HSPA8 regulates anti-bacterial autophagy through liquid-liquid phase separation

ABSTRACT

HSPA8 (heat shock protein family A (Hsp70) member 8) plays a significant role in the autophagic degradation of proteins, however, its effect on protein stabilization and anti-bacterial autophagy remains unknown. Here, it is discovered that HSPA8, as a binding partner of RHOB and BECN1, induce autophagy for intracellular bacteria clearance. Using its NBD and LID domains, HSPA8 physically binds to RHOB residues 1–42 and 89–118 as well as to BECN1 ECD domain, preventing RHOB and BECN1 degradation. Intriguingly, HSPA8 contains predicted intrinsically disordered regions (IDRs), and drives liquid-liquid phase separation (LLPS) to concentrate RHOB and BECN1 into HSPA8-formed liquid-phase droplets, resulting in improved RHOB and BECN1 interactions. Our study reveals a novel role and mechanism of HSPA8 in modulating anti-bacterial autophagy, and highlights the effect of LLPS-related HSPA8-RHOB-BECN1 complex on enhancing protein interaction and stabilization, which improves the understanding of autophagy-mediated defense against bacteria.

Introduction

HSPA8 (heat shock protein family A (Hsp70) member 8), known as a cognate protein of the HSPA/HSP70 family, is a conserved and constitutively expressed chaperone in the cell membrane, cytosol and nucleus. It regulates the cellular proteostasis by mediating protein assembly, refolding, trafficking and degradation [1,2]. Four functional domains of HSPA8 have been identified: the nucleotide binding domain (NBD, residues 1–387) undergoes conformational changes through binding of ATP [3]; the substrate binding domain (SBD, residues 388–509) dominates the binding or release of the client peptide/proteins [4]; the LID domain (residues 510–606) located at the end of SBD raises its binding affinity and reduces the exchange rate [5]; C-terminal domain (CTD, residues 606–646) holds EEVD motif and mediates the interactions with accessory and co-chaperones proteins as a flexible region [6]. The interacting regions of HSPA8 with client proteins are diverse, leading to the distinct roles that affect the client turnover [2,7,8]. However, the role and mechanism of HSPA8 in facilitating protein stabilization are still unclear.

Autophagy, known as a degradation process that maintains cellular homeostasis, is involved in anti-bacterial defense [9]. Downstream factors are induced upon bacterial infections and regulate autophagic pathways, resulting in recognition and elimination of invasive bacteria, such as uropathogenic Escherichia coli (UPEC) [10], Salmonella typhimurium [11–13], Mycobacterium tuberculosis [14], Shigella flexneri [15]. We previously found that RHOB, an infection-induced small GTPase, played a role in autophagic degradation against UPEC [10]. It is known that HSPA8 recognizes and delivers KFERQ-motif containing proteins with helps of co-chaperones (including J-domain [16], tetratricopeptide-repeat (TPR) domain [17], and the nucleotide exchange factors (NEFs) [18]] for degradation via macroautophagy [19], chaperone-mediated autophagy (CMA) [20] or endosomal microautophagy (eMI) pathways [21,22]. Recent studies indicate that HSPA8 binds to ubiquitin ligases in the process of ubiquitin-proteasome degradation of client proteins [19,23,24]. However, the effect of HSPA8 in regulating anti-bacterial autophagy and how HSPA8 manipulates autophagy factors are unknown.

Recently, the role of liquid – liquid phase separation (LLPS) in assembling concentrated functional molecules in liquid-like droplets, was highlighted [25–28]. Several results indicate that LLPS represents a suitable environment, in which the dynamic turnover or reconstruction of protein assembly is rapidly modulated for downstream signal transduction [25,29]. There is currently no report regarding the role of HSPA8 in driving LLPS.

HSPA8, a novel binding partner of RHOB, was identified in this study, which promotes anti-bacterial autophagy in vitro and in vivo. HSPA8 physically associated with RHOB and BECN1/Beclin 1, concentrating the BECN1 into liquid-like droplets formed by HSPA8 to prevent their degradation. The interacting domains of HSPA8-RHOB-BECN1 were characterized in detail, as were the domains of LLPS modeling. Our study reveals a novel mechanism by which HSPA8 modulates anti-bacterial autophagy, and highlights the LLPS-related physical association between HSPA8, RHOB and BECN1 to achieve protein stabilization and autophagy effects.

Results

HSPA8 physically interacts with RHOB and promotes autophagy

To explore proteins involved in the RHOB-related complex that promotes intracellular UPEC clearance, we applied affinity purification and mass spectrometry analysis to identify the RHOB-interacted proteins. FLAG-tagged RHOB was overexpressed in HEK293 cells and purified with an anti-FLAG M2 affinity gel. The bound proteins were subjected to SDS-PAGE for silver staining and analyzed by mass spectrometry (Table S1). Among the proteins co-immunoprecipitated with RHOB, HSPA8 [2,30,31], PHGDH (phosphoglycerate dehydrogenase) [32], PRDX2 (peroxiredoxin 2) [33,34] and S100A9 (S100 calcium binding protein A9) [35–37], have been reported to be related with autophagy (Figure 1A). We further found that overexpression of HSPA8 but not PHGDH, PRDX2 and S100A9, induced accumulation of LC3-II and BECN1 in HEK293 cells using western blotting (Figure 1B and S1A). We also observed that HSPA8 overexpression promoted autophagosomes, autolysosomes, and the autophagy flux, compared to that of vector, PHGDH, PRDX2, and S100A9 in mCherry-EGFP-LC3-expressing 293 cells using fluorescence microscopy analysis (Figure S1B). The treatment with bafilomycin A₁ (BafA₁), an inhibitor of autophagosome-lysosome fusion, further increased the number of autophagosomes in the HSPA8 group, suggesting that HSPA8 promotes autophagy [38,39]. In addition, HSPA8 dosage-dependently accelerated BECN1 and LC3-II upregulation with or without CQ (an inhibitor of autolysosomes) treatment (Figure 1B). Moreover, we examined the role of HSPA8 in autophagy in SW480 cells (human colon epithelial cells) infected by Salmonella typhi LT2 and 5637 cells (bladder epithelial cells) infected by uropathogenic Escherichia coli CFT073, which induce autophagy as previously reported [10] (Figure S1C). We observed that HSPA8 knockdown decreased the levels of LC3-II and BECN1, but increased the SQSTM1 level in bacteria-infected cells (Figure 1C and S1D). Accordingly, HSPA8 knockdown decreased the number of autophagosomes and autolysosomes in LT2-infected SW480 cells and CFT073-infected 5637 cells with or without the treatment of BafA1 (Figure S1E-F). Taken together, these results suggest that HSPA8 induces autophagic flux during bacterial infection.

Figure 1.

HSPA8 is physically associated with RHOB and promotes autophagy. (A) Immunoprecipitation and mass spectrometry analysis of RHOB-containing protein complexes. 3×FLAG-RHOB was transiently expressed in HEK293 cells and immunopurified with anti-FLAG affinity gel, followed by elusion with 3×FLAG peptide. 3×FLAG-vector served as the negative control. The eluate was then subjected to SDS-PAGE and silver staining. The protein bands arrowed were analyzed by mass spectrometry. (B) HSPA8 overexpression upregulates levels of BECN1 and LC3-II in HEK293 cells in a dose-dependent manner. HEK293 cells were transfected with 3×FLAG-HSPA8 or vector, followed by treatments of 10 μM chloroquine (CQ) overnight at 24 h post transfection. Cell lysates were made and subjected to western blotting analysis. Signal densities of LC3-II and BECN1 were normalized to that of ACTB. The relative density of vector-transfected cells was set to 100%. n = 3 independent experiments. LC3-II:ACTB: 0 vs. 500 P = 0.9403, 0 vs. 1500 P = 0.0011, CQ 0 vs. CQ 500 P = 0.0275, CQ 0 vs. CQ 1500 P = 0.0117; BECN1:ACTB: 0 vs. 500 P = 0.0228, 0 vs. 1500 P < 0.0001, CQ 0 vs. CQ 500 P = 0.0017, CQ 0 vs. CQ 1500 P = 0.0001. (C) Expression of autophagic proteins in HSPA8 knockdown SW480 cells with LT2 infection. SW480 cells were transfected with HSPA8 siRnas or scramble siRNA for 48 h followed by infection of LT2 at MOI of 100 for 4 h. Signal densities of LC3-II, SQSTM1 and BECN1 were normalized to that of ACTB. The relative density of scramble siRNA-transfected cells was set to 100%. n = 3 independent experiments. LC3-II:ACTB: siScr vs. siHSPA8 #1 P = 0.0347, siScr vs. siHSPA8 #2 P = 0.0019, siScr vs. siHSPA8 #3 P = 0.0256; SQSTM1:ACTB: siScr vs. siHSPA8 #1 P = 0.1574, siScr vs. siHSPA8 #2 P = 0.0076, siScr vs. siHSPA8 #3 P = 0.6113; BECN1:ACTB: siScr vs. siHSPA8 #1 P = 0.0597, siScr vs. siHSPA8 #2 P = 0.0057, siScr vs. siHSPA8 #3 P = 0.0536. (D and E) the interaction of overexpressed HSPA8 with RHOB in HEK293 cells. HEK293 cells were co-transfected with MYC-RHOB, 3×FLAG-HSPA8 or vector. At 48 h post transfection, cell lysates were prepared and subjected to immunoprecipitation with anti-FLAG (D) or anti-MYC (E) gels followed by western blotting analysis. n = 3 independent experiments. (F and G), Reciprocal immunoprecipitation of endogenous HSPA8 and RHOB in SW480 cells. SW480 cells were treated with LPS for 2 h to increase RHOB expression and lysed. Cell lysates were subjected to immunoprecipitation with anti-RHOB (F) or anti-HSPA8 (G) antibodies plus protein A/G agarose, and then analyzed by western blotting. Normal IgG served as a negative control. n = 3 independent experiments. (H and I) HSPA8 physically interacts with RHOB. Purified HA-HSPA8 and FLAG-RHOB were mixed and immunoprecipitated with anti-FLAG (H) or anti-HA (I) affinity gels, and then analyzed by western blotting. n = 3 independent experiments. Data are shown as the mean ± SD, *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001, one-way ANOVA (B and C).

To confirm the association between HSPA8 and RHOB, the co-immunoprecipitation assay was performed using HEK293 cells overexpressing MYC-tagged RHOB, FLAG-tagged HSPA8, or vector. The results showed that HSPA8 was efficiently co-immunoprecipitated with RHOB, but not with vector, and vice versa (Figure 1D,E). Similar results were also observed in 5637 cells (Figure S1G). This in vivo association between HSPA8 and RHOB was then verified with endogenous proteins from LPS-treated SW480 cells in a reciprocal way (Figure 1F,G). To determine whether HSPA8 directly interacts with RHOB, purified HA-HSPA8 and FLAG-RHOB were subjected to the co-immunoprecipitation assay, and RHOB was found to efficiently co-immunoprecipitate with HSPA8 and vice versa (Figure 1H,I and S1H).

HSPA8 restricts intracellular bacteria through autophagy

Selective autophagy has been found important for host defending against bacterial infection. To explore the role of HSPA8 during bacterial infection, endogenous HSPA8 was knocked down by specific siRNAs, which led to a significant increase of intercellular LT2 in SW480 cells and CFT073 in 5637 cells (Figure 2A and S2A). Immunofluorescent staining showed a higher number of LT2 within HSPA8 knockdown SW480 cells, whereas similar results were also obtained in 5637 cells infected by CFT073 (Figure 2B and S2B). In addition, HSPA8-overexpression in SW480 or 5637 cells decreased the number of intracellular LT2 or CFT073 (Figure 2C and S2C). However, the bacterial entry and expulsion assays indicated that HSPA8 knockdown had no effect on the entry of LT2 or CFT073 into host cells, but increased extracellular LT2 or CFT073 (Figure 2D-E and S2D-E). Overall, these results revealed that HSPA8 restricts intracellular bacteria after bacterial entry.

Figure 2.

HSPA8 restricts LT2 infection. Scramble or HSPA8 siRnas-transfected SW480 cells were infected by LT2, and treated with gentamycin and/or inhibitors as described in Materials and Methods, MOI = 100. (A) Knockdown of HSPA8 increases intracellular LT2 in SW480 cells. n = 3 independent experiments. siScr vs. siHSPA8 #1 P = 0.3144, siScr vs. siHSPA8 #2 P = 0.0013, siScr vs. siHSPA8 #3 P = 0.5591. (B) Representative image (left) and quantification (right) of internalized LT2 in SW480 cells transfected with HSPA8 siRNA at 4 hpi. n = 4 random areas assessed over 3 independent experiments (more than 100 cells for each group). Scale bar: 10 μm. P = 0.0286. (C) HSPA8 overexpression decreases intracellular LT2 in SW480 cells at 4 hpi. n = 4 independent experiments. P = 0.0286. (D) Effect of HSPA8 on entry of LT2 in SW480 cells. Infected cells were incubated in 37°C for 30 min followed by washing and treatments with gentamycin. At 30 min post gentamycin treatment, infected cells were lysed and CFU of intracellular bacteria were measured. P = 0.1212. n = 6 independent experiments. (E) Effect of HSPA8 on bacterial expulsion of LT2 from SW480 cells. At 4 hpi, the number of extracellular LT2 in culture media was measured. n = 6 independent experiments. P = 0.0087. (F and G) Representative confocal images of GFP-LT2 colocalization with SQSTM1, ubiquitin or mCherry-LC3 in HSPA8 siRNA-transfected SW480 cells. Infected cells were pre-transfected with mCherry-LC3 and HSPA8 siRnas, and then infected with GFP-expressing LT2, followed by immunostaining with anti-SQSTM1 (F) or anti-ubiquitin antibodies (G) at 4 hpi. Scale bars, 10 μm. Quantification of SQSTM1-, ubiquitin- or LC3-positive LT2 in SW480 cells was shown on the right. n = 3 independent experiments. (F) SQSTM1⁺ P = 0.9999, LC3⁺ P = 0.0025, (G) UB⁺ P = 0.7960, LC3⁺ P = 0.0124. Data are shown as the mean ± SD, *P < 0.05, **P < 0.01, one-way ANOVA (A), two-tailed unpaired Student’s t test (B-E), two-way ANOVA (F and G).

To investigate whether HSPA8 promotes bacterial clearance through autophagy, we examined the role of HSPA8 in bacterial LC3-decoration. Interestingly, we observed less LC3-decorated LT2 in HSPA8 knockdown cells (Figure 2F,G). However, the colocalization of ubiquitin or SQSTM1 with intracellular bacteria was not significantly affected, suggesting HSPA8 has no effect on recognition of invading bacteria (Figure 2F,G). We also knocked down BECN1, SQSTM1, ATG5, RUBCN, or LAMP2 by specific siRNAs to reduce the formation of autophagosomes, LC3-associated phagocytosis (LAP) or chaperone-mediated autophagy (CMA), respectively, in LT2 or CFT073 infected cells. We found that knockdown of BECN1, ATG5, or SQSTM1 attenuated HSPA8-mediated clearance of intracellular LT2, compared to that of scramble siRNAs, while RUBCN and LAMP2 knockdown had no effect (Figure S3A-G). These results were also confirmed in CFT073 infected 5637 cells overexpressing FLAG-HSPA8 (Figure S3H – K). Therefore, it is likely that HSPA8 promotes intracellular LT2 and UPEC clearance through macroautophagy pathways.

HSPA8 physically interacts with BECN1 and plays a key role in RHOB-mediated autophagy

As we found that HSPA8 dosage-dependently upregulated the level of BECN1 in HEK293 cells (Figure 1B), we examined whether HSPA8 interacts with BECN1. Immunoprecipitation assays with overexpressed MYC-BECN1 and FLAG-HSPA8 indicated that HSPA8 was efficiently co-immunoprecipitated with BECN1, but not with vector, and vice versa (Figure 3A,B). The in vivo association of HSPA8 and BECN1 was also reciprocally verified with endogenous proteins from SW480 cells (Figure 3C,D). Moreover, purified GST-fused BECN1 and HA-HSPA8 confirmed that HSPA8 could directly interact with BECN1, and vice versa (Figure 3E,F). In addition, immunofluorescent staining showed that colocalization of HSPA8 and BECN1 was observed in SW480 cells with or without LT2 infection, and LT2 infection did not affect the expression of HSPA8 (Figure 3G). These results suggest that HSPA8 indeed interacts with BECN1.

Figure 3.

HSPA8 physically interacts with BECN1 and governs its stabilization. (A and B) the interaction of overexpressed HSPA8 with BECN1 in HEK293 cells. HEK293 cells were co-transfected with 3×FLAG-HSPA8, MYC-BECN1 or vector. At 48 h post transfection, cell lysates were prepared and subjected to immunoprecipitation with anti-FLAG (A) or anti-MYC (B) affinity gel followed by western blotting analysis. n = 3 independent experiments. (C and D) Reciprocal immunoprecipitation of endogenous BECN1 and HSPA8 in SW480 cells. Lysates of SW480 cells were immunoprecipitated with anti-BECN1 (C) or anti-HSPA8 (D) antibodies plus protein A/G agarose and analyzed by western blotting. IgG served as a negative control. n = 3 independent experiments. (E and F) HSPA8 directly interacts with BECN1. HA-HSPA8 and GST-MYC-BECN1 were expressed in E. coli and subjected to immunoprecipitation assay with anti-HA affinity gel (E) or anti-GST Sepharose (F), which was then analyzed by western blotting. n = 3 independent experiments. (G) Representative confocal images of LT2-infected SW480 cell stained for endogenous HSPA8 and BECN1 (left). Quantification of HSPA8 and BECN1 levels were shown (right bottom). Fluorescence intensity profile along the white line were shown (right top). SW480 cells were infected with LT2 for 4 h, and then fixed, permeabilized and stained with antibodies against HSPA8 and BECN1. Images were taken by confocal microscope. Scale bars, 10 μm. HSPA8: Control vs. LT2 P = 0.8825, siScr vs. siHSP8 P < 0.0001; BECN1: Control vs. LT2 P = 0.0012, siScr vs. siHSPA8 P = 0.0113. n = 3 random areas assessed over 3 independent experiments. (H) Effect of HSPA8 knockdown on RHOB-mediated upregulation of BECN1 and LC3 lipidation. SW480 cells were co-transfected with 3×FLAG-RHOB and HSPA8 siRNA or scramble siRNA for 48 h. (I) Effect of RHOB knockdown on HSPA8-mediated upregulation of BECN1 and LC3 lipidation. SW480 cells were co-transfected with 3×FLAG-HSPA8 and RHOB siRNA or scramble siRNA. At 48 h post transfection, cell lysates were prepared and subjected to western blotting analysis (H and I). (J) Effect of HSPA8 knockdown or HSPA8 overexpression on RHOB-mediated clearance of intracellular LT2 in SW480 cells. The number of intracellular bacteria in scramble siRNA or vector-transfected cells was set to 100%. siScr: siScr vs. siRHOB P = 0.0005; siHSPA8: siScr vs. siRHOB P = 0.9785; Vector: siScr vs. siRHOB P = 0.0497; HSPA8: siScr vs. siRHOB P = 0.0077. (K) Effect of HSPA8 knockdown on LT2-inducted upregulation of RHOB, BECN1 and LC3 lipidation. SW480 cells were transfected with HSPA8 siRNA or scramble siRNA for 24 h, and then infected by LT2 for indicated times. MOI = 100. The protein expression at indicated time point was analyzed by western blotting. n = 3 independent experiments. (L and M) Fluorescent staining of RHOB, BECN1 and HSPA8 in LT2-infected SW480 with HSPA8 (L) or RHOB (M) siRnas. SW480 were pre-transfected with HSPA8, RHOB or scramble siRnas for 48 h, and infected by LT2 for 4 h. Infected cells were then fixed and subjected to immunostaining with anti-RHOB (L), anti-HSPA8 (M), and anti-BECN1 antibodies. Quantification data were analyzed using Image Pro Plus software (right). Fluorescence intensity profiles along the white line were shown (bottom). n = 3 random areas per group assessed over 3 independent experiments. Scale bar: 10 μm. (L) RHOB P = 0.0103, BECN1 P = 0.0099. (M) HSPA8 P = 0.5673, BECN1 P = 0.0064. Data are shown as the mean ± SD, *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001, two-way ANOVA (G, L-M), one-way ANOVA (J).

Then we stepped forward to investigate whether HSPA8 participated in RHOB-mediated upregulation of BECN1. Interestingly, HSPA8 knockdown greatly attenuated the RHOB-overexpression-induced BECN1 upregulation and LC3 lipidation in SW480 cells (Figure 3H). Conversely, HSPA8 overexpression in HEK293 cells enhanced the effect of RHOB-overexpression on BECN1 upregulation and LC3 lipidation (Figure S4A). However, RHOB knockdown seems to generally decrease the level of BECN1 and LC3 lipidation, but not influence the effects of either HSPA8 overexpression or knockdown on BECN1 level and LC3 lipidation in SW480 cells (Figure 3I and S4B). Similarly, the RHOB effect on bacterial clearance was abolished by HSPA8 knockdown but slightly enhanced by HSPA8 overexpression (Figure 3J). Of note, the level of transfected RHOB was also decreased in HSPA8 knocked down cells but increased in HSPA8 overexpressed cells (Figure 3H and S4A), suggesting that both RHOB and BECN1 are under the protection of HSPA8. Overall, these results suggest that HSPA8 could increase BECN1 level and LC3 lipidation independently of RHOB, but the effects of RHOB on BECN1 and LC3 lipidation depend on HSPA8.

HSPA8 increases BECN1 and RHOB stabilization through blockage of their K48-mediated ubiquitination

We further examined the role of HSPA8 in BECN1 and RHOB levels. HSPA8 overexpression dosage-dependently increased levels of BECN1, RHOB, and LC3 lipidation with or without CQ treatment in SW480 cells (Figure S4C). Accordingly, HSPA8 knockdown decreased the induction of BECN1, RHOB, and LC3 lipidation raised by LT2 infection in SW480 cells (Figure 3K). Of note, while RHOB is induced during LT2 infection, HSPA8 is highly expressed with or without LT2 infection (Figure 3K). In addition, immunofluorescent assays in LT2 infected SW480 cells revealed that HSPA8 knockdown reduced RHOB and BECN1 levels, but RHOB knockdown did not affect HSPA8 level (Figure 3L,M). Therefore, we asked how HSPA8 affected levels of BECN1 and RHOB. No significant change was observed in both BECN1 and RHOB mRNA levels when HSPA8 was overexpressed or knocked down in HEK293 or SW480 cells (Figure S4D). The degradation of BECN1 and RHOB was greatly accelerated in HSPA8 knockdown SW480 cells when protein synthetization was inhibited by cycloheximide (CHX), compared to those of control cells (Figure S4E). Conversely, inhibition of proteasomal degradation by MG132 sufficiently restored levels of RHOB and BECN1 in HSPA8 knockdown SW480 cells (Figure S4F). HSP90 (Heat shock 90kDa proteins) was reported to participate in the maintenance of BECN1 [10,40]. However, geldanamycin (GA; an inhibitor of HSP90) had no effect on HSPA8-mediated BECN1 reduction or upregulation (Figure S4G-H). Interestingly, blockage of ubiquitination by TAK−243 (an inhibitor of the ubiquitin activating enzyme) abolished the reduction of RHOB and BECN1 caused by HSPA8 knockdown (Figure S4I). Moreover, we also observed that the K48- but not K11-mediated ubiquitination of RHOB and BECN1 was significantly increased in HSPA8 knockdown HEK293 cells (Figure S4J-M). As a summary, these results indicate that HSPA8 prevents both BECN1 and RHOB degradation but has no effect on their transcription or synthesis.

Molecular interactions between HSPA8, BECN1 and RHOB

To map the interacting domains between HSPA8, BECN1 and RHOB, truncation mutants of HSPA8 (Figure 4A), BECN1 (Figure 4B) and RHOB (Figure 4E) were generated. GST-pull down and immunoprecipitation assays with purified GST-tagged proteins indicated that HSPA8 1–387 (the NBD domain) is required for binding to BECN1 (Figure 4C), whereas BECN1 residues 244–337 (the ECD domain) are essential (Figure 4D). Reciprocally, domains of HSPA8 (1–387 and 510–605, the NBD and LID domain) bound to domains of RHOB (1–42 and 89–118) (Figure 4F,G). These interactions results were then confirmed by isothermal titration calorimetry (ITC) experiments (Figure 5A-E, and S5A-B). Accordingly, both IP and ITC results pointed out that HSPA8 1–387 binds to RHOB 1–42 or BECN1 244–337, whereas HSPA8 510–605 binds to RHOB 89–118 (Figure 5A-H). The presence of the protein complex HSPA8-RHOB-BECN1 was observed in LPS-treated 5637 cells (Figure 5I). Overexpression of these mutants containing HSPA8 1–387 significantly increases the formation of autophagosomes and autolysosomes in HEK293 cells (Figure S5C). Accordingly, these mutants decreased intracellular LT2 in SW480 cells compared to others (Figure S5D). Overall, these findings suggest that HSPA8, BECN1 and RHOB form a complex through direct interactions and regulate bacterial clearance activities in an autophagy-dependent way.

Figure 4.

Determination of interacted domains between HSPA8, BECN1 and RHOB. (A and B) domain architecture of truncated GST-HSPA8 (A) and GST-BECN1 (B) constructs. GST: GST-tag; NBD: nucleotide-binding domain; SBD, substrate-binding domain; LID, lid subdomain; CTD: C-terminal domain; BH3: Bcl −2 homology − 3 domain; CCD, coil-coil domain; ECD, evolutionarily conserved domain. (C and D) GST-pull down experiments with truncated GST-fused HSPA8 and BECN1. Indicated HSPA8 and BECN1 proteins were purified, mixed and subjected to immunoprecipitation with anti-GST Sepharose. Bound BECN1 and HSPA8 were analyzed by immunoblotting with anti-BECN1, GST or HSPA8 antibodies (shown as the slanting labels). Signal densities of bound proteins was normalized in IP group. Percentage of bound wild-type protein was set to 100%. n = 3 independent experiments. (C) BECN1:GST-HSPA8: 1–605 P = 0.9971, 1–509 P = 0.6818, 1–396 P = 0.4441, 1–387 P = 0.9996, 388–646 P = 0.0482, 397–646 P = 0.0288, 510–646 P = 0.0274, 606–646 P = 0.0230. (D) HSPA8:GST-BECN1: 1–88 P = 0.0002, 1–150 P = 0.0002, 1–244 P = 0.0002, 1–337 P = 0.9996, 88–450 P = 0.9999, 150–450 P = 0.2752, 244–450 P = 0.1446, 337–450 P = 0.0002. (E) domain architecture of truncated GST-RHOB constructs. GST: GST-tag; C, C-terminal domain. (F and G) Reciprocal immunoprecipitation assay of truncated RHOB and HSPA8 constructs. Indicated proteins were purified, mixed and subjected to immunoprecipitation with anti-FLAG gel. Bound proteins were exchanged with 3×FLAG peptides and analyzed by immunoblotting with anti-GST, HSPA8 or FLAG antibodies (shown as the slanting labels). Signal densities of HSPA8 constructs were normalized by those of RHOB constructs in IP group. Percentage of bound HSPA8 wild-type to RHOB wild-type was set to 100%. n = 3 independent experiments. (F) GST-HA-HSPA8:FLAG-RHOB: 1–605 P = 0.4312, 1–509 P = 0.9614, 1–396 P = 0.9769, 1–387 P = 0.8581, 388–646 P = 0.9682, 397–646 P = 0.0941, 510–646 P = 0.2496, 606–646 P = 0.0058. (G) HA-HSPA8:GST-FLAG-RHOB: 1–42 P = 0.3772, 42–89 P < 0.0001, 89–118 P = 0.9966, 118–141 P < 0.0001, 141–196 P < 0.0001. n = 3 independent experiments. Data are shown as the mean ± SD, *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001, one-way ANOVA (C-D, F-G).

Figure 5.

Characterization of HSPA8-BECN1-RHOB complex formation. (A-E) Representative diagram of Isothermal titration calorimetry (ITC) titration and fitting curves of HSPA8 with different BECN1 and RHOB constructs. The corresponding proteins and binding affinities are indicated at bottom. (F-H) GST-pull down experiments with binding-corresponding domains of HSPA8, BECN1 and RHOB. Purified proteins were mixed and subjected to immunoprecipitation with anti-HA (F) or anti-FLAG (G-H) gel. Bound HSPA8, BECN1 and RHOB were analyzed by immunoblotting with anti-HA, -MYC, -FLAG or -HSPA8 antibodies (shown as the slanting labels). Signal density of bound protein was normalized in IP group. Percentage of bound GST-MYC-BECN1 244–337 to GST-HA HSPA8 wild-type was set to 100% (F). Percentage of bound GST-HA-HSPA8 mutants to GST-FLAG-RHOB wild-type was set to 100% (G-H). n = 3 independent experiments. (F) GST-MYC-BECN1 244–337:GST-HA-HSPA8: 1–387 P = 0.9205, 388–646 P = 0.0391. (G) GST-HA-HSPA8 1–387:GST-FLAG-RHOB: 1–42 P = 0.1555, 42–89 P < 0.0001, 89–118 P < 0.0001. (H) GST-HA-HSPA8 510–646:GST-FLAG-RHOB: 1–42 P < 0.0001, 42–89 P < 0.0001, 89–118 P = 0.0549. (I) the interaction of HSPA8, BECN1 and RHOB in LPS-treated 5637 cells. 5637 cells were treated with 1 μg/ml LPS for 2 h. After LPS treatments, cell lysates were prepared and subjected to immunoprecipitation with anti-RHOB antibody plus protein A/G agarose, and analyzed by western blotting. IgG served as a negative control. n = 3 independent experiments. Data are shown as the mean ± SD, *P < 0.05, **P < 0.0001, one-way ANOVA (F-H).

HSPA8 drives liquid-liquid phase separation (LLPS), and forms liquid-phase condensates with BECN1 and RHOB via its residues 243–283

As reported, HSPA8 participates in multiple protein-associated processes, including protein complex formation or dissociation, folding and proteolysis of misfolded proteins. A sequence analysis with VSL2 algorithms indicated that HSPA8 contains predicted intrinsically disordered regions (IDRs), a key feature of proteins involved in liquid-liquid phase separation (LLPS) (Figure 6A).

Figure 6.

HSPA8 drives liquid-liquid phase separation via its residues 243–283. (A) Diagram of intrinsically disordered regions (IDRs) of HSPA8 calculated by VSL2 algorithms. The corresponding IDRs are bold. (B) Representative images of liquid droplets formation with increased-concentrations of HSPA8-Megfp,, BECN1-Megfp and RHOB-Megfp.. Indicated proteins were purified and mixed into the droplet formation buffer (120 mM NaCl and 10% PEG − 8000). Images were taken by fluorescent microscope. Scale bar: 10 μm. (C) Schematic diagram of HSPA8 fragments designed for studying HSPA8-drived LLPS. IDR: HSPA8 residues 243–283; ΔIDR: HSPA8 with residues 243–283 removed. (D) Representative images of liquid-like droplets formed by truncated HSPA8 proteins. n = 3 independent experiments. Scale bar: 10 μm. (E) Representative images of liquid-like droplets formed by HSPA8 in different concentrations of NaCl. Scale bar: 10 μm. (F) Representative confocal fluorescence microscope and differential interference contrast (DIC) microscope images of HSPA8 droplets fusion over time. n = 3 independent experiments. Scale bar: 10 μm. (G) FRAP measurements of HSPA8 droplets at the indicated time points (top). Normalized FRAP intensity curve of HSPA8 droplets was shown at the bottom (n = 3 independent experiments). Scale bar: 5 μm. (H and I) Recruitments of BECN1-mCherry or RHOB-mCherry into HSPA8-Megfp formed droplets with or without treatments of high concentrated NaCl. Scale bar: 10 μm. (J) Puncta formation in HEK293 cells overexpressing EGFP-HSPA8, EGFP-BECN1 or EGFP-RHOB. HEK293 cells were transfected with plasmids encoding EGFP-HSPA8, EGFP-BECN1 or EGFP-RHOB. Live cell images were taken by fluorescent microscope at 24 h post transfection. n = 3 independent experiments. Scale bar: 10 μm. (K) HSPA8 induces puncta formation in HEK293 cells via its IDR domain. HEK293 cells were transfected with plasmids encoding EGFP-HSPA8 wild-type or ΔIDR. At 24 h post transfection, live cells were treated with 3% 1, 6-hexanediol for 5 min and imaged by fluorescent microscope. n = 3 independent experiments. Scale bar: 10 μm. (L) Recruitment of mCherry-BECN1 or mCherry-RHOB into GFP-HSPA8 formed droplets with or without treatments of 3% 1, 6-hexanediol. HEK293 cells were transfected by EGFP-HSPA8 along with mCherry-BECN1 or mCherry-RHOB. At 24 h post transfection, live cells were imaged by fluorescent microscope in the present or absent of 3% 1, 6-hexanediol. n = 3 independent experiments. Scale bar: 10 μm. (M and N) GFP-RHOB and mCherry-BECN1 were concentrated in HEK293 and SW480 HSPA8-KO cells with overexpression of HSPA8 wild-type, 243–283 but not ΔIDR. HEK293 and SW480 HSPA8-KO cells were co-transfected with EGFP-RHOB and mCherry-BECN1 in the present of 3×FLAG-HSPA8 wild-type, 243–283 or ΔIDR. Scale bar: 10 μm. (O) Survival of intracellular LT2 in SW480 cells overexpressing HSPA8, 243–283 or ΔIDR. SW480 cells were pre-transfected with 3×FLAG-HSPA8 wild-type, IDR or ΔIDR before LT2 infection. n = 6 independent experiments. vector vs. HSPA8 P = 0.0497, vector vs IDR P = 0.0325, vector vs ΔIDR P = 0.2991. Data are shown as the mean ± SD, *P < 0.05, one-way ANOVA (O).

Droplet formation experiments indicated that mEGFP-fused HSPA8 forms phase-separated droplets in a dose-dependent manner. Nevertheless, RHOB, which has no predicted IDRs (Figure S6A), failed to form phase-separated droplets at a similar concentration (Figure 6B). BECN1, which also has two predicted IDRs (Figure S6B), only exhibited weak activity in droplet formation (Figure 6B). To map the key domains of droplet formation, truncated mutants with two IDRs in HSPA8 were individually generated, purified, and subjected to droplet formation experiments. Results revealed that HSPA8 243–283 (the first IDR) formed liquid droplets in vitro, whereas HSPA8 492–646 (the second IDR) and other mutants did not (Figure 6A,C,D). In addition, HSPA8 droplet formation was promoted by low salt concentrations but disassembled by higher salt concentrations (Figure 6E). As expected, the HSPA8 droplets showed liquid-like properties in fluorescence recovery after photobleaching (FRAP) experiments with a signal recovery time of 2 min (Figure 6G). The droplet fusion could be observed in a few minutes (Figure 6F). These results indicated that HSPA8 243–283 can form dynamic and liquid droplets.

To determine whether HSPA8 induces BECN1 and RHOB into its liquid-phase condensate, mCherry-fused BECN1 (BECN1-mCherry) and RHOB (RHOB-mCherry) were thus constructed, purified, and mixed with HSPA8-mEGFP. In either the HSPA8-BECN1 or HSPA8-RHOB group, double positive droplets were formed and could be disassembled by high salt treatments (Figure 6H,I), suggesting that HSPA8 can incorporate BECN1 or RHOB into its liquid-phase condensate. However, the incorporation of either BECN1 or RHOB did not significantly increase the size of HSPA8-formed liquid droplets (Figure S6C-D). To test whether HSPA8 facilitates BECN1 and RHOB to concentrate in vitro, we combined RHOB-mEGFP and BECN1-mCherry together in the presence of HSPA8. Interestingly, double-positive droplets formed after adding HSPA8 wild-type or 1–387, which is responsible for both droplet formation and binding to BECN1 and RHOB (Figure S6E). In contrast, no clear double-positive droplet formed when control or HSPA8 388–646 was added (Figure S6E). In addition, we also examined whether interactions between BECN1 and RHOB affect their liquid-phase condensate. Results indicated that RHOB 90–196 (including the binding domain to BECN1), but not 1–89 (not binding to BECN1), induced the droplet formation by BECN1, RHOB and HSPA8 (Figure S6F), suggesting physical interactions enhance liquid-phase condensate.

We next confirmed these results in live HEK293 cells expressing GFP-fused HSPA8, BECN1 and RHOB. Only EGFP-HSPA8 but not BECN1 or RHOB formed puncta in HEK293 cells (Figure 6J), whereas the HSPA8 mutant without the IDR domain (HSPA8Δ243–283) did not form puncta (Figure 6K). These formed puncta could be abolished by treatment with 1, 6-hexanediol, a common inhibitor of liquid-phase condensate used in live cells (Figure 6K). The fusion of HSPA8 droplets in SW480 cells was observed in 5 min (Figure S6G). Images in the FRAP assay indicated that the HSPA8 droplets in HEK293 cells also exhibit the liquid-like property (Figure S6H). In line with these results, HSPA8 also individually concentrated BECN1 or RHOB to form puncta in HSPA8-knockout (KO) HEK293 cells (Figure 6L). The amount of BECN1 and RHOB in these puncta was greatly increased in HSPA8-knockout SW480 or HEK293 cells when HSPA8 or HSPA8 243–283 (IDR domain) was overexpressed, but not for the overexpressed HSPA8Δ243–283 (Figure 6M,N). Accordingly, HSPA8-knockout SW480 cells transfected with HSPA8 wild-type or IDR showed less intracellular LT2 after infection, compared to vector and HSPA8Δ243–283 (Figure 6O). Moreover, results in the immunoprecipitation assay further confirmed that HSPA8 but not its IDR-removed mutant increased the stability of the class III phosphatidylinositol 3-kinase (PtdIns3K) complex I (PtdIns3K-CI) in HSPA8-knockout 293 cells, as well as its kinase activity (Figure S6I-J). Taken together, these results suggest that HSPA8 concentrates BECN1 and RHOB to enhance HSPA8-BECN1-RHOB interaction via LLPS, and promotes intracellular bacterial clearance.

HSPA8 promotes bacterial clearance in mice with acute urinary tract infection (UTI)

To investigate the role of HSPA8 in bacterial infection in vivo, we employed the Accell siRNAs to knock down Hspa8 in murine bladders with acute UTI. As expected, a strong reduction of HSPA8 expression was observed in bladder homogenates from CFT073-infected mice with the treatment of Accell siRNA targeting Hspa8, along with significantly decreased BECN1 and LC3 expression (Figure 7A). Immunohistofluorescence (IHF) staining of bladder tissue sections also confirmed the decreased levels of HSPA8, BECN1, and LC3 in Hspa8 knocked-down mice (Figure 7B). As a result, a higher bacterial titer was found in the bladders of CFT073-infected mice treated with Accell siRNA targeting Hspa8, compared to that of the scramble siRNA group (Figure 7C). Of note, no physiological changes were observed in the bladders of mice by H&E staining when treated with Accell siRNA targeting Hspa8 or scramble siRNA (Figure 7D). These results suggest that HSPA8 plays a key role in protecting mice against bacterial infections (Figure 7E).

Figure 7.

HSPA8 promotes bacterial clearance in mice with acute UTI. (A) Western blotting analysis of BECN1 and LC3 expression in the bladder of Hspa8 knockdown mice infected by CFT073. Female C57BL/6J mice were transurethrally injected with Accell siRNA targeting Hspa8 or non-targeting control before inoculation with 10⁸ CFU of CFT073. Quantification was shown on the right (n = 3 mice per group). Level of indicated protein in control group was set to 100%. HSPA8: siScr vs siHspa8 P < 0.0001; BECN1: siScr vs siHspa8 P < 0.0001; LC3: siScr vs siHspa8 P = 0.0009. (B) Immunohistofluorescent staining of the bladder of Hspa8 knockdown mice infected by CFT073. Representative images of HSPA8, BECN1 and LC3 expression were shown (left). Signals of indicated proteins was measured and quantified by Image Pro Plus software (right). Level of indicated protein in control group was set to 100%. n = 3 images assessed from 3 mice per group. Scale bar: 25 μm. HSPA8: siScr vs siHspa8 P = 0.0030; BECN1: siScr vs siHspa8 P = 0.0104; LC3: siScr vs siHspa8 P = 0.0022. (C) Bacterial titers in bladders of Hspa8 knockdown mice infected by CFT073 (n = 5 mice per group). siScr vs. siHspa8 P = 0.0079. (D) H&E staining of the bladder of Hspa8 knockdown mice. Scale bar: 200 μm. n = 3 mice. (E) Schematic diagram illustrating roles of HSPA8 in anti-bacterial autophagy. During bacterial infections, levels of RHOB and BECN1 are upregulated upon recognition of invading pathogens through receptors in membrane. Constitutively expressed HSPA8 explored in cytoplasm undergoes liquid-liquid phase separation, whereas the binding of RHOB and BECN1 to HSPA8 transports them into this liquid-like condensate, resulting in the assembly of HSPA8-RHOB-BECN1 complex and raised protein stability. Increased level of BECN1 thereafter drives formation of autophagosomes through macroautophagy, leading to clearance of intracellular bacterial pathogens. Data are shown as the mean ± SD, *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001, two-way ANOVA (A and B), two-tailed unpaired Student’s t test (C).

Discussion

We previously reported an infection-induced RHOB-BECN1-HSP90 complex that promoted LC3 lipidation and intracellular UPEC clearance. Here, through mass spectrometry analysis, we identified new candidates of RHOB bound proteins that may contribute to the autophagic anti-bacterial defense. Among those, we found HSPA8, which induced both LC3 lipidation and clearance of intracellular UPEC and LT2 in bladder and colon epithelial cells, is likely involved in this process. However, less is known about how HSPA8 acts as an anti-bacterial factor during infection.

One study reported that HSPA8 is upregulated in mesenteric lymph nodes (MLN) during immune responses to intestinal colonization of S. Typhimurium [41]. Others also showed that HSPA8 could also be utilized as a receptor on the cell surface for viral [42] or bacterial infections [43]. In our case, HSPA8 was constantly expressed in both colonic and bladder epithelial cells with or without infection, and knockdown of HSPA8 attenuated the bacteria-induced RHOB and BECN1 upregulation. As an anti-bacterial factor, HSPA8 had no effects on bacterial recognition at the early stage of infection. However, downstream macroautophagy factors affected HSPA8-mediated bacterial clearance, suggesting that the anti-bacterial activity of HSPA8 relies on macroautophagy, and functions similarly to RHOB [10].

We confirmed that the presence of HSPA8 increased BECN1 and RHOB stabilization by preventing them from proteasomal degradation. Additionally, we found the K48-mediated ubiquitination of both BECN1 and RHOB was enhanced in the absence of HSPA8. HSPA8 functions to stabilize both of these factors in autophagic bacterial defense rather than induce protein degradation. This differs from numerous studies showing that HSPA8 targets KFERQ motif containing proteins as cargo for autophagic degradation. Interestingly, in the absence of RHOB, HSPA8 still mildly upregulated the BECN1 level and clearance of intracellular bacteria, suggesting HSPA8 could independently target BECN1 or RHOB as a client protein for stabilization. As a chaperone protein, HSPA8 has a high intracellular level and many binding proteins. It has been reported that HSPA8 serves as a membrane chaperone that drives HSP90 into liposomes [44], but our results show that the autophagy-promoting effect of HSPA8 does not depend on HSP90, which also suggests that the stabilizing effect of HSPA8 on BECN1 has a greater influence on autophagy.

In this study, we provided evidence from IP and ITC, mapping the detailed interacting domains between HSPA8-RHOB-BECN1, and deciphering their way of the protein complex assembly. The NBD and LID domains of HSPA8 were found to interact with RHOB residues 1–42 and 89–118, respectively. On the other hand, HSPA8 only binds to BECN1 residues 244–337 through the NBD domain. HSPA8 domains that bind to RHOB or BECN1 are different from RHOB-BECN1 interactions, implying the formation of a HSPA8-RHOB-BECN1 complex.

We identified the IDR region in the NBD domain of HSPA8 that drives LLPS, the way of which is reported by recent studies for controlling protein-protein interactions, aggregation and refolding in cellular processes [28]. This activity of HSPA8 enhanced the association of RHOB and BECN1 by concentrating them into HSPA8-formed liquid-phase droplets, whereas RHOB colocalizes with BECN1 through their interactions. By stabilizing effects of HSPA8, BECN1 is increased and in turn facilitates the assembly and activities of PtdIns3K-CI, a large protein complex that promotes PtdIns3P generation in autophagy initiation. Removal of the IDR domain not only abrogated the HSPA8-mediated RHOB-BECN1 droplet formation but also attenuated intracellular bacterial clearance. Since the NBD domain of HSPA8 is considered as the site for ATP/ADP exchange [2], LLPS induced by HSPA8 represents a distinct function of chaperons in protein-protein interactions [45,46].

In summary, we identified the protein stabilizing activity of HSPA8 and its LLPS-related interaction with RHOB and BECN1, providing new insight into the anti-bacterial autophagy mechanism of HSPA8 during uropathogenic or intestinal pathogenic bacterial infections.

Materials and methods

Cell lines and reagents

HEK293 (human kidney epithelial cells; American Type Culture Collection [ATCC], CRL−1573), 5637 (human bladder epithelial cells; ATCC, HTB−9) and SW480 cells (human colon epithelial cells; ATCC, CCL−228) were cultured in Dulbecco’s modified Eagle’s medium (DMEM) or Roswell Park Memorial Institute 1640 (RPMI1640) containing 10% FBS and 100 U/ml penicillin-streptomycin at 37°C with 5% CO₂. For bacterial infection, cells were pre-cultured in antibiotic-free medium with 10% FBS.

Anti-FLAG M2 affinity gel (A2220), 3×FLAG peptide (F4799), anti-MYC/c-Myc agarose affinity gel (A7470), MYC/c-MYC peptide (M2435), Influenza hemagglutinin (HA) peptide (I2149), lipopolysaccharides (LPS; L6529), poly (ethylene glycol) 8000 (P5413), and 1, 6-hexanediol (H11807) were purchased from Sigma-Aldrich. Anti-HA Affinity Gel (P2287) was purchased from Beyotime Biotechnology. Pierce protein A/G plus agarose (20423) and Pierce™ Silver Stain Kit (24612) were purchased from Thermo Scientific. Glutathione Sepharose 4B (17075601) was purchased from Cytiva. L-glutathione reduced (G8180) was purchased from Solarbio. Bafilomycin A₁ (BafA₁, ab120497) was obtained from Abcam. Gentamycin sulfate (A620217) was obtained from BBI-lifesciences. Rhodamine Phalloidin (YP0063S) was purchased from UElandy. Methyl α-D-mannopyranoside (HY-W039897), Trimethoprim (HY-B0510), Sulfamethoxazole (HY-B0322), MG132 (HY−13259C), geldanamycin (GA, HY−15230), chloroquine phosphate (CQ, HY−17589) and cycloheximide (CHX, HY−12320) were obtained from MedChemExpress. Class III PI3-Kinase kit (K−3000) was obtained from Echelon Biosciences.

DNA constructs

The complementary DNA encoding full-length human HSPA8 was amplified by PCR and cloned into p3×FLAG-CMV−10 vector with an N-terminal 3×FLAG tag (3×FLAG-HSPA8) using primers (Table S2) and standard procedure [10]. Full-length human RHOB was cloned into pCDNA 3.1 (-) vector with a MYC tag at the N terminus (MYC-RHOB). Truncation mutants of HSPA8 were generated with p3×FLAG-CMV−10 vector, using standard procedure or Seamless Cloning according to manufacturer’s instruction. For evaluation of LLPS in cells, HSPA8, RHOB, BECN1 and related mutants were cloned into pEGFP-C1 or pmCherry-C1 vector. The mCherry-EGFP-LC3 (P0446), mCherry-LC3 (P1026) and GFP-LC3 (P0199) constructs were obtained from MiaoLingPlasmid.

Bacterial strains and invasion in cultured cells

Salmonella Typhimurium LT2 (ATCC, 700720) or Escherichia coli CFT073 (ATCC, 700928) were grown in Luria-Bertani (LB) medium for 12 h with constant shaking at 37°C and employed in the following experiments with previously described methods [10]. For bacterial invasion, bacterial strains of log-phase in PBS (Solarbio, P1020) were added into cells pre-cultured in 24-well plates at the indicated MOI. A short spin at 500 × g was applied to increase the efficiency of invasion if needed. Infected cells were incubated for 30 min at 37°C, and then treated with 100 ug/ml gentamycin for additional 2 h and lysed with 0.2% Triton X−100 (Solarbio, T8200) in PBS buffer. Colony-forming units (CFU) of intracellular bacteria were counted by plating cells lysate on LB agar. For evaluating bacterial entry, infected cells were treated with gentamycin for 30 min before CFU assessment. For evaluating bacterial expulsion, infected cells were treated with media containing 100 ug/ml gentamycin (BBI-lifesciences, A620217) and 100 mM methyl α-D-mannopyranoside (MedChemExpress, HY-W039897) for 1 h at 1 hpi to block bacterial re-invasion. Infected cells were thoroughly washed, and incubated with culture media containing bacterial inhibitors (100 mM methyl α-D-mannopyranoside, 25 μg/mL trimethoprim (MedChemExpress, HY-B0510) and 125 μg/mL sulfamethoxazole (MedChemExpress, HY-B0322)) for additional 4 h. CFU of expelled bacteria were measured with culture media from infected cells.

Antibodies and western blotting

Anti-HSPA8 (sc−7298), Anti-RHOB (sc−8048), Anti-LAMP2 (sc−18822), Alexa Fluor 488-conjugated anti-RHOB (sc−8048 AF488), and normal mouse IgG and IgG1 (sc−2025, sc−3877) antibodies were obtained from Santa Cruz Biotechnology. Anti-ACTB/β-actin (A1978), anti-LC3B (L7543), anti-FLAG (F3165) and anti-ubiquitinylated proteins (FK2; 04–262) antibodies were obtained from Sigma-Aldrich. Anti-MYC (2278), anti-ATG5 (9980 and 2630), anti-BECN1 (3495 and 4122), anti-HA (2367 and 3724), anti-RUBCN (8465), anti-SQSTM1 (8025 and 7695), anti-PtdIns3K/PI3 Kinase class III (D9A5) rabbit mAb (4263), anti-GST-Tag (91G1) Rabbit mAb (2625) and normal rabbit IgG (2729) antibodies were obtained from Cell Signaling Technology. Anti-PIK3R4 polyclonal antibody (17894–1-AP), CoraLite 594 conjugated goat anti-mouse (SA00013–3) and goat anti-rabbit (SA00013–4), CoraLite 488 conjugated goat anti-mouse (SA00013–1) and goat anti-rabbit (SA00013–2) antibodies were obtained from Proteintech. Anti-Atg14 polyclonal antibody (PD026) was purchased from Medical & Biological Laboratories. Alexa Fluor 405-conjugated goat anti-rabbit antibody (A31556) was obtained from Invitrogen. For detecting protein expression, cells lysates were made with RIPA lysis buffer (Solarbio, R0020) containing freshly added protease inhibitors (Roche, 05892791001). Samples were boiled with 5×SDS loading buffer, loaded to SDS-PAGE electrophoresis and transferred onto PVDF membrane, followed by detection with indicated antibodies by Amersham Imager 600 (GE Healthcare Life Sciences, Buckinghamshire, UK). Signals were analyzed using NIH ImageJ software (National Institutes of Health).

Immunoprecipitation (IP) assays

The immunoprecipitation assays were performed as described before [47]. Cell lysates were made with mild RIPA buffer (0.2 mM EDTA, 50 mM Tris-HCl, pH 7.4, 150 mM NaCl, 1% NP−40 [Sigma-Aldrich, 74385], Roche protease inhibitor cocktail) on ice, followed by centrifugation at 13,500 × g, for 10 min to remove cell debris. For IP of 3×FLAG-, HA- or MYC-tagged proteins, cell lysates were mixed with PBS-washed anti-FLAG affinity gel (Sigma-Aldrich, A2220), anti-HA affinity gel (Beyotime, P2287) or anti-MYC affinity gel (Sigma-Aldrich, A7470), respectively, and constantly rotated overnight at 4°C. Bound proteins were washed with binding buffer (2 mM EDTA, 50 mM Tris-HCl, pH 7.4, 150 mM NaCl, 0.1% NP−40, 30% glycerol) for 5 times with 500 × g centrifugation for 5 min, and eluted by 3×FLAG, HA- or MYC peptides (Sigma-Aldrich, M2435) in TBS buffer (20 mM Tris-HCl, 140 mM NaCl, pH 7.6) on ice for 4 h. The eluents were then collected for detection with WB. For IP of endogenous proteins, 1 μg/ml LPS stimulation was applied in cells for 2 h if needed. Cell lysates were made and mixed with antibodies or immunoglobin G (IgG) overnight at 4°C, followed by incubation with pre-washed Protein A/G agarose (Thermo Scientific, 20423) with constant rotation for 4 h at 4°C. After several washes, bound protein complexes were boiled within 1×SDS loading buffer (CWBIO, CW0027S) and subjected to western blotting analysis.

For evaluating ubiquitination of RHOB and BECN1, HEK293 cells were co-transfected with FLAG-RHOB/MYC-BECN1 and HA-tagged ubiquitin constructs (HA-UB K11 or HA-UB K48, with arginine substitutions of all lysine residues except K11 or K48). Transfected cells were treated with MG132 (10 μM) overnight and lysed for immunoprecipitation and western blotting analysis with anti-HA antibody.

For LC-MS/MS analysis [47], the eluents from IP were loaded onto electrophoresis and treated with silver staining kit (Thermo Scientific, 24612) according to manufacturer’s instruction. The bands of interest were retrieved and analyzed using a full-scan mass spectrum (300 to 1800 m/z) with Proteome Discoverer software (version 1.4.0.288, Thermo Scientific).

For detecting protein interaction in SDS-PAGE gels, Coomassie brilliant blue (CBB) fast staining solution [48] (Solarbio, P1300) was used according to the manufacture protocol. In briefly, SDS-PAGE gels after electrophoresis were soaked in distilled water, shortly boiled and stained with heated Quick dyeing solution for 30–60 s. Decolorization was then performed with boiling water again on a shaker for 30–60 s to make protein bands visible.

In vitro kinase assay of PtdIns3K-CI

The PtdIns3K-CI complex was expressed and examined for its kinase activity as previously reported [49,50]. In brief, HSPA8-KO HEK293 cells were transfected with plasmids encoding ATG14–3×FLAG (MiaoLingPlasmid, P32426), MYC-BECN1, MYC-PIK3C3 (MiaoLingPlasmid, P1799) and PIK3R4–3×MYC (MiaoLingPlasmid, P28214) in the present of HSPA8 WT or IDR domain-removed mutant. After 48 h, cells lysates were made and subjected to immunoprecipitation assay with anti-FLAG affinity gel for purification of ATG14 associated protein complex. The enzyme activity of purified protein complex was further detected using Class III PI3-Kinase kit (Echelon Biosciences, K−3000) according to manufacture.

Recombinant protein purification and Glutathione S-transferase (GST) affinity-isolation assay

For recombinant proteins expression, pGEX4T–3 and pGEX6P–1 vectors (gifts from Professor Yupeng Chen’s Lab) were used to generate constructs encoding GST-, mEGFP- or mCherry-fused proteins, including full-length HSPA8, RHOB, BECN1 and related truncated mutants. These constructs were transformed into BL21 (DE3; Sigma-Aldrich, CMC0014) with antibiotics, and treated by 100 μM IPTG (Sigma-Aldrich, I6758) at 16°C to allow protein expression. Expressed proteins were extracted from cells by centrifugation after lysis buffer treatments (50 mM Tris, 100 mM NaCl, 1 mg/ml lysozyme (BBI, A610308–0005), 1% PMSF (Solarbio, P0100), pH 7.9). The mixture was then sonicated for 60 cycles on ice, and treated with a second lysis buffer (10 mM MgCl₂, Benzonase [250 units per 15 ml lysis buffer; HaiGene, C2001], 1% Triton X−100). After centrifugation at 8000 × g, for 5 min to remove cell debris, GST-fused proteins were affinity isolated with glutathione Sepharose 4B (Cytiva, 17075601) and eluted with reduced L-glutathione according to manufacturer’s instruction. For removal of the GST tag, Thrombin (Solarbio, T8021) was applied in purification procedure according to the manufacturer’s instruction. Protein concentration was determined using the Pierce BCA Protein Assay Kit (Thermo Scientific, 23227).

For GST affinity-isolation assays, 1 to 5 μg of proteins indicated was mixed in binding buffer (0.8% BSA [BBI, A600332] in PBS with fresh PMSF) with constant rotation at 4°C overnight. After binding, mixture was next incubated with anti-HSPA8, anti-RHOB or anti-BECN1 antibodies (IgG served as negative control) plus protein A/G agarose overnight at 4°C with constant rotation. For GST-fused proteins, mixture was incubated with glutathione Sepharose 4B. For IP of FLAG-, MYC- or HA-tagged proteins, mixture was incubated with anti-FLAG-, anti-MYC- or anti-HA-affinity gels. Bound proteins were washed 5 times with binding buffer and boiled in SDS sample buffer, followed by western blotting analysis.

Isothermal titration calorimetry (ITC)

The ITC experiments were performed using a modified method as described previously [51]. Briefly, purified proteins were dialyzed in buffer (20 mM Tris-HCl, 150 mM NaCl, pH 8.0) and concentrated by ultrafiltration. For probing HSPA8-BECN1 complex formation, BECN1 (200 μM, in the syringe) was titrated into the cell containing HSPA8 (20 μM). Titrations were performed through 13 injections of 3 μL at 120 second intervals from syringe. An initial injection of 0.4 μL was made and excluded from data analysis. Data was analyzed using MicroCal PEAQ-ITC Analysis Software (Version 1.22, Malvern Panalytical).

Immunofluorescence staining of cells

Cells grown in LAB-TEK 4-well chamber slides (Thermo Scientific, 155383) were infected with LT2 as described above. Infected cells were washed with sterilized PBS and fixed with 4% paraformaldehyde for 30 min at room temperature. For detecting protein colocalization, cells were permeabilized with 0.5% Triton X−100 in PBS and blocked with 5% BSA in PBS containing 10% goat serum (Solarbio, SL038). For detecting LC3 decoration of GFP-LT2, cells were treated with 0.1% Triton X−100 in PBS instead. Target proteins were stained using specific primary antibodies plus CoraLite 488- or CoraLite 594-conjugated secondary antibodies. For intracellular bacterial staining, F-actin of infected cells was labeled with Rhodamine Phalloidin, and nuclei were counterstained with DAPI. Signals were detected and analyzed by Leica confocal fluorescence microscope (Leica TCS-SP8, Leica Microsystems, Germany) and Image-Pro Plus software (Media Cybernetics, Inc.).

mCherry-EGFP-LC3 transient transfection

HEK293, 5637 or SW480 cells were transfected with the plasmid encoding mCherry-EGFP-LC3B along with indicated proteins or siRNAs according to the manufacturer’s instruction. At 48 h post transfection, cells were infected with LT2 or CFT073 for 4 h if needed. Images of mCherry positive (Red), GFP positive (Green) and both positive (Yellow) puncta in each cell were taken under a Leica confocal fluorescence microscopy. Numbers of autophagosomes (Yellow puncta) and autolysosomes (Red-only puncta) were manually counted. The ratio of autolysosomes to autophagosomes (referred as autophagy flux) was analyzed as described before [38].

For detecting colocalization of GFP-LT2 with SQSTM1, ubiquitin and LC3, mCherry-LC3, HSPA8 siRNAs or scramble siRNAs were transfected into SW480 cells grown in LAB-TEK 4-well chamber slides for 48 h. Cells were then infected with GFP-LT2 for 4 h at MOI of 100. BafA₁ was added in culture medium of infected cells along with bacteria if needed. Infected cells were then washed, fixed and permeabilized with 0.1% Triton X−100 in PBS, followed by immunostaining with anti-SQSTM1 or anti- ubiquitin antibodies plus Alexa 405-conjugated secondary antibodies. Images were examined by Leica confocal fluorescence microscopy.

RNA extraction and Qrt-PCR

RNA of indicated proteins was prepared by a Total RNA Extraction Kit (Solarbio, R1200) according to the manufacturer’s instructions. The cDNA was then amplified by PCR with specific primers (Table S2) and MonScript™ 5× RTIII All-in-One Mix (Monad, MR05001M). qRT-PCR reactions were performed using UltraSYBR Mixture (CWBIO, CW2601H) with regular setting: preincubation at 95°C for 10 min, 40 cycles of 3 step amplification (95°C for 20 s, 61°C for 20 s, 72°C for 20 s, 78°C for 1 s), melting (95°C for 10 s, 65°C for 60 s, 97°C for 1 s), and cooling at 37°C for 30 s. GAPDH served as control. All data were analyzed by the comparative critical threshold cycle 2^−∆∆Ct method.

In vitro droplet assay

GST-tagged mEGFP or mCherry fusion proteins were constructed with pGEX−6p−1 vector, and purified by methods as previously described [52]. The recombinant proteins were mixed in the droplet formation buffer solution (120 mM NaCl and 10% polyethylene glycol 8000 (PEG−8000; Sigma-Aldrich, P5413)). The mixture was then loaded onto a glass slide with coverslips to form droplets, and imaged by using Leica confocal fluorescence microscope or differential interference contrast (DIC) microscopy (ZEISS Axio-Imager LSM−800, Oberkochen, Germany).

Images of LLPS in cells

HEK293 cells were transfected with plasmids encoding mEGFP or mCherry fused indicated proteins for 24 h. Images were taken by Leica confocal fluorescence microscope. For inhibitor studies, transfected cells were treated with 3% 1, 6-hexanediol (Sigma-Alrich, H11807) in PBS for 5 mins as described previously [53]. Images were taken by confocal fluorescence microscopy before and after cells were treated.

FRAP analysis

Formed droplets of indicated proteins or fluorescent puncta in transfected cells were bleached using the 488-nm laser from the confocal fluorescent microscope (ZEISS Axio-Imager LSM−800, Oberkochen, Germany). Images of fluorescence recovery were taken by confocal fluorescent microscope every 10 s after photobleaching.

Mouse model with acute UTI

Six- to eight-week-old female C57BL/6J mice were purchased from the Academy of Military Medical Science and used in experiments following the Guide for the Care and Use of Laboratory Animals with approvals of the Animal Ethics Committee at Tianjin Medical University. Mice were routinely maintained under specific pathogen free conditions with 30% humidity and 12 h light/12 h dark cycle at 25°C.

For in vivo RNA silencing, Accell siRNA targeting Hspa8 (SMARTPool; Horizon, E−062625) or non-targeting control (Horizon, D−001910) were transurethrally injected into female mice according to the manufacturer’s protocol as describe before [54].

For acute UTI, 10⁸ CFU of pre-cultured CFT073 were resuspended in 50 μl sterilized PBS and transurethrally injected into anesthetized female mice. Infected mice were euthanized at 12 hpi, and the bladders were then aseptically collected.

For bacterial titer measurements, the bladder homogenate was made in 0.025% Triton X−100 in PBS, and plated on to LB agar plates for CFU assessment. For detecting the expression of indicated proteins, RIPA lysis buffer was used instead. Samples were boiled in SDS loading buffer before western blotting analysis.

For immunofluorescent staining, bladder tissues collected from infected mice were embedded in OCT compound (SAKURA, 4583) and frozen in liquid nitrogen. Sections were then cut, loaded onto glass slides, and fixed with acetone. After that, fixed sections were subsequently treated with pre-chilled methanol for 20 min, merged in 3% hydrogen peroxide in methanol for 10 min, and rehydrated with PBS. Sections were blocked with 5% BSA in PBS at room temperature for 1 h, and incubated with proper primary antibody overnight at 4°C. After wash with PBS, sections were stained with CoraLite 488- or 594-conjugated secondary antibodies at room temperature for 1 h and mounted with DAPI (SouthernBiotech, 0100–20) under coverslips. Images were taken by using Leica confocal fluorescence microscope. Protein expression was quantified using Image Pro Plus software (Media Cybernetics).

H&E staining

Paraformaldehyde (4%)-fixed paraffin-embedded bladders were used for making tissue sections of 5 μm and stained with hematoxylin and eosin (H&E). Images were taken using Olympus microscope (Tokyo, Japan).

RNA interference

To knockdown genes in cells, scrambled siRNA and siRNAs of the indicated genes were purchased from GenePharma. Transfection of siRNAs was performed using Lipofectamine 3000 (Invitrogen, L3000150) in SW480, 5637 or HEK293 cells with centrifugation at 1680 ×g for 1 h. Cells were then incubated for another 48 h before use. Efficiency of each siRNA was determined by western blotting.

Statistical analysis

All data were shown as the mean ± SD. Statistical significance was analyzed using Student’s t test (non-parametric Mann-Whitney test) or ANOVA analysis.

Funding Statement

This study was supported by grants from the National Natural Science Foundation of China (NSFC) Programs (32170186, 31970133), Tianjin Science and Technology Commissioner Project (22JCZDJC00490).

Disclosure statement

No potential conflict of interest was reported by the authors.

Data availability statement

All data needed to evaluate the conclusions in the paper are present in the paper and/or the supplementary materials. Additional data related to this paper may be requested from the authors.

Supplementary data

Supplemental data for this article can be accessed online at https://doi.org/10.1080/15548627.2023.2223468.