR partially colocalized with other endocytic compartments, such as early endosomes (Figure 4—figure supplement 1c) and recycling endosomes (Figure 4—figure supplement 1d) in Hep G2 cells. Therefore, Hep G2 cells were more suitable than mouse primary hepatocytes for investigating the recycling of LDLR via endocytic compartments.

Figure 4 with 6 supplements see all

Download asset Open asset

Figure 4—source data 1

KIF13A transport LDLR to cell periphery in both control and cKO hepatocytes.

(a) Representative confocal image showing the periphery accumulation of LDLR (red) after KIF13A-GFP (green) expression in control primary hepatocytes. (b) Representative confocal image showing the periphery accumulation of LDLR (red) after KIF13A-GFP (green) expression in cKO primary hepatocytes. Magnified insets of boxed areas are shown on bottom. Bar = 10 µm.

https://cdn.elifesciences.org/articles/58069/elife-58069-fig4-data1-v2.pptx

Download elife-58069-fig4-data1-v2.pptx

Expression of KIF13A-GFP in primary hepatocytes showed the majority of LDLR were enriched at some peripheral domains (Figure 4b), while the KIF13A truncation lacking the motor region (KIF13A-ST) retained LDLR on dispersed vesicles in the cytosol (Figure 4c), suggesting that KIF13A participates in the transport of LDLR to the cell periphery, and dysfunction of KIF13A disturbs the normal trafficking of LDLR.

As BLOS1 is seen on endosomal structures in Hep G2 cells (Figure 3), LDLR was localized to the cell periphery in full-length KIF13A-FLAG overexpressed Hep G2 cells, and these LDLR signals colocalized with the RE marker TfR (Figure 4d and Figure 4—source data 1), suggesting that KIF13A functions on REs during LDLR recycling. Besides, the rigor mutant of KIF13A (KIF13A-R), an ATPase activity-lacking point mutant which binds to microtubule but is unable to move along the microtubule (Guardia et al., 2016; Nakata and Hirokawa, 1995), anchored LDLR and REs on KIF13A-R positive microtubules (Figure 4e, Figure 4—figure supplement 2a and Figure 4—video 1). This suggests a critical role of KIF13A in RE-dependent LDLR transport. As a negative control, neither full-length KIF13A nor KIF13A-R mutant affects the cellular distribution of lysosome/MVB marker CD63 (Figure 4—figure supplement 2b,c and Figure 4—video 2).

In agreement with a previous report that KIF13A functions in RE tubule morphogenesis (Delevoye et al., 2014), overexpressed full-length KIF13A elongated tubular REs (labeled by EHD3), while KIF13A-ST truncation had no this effect (Figure 4f,g). We noticed that the majority of actively moving Rab11A-positive recycling endosomes were also labeled by LDLR in live-cell imaging (Figure 4—video 3 and Figure 4—figure supplement 1d). Live-cell imaging revealed that, in contrast to the re-distribution of REs (labeled by RAB11A) to the cell periphery by KIF13A (Figure 4—video 4), expression of KIF13A-R affects the movement of REs (labeled by RAB11A), most REs distributed along the KIF13A-R positive microtubules and were almost static during the 2 min imaging period (Figure 4h and Figure 4—figure supplement 2d and Figure 4—video 1), indicating that the LDLR anchored on microtubules through KIF13A-R on REs.

Furthermore, we found that KIF13A interacted with LDLR in the co-IP assays (Figure 4i), and knockdown of KIF13A resembles the reduction of LDLR and TfR occurred in BLOS1 deficiency (Figure 4j). From these results, we firstly report that LDLR in REs is a cargo transported by KIF13A, and dysfunction of KIF13A affects the cellular distribution and homeostasis of LDLR on plasma membrane.

In both Hep G2 cell lines (Figure 5a) and mouse primary hepatocytes (Figure 5—source data 1), KIF3A or KIF3B was evenly distributed in cells without any enrichment of LDLR in specific regions or puncta. Similarly, The expression of KIF3B rigor mutant (KIF3B-R) did not affect the cellular localization of LDLR on REs (labeled by TfR) (Figure 5b). However, by live-cell imaging analysis, we found that, unlike the active movement in control cells, most of the RE puncta in KIF3B-R expressing cells were anchored at dispersed sites on KIF3B-R-positive microtubules (Figure 5c and Figure 5—video 1), suggesting that KIF3B may play a role in the trafficking of REs. In our observations, non-specific blocking of all microtubule-based vesicle transport by KIF3B-R were excluded as we can observe the transport of lysosomes (labeled by dextran-TMR) using live-cell imaging (Figure 5—video 2).

Figure 5 with 2 supplements see all

Download asset Open asset

Figure 5—source data 1

KIF3A distributes evenly in primary hepatocytes.

Representative confocal images of non-puncta and even distribution pattern of overexpressed KIF3A-Scarlet (red) in mouse primary hepatocytes. Bar = 10 µm.

https://cdn.elifesciences.org/articles/58069/elife-58069-fig5-data1-v2.pptx

Download elife-58069-fig5-data1-v2.pptx

In addition, we observed that BLOS1 could be recruited to microtubules by KIF3B-R, and the rigor mutants of other two KIF proteins of kinesin-2 (KIF3A or KIF3C) both had the same effect, while KIF13A-R, which also interacted with BLOS1, or KIF5B-R (a rigor mutant of a kinesin-1 component) did not affect BLOS1 localization (Figure 5d). Furthermore, when KIF3A and KIF3B, which forms a heterodimer, were co-transfected with BLOS1 in Hep G2 cells, these two KIF proteins were redistributed to BLOS1 positive puncta structures (Figure 5e). One possible explanation for these results is that BLOS1 may form a tight protein complex with KIF3A/B or KIF3A/C heterodimers.

Due to the lack of a cargo-binding region in KIF proteins of kinesin-2, adaptor proteins are required to form a functional kinesin-2 complex for cargo transport. Therefore, we cloned a known kinesin-2 adaptor protein gene KAP3 and found that when co-expressed with KIF3B-R or KIF3A/B heterodimer, KAP3 show similar localization pattern as BLOS1 (Figure 5f,g), indicating that BLOS1 may act as a new adaptor protein of kinesin-2.

To avoid a possible redundant function between KIF3B and KIF3C, we chose the common subunit KIF3A as the target to be knocked down. Immunoblotting assays revealed that KIF3A knock-down also caused decreased LDLR levels in Hep G2 cells (Figure 5h). Together, our data indicate that BLOS1 may be a new adaptor protein participating in the assembly of functional kinesin-2 complexes, which regulate RE trafficking at dispersed sites on microtubules.

To explore the detailed mechanism underlying the regulation of RE trafficking by kinesin-2, we labeled REs in both control and KIF3A-KD cells with RAB11A-GFP and then analyzed their motion using live-cell imaging. We found that in these cells, the movement of REs could be categorized into two groups, a portion of REs was static or showed slightly local motion, whereas the other REs moved quickly in anterograde or retrograde transport directions. We named these quickly moving REs as ‘active REs’. Further analysis revealed that the motion of anterograde transported active REs in KIF3A-KD cells was significantly different from that of control cells (Figure 6a–d). In control cells, active REs in anterograde transport sometimes paused and then usually continued their motion in their original anterograde direction (Figure 6a and Figure 6—video 1). In contrast, after the knockdown of KIF3A (Figure 6b and Figure 6—video 1) or BLOC1S1 (Figure 6c and Figure 6—video 2), active REs frequently moved backward after paused anterograde transport. Most of these REs moved back to the perinuclear endocytic recycling compartment (ERC) of their origin by following the backward movement (Figure 6—video 1), suggesting a reduction in the percentage of REs that could eventually be destined to the cell periphery.

Figure 6 with 6 supplements see all

Download asset Open asset

The above results showed that both kinesin-2 and kinesin-3 participate in the transport of REs. We wondered whether they cooperate in this process. To test whether kinesin-2 is required for the normal function of kinesin-3, we expressed KIF13A-GFP in both control and KIF3A-KD cells and then analyze the behaviors of these KIF13A-positive tubular structures before and after KIF3A deficiency. We found that, unlike the successful extension to the cell periphery in control cells (Figure 6d and Figure 6—video 3), most of the tips of KIF13A-positive tubules in KIF3A-KD cells showed interrupted movement when they encountered other KIF13A tubules (Figure 6e and Figure 6—video 3). Some KIF13A tubule tips moved forward and backward repeatedly on the same track (Figure 6f, arrow 2), while the others paused and then moved backward on another track resembling the behavior of REs in KIF3A-KD cells (Figure 6f, arrow 1). Similar abnormal movement of KIF13A-positive tubules could be observed when the dominant-negative mutant KIF3B-R were expressed (Figure 6—video 4). In addition, KIF3A-KD cells showed normal microtubule architecture (Figure 6—figure supplement 1a) and dynamics (indicated by motility of EB1-GFP positive microtubule tips) (Figure 6—figure supplement 1b and Figure 6—video 5), suggesting unaffected microtubule network after KIF3A deficiency.

In line with the above results, when KIF13A-R and KIF3B-R were co-expressed, the distribution of LDLR is determined by KIF13A-R, indicating that KIF13A could function upstream of KIF3B (Figure 6—figure supplement 1c). However, when KIF3A, KIF3B, and BLOS1 were co-overexpressed to form an intact kinesin-2 complex, peripheral accumulation of LDLR which occurred in the case of KIF13A expression, was not observed (Figure 6—figure supplement 1d), suggesting that kinesin-2 itself may not drive the long-range anterograde transport of REs, and KIF3-BLOS1 and KIF13A may play different roles in this process.

To further determine the distribution of KIF13A-R and KIF3B-R on microtubules, we co-stained these two rigor mutants with antibodies to acetylated and tyrosinated tubulin, as it has been shown that tyrosinated tubulin has a broader distribution than centrally located acetylated tubulin (Guardia et al., 2016), which resembles the distribution pattern of KIF3B-R and KIF13A-R. We found that both KIF13A-R and KIF3B-R located on acetylated microtubules (Figure 6—figure supplement 1e,f, top), and KIF3B-R but not KIF13A-R was additionally found on peripherally distributed tyrosinated microtubules (Figure 6—figure supplement 1e,f, bottom). These observations indicated that KIF13A and KIF3B may walk along the same set of microtubule tracks with acetylated tubulin and post-translational modifications of tubulin may contributes to the association of KIF13A and KIF3B with microtubules.

It has been reported that cargoes (lysosomes) pause at intersection points between multiple microtubules and switch to another microtubule track at some specific intersections, and these microtubule intersections serve as important sites for direction determination during cargo transport (Bálint et al., 2013; Bergman et al., 2018; Verdeny-Vilanova et al., 2017). We found that, after knockdown of KIF3A, when encountering other KIF13A-GFP positive tubules, tips of KIF13A tubules were unable to pass through and move on (Figure 7a,b). As the KIF13A-positive tubules move along microtubules, the site where a tip of KIF13A tubule encounters another KIF13A tubule may represent the intersection point between these two microtubules (Figure 7a). This raised the possibility that the BLOS1-dependent kinesin-2 complex may function at specific microtubule intersections.

Figure 7 with 4 supplements see all

Download asset Open asset

To test this idea, we first detected the distribution of BLOS1 on microtubules using super-resolution microscopy and found that a certain percentage of BLOS1 puncta (77.6 ± 4.6%, n = 8 cells, mean ± SD) located at the intersections of microtubules in Hep G2 cells (Figure 7c, see also Figure 7—video 1 for z-stack reconstruction movie), which could also be observed in mouse primary hepatocytes (Figure 7—figure supplement 1a,b), indicating that kinesin-2 may function at these sites. Then we observed the behavior of recycling endosomes (indicated by RAB11A-GFP) on microtubule tracks (stained by Tubulin Tracker Deep Red) in live cells using Zeiss LSM 880 system’s Airyscan super-resolution module. We found that a large proportion of pausing events of recycling endosomes occurred at microtubule intersection sites, and in most cases, recycling endosomes moved on without changing their motion polarity after the pausing (Figure 7—video 2). But under certain conditions, recycling endosomes reversed their moving direction after their pausing at specific microtubule intersections (Figure 7—video 3). Furthermore, statistic results showed that the frequency of reversed movement of recycling endosomes after paused anterograde transport at specific microtubule intersections was increased by knockdown of either BLOC1S1 (77.3 ± 8.4%) or KIF3A (78.7 ± 6.6%) as compared to control cells (10.1 ± 1.8%) (Figure 7d), suggesting that dysfunction of BLOS1-dependent kinesin-2 complex alters the behavior of recycling endosomes at specific microtubule intersections by changing their motion polarity and finally reduces the recycling of LDLR.

In summary, our observations suggest that BLOS1 acts as an adaptor for kinesin-2 to assist the LDLR cargo transportation driven by kinesin-3 at specific microtubule-microtubule intersections where hurdles may preclude the cargo for smooth long-range anterograde transport (Figure 8). Our results reveal a novel function of BLOS1 in mediating a kinesin switch at microtubule intersections.

Figure 8 with 2 supplements see all

Download asset Open asset

All resources used in this study are listed in Appendix 1.

The Bloc1s1 floxed mice (loxp mice) and Kxd1-KO mice were originally generated in our lab (Yang et al., 2012; Zhang et al., 2014) and subsequently bred locally in the animal facility (specified pathogen free) of Institute of Genetics and Developmental Biology, Chinese Academy of Sciences. pa mice were derived from The Jackson Laboratory. Alb-Cre tool mice were supplied by Model Animal Research Center of Nanjing University. To generate liver-specific Bloc1s1 knockout mice (cKO mice), Alb-Cre mice were crossed with loxp mice. Colonies were maintained by breeding Alb-Cre; loxp/loxp mice with loxp/loxp mice. Control loxp mice and cKO mice were littermates. Male mice of 3 months were used unless is stated. Mice were kept under a 12 hr dark-light period and provided a standard chow diet. For starvation treatment, mice were fasted for 12 hr overnight, and all mice livers used for Oil Red O staining were obtained at the next morning. All animal work were approved by the Institutional Animal Care and Use Committee of the Institute of Genetics and Developmental Biology, Chinese Academy of Sciences (mouse protocol KYD2005-006).

A two-step collagenase perfusion technique was used to isolate primary hepatocytes from loxp and cKO mice liver as described (Li et al., 2010). Mice were euthanized by CO₂ inhalation, the abdominal cavity was opened, and then the precaval vein was closed with a vascular clamp. Liver perfusion from the portal vein was initiated with 50 mL pre-warmed (37°C) HBSS (pH 7.4, no calcium and magnesium) containing 10 mM HEPES and 200 µM EDTA for 10 min. After the liver turned pale, change perfusion medium and perfuse with 50 mL pre-warmed 50 HBSS (pH 7.4, with calcium and magnesium) containing 0.5 mg/mL collagenase IV and 20 mM HEPES for 10–15 min. Then remove the entire liver to a petri dish containing HBSS (with calcium and magnesium) on ice, and dissociate the liver lobs by tearing with forceps. The resulting cell suspension was filtered through a 70 µm mesh filter and washed three times with pre-cooled HBSS (centrifuged at 50 × g for 2 min). Cells were then suspended in 10 mL William's E Medium (Gibco) supplemented with 10% (v/v) fetal bovine serum (FBS, Gibco), 1% (v/v) 200 mM GlutaMAX (Gibco) and 1% (v/v) Nonessential amino acids (NEAA, Gibco), and then seeded at a final density of 0.4 × 10⁶ cells per mL onto collagen I-coated dishes (coverslips). Cells were incubated with 5% CO₂ at 37°C for 2 hr and then remove the medium containing attached cells and replaced with fresh culture medium. Transfections were done 4 hr after attachment using jetPEI-Hepatocyte reagent (Polyplus).

Hep G2 cells were obtained from the cell bank of the Chinese Academy of Sciences (Shanghai, China). All cell lines used in experiments were maintained at 37°C with 5% CO₂. Hep G2 cells were cultured in Minimal Essential Medium (MEM, Hyclone) supplemented with 10% (v/v) FBS, 1% 100 mM sodium pyruvate (Gibco) and 1% NEAA (Gibco).

The strain used for protein expression and purification was Escherichia coli (E. coli) BL21 (DE3) (Vazyme), and the expressing vector was PGEX-4T-1 (Pharmacia). First, BL21 competent cells were transformed with the expression plasmid and selected on LB plates supplemented with 100 µg/mL ampicillin overnight at 37°C. Single colonies were picked and grown overnight in 5 mL LB medium supplemented with 100 µg/mL ampicillin at 37°C with shaking at 300 rpm. On next morning, the overnight culture was diluted 1:100 into fresh 100 mL LB medium and cultured at 37°C (about 2.5 hr) until the optical density at 600 nm (OD₆₀₀) reached 0.5 to 0.7. Isopropyl-1-thio-β-D-galactopyranoside (IPTG) was added to final concentration of 0.4 mM and fusion protein expression induced at 25°C for 8 hr.

For the expression of N-terminal tagged FLAG/Myc-fusion proteins (BLOS1-FLAG, BLOS1-Myc), pCMV-Tag-2B/3B vectors (Agilent) were used with cDNA fragments inserted into multiple cloning sites. pEGFP-C2 (Clontech) was used to generate BLOS1-GFP-C2 and RAB11A-GFP plasmids. pEGFP-N2 (Clontech) was used for KIF13A-GFP, KIF13A-ST-GFP expression. For the red fluorescent protein Scarlet (Bindels et al., 2017) expression, ORF of GFP was substituted by the coding sequence of Scarlet by modifying pEGFP-C2 and pEGFP-N2 vectors without changing other DNA elements. Similarly, GFP ORF in the pEGFP-N2 vector was replaced by FLAG/Myc/HA coding sequences to generate plasmid for C-terminal fusion proteins (BLOS1-HA, LDLR-HA, KIF13A-FLAG and all other KIF proteins with corresponding FLAG/Myc/HA tags) expression. The expressing plasmids of KIF protein rigor mutants were generated by site-directed mutagenesis using primers shown in Appendix 1.

The shRNA expressing plasmids used in stable knockdown cell line (BLOC1S1-KD, KIF13A-KD, and KIF3A-KD cells) construction were modified from pSilencer 5.1-H1 Retro vector (Ambion) by introducing a MfeI cleavage site (same overhang with EcoRI) between 2166 to 2172 just after the XhoI site and upstream of the expressing element of shRNA. Take KIF3A knockdown plasmid as an example, three separate shRNA expressing plasmids were generated by annealing the templates (Appendix 1) and inserting into the modified plasmids. One plasmid was digested with XhoI and EcoRI to get the whole shRNA expressing element, another was digested with XhoI and MfeI. Then ligation was performed to generate a new plasmid containing two sets of shRNA expressing elements. This procedure was repeated until all templates for a specific gene were introduced on one plasmid.

ORF of KIF5B was amplified from the cDNA of Hep G2 cells, and all the other ORFs were amplified from the cDNA obtained by reverse transcription of total mRNA of mouse liver. For the construction of KIF3A-FLAG-KIF3B-Myc-BLOS1-HA co-overexpression plasmid, the expression cassette of KIF3B-Myc was amplified and then introduced into BspTI digested KIF3A-FLAG plasmid by recombination (ClonExpress II One Step Cloning Kit, Vazyme) after which the BspTI site was kept intact, and the expression cassette of BLOS1-HA was inserted by a similar procedure.

Homogenates of either cell cultures or tissues were used in western blot analysis. Cells in one well of 6 well plates were collected in ice-cold PBS (Gibco) and lysed in 200 µL lysis buffer (50 mM Tris-HCl pH 7.0, 150 mM NaCl, 1 mM EDTA, 1% Triton X-100) supplemented with protease inhibitor cocktail. Tissue samples (about 50 mg) were homogenized with a micro tissue grinder in 500 µL lysis buffer. All homogenates were rotated for 1 hr at 4°C in a vertical rotator and centrifuged at 13000 × g for 15 min. Supernatants with soluble proteins were mixed with 6 × Laemmli loading buffer with β-mercaptoethanol and boiled for 5 min. Protein samples were resolved by SDS-PAGE on 10% or 15% Tris-Glycine buffered polyacrylamide gels and transferred to PVDF membranes using mini trans-blot module (Bio-Rad).

The blots were blocked in 5% (w/v) non-fat milk (BD) at room-temperature for 1 hr in PBS/0.1% (v/v) Tween 20 (PBST). After a brief wash in PBST, membranes were then incubated in primary antibody diluted in 3% (w/v) BSA dissolved in PBST overnight at 4°C. On the following day, the membranes was washed three times, each time for 10 min in PBST with shaking and then incubated with 1:5000 HRP-conjugated secondary antibodies (Zsbio) diluted in blocking buffer at room temperature for 1 hr. Three more 10 min washes with PBST were then performed before detection using Chemiluminescent Substrate (Thermo) and imaging with a Minchemi system (Sage Creation). Quantification was performed using Fiji (Schindelin et al., 2012). The following primary antibodies were used (additional information was shown in the Appendix 1): anti-LDLR (1/5000 for ab52818; 1/2000 for ab30532), anti-TfR (1/2000), anti-β-Actin (1/50000), anti-KIF13A (1/1000), anti-KIF3A (1/1000), anti-Pallidin (1/2000), anti-Dysbindin (1/20000), anti-GST (1/5000), anti-FLAG (Sigma, Cat#F3165, 1/5000), anti-Myc (MBL, Cat#562, 1/2000), anti-HA (Abcam, 1/5000).

All immunoprecipitations were performed in the absence of cross-linking reagents. Cells in six well plates were transfected with FLAG empty vector/FLAG fusion protein plasmid and corresponding candidates of interacting protein plasmids for 48 to 72 hr, and cell lysates were prepared in cell lysis buffer as described above. 40 µL anti-FLAG M2 agarose beads (Sigma) were used for the enrichment of FLAG-fusion proteins of each well by incubating with the supernatants overnight at 4°C on a vertical rotator. Sufficient washes were carried out at 4°C using lysis buffer with a total time of 30 min on a vertical rotator (5 to 6 washes), beads were pelleted by centrifuging at 5000 × g for 30 s after each wash. Proteins were eluted by boiling with 50 µL 2 × Laemmli loading buffer for 5 min and subsequent western blot was performed as described above. For the detection of eluted FLAG-fusion proteins, a particular HRP-conjugated secondary antibody (Abcam, Cat# ab131368) which preferentially detects the non-reduced form of mouse IgG over the reduced, SDS-denatured forms was used to eliminate the IgG band results from the anti-FLAG M2 beads.

The full length and truncated BLOS1s were expressed in E. coli BL21 as described above, and GST protein which was induced at 37°C for 4 hr served as control. After induction, bacteria were collected by centrifuge at 4000 × g, 4°C for 15 min. The bacteria pellet was resuspended in 10 mL ice-cold lysis buffer (50 mM Tris-HCl pH 8.0, 150 mM NaCl, 5 mM EDTA, 1% Triton X-100) supplemented with protease inhibitor and 1 mM DTT. The suspension was then sonicated on ice using a probe-tip sonicator using the following parameters: 10 s each with 30 s rest at 200 W for a total time of 15 min. The supernatant was collected and stored at −20°C after centrifugation at 12,000 × g, 4°C for 15 min.

For GST pull-down assay, 50 µL GST protein-containing supernatant diluted in 450 µL lysis buffer was used to incubate with 50 µL slurry of glutathione-Sepharose beads at 4°C for 6 hr, and the dosage of supernatants of other GST fusion proteins was determined by Coomassie brilliant blue staining. The beads were pelleted at 5000 × g for 30 s and washed five to six times with lysis buffer before the incubation with 500 µL mouse liver lysate overnight at 4°C. After incubation, beads were washed with tissue lysis buffer for another 5 to 6 times, and binding proteins were eluted with 2 × Laemmli loading buffer by boiling for 5 min. Immunoblots were performed as described above.

Freshly collected liver lobes of control (loxp) and cKO mice fed in chow diets or under starvation were embedded in OCT (Tissue-Tek) and flash-freezed immediately with liquid nitrogen. Samples were left in the chamber of the freezing microtome with the temperature set at −20°C for at least 1 hr before the sectioning. Three consecutive 10-µm-thick sections of the liver were collected onto one slide and fixed immediately in ice-cold 4% (w/v) paraformaldehyde dissolved in PBS (pH 7.4) for 10 min and then rinsed in three changes of distilled water. Slides were dehydrated in 60% isopropyl alcohol for 2 min before the immersed by Oil Red O staining solution (1.5 part of Oil Red O saturated isopropyl alcohol solution mixed with one part of distilled water) at room temperature for 15 min. Slides were rinsed in 60% isopropyl alcohol for two times and then placed in distilled water before the next step. Counterstaining of the nucleus with Mayer′s hematoxylin was done by submerging the sections in hematoxylin for 10 s and thereafter dipping the sections in distilled water for three times before bluing the stain in PBS (pH 7.2) for 5 min. Slides were mounted in water-soluble mounting medium (PBS: glycerol = 1:9), and coverslip edges were sealed with nail polish. Micrographs were acquired using an Olympus DP71 imaging system, and Fiji was used in the quantification of Oil Red O staining areas.

Coverslips were pre-coated with collagen I (Sigma) for Hep G2 attachment. Cells attached on coverslips were fixed in 4% (w/v) formaldehyde dissolved in PBS and immunocytochemistry (ICC) was performed according to a general ICC protocol (Abcam). Permeabilization of cells was done by incubating in PBS containing 0.1% (v/v) Triton X-100 for 15 min. Blocking was done with 3% (w/v) BSA in PBS. The following primary antibodies were used (additional information was shown in Appendix 1): anti-mouse LDLR (R and D, Cat#AF2255, 1/100), anti-human LDLR (R and D, Cat#AF2148, 1/100), anti-TfR (1/200), anti-EEA1 (1/500), anti-Cytochrome C (1/250), anti-CD63 (1/200), anti-FLAG (Sigma, Cat#F3165, 1/1000; Sigma, Cat#F7425, 1/1000), anti-Myc (MBL, Cat#562, 1/500; MBL, Cat#M192-3S, 1/500), anti-HA (Santa Cruz, Cat#sc-7392, 1/100; Abcam, Cat#ab9134, 1/1000) and anti-α-Tubulin (Abcam, Cat#ab7291, 1/500; Abcam, Cat#ab18251, 1/1000). Coverslips were incubated with primary antibodies diluted in blocking buffer overnight at 4°C and wash buffer used after antibody incubation was PBS containing 0.1% (v/v) Tween 20. All Alexa Fluor-conjugated secondary antibodies (Invitrogen/Abcam) were diluted 1/1000 and used with an incubating time of 1 hr at 37°C in the dark. After the wash, coverslips were mounted on glass slides with Prolong Gold Antifade Mountant (Invitrogen), sealed with nail polish and stored in the dark at 4°C before imaging. Confocal imaging was carried out using a Nikon Eclipse Ti Confocal Laser Microscope System with NIS-Elements Software (Nikon, Japan). Super-resolution images were acquired in a Zeiss LSM 880 system with Airyscan module. Images were analyzed and quantified were Fiji software with raw data imported through Bio-Formats Importer.

Lipoproteins in plasma samples were pre-stained with Sudan Black B (SBB) staining solution (SBB saturated thanol solution) by mixing 2 µL of SBB staining solution with 30 µL plasma and incubating at 37°C for 30 min. A 4% to 15% polyacrylamide gradient mini gel prepared with Tris-HCl buffer (pH 8.3) was used to separate lipoproteins, and the running buffer (0.6 g/L Tris, 2.88 g/L Glycine, pH 8.3) was changed several times during the overnight electrophoresis at 4°C with a voltage of 70 V. For the Oil Red O staining, unstained plasma samples were separated by electrophoresis and then stained with 0.2% (w/v) Oil Red O methanol solution for 30 min with shaking.

LDL of pooled mouse plasma were isolated by sequential ultracentrifugation as described (Havel et al., 1955). High-density salt solution (1.346 g/mL) was prepared by dissolving 153 g NaCl and 354 g KBr in Milli-Q water to a total volume of 1L. Low-density salt solution (1.005 g/mL) contained 0.15 M NaCl. Salt solutions of other densities were prepared by mixing the high-density and low-density salt solutions at different ratios. 5 mL pooled plasma with the density of 1.006 g/mL was mixed with 1 mL of 1.085 g/mL salt solution to reach a total density of 1.019 g/mL and ultracentrifuged at 180,000 × g, 4°C for 12 hr. The top 1.5 mL layer containing lipoproteins with a density less than 1.019 g/mL (mainly VLDL) was collected, and additional 1.5 mL salt solution (1.200 g/mL) was mixed with the remaining plasma to reach a density of 1.063 g/mL. Ultracentrifugation was performed as above, and the top 1.5 mL solution containing lipoproteins with the density between 1.019 g/mL and 1.063 g/mL (LDL) was collected. All collected lipoproteins were dialyzed in 500 mL 0.15 M NaCl twice at 4°C before electrophoresis or DiI labeling.

For DiI labeling, 1 mg of purified LDL were mixed with 50 µL DiI stock solution (3 mg/mL in DMSO; Sigma) and incubated at 37°C for 8 hr. The density was adjusted to 1.063 g/mL, and ultracentrifugation was performed. Collected LDL-DiI on the top layer was dialyzed with 0.15 M NaCl and further filter-sterilized (0.1 µm, Millipore). LDL-DiI were added to the medium of primary hepatocytes at the final concentration of 10 µg/mL, and binding of LDL-DiI to the cell surface was done at 4°C for 30 min. After the binding, cells at different time points post endocytosis were fixed and visualized.

Hep G2 cells in 24 well plates were transfected with targeted gene shRNA or scrambled negative control shRNA expressing plasmids and split into 60 mm dishes at the concentration of 10% confluent 24 hr post-transfection. The selection was started the day after splitting using 2 µg/mL puromycin (concentration determined by preliminary experiment). Culture medium was changed every day in the first week and every 3 days after that. Cells were grown for 3 weeks under selection pressure, and single colonies were picked and expanded in 24-well plates with puromycin added. Western blot of target proteins was performed to identify positive colonies.

Hep G2 cells were plated in 12-well plates at a seeding density of ~50% confluency and were cultured for 12 hr before leupeptin treatment. Leupeptin (Sigma) dissolved in Milli-Q water was added to a final concentration of 50 µM in the culture medium. Cells were collected 24 hr later, and western blot was performed with indicated antibodies as previously described.

Hep G2 cells were cultured on collagen I-coated 35 mm µ-Dish (ibid) with a high-performance glass bottom (170 ± 5 µm). Hep G2 cells were transfected with GFP or/and Scarlet fusion protein expressing plasmids using lipofectamine 3000 reagent 24 hr before the imaging. Live cell imaging was carried out on a Nikon Eclipse Ti Confocal Laser Microscope System equipped with 405 nm (20 mW), 488 (50 mW) and 561 (100 mW) laser lines, temperature controller, a 100 × oil immersion objective (Nikon, NA = 1.40) and appropriate filter sets. Before imaging, the culture medium was replaced by CO₂-independent Live Cell Imaging Solution (Invitrogen) and adapted for 15 min. Image series were captured using the 488 laser line (~10% laser power) and 561 laser line (~5% laser power) with 1 s interval and total imaging time of 120 s under the ‘live’ mode.

Super-resolution microscopy of microtubules was carried out on Zeiss LSM 880 system equipped with Airyscan module and cell incubation chamber. Cells transfected with RAB11A-GFP were first stained with Tubulin Tracker Deep Red according to the manufacturer’s instruction, and image series were then captured using the 488 laser line and 633 laser line using Airyscan module with a interval time of 0.32 s and zoom factor of 3. Subsequent imaging processing was accomplished using ZEN 2.3 (Zeiss) and Fiji software.

Data were presented as Mean ± SEM unless stated, the statistical significance of mean differences was determined using two-tailed Student’s t test as indicated in the figure legends. The sample size (n) was also indicated in the corresponding figure legends, which represents the number of identically-treated replicates. Statistical significance is defined as, n.s., not significant, *p<0.05, **p<0.01, ***p<0.001. Statistical analyses were performed using GraphPad Prism.

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

We thank Prof. Richard T Swank for his proofreading this manuscript and comments. This work was partially supported by grants from the Ministry of Science and Technology of China (2019YFA0802104), and the National Natural Science Foundation of China (31830054; 91954104; 91539204; 81670789), and the Chinese Academy of Sciences (XDA12030211).

Animal experimentation: All animal work was approved by the Institutional Animal Care and Use Committee of the Institute of Genetics and Developmental Biology, Chinese Academy of Sciences (mouse protocol KYD2005-006).

© 2020, Zhang et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.