The N terminus of the small heat shock protein HSPB7 drives its polyQ aggregation–suppressing activity

Heat shock protein family B (small) member 7 (HSPB7) is a unique, relatively unexplored member within the family of human small heat shock proteins (HSPBs). Unlike most HSPB family members, HSPB7 does not oligomerize and so far has not been shown to associate with any other member of the HSPB family. Intriguingly, it was found to be the most potent member within the HSPB family to prevent aggregation of proteins with expanded polyglutamine (polyQ) stretches. How HSPB7 suppresses polyQ aggregation has remained elusive so far. Here, using several experimental strategies, including in vitro aggregation assay, immunoblotting and fluorescence approaches, we show that the polyQ aggregation-inhibiting activity of HSPB7 is fully dependent on its flexible N-terminal domain (NTD). We observed that the NTD of HSPB7 is both required for association with and inhibition of polyQ aggregation. Remarkably, replacing the NTD of HSPB1, which itself cannot suppress polyQ aggregation, with the NTD of HSPB7 resulted in a hybrid protein that gained anti-polyQ aggregation activity. The hybrid NTDHSPB7–HSPB1 protein displayed a reduction in oligomer size and, unlike WT HSPB1, associated with polyQ. However, experiments with phospho-mimicking HSPB1 mutants revealed that de-oligomerization of HSPB1 alone does not suffice to gain polyQ aggregation–inhibiting activity. Together, our results reveal that the NTD of HSPB7 is both necessary and sufficient to bind to and suppress the aggregation of polyQ-containing proteins.

Small heat shock proteins (HSPBs) 2  Most of these form highly dynamic oligomeric complexes (3,4). Besides forming homo-oligomeric complexes, some HSPBs can also form hetero-oligomers with specific other HSPBs (4,5). Exceptions to this paradigm are HSPB8, which forms a chaperone complex with BAG3 and Hsp70 (6,7), and HSPB7, which has no other known chaperone partners and does not form large oligomers (7,8).
Unlike the human oligomeric HSPB family members, HSPB7 overexpression does not enhance the cellular capacity to refold heat-unfolded luciferase (8) nor leads to suppression of the heat-induced aggregation of cytosolic proteins (9). Instead, within the human HSPB family, HSPB7 is the strongest suppressor of aggregation of polyglutamine-containing proteins (8) that cause diseases like spinocerebellar ataxia's, Kennedy's disease, and Huntington's disease (HD) (10,11). These diseases are caused by expansions of CAG triplet repeats coding for abnormally long polyglutamine (polyQ) stretches (11). The size of these expansions are inversely correlated with the onset age of these diseases and to the aggregation propensity of the respective proteins (10). Unlike heat-induced aggregates that are generally driven by hydrophobic interactions (12), polyQ aggregation is suggested to be driven by hydrogen bonding and ␤-hairpin structures (13,14) potentially explaining the different dependence on chaperone activity for suppression of polyQ aggregation. However, why HSPB7 stands out as the most potent suppressor of polyQ aggregation within the HSPB family has remained an enigma. Previously, we showed that HSPB7 requires (macro) autophagy for its suppressive action on polyQ aggregation (8). However, HSPB7 differs from other autophagy-related suppressors like Alfy, as it cannot act on preformed aggregates to fulfill its function (15). One model that explains this finding predicts that HSPB7 is present during aggregate formation and binds these protofibrils to maintain early aggregates in a state competent for processing by the autophagic machinery (16).
In this article, we investigated which characteristics of HSPB7 are required to chaperone polyQ proteins and prevent their aggregation. HSPB7 acts directly on the aggregation process as recombinant HSPB7 can delay in vitro polyQ aggrega-tion, whereas HSPB1 cannot. HSPB7 acts mainly on regions flanking the polyQ and not on the amyloid core itself. In cells, this remarkable suppression depends on the N terminus of HSPB7 that is predicted as an intrinsically disordered region. Although the N terminus by itself is not capable of suppressing aggregation, fusion to the ␣-crystallin domain of HSPB1 results in a hybrid protein with polyQ anti-aggregation properties. In summary, the full and unique N-terminal domain is key to the specific ability of HSPB7 to inhibit polyQ aggregation.

The full NTD is required for HSPB7 to prevent polyQ aggregation
Within the HSPB family, cell-based analyses had revealed that HSPB7 stands out as a chaperone that is the most capable to prevent aggregate formation by amyloidogenic polypeptides including polyQ containing proteins (8). First, we asked whether this protective function of HSPB7 against polyQ aggregation seen in cells is related to a direct action of the chaperone on the substrate. Hereto, we incubated a purified exon-1 fragment of mutant Huntingtin with 48 glutamines (mHttQ48) with either purified HSPB7 or HSPB1 (with HSPB1 acting as control as it is unable unable to protect against polyQ aggregation in cells (8)). Analyzing the ratio of soluble:insoluble mHttQ48 by gel electrophoresis (9), we found that with time, the amount of soluble mHttQ48 declined and appeared as aggregated material in the stacking gel (Fig. 1, A and B). The rate of decline in the soluble:insoluble ratio of mHttQ48 was unaffected by co-incubation with HSPB1 but clearly slowed down by the presence of HSPB7 (Fig. 1, A and B), meaning that HSPB7 acts directly on the aggregation process.
Next, we asked what structural characteristics in HSPB7 specifies this activity toward suppressing aggregation of polyQ proteins. The NTDs and CTDs of HSPBs display the highest level of variability and have been suggested to be crucial for their differential function (17,18). Our earlier data had revealed that the CTD of HSPB7 is not required for its activity to prevent polyQ aggregation (8). So, we focused our attention on the NTD of HSPB7 and noted a serine-rich stretch (SRS) (Fig. 1C, Fig.  S1A) that is not found in other HSPB members. Interestingly, a serine-rich stretch has been identified as a crucial segment in the ability of DNAJB6, a Hsp70 co-chaperone, to suppress polyQ aggregation (19,20). Although structurally distinct (Fig. S1A) we first set out to test the importance of the NTD in HSPB7, and in particular its SRS, in its action to suppress polyQ aggregation by generating a series of NTD truncations of HSPB7 mutants (Fig. 1C) that we co-expressed in cells together with an exon-1 fragment of Huntingtin with 74 glutamines tagged to GFP (mHttQ74 -GFP). In contrast to our speculations, deleting the SRS from HSPB7 (v5-HSPB7 ⌬S17-29 ) did not result in a loss of its anti-aggregation activity (Fig. 1, D-F), despite its relatively lower expression as compared with that of WT HSPB7 (Fig. S1B). However, a gradual decline in anti-aggregation activity was found upon progressive NTD deletion with the complete deletion of the NTD (v5-HSPB7 ⌬73 ), being inactive ( Fig. 1, D-F, Fig. S1B). None of the HSPB7 variants caused an induction of the heat shock response, evidenced by the absence of induced expression of HSPA6, an exclusive HSF-1-regulated chaperone (21,22) (Fig. S1C). Also the HSF-1-regulated chaperone DNAJB1, which when overexpressed alone, can also lead to suppression of polyQ aggregation in these cells (23,24) was not elevated by the expression of any of the HSPB7 variants (Fig. S1C), implying that it is unlikely that their (remaining) protective effects are due to up-regulation of endogeneous stress responses. Finally, endogenous expression of HSPB7 is undetectable in these cells irrespective of expression of the HSPB7 variants (not shown) implying that protective effects of the mutants cannot be attributed to formation of complexes between full-length endogenous and the ectopically expressed HSPB7 variants.

Fusing the NTD of HSPB7 to polyQ does not prevent polyQ aggregation
Three algorithms designed to predict intrinsically disordered regions (25,26) suggest that the NTD of HSB7 is highly disordered ( Fig. S1D) and reminiscent of proteins involved in phase separation (27)(28)(29). In fact, HSPB7 has been shown to be associated with intranuclear RNA speckles (SC35 speckles) and the fusion of its NTD to other small HSPBs like HSPB1 (of which the NTD is less disordered: Fig. S1, D and E) suffices to drive their association with these SC35 speckles, suggesting it might induce phase separation of the proteins it interacts with to prevent amyloidogenesis (30). Whereas for polyQ proteins, a protective role for chaperone-mediated phase separation has not been established, it has been speculated that in yeast compartmentalization of polyQ proteins into so-called "insoluble protein deposit" may have physiological, cell protective relevance (31, 32). In HEK293 cells, WT HSPB7 (v5-HSPB7 WT ) indeed colocalizes with mHttQ74 -GFP inclusions and this co-localization is much reduced for the NTD deletion mutant (v5-HSPB7 ⌬73 ) (Fig. 1G). Moreover, whereas full-length HSPB7 was bound to mHttQ74 -GFP in co-immunoprecipitation experiments, the NTD deletion mutant (v5-HSPB7 ⌬73 ) was not (Fig. 1H). This could suggest that binding of HSPB7 to polyQ proteins drives its phase separation and hence prevents its aggregation into SDS-insoluble amyloids. Our initial attempts to co-express mHttQ74 -GFP with only the NTD of HSPB7 (NTD HSPB7 ) failed as the NTD alone is not stable when expressed in cells (Fig. S2A). To test whether the predicted major IDR fragment of HSPB7 suffices to prevent polyQ aggregation, we therefore tagged the polyQ fragment directly with the NTD of HSPB7 (Fig. 2, A-D, Fig. S2B). However, for both fragments with either 43 or 74 CAG repeats, tagging with the NTD HSPB7 did not prevent polyQ aggregation, but rather enhanced it (Fig. 2, A-D). These data imply binding of HSPB7 to polyQ proteins and a putative subsequent phase separation driven by its NTD as such is insufficient to prevent polyQ aggregation. Direct fusion of the full-length HSPB7 to polyQ-74 did prevent aggregation indicating that the NTD in these hybrid fusion proteins is functional (Fig. 2, E and F, Fig. S2C).
Since we previously showed that the CTD of HSPB7 is dispensable for its action on polyQ (8), the combined data suggest that interaction of a (monomeric) HSPB7 via its NTD to polyQ proteins and the presence of the ACD of HSPB7 suffice to prevent mHtt aggregation.

Fusion of the NTD of HSPB7 to HSPB1 is sufficient to convey HSPB1 with anti-polyQ aggregation activity
We next hypothesized that fusing the NTD of HSPB7 to the ACD of other HSPBs may turn these into polyQ aggregation preventive chaperones. As mentioned above, HSPB7 does not form homo-nor hetero-oligomers with any of the HSPBs, so targeting the other HSPBs to polyQ aggregates via hetero-olig-omerization is unlikely to occur. To test whether the NTD of HSPB7 could turn another HSPB protein into an effective inhibitor of polyQ aggregation we choose HSPB1, as neither its presence in vitro (Fig. 1, A and B) nor its overexpression in cells (Fig. 3, B-D, Fig. S3A) could inhibit polyQ aggregation. Strikingly, expression of a fusion of the NTD of HSPB7 with the ACD and C-terminal domain of HSPB1 (NTD HSPB7 -HSPB1)

Role of NTD in HSPB7 function
lead to reduced mHttQ74 -GFP aggregation comparable with the effect of HSPB7 WT expression (Fig. 3, A-D). This result indicated that the conserved ACD is a necessary structural requirement for HSPB7 to prevent polyQ aggregation. Strikingly, this requirement is not restricted to the ACD of HSPB7 as also the ACD of HSPB1 can exert this function. Indeed, a fusion of the NTD HSPB7 to mRFP (as a control) did not generate a hybrid protein that could prevent mHttQ74 aggregation (Fig.  S3, B and C).
In line with the biochemical data, we found that although WT HSPB1 did not co-localize with mHttQ74 -GFP, the NTD HSPB7 -HSPB1 hybrid did (Fig. 3E). Inversely, replacing the NTD of HSPB7 with the NTD of HSPB1 resulted in a hybrid protein (NTD HSPB1 -HSPB7) that had lost its anti-polyQaggregation activity (Fig. 3, F and G, Fig. S3D) and no longer co-localized with the polyQ aggregates (Fig. 3H).
The co-localization of polyQ with HSPB7 and NTD HSPB7 -HSPB1 suggested that the NTD of HSPB7 is both required and A and B, HSPB7 prevents mHttQ48 aggregation in vitro. A purified exon-1 fragment of Huntingtin (mHttQ48) was incubated without chaperones or with either HSPB7 or HSPB1 for 0 -5 h at 37°C and processed on SDS-polyacrylamide gels. A, Soluble mHttQ48 was detected by Western blotting using anti-polyQ antibody (upper row). Soluble and insoluble mHttQ48 as well as HSPBs levels were also detected by SYPRO Orange staining of the SDS-polyacrylamide gels (lower row). C means control in which HSPB1 or HSPB7 were incubated at 37°C for 5 h without mHttQ48. B, quantification of the ratio of soluble to insoluble mHttQ48. C, schematic representation of HSPB7 NTD-truncated mutants. D, FTA showing the effects of HSPB7 WT or NTD-truncated mutants on polyQ (mtHttQ74) aggregation in HEK293 cells (left). Control samples were transfected with mRFP. Quantifications are shown on the right, normalized to control (control ϭ 1.0). E, quantification of cells displaying mHttQ74 -GFP aggregates when expressed together with mRFP (control ϭ 1.0), HSPB7 WT , or its mutants. F, representative immunofluorescence pictures of HEK293 cells expressing mHttQ74 -GFP with control plasmid (mRFP) or together with HSPB7 WT or its mutants (bar ϭ 100 m, zoom in upper right, dotted lines indicate the cell outlines). G, immunofluorescence staining of HEK293 cells co-expressing mHttQ74 -GFP (green) with v5-HSPB7 WT (upper row in red) or v5-HSPB7 ⌬73 (lower row in red). Bar ϭ 10 m. H, co-immunoprecipitation of HSPB7 WT but not HSPB7 ⌬73 with mHttQ74 -GFP. Western blotting with the indicated antibodies is shown. All experiments were repeated at least three times. Data are presented as mean Ϯ S.D. *, p Ͻ 0.05; **, p Ͻ 0.01.

Role of NTD in HSPB7 function
sufficient for association of small HSPB proteins with polyQ proteins. This was confirmed by GFP pulldown experiments with mHttQ74 -GFP. Whereas HSPB7 WT was clearly co-precipitating with mHttQ74 -GFP, HSPB7 ⌬73 did not (Fig. 1H). Inversely, whereas HSPB1 WT did not co-IP with mHttQ74 -GFP, NTD HSPB7 -HSPB1 did (Fig. 3I). These data confirm that the NTD of HSPB7 drives the interaction between HSPB7 and polyQ proteins. Because the interaction in our co-IP experiments is performed on soluble material, these data further suggest that the N terminus of HSPB7 can already bind to polyQ prior to its insolubilization into high molecular weight aggregates. This mode of action is analogous to those of other small HSPs that need to be present prior to aggregation to be most active as well (7,33). Our data suggest that the flanking regions, here the NTD, are required and sufficient for targeting small HSPs to their substrates and that the ACD enables the suppression of aggregation.

Shifting the HSPB oligomeric status is not sufficient for polyQ aggregation prevention
How can the NTD of HSPB7 turn HSPB1 into a protein that now can bind to polyQ proteins and suppress their aggregation? Because, HSPB7 does not form oligomers like the canonical small HSPs including HSPB1 (8), we determined the role of the HSPB7-NTD in the oligomerization process. Sucrose gradient centrifugation of whole cell extracts confirmed that WT HSPB1 is polydispersed (Fig. 4A), reminiscent of being present as dynamic oligomers in cells. In contrast, WT HSPB7 predominantly runs in a monodispersed peak, likely representing dimers. (Fig. 4A). Deletion of the NTD from HSPB7 (HSPB7 ⌬73 ) does not change this behavior (Fig. 4A), implying that loss of its anti-aggregation activity is not related to a change into a more polydispersed form. Importantly, fusion of the HSPB7-NTD to HSPB1 (NTD HSPB7 -HSPB1) has a strong impact on the oligomeric properties of HSPB1. The NTD HSPB7 -HSPB1 chimera was found to sedi-

Role of NTD in HSPB7 function
ment in the same fractions as HSPB7, representing smaller oligomeric species such as dimers (Fig. 4A). These data indicate that the presence of HSPB7-NTD inhibits the association of the HSPB to higher oligomeric assemblies and might point to an association between the oligomeric status of HSPBs and their ability to prevent polyQ aggregation.
To test whether the shift in oligomerization behavior suffices to turn HSPB1 into a polyQ-aggregation preventing chaperone, we used HSPB1-phosphorylation mutants that also impede the oligomeric status. HSPB1 can be phosphorylated at three serines, which drives its de-oligomerization (34,35). Mutation of these three serines (S) into aspartic acid (D) (HSPB1 DDD ) simulates the oligomeric behavior of this phosphorylated form of HSPB1 (34) (Fig. 4B, Fig. S3E), whereas mutating these serines into alanines (HSPB1 AAA ) simulate the un-phosphorylated form of HSPB1 forming, on average, larger oligomeric complexes (34) (Fig. 4B, Fig. S3F). However, expression of neither HSPB1 DDD nor HSPB1 AAA was found to reduce mHttQ74 -GFP aggregation (Fig. 4C, Fig. S3F). These results suggest that de-oligomerization alone is not sufficient for HSPB1 to prevent polyQ aggregation and pinpoints the need for the presence of the HspB7-NTD for polyQ protein recognition.

The ability of HSPB7 to prevent polyQ aggregation depends on the regions flanking the polyQ stretch
The polyQ stretch in (mutant) huntingtin is flanked by a 17-amino acid long N-terminal (N17) domain (36) and a proline-rich region (PRR) C-terminal of the polyQ stretch. These flanking regions not only affect the propensity of aggregation of the polyQ protein, but also affect how chaperones may or may not be able to delay the onset of aggregation (10,36,37). Whereas some chaperones like DNAJB6 can prevent initiation of aggregation independently of such flanking regions by directly interacting with the amyloidogenic polyQ region (10), other chaperones like TRiC mediate their anti-aggregation effects via binding to the N17 domain of Huntingtin (10,38). To test how HSPB7 acts, we co-expressed it with 4 different mHtt constructs (Fig. 5, A-D) one containing both the intact N17 domain and PRR, one with a truncated N17, one with a truncated PRR, or one in which both the N17 and PPR are truncated. HSPB7 could suppress the aggregation of all these variants, except the one with a truncation of both the N17 domain and the PPR constructs (Fig. 5, A-D). This implies that HSPB7 does not directly act on the amyloidogenic polyQ region but requires either the N17 domain and/or the PPR to mediate its anti-aggregation effects.

Discussion
Here we show that HSPB7, a noncanonical member of the human HSPB family of small heat shock proteins, prevents aggregate formation by directly acting on polyQ containing proteins via its unique N-terminal domain. This NTD facilitates binding of HSPB7 to polyQ proteins. Intriguingly, the NTD is sufficient to provide another member of the HSPB family, HSPB1, with the capacity to bind to and suppress aggregation of polyQ proteins.
HSPB7, unlike canonical HSPBs, is not forming large oligomeric species (8). Although replacing the NTD of HSPB1 by the NTD of HSPB7 leads to a reduction in the average oligomeric size of the hybrid protein, de-oligomerization alone was found insufficient to confer anti-polyQ aggregation activity to HSPB1. Rather, the NTD of HSPB7 seems to confer substrate binding capacity to HSPBs as its deletion from HSPB7 (HSPB7 ⌬73 ) lead to a loss of binding and suppression capacity. Currently it is unclear to which polyQ species HSPB7 is binding. Our co-IP experiments, performed using soluble material, suggest that HSPB7 binds to early profibrillar species. The fact that HSPB7 is associated with the (later) polyQ inclusions (IF) confirms its capacity to bind to polyQ species, but may also reflect a failed function.

Role of NTD in HSPB7 function
Interestingly, the NTD of HSPB7 is intrinsically disordered (27), characteristic of proteins involved in phase separation (39,40). Although we did not observe any HSPB7-mediated demixing, our results do not rule out a role of the NTD in phase separation. The unique properties of the NTD of HSPB7 in the context of an ACD are clearly sufficient to drive inhibition of polyQ aggregation. The notion that this suppressive nature gradually declines with the progressive shortening of the NTD (Fig. 1) is arguing against the presence of specific structural motifs within the NTD. Indeed for other intrinsically disordered regions, it has been shown that rather than amino acid order and chemical composition, function is determined by the length of the IDR (40). These results further underscore the possible IDR-like characteristics of the NTD of HSPB7 being crucial for its anti-amyloidogenic activity. Yet, these features alone do not suffice to confer such activity as fusion of the NTD of HSPB7 to polyQ proteins directly did not reduce but rather enhanced polyQ aggregation (Fig. 2, A-D).
Our combined data suggest that the NTD of HSPB7 drives a de-oligomerization of HSPB proteins that causes exposure of its IDR-like features that next drives binding of the HSPB to early intermediate aggregates of polyQ proteins, by interacting with regions flanking the amyloidogenic polyQ core. Through such interactions further oligomerization into amyloid like fibers is delayed (41). Alternatively early formed protofibrils are stabilized by HSPB7 to inhibit the rate of the aggregation process (41,42). Irrespective of either one of these mechanisms, aggregate growth is delayed, which may facilitate their engulfment into autophagosomes to facilitate aggrephagy (8,15,16).

In vitro aggregation of mHtt
Purified HSPB1 and HSPB7 (9) and a mHttQ48 exon-1 fragment containing the N17 domain, 48 glutamines, and the PPR protein (43) were used for in vitro aggregate formation (9). In short, 20 M mHttQ48 was incubated with 20 M sHSPB at 37°C for 0, 1, 2, 3, 4, and 5 h in PBS (pH 7.4) containing 1 mM DTT. After incubation, 8 l of each sample was mixed with the same amount of denaturing buffer (180 mM Tris-HCl, pH 6.8, 30% glycerol, 15% ␤-mercaptoethanol, 6% SDS) (43) and then immediately boiled at 95°C for 5 min. Samples were stored at Ϫ140°C until use. Samples were boiled before loading onto a 12% SDS-PAGE gel. After electrophoresis, the gel was stained with SYPRO Orange (Sigma). Images were analyzed using a ChemiDoc Touch Imaging system (Bio-Rad). Hereafter, the gel was washed in washing buffer (25 mM Tris base, 190 mM glycine, and 0.1% SDS, pH 8.3) three times, followed by transfer onto a nitrocellulose membrane for regular Western blotting analyses.

Sucrose gradient centrifugation
Sucrose gradient analysis was done as reported before (8). Briefly, transfected HEK293 cells were washed with PBS and lysed in 200 l of Nonidet P-40 lysis buffer (50 mM HEPES, 60 mM KCl, 2 mM MgCl 2 , 10% glycerol, 0.4% Nonidet P-40). Cell lysates were cleared through a 26-gauge syringe 10 times, followed by centrifugation at 100,000 ϫ g for 16 h using 20 -60% sucrose gradient column in a ultraspeed centrifuge using a SW41 rotor. 700-l fractions were precipitated by adding equal amounts of 25% trichloroacatic acid. The pellets were washed twice using 80% acetone (Ϫ20°C) and allowed to dry. The pellet was dissolved in 20 l of 0.1 M NaOH containing 1% SDS. An equal amount of 2ϫ Laemmli buffer was added and samples were boiled for 5 min before Western blot analysis.

Immunoblotting
For Western blotting, cells were lysed in FTA sample buffer (10 mM Tris-HCl, 150 mM NaCl, 2% SDS, pH 8.0) and boiled with an equal amount of 2ϫ Laemmli buffer. Samples were then loaded and run on a 12% SDS-PAGE gel and transferred (Bio-Rad). After transfer, membranes were blocked and incubated with primary and secondary antibodies followed by detection using the ChemiDoc Touch Imaging system (Bio-Rad).

Role of NTD in HSPB7 function
plates. Cells in each well were lysed in 200 l of FTA sample buffer (10 mM Tris-Cl, pH 8.0, 150 mM NaCl, 2% SDS). 100-, 20-, and 4-g samples were diluted in FTA sample buffer supplied with 50 mM DTT. Boiled samples were then applied onto a pre-washed (by FTA wash buffer, 10 mM Tris-Cl, pH 8.0, 150 mM NaCl, 0.1% SDS) 0.2 M cellulose acetate filter (GE Water and Process Technologies, Trevose, PA) with 2 Whatman papers (Bio-Rad) in a Bio-Dot microfiltration apparatus (Bio-Rad). Gentle suction was applied to filtrate the samples followed by three washes using FTA wash buffer. After blocking the membrane for 1 h in 5% milk, incubation with primary (overnight) and secondary antibodies (1 h), the membrane was exposed using the ChemiDoc Touch Imaging system (Bio-Rad).
Immunofluorescence 48 h after transfection, cells on coverslips were fixed in 2% formaldehyde for 15 min, washed 2 times with PBS (pH 7.5), and permeabilized with PBS-Triton X-100 (0.1%). Then cells were washed with PBS and with PBS plus (PBS with 0.5% BSA, 0.15% glycine), followed by primary antibody incubation overnight at 4°C. Then cells on coverslips were washed 4 times with PBS plus and incubated with secondary antibody. After washing with PBS plus and PBS, DNA was stained with 4Ј,6-diamidino-2-phenylindole (Thermo Fisher Scientific) for 5 min followed by washing with PBS. Coverslips were mounted in Vectashield (Agar Scientific). Mouse anti-v5 primary antibody (Invitrogen) at a 1:10,000 dilution, Alexa 594 anti-mouse (Invitrogen) secondary antibody was diluted as 1:1,500. Images were taken by a Leica sp8 confocal microscope and edited using Fiji.

Statistical analysis
The software Curve Expert 1.3 was used to make a standard curve for measuring protein concentration. Differences among the experimental groups were analyzed by one-way analysis of variance (p Ͻ 0.05 and p Ͻ 0.01). Values were considered to differ significantly when p Ͻ 0.05. All data are expressed as the mean Ϯ 1 S.D. All experiments were repeated at least three times.