The mRNA binding-mediated self-regulatory function of small heat shock protein IbpA in γ-proteobacteria is conferred by a conserved arginine

Abstract

Bacterial small heat shock proteins, such as inclusion body-associated protein A (IbpA) and IbpB, coaggregate with denatured proteins and recruit other chaperones for the processing of aggregates thereby assisting in protein refolding. In addition, as a recently revealed uncommon feature, Escherichia coli IbpA self-represses its own translation through interaction with the 5′-untranslated region of the ibpA mRNA, enabling IbpA to act as a mediator of negative feedback regulation. Although IbpA also suppresses the expression of IbpB, IbpB does not have this self-repression activity despite the two Ibps being highly homologous. In this study, we demonstrate that the self-repression function of IbpA is conserved in other γ-proteobacterial IbpAs. Moreover, we show a cationic residue-rich region in the α-crystallin domain of IbpA, which is not conserved in IbpB, is critical for the self-suppression activity. Notably, we found arginine 93 (R93) located within the α-crystallin domain is an essential residue that cannot be replaced by any of the other 19 amino acids including lysine. We observed that IbpA-R93 mutants completely lost the interaction with the 5′ untranslated region of the ibpA mRNA, but retained almost all chaperone activity and were able to sequester denatured proteins. Taken together, we propose the conserved Arg93-mediated translational control of IbpA through RNA binding would be beneficial for a rapid and massive supply of the chaperone on demand.

Small heat shock proteins (sHsps) are characterized by their low molecular weights (12 ∼ 43 kD) in the subunits and conserved α-crystallin domains (ACDs) franked by disordered N-terminal domains (NTDs) and C-terminal domains (CTDs) (1, 2, 3, 4). As molecular chaperones, sHsps are widely conserved in all kingdoms of life to protect cellular protein homeostasis (proteostasis) by binding with and then sequestering misfolded proteins in an ATP-independent manner, which is termed as a “sequestrase” activity (4). Other chaperones, such as Hsp70 family, are required for later refolding or degradation to disassociate coaggregates between sHsps and substrate proteins since sHsps are not refolding active (1, 2, 4).

sHsps assemble into dimers and large oligomers (1, 2, 3, 4). A basic scaffold of sHsps is formed through the ACD dimerization with an ACD interacting with a partnering ACD to form homodimers or heterodimers (1, 2, 3, 4). The higher-order oligomers are then regulated by intrinsically disordered NTDs and CTDs (1, 2, 3, 4). The CTD tails harbor a short yet highly conserved functional sequence motif, the IXI/V motif, which plays a crucial role in the oligomerization and chaperone activity of sHsps (1, 2, 3, 4, 5). In bacterial sHsps, the IXI/V motif interacts with a β4/β8 groove of the neighboring ACD to assemble two dimers (2, 4, 5). Mutations in the IXI/V motif, such as IXI-to-AEA, usually cause defects in the oligomerization ability and chaperone activity (2, 6). Notably, sHsp subunit associations are relatively weak, and oligomeric sHsps are primarily in a dimer-based and dynamic subunit exchange (4, 5). This property enables sHsps to cope with environmental changes such as heat stress, pH, and oxidative stress (4, 5).

sHsps can rapidly respond to environmental stresses, as evidenced by the significant upregulation of expression (7). Bacterial sHsps are upregulated by a sigma factor σ³² at the transcriptional level upon heat shock (8). Moreover, in α- and γ-proteobacteria, sHsps are translationally regulated by a thermosensitive mRNA structure (RNA thermometer, [RNAT]) in the 5′-untranslated region (5′ UTR), which contains the element of heat shock gene expression repression (9, 10). The elements, ranging in length from 60 to over 100 nucleotides, typically consist of two to four stem-loops, which play a critical role in the modulation of sHsp translation (11).

The number of sHsp members in organisms varies from one or two in prokaryotes to even over ten in eukaryotes (4). In Escherichia coli, two sHsps, inclusion body-associated protein A (IbpA_Ec) and B (IbpB_Ec), encoded in the ibpAB operon, share high similarity amino acid sequences (∼50% identity, Fig. 1A) but are specialized in different functions during the antiaggregation process (12). IbpA_Ec and IbpB_Ec interact with misfolded client proteins to form coaggregates (so-called holdase activity) but are not involved in the subsequent refolding and degradation processes (13). The coaggregates are recognized by other chaperones such as DnaK/DnaJ and ClpB for refolding/degradation, at which point IbpAB are released from the substrates (14). IbpA_Ec is more efficient for associating with client proteins to form small coaggregates, while IbpB_Ec, forming functional complexes with IbpA_Ec, is more competent to assist the disassembly of sHsps from the coaggregates (12, 15). IbpA_Ec and IbpB_Ec can form heterodimers and heterooligomers (16). In the absence of IbpB_Ec, IbpA_Ec tends to form fibril-like structures both in vivo and in vitro, which might be mediated by NTD and CTD (6, 17). The fibrillation of IbpA_Ec can be blocked by its substrate proteins as well as IbpB_Ec (17).

Figure 1.

Identification of a crucial residue, Arg93, in Escherichia coli IbpA (IbpA_Ec) for discriminating between IbpA_Ecand IbpB_Ecin IbpA-mediated translation suppression function.A, alignment of E. coli IbpA (IbpA_Ec) and IbpB (IbpB_Ec). NTDs, ACDs, and CTDs are colored blue, red, and green, respectively. IbpA_Ec tyrosine 34 (Y34) and arginine 93 (R93) are indicated by blue and red arrowheads, respectively. The IXI/V motif, located in CTDs, are marked by a black box. Locations of β strands, based on an AlphaFold2-predicted IbpA_Ec monomer structure, are also shown. B, IbpA_Ec-mediated translation suppression using a GFP reporter assay. The translation of GFP reporter from gfp mRNA harboring the 5′ UTR of ibpA_Ec in E. coli was inhibited by excess amounts of IbpA. C and D, systematic chimera and mutagenesis to identify a critical residue in IbpA_Ec to differentiate between IbpA_Ec (blue) and IbpB_Ec (orange) using the GFP reporter assay. Upper: schematic of an IbpA_Ec/IbpB_Ec chimera (C) and subsequent single-residue IbpA_Ec mutations (D). Lower: Western blotting analysis evaluating the effects of the ACD substitution (C) and individual alanine substitution among IbpA_Ec residues 92∼95 (D) on the reporter GFP translation level in the E. coli BW25113 strains (WT: BW25113 WT strain; ΔAB: ibpAB operon-deleted BW25114 strain). The cells labeled as “−” only expressed the GFP reporter. The expression of FtsZ was used as a control of the constitutive expression level. Note: other chimera and mutation analyses are shown in Figs S1 and S2. E, reconstituted E. coli cell-free translation system (PURE system) in the absence or presence of purified IbpA_Ec-WT or the R93A mutant. The gfp reporter mRNAs carrying the pBAD30 vector 5′ UTR or the ibpA_Ec 5′ UTR were used as templates for translation. Upper: the fluorescence intensity of the translated GFP; lower: the fold of quantified GFP fluorescence level. GFP expression levels without IbpA_Ec (denoted as “−”) were used for the normalization and set to 1. The data represent the means (±SD) of three independent experiments and were analyzed by one-way ANOVA within the comparison with each -IbpA_Ec group (ns: nonsignificant; ∗p < 0.0332; ∗∗p < 0.0021; ∗∗∗p < 0.0002; ∗∗∗∗p < 0.0001, wherever shown). ACD, α-crystallin domain; CTD, C-terminal domain; Ibp, inclusion body-associated protein; NTD, N-terminal domain; UTR, untranslated region.

Although IbpA_Ec and IbpB_Ec are highly similar in sequence and closely collaborate in both structures and chaperone activity, IbpA_Ec possesses a unique posttranscriptional regulation mechanism that IbpB_Ec lacks (18). IbpA_Ec is upregulated at the posttranscriptional level even without heat stress by the overexpression of aggregation-prone proteins (18). Further analysis revealed that IbpA_Ec directly interacts with the 5′ UTR of ibpA (and ibpB) mRNA and inhibits the translation of ibpA (and ibpB). The IbpA-mediated translation suppression is relieved by aggregation-prone client proteins that recruit IbpA_Ec, thereby reducing the amount of IbpA_Ec involved in self-suppression (18). In addition, the IbpA-mediated regulation in cells is independent of proteolysis (18). The nonconventional function of IbpA_Ec as an aggregation-sensor tightly suppresses IbpA_Ec expression under aggregation-free conditions but enables cells to rapidly upregulate the IbpA_Ec levels upon acute aggregation stress, such as heat shock (10, 18). Moreover, a recent study has demonstrated that IbpA_Ec also downregulates the expression of other Hsps by inhibiting the translation level of the heat shock transcription factor σ³², which highlights the general importance of IbpA in the heat shock response regulation (19).

Elucidating the molecular mechanism underlying IbpA_Ec self-suppression on translation is of great interest, but many questions remain. For example, why does IbpB_Ec lack self-regulation activity? Is this translational repression by IbpA conserved in other bacterial IbpAs? Here, we revealed the conservation of the IbpA-mediated self-regulation in other γ-proteobacterial IbpAs and found that a highly conserved residue, Arg93 (R93), located within the IbpA ACD and absent in IbpB, is irreplaceably important for discriminating between IbpA and IbpB in the translation suppression function.

Results

The residue Arg93, located within the ACD, plays a crucial role in the self-regulation of IbpA_Ec

To explore the region responsible for the notable difference between IbpA_Ec and IbpB_Ec, we conducted a reporter assay to assess the translation regulation activity of IbpA_Ec (18). In brief, the translation of a gfp reporter, containing the ibpA_Ec 5′ UTR sequence in an arabinose-inducible vector (pBAD30), is suppressed in E. coli cells with IbpA_Ec overexpressed from another vector (pCA24N) (18) (Fig. 1B). In this assay, the expression level of the GFP reporter in E. coli WT cells was slightly lower than that in ibpAB operon-deleted (ΔAB) cells due to the endogenous IbpA_Ec (Fig. 1C, lanes 1 and 2). Overexpression of exogenous IbpA_Ec, about 10-fold compared to endogenous IbpA_Ec (18), markedly reduced the GFP levels both in E. coli WT and ΔAB cells (Fig. 1C, lanes 3 and 4), confirming the suppressive effect of IbpA_Ec on the reporter translation, as shown previously (18).

We overexpressed chimeric IbpABs that included ACD, NTD, and CTD domain substitutions in both E. coli WT and ΔAB strains, along with reporter plasmids (Figs. 1C and S1). The chimeric IbpABs with substitutions in both NTD and CTD had no significant effects on self-suppression (Fig. S1). In contrast, the A_ACD::B_ACD chimera, in which the IbpA_Ec ACD was replaced with that of IbpB_Ec, significantly abolished the translation suppression (Fig.1C, lanes 5 and 6), indicating that the IbpA_Ec ACD includes a critical region for the self-regulation activity of IbpA_Ec. Following a series of mutation experiments in ACD (Fig. S2), we narrowed the critical region down to IbpA_Ec residues from 92 to 95 and then individually mutated these residues with alanine. Among these mutants, mutation of Arg93 to Ala (IbpA_Ec-R93A) greatly increased the translation level of the reporter gfp, compared to the WT and the other alanine mutants (Fig. 1D, lanes 13 and 14), indicating that the point mutation R93A is sufficient to lose the translation suppression activity. The overexpression of IbpB_Ec did not change the reporter level between WT and ΔAB cells. It is consistent with the previous study (18) and indicates the heterooligomerization between IbpA_Ec and IbpB_Ec (Fig. 1C, lanes 7 and 8). The same tendency was observed in the cells overexpressing A_ACD::B_ACD and R93A (Fig. 1C, lanes 5 and 6; Fig. 1D, lanes 13 and 14), suggesting that the mutants form heterooligomers with the endogenous IbpA_Ec.

Next, we evaluated the effect of IbpA_Ec-R93A using a chaperone-free, reconstituted cell-free translation system (PURE system) (20), as previously used for the IbpA_Ec-mediated translation suppression (18). Purified IbpA_Ec-WT suppressed the translation of the gfp reporter in an ibpA_Ec 5′ UTR-dependent manner (18), whereas purified IbpA_Ec-R93A had no suppressive effect on the translation level (Fig. 1E), providing direct evidence that the IbpA_Ec-R93A lost the ability to suppress its own translation.

Conservation of self-regulation activity and the significance of R93 in other γ-proteobacterial IbpAs

We noticed that R93 is commonly found in both α- and γ-proteobacterial IbpAs (Figs. 2A and S3). However, R93 is not conserved in gram-positive bacterial IbpAs (or annotated as Hsp20) (Fig. S3). Given the prevalent occurrence of RNAT-mediated sHsp translation regulation in α- and γ-proteobacteria but not in gram-positive bacteria (9, 10), R93 potentially exhibits a comparable significance in both α- and γ-proteobacterial IbpAs as it does in IbpA_Ec.

Figure 2.

Conservation of self-translation regulation and the critical residue in other bacterial IbpAs.A, multiple sequence alignment of Escherichia coli IbpA (IbpA_Ec) and two IbpAs in γ-proteobacteria—Cedecea neteri IbpA (IbpA_Cn), and Vibrio harveyi IbpA (IbpA_Vh). NTDs, ACDs, and CTDs are represented in blue, red, and green, respectively. The amino acids Y34 and R93 are indicated by blue and red arrowheads, respectively. Arg/Lys-rich regions in IbpA ACDs are highlighted with a black box. IbpA_Cn and IbpA_Vh share 93% and 60% amino acid sequence identity with IbpA_Ec, respectively. B, translation suppression activity of IbpA_Cn (left) and IbpA_Vh (right) and their respective R93 mutants evaluated by the GFP reporter assay. The gfp mRNA harboring the 5′ UTRs of ibpA_Cn (left) or ibpA_Vh (right) were expressed in E. coli (WT: BW25113 WT strain; ΔAB: ibpAB operon-deleted BW25114 strain). FtsZ expression level was used as a loading control. C, cell-free translation in the presence or absence of purified IbpA_Cn (left), IbpA_Vh (right), and their respective R93 mutants. The gfp reporter mRNAs carrying either the ibpA_Cn or the ibpA_Vh 5′ UTR were translated in the PURE system. The fluorescence intensity of translated GFP (upper) was quantified and normalized to the control group without IbpAs (denoted as “−”). The data represent the means (±SD) of three independent experiments and were analyzed by one-way ANOVA within the comparison with each -IbpA group (ns: nonsignificant; ∗p < 0.0332; ∗∗p < 0.0021; ∗∗∗p < 0.0002; ∗∗∗∗p < 0.0001, wherever shown). ACD, α-crystallin domain; CTD, C-terminal domain; Ibp, inclusion body-associated protein; NTD, N-terminal domain; UTR, untranslated region.

To investigate the conservation of translational repression activity and the importance of R93 in IbpA_Ec across other γ-proteobacterial IbpAs, we selected two IbpAs from Cedecea neteri (IbpA_Cn) and Vibrio harveyi (IbpA_Vh) (15). Although the homology of these two IbpAs differs from IbpA_Ec (15), R93 is conserved in all IbpAs (Fig. 2A). We overexpressed IbpA_Cn and IbpA_Vh or the corresponding R93 mutants (IbpA_Cn-R93A and IbpA_Vh-R94A) in E. coli with GFP-reporter plasmids harboring the ibpA_Cn and ibpA_Vh 5′ UTRs, respectively, which conform to the structural properties of RNAT (Fig. S4 and Table S1) (11). We found that the overexpression of IbpA_Cn and IbpA_Vh in E. coli suppressed the expression of GFP, which was relieved by the corresponding R93 mutants (Fig. 2B). Furthermore, in the PURE system analysis, purified IbpA_Cn and IbpA_Vh WT suppressed the translation of the gfp reporter, whereas the corresponding R93 mutants had no suppression effect (Fig. 2C). Together, the results show the prevalence of the IbpA-mediated self-translation inhibition activity and the general importance of R93 in IbpAs.

Arg93 in IbpA has irreplaceable importance on IbpA self-suppression

Previous studies on human sHsps have shown that a region in ACD enriched with positively charged residues is associated with human congenital diseases (21, 22). For example, the R116C mutant of human HspB4 is linked to cataract disease, resulting in nonnegligible changes in the structures, chaperone activity, and oligomerization trend (23), while the R116K mutant retains similar properties to HspB4 WT, indicating that the basic amino acid in that position is essential for maintaining HspB4 activities (24). We noticed that the positive charge cluster in ACD is conserved in E. coli, C. neteri, and V. harveyi IbpAs (R83-K98 in E. coli IbpA, as shown in Figs. 2A and 3A) but not in IbpB. Consequently, we investigated the role of conserved basic residues in IbpA-mediated self-suppression. Reporter assays using alanine-substituted mutants (IbpA_Ec-R83A, R97A, and K98A) demonstrated that R97A and K98A, but not R83A, lost IbpA-mediated translation suppression activity similar to R93A (Fig. 3B). We then mutated IbpA_Ec-R93/R97/K98 to K/R to maintain the positions positive charge. We found that R97K and K98R mutants maintained translation suppression activity comparable to IbpA_Ec-WT (Fig. 3C, lanes 17–20), whereas R93K was repression-defective like R93A (Fig. 3C, lanes 15 and 16). The results demonstrate that the positive charges at positions 97 and 98 are sufficient to preserve IbpA suppression activity, but R93 is not replaceable with Lys. To further investigate the exclusive role of Arg in the 93rd position in IbpA self-suppression, we mutated R93 in IbpA_Ec with all other amino acid residues individually. The reporter assay results showed that all other R93 mutants lost the suppression ability (Fig. S5), emphasizing the irreplaceable importance of R93 in IbpA self-regulation.

Figure 3.

The effect of cationic amino acids near Arg93 on IbpA_Ec-mediated self-regulation function.A, partial alignment of IbpA_Ec, IbpB_Ec, and human sHsps (HspB1, HspB4, and HspB5). The region enriched in cationic residues (R/K/H) is enclosed in a box. The conserved Arg/Lys residues in the boxed area are indicated below. B, the effect of Ala mutants of R83, R93, R97, and K98 in IbpA_Ec on the GFP reporter harboring the ibpA_Ec 5′ UTR. WT: Escherichia coli BW25113 WT strain; ΔAB: ibpAB operon-deleted BW25114 strain. C, effect of Arg-to-Lys or Lys-to-Arg mutants of R93, R97, and K98 in IbpA_Ec on the GFP reporter. The experimental details are the same as in (B). D, a hexamer structure of IbpA_Ec-WT predicted by AlphaFold2 using MMseqs2 (34). The structure depicts one of the subunits in light blue for clarity. A zoom-in figure of the subunit is shown below, and the residues Y34, R83, R93, R97, and K98 are colored pink, orange, red, yellow, and green, respectively. The oligomerization motif–IEI in the CTD tail is colored purple. Tyr34 in the NTD loop, which might interact with R93, is marked by a blue arrowhead. E, the effect of IbpA_Ec Y34 mutants (Y34A, Y34R, Y34W, Y34F, and Y34H) on the GFP reporter was evaluated in E. coli, as described in (B). CTD, C-terminal domain; Ibp, inclusion body-associated protein; NTD, N-terminal domain; UTR, untranslated region.

What distinguishes R93 from R97/K98? The AlphaFold2-predicted structure of IbpA_Ec provides a possible explanation; while R93 is located in a flexible loop connecting β6 and β7 sheets, R97/K98 are positioned in the β7 sheet (Fig. 3D). Moreover, R93 resides in close proximity to a disordered NTD loop, implying a potential interaction with NTD. Similar interactions are suggested in the predicted structures of IbpA_Cn and IbpA_Vh (Fig. S6), indicating a common structural feature of R93. We then mutated Y34 in IbpA_Ec, which is the closest residue to R93 in the predicted structure (Fig. 3D) and is highly conserved in IbpAs (Fig. 2A), to several amino acids. Notably, IbpA_Ec-Y34A, Y34R, and Y34H mutants showed a loss of translation suppression activity, while Y34W and Y34F retained this activity (Fig. 3E), suggesting that aromatic residues at position 34 are crucial for the self-regulation function of IbpA. Taken together, these results suggest that the potential interplay between R93 and the NTD loop contributes to the self-suppression ability of IbpA.

The self-regulation of IbpA is not solely dependent on its oligomer size

Given that the function of IbpA relies on its oligomeric structure, we investigated whether R93 mutation could induce conformational changes in IbpA_Ec. Far-UV CD spectra of the purified IbpA_Ec-R93A and IbpA_Ec-WT were almost identical in the wavelength region that determines the secondary structure (Fig. S7, A and B), but sucrose density gradient (SDG) centrifugation analysis revealed that IbpA_Ec-R93A was mainly found at the bottom fraction compared to IbpA_Ec-WT (Fig. 4A), indicating that IbpA_Ec-R93A forms larger assemblies. Transmission electron microscopy (TEM) analysis showed that IbpA_Ec-WT formed fibril-like structures (Fig. 4B), consistent with previous findings (17). IbpA_Ec-R93A formed much longer fibrils than WT (Fig. 4B), in agreement with the tendency to form larger assemblies in the SDG analysis.

Figure 4.

The self-regulation of IbpA is not solely dependent on oligomer size.A, oligomeric states of IbpA_Ec were analyzed using sucrose density gradient (SDG) centrifugation. Purified IbpA_Ec-WT or R93A were applied to 10% to 30% (w/v) sucrose gradient solutions, followed by SDS-PAGE and Western blotting. The distribution of IbpA_Ec was probed with an anti-IbpA antibody. B, representative TEM images of purified IbpA_Ec-WT (left) and R93A (right) are shown, with scale bars of 100 nm. C and D, oligomeric states of IbpA_Cn (C) and IbpA_Vh (D) (WT and their respective R93 mutants) were evaluated using SDG centrifugation and SDS-PAGE with CBB staining to visualize the IbpA distribution, as above. CBB, Coomassie brilliant blue; Ibp, inclusion body-associated protein; SDG, sucrose density gradient; TEM, transmission electron microscopy.

We also examined the oligomeric states of IbpA_Cn, IbpA_Vh, and their corresponding R93 mutants via SDG centrifugation. Unlike IbpA_Ec, there was no significant difference in the oligomer distribution between the WTs and the mutants (Fig. 4, C and D), suggesting that IbpA self-regulation does not solely rely on the oligomer size.

IbpA-R93A is impaired in the interaction with the ibpA 5′ UTR mRNA

We investigated the role of R93 in chaperone activity and mRNA interaction. Under conditions where all heated luciferases sedimented to the bottom in the SDG centrifugation, IbpA_Ec-WT formed oligomers with heat-denatured luciferase in middle fractions (Fig. 5A). IbpA_Ec-R93A also formed oligomers with luciferase in the middle fractions, although less efficiently than IbpA_Ec-WT, indicating that the mutation of R93 partially impairs the chaperone functions. We also evaluated the chaperone activity of IbpA_Cn-R93A and IbpA_Vh-R94A and found that they exhibited client-binding activity indistinguishable from the corresponding WT (Figs. 5A and S8). These results suggest that the R93 mutation may weaken the chaperone activity but is not critical.

Figure 5.

Interaction between IbpA-R93A mutants and denatured proteins or ibpA 5′ UTR mRNAs.A, the interaction between IbpAs and denatured luciferase was analyzed using SDG centrifugation. Luciferase was thermally denatured in the presence or absence of IbpA-WTs or respective R93 mutants, and then the mixtures were applied to 10% to 50% (w/v) sucrose gradient solutions, followed by SDS-PAGE and CBB staining. B, the interaction between IbpA_Ec or the R93A mutant and biotin-labeled ibpA_Ec 5′ UTR-gfp mRNA was evaluated using a filter binding assay. The mRNA and IbpA mixture was applied through double membranes consisting of a nitrocellulose membrane (upper) and a positively charged nylon membrane (lower), to capture IbpA-bound mRNA and free mRNA, respectively. mRNA intensity was detected by streptavidin-HRP. C and D, the interaction of IbpA_Cn-ibpA_Cn 5′ UTR mRNA (C) and IbpA_Vh-ibpA_Vh 5′ UTR mRNA (D) was evaluated by the filter binding assay as described above. The applied amounts of the corresponding IbpA-R93 mutants were the same as the respective WTs. mRNA intensity was detected by the streptavidin Alexa Fluor 647 conjugate. E, the filter binding assay was used to examine the interaction between IbpA_Ec-WT and ibpA 5′ UTR-gfp mRNA in the presence of varying amounts of IbpA_Ec-R93A. mRNA intensity was detected by streptavidin Alexa Fluor 647 conjugate. CBB, Coomassie brilliant blue; HRP, horseradish peroxidase; Ibp, inclusion body-associated protein; SDG, sucrose density gradient; UTR, untranslated region.

Next, we tested the effect of R93 on the interaction with the ibpA 5′ UTR mRNA using a filter binding assay. When the mixture of proteins and RNAs is passed through a positively charged nylon membrane covered with a nitrocellulose membrane, the protein-bound RNAs and free RNAs are trapped in the nitrocellulose and nylon membranes, respectively (25). The filter binding assay confirmed the interaction between purified IbpA_Ec-WT and the biotin-labeled mRNA including the ibpA 5′ UTR in a dose-dependent manner, while there was no interaction of IbpA_Ec-R93A with the mRNA (Fig. 5B). Combined with the results on the complete loss of the mRNA binding in IbpA_Cn-R93A and IbpA_Vh-R94A (Fig. 5, C and D), we conclude that the R93 mutation loses the ability to bind the mRNA.

Furthermore, the addition of IbpA-R93A mutants weakened the mRNA binding ability of IbpA-WTs (Fig. 5, C–E), suggesting the formation of a heterooligomer between IbpA-WTs and the R93A mutants causes the impaired interaction of IbpA-WTs with the mRNAs. This is consistent with the GFP reporter results of IbpA_Ec-R93A; Overexpressed IbpA_Ec-R93A in E. coli WT cells eliminated the suppression effect of endogenous IbpA_Ec-WT at the reporter translation level (Fig. 1D, lanes 13 and 14).

IbpB enhances the suppression ability of IbpA-WT but has no effect on IbpA-R93A

In E. coli, IbpA_Ec and IbpB_Ec prefer the formation of heterospecies over homospecies (16) to mitigate aggregation-induced stress (4, 6). Subsequently, we investigated the impact of IbpB_Ec-WT on IbpAB hetero-oligomer formation and IbpA-mediated translation suppression. SDG analysis showed that both IbpA_Ec-WT and R93A formed heterocomplexes with IbpB_Ec, which were smaller than corresponding homo-oligomers, particularly IbpA_Ec-R93A (Fig. 6A). TEM analysis corroborated this trend. The addition of IbpB_Ec hindered the formation of fibril-like structures of IbpA_Ec-WT, as demonstrated previously (17), and resulted in a shorter fibril formation in IbpA_Ec-R93A (Fig. 6B).

Figure 6.

IbpB_Ecenhances the translation suppression ability of IbpA_Ec-WT but exerts no effect on IbpA_Ec-R93A.A, interaction between IbpA_Ec (WT or R93A) and IbpB_Ec assessed by SDG centrifugation. Purified IbpA_Ec (3 μM) in the presence or the absence of IbpB_Ec (7 μM) was subjected to 10% to 30% (w/v) sucrose gradient solutions, as shown in Figure 4. Collected fractions were analyzed by SDS-PAGE with 6 M urea to separate IbpA_Ec and IbpB_Ec, and visualized by CBB staining. B, representative TEM images of purified IbpA_Ec-WT (upper) and R93A (lower) in the presence of IbpB_Ec. The scale bars represent 100 nm. C, effect of IbpA_Ec on the cell-free translation of the ibpA_Ec 5′ UTR-gfp reporter using the PURE system in the absence or presence of IbpB_Ec. Upper: the fluorescence intensity of the translated GFP; lower: the fold-change in GFP translation level. GFP expression level without both IbpA_Ec and IbpB_Ec was used for the normalization and set to 1. D, IbpA_Ec-mRNA interaction in the presence or absence of IbpB_Ec assessed by filter binding assay. Upper: the biotin-labeled mRNA intensity detected by streptavidin Alexa Four 647 conjugate; lower: the quantified intensity fold of IbpA_Ec-bound mRNA compared to that in the absence of IbpB_Ec. All data represent the means (±SD) of three independent experiments and were analyzed by one-way ANOVA. CBB, Coomassie brilliant blue; Ibp, inclusion body-associated protein; SDG, sucrose density gradient; TEM, transmission electron microscopy; UTR, untranslated region.

Furthermore, we assessed the effect of IbpB_Ec on IbpA-mediated translation suppression. We validated that IbpB_Ec had no impact on the translation of ibpA 5′ UTR-gfp in the PURE system, whereas it enhanced the IbpA WT-mediated translation suppression (Fig. 6C). However, the absence of translation repression by IbpA_Ec-R93A remained unchanged in the presence of IbpB_Ec (Fig. 6C). In the filter binding assay, we observed that the presence of IbpB_Ec enhanced the interaction of IbpA_Ec-WT with the ibpA 5′ UTR mRNA, while IbpB_Ec did not alter the property of IbpA_Ec-R93A to not bind to the mRNA (Fig. 6D).

Discussion

After discovering the nonconventional function of IbpA_Ec in suppressing its own translation (18), there remains several fundamental questions to address. This study sheds light on some of the mechanisms underlying the unique function of IbpA that its paralog, IbpB, lacks. First of all, it is worth noting that not only IbpA_Ec but also other γ-proteobacterial IbpAs, C. neteri and V. harveyi IbpAs (IbpA_Cn and IbpA_Vh), have the function to suppress their own translation through the interaction with their 5′ UTR mRNAs, indicating that the IbpA-mediated self-regulation function is evolutionally conserved.

In exploring the difference between IbpA_Ec and IbpB_Ec, we found that a cluster of positively charged amino acids within the ACD, which is not conserved in IbpB_Ec, is crucial for the translation suppression activity (Fig. 3B). Within the cluster, the R93, R97, and K98 residues in the IbpA_Ec are conserved in those in other γ-proteobacterial IbpAs, suggesting a key role of these R/K residues in discriminating between IbpA and IbpB. Importantly, human sHsps (HspB4 and HspB5) also possess the conserved positively charged residues, mutations of which are associated with diseases (21, 22). Although there is no evidence for the translation suppression activity in human sHsps, the positive charge clusters in the ACD are likely of common importance for sHsps in both prokaryotes and eukaryotes.

Among the conserved positive charge residues, the R93 position in IbpA_Ec has a unique role since it cannot be replaced by any of the other 19 amino acids, including Lys (Fig. S5). In contrast, R97 and R98 can be replaced by Lys, indicating that the cationic property of these residues is sufficient. It is noteworthy that this interconversion between K and R is largely the same trend observed in the positive charge cluster in human sHsps (24). An AlphaFold2-predicted hexamer structure of IbpA_Ec revealed that R93, R97, and K98 are in close proximity to NTD (Fig. 3D), suggesting a potential interplay between the region and NTD. Unlike R97 and K98, which are located in a stable β7 sheet, R93 is situated in a flexible loop that bridges β6 and β7. Mutants of Y34 in a disordered loop of NTD, which is closest to R93 (Fig. 3D), abolish the translation suppression activity, unless the substitution is to aromatic residues (Fig. 3E) (see below). These results support the notion that the potential interaction between R93 and Y34 is responsible for the activity. Similar potential interactions are also apparent in the predicted IbpA_Cn and IbpA_Vh structures (Fig. S6), further supporting this notion. Moreover, R93 and Y34 are also well-conserved in α-proteobacterial IbpAs (Fig. S3A), suggesting conservation of self-suppression in α-proteobacterial IbpAs. We note that IbpB_Ec also has an aromatic residue (F32) in the position corresponding to Y34 in IbpA_Ec, which is consistent with the observation that the NTD-substituted chimera retained the self-suppression activity (Fig. S1A).

Why is it only possible for Arg to be in position 93, but not Lys? Although Arg and Lys are typically considered equivalent in their positive charge, there are notable differences in their properties. Among these differences, it is worth noting that Arg is abundant in RNA-binding proteins, as observed in previous studies (26, 27, 28, 29). Furthermore, Arg is found more frequently at protein–protein interfaces than Lys (30, 31), which makes it a “stickier” amino acid. It is plausible that the “stickiness” of Arg, which could be involved in RNA binding, contributes to the ability of IbpAs to bind to mRNA, thus repressing self-translation.

When considering the irreplaceable nature of R93 and the importance of aromatic residues at the Y34 position in NTD, it is tempting to consider the possibility of a cation-π interaction in the mRNA binding activity. IbpA does not bind any mRNA indiscriminately, but rather seems to bind RNAs with secondary structure, such as RNATs (18). The cation-π interaction involving a specific RNA structure may be crucial to the nonconventional function of IbpA. Additionally, since the cation-π interaction is one of the driving forces for undergoing liquid-liquid phase separation (LLPS) (26, 27, 28, 29), there is a possibility of LLPS in the IbpA-mediated translation suppression function, although our rationale for investigating the Tyr residue is based on an AlphaFold2-predicted structure that suggests intramolecular interaction. Nonetheless, a previous report on sHsps, which suggests HspB2 concentration-dependent LLPS formation in cells (32), implies a possible involvement of LLPS in the IbpA-mediated self-regulation activity.

We previously showed that IbpA_Ec’s self-translation suppression relies on its interaction with the ibpA 5′ UTR mRNA using a gel shift assay (18). In this study, we developed a filter binding assay using purified protein and mRNA to analyze the IbpA_Ec-mRNA interaction (Figs. 5 and 6). The filter binding assay revealed that IbpA_Ec-R93A completely lost the ability to interact with the ibpA 5′ UTR mRNA, as expected from the in vivo reporter assay and the PURE system analysis. In contrast, the chaperone activity of IbpA_Ec-R93A, as assessed by binding to heat-denatured proteins, showed only a small decrease. Furthermore, the chaperone activity of the R93 mutants in other γ-proteobacterial IbpA (IbpA_Cn and IbpA_Vh) was indistinguishable from that of the WTs. Therefore, the R93 mutation mainly abolishes the mRNA binding activity of IbpAs, indicating that the mRNA-binding activity of IbpAs and its chaperone activity to sequester denatured proteins are distinguishable.

The AlphaFold2-based predicted structures indicate no apparent difference in dimer structures between the WT and the R93 mutants for all IbpAs examined in this study (Fig. S9). This would explain why the R93 mutants possess the same or similar chaperone activity (Figs. 5A and S8). IbpA_Ec-R93A tends to form higher-order oligomers as well as longer fibril-like structures (Fig. 4, A and B). It is possible that the chaperone activity of IbpA_Ec-R93A is slightly defective due to its propensity to form larger oligomers and fibril-like structures, which are known to be inactive as chaperones, as previously reported (17).

The self-suppression function of IbpAs relies on the oligomers, not the dimers. Although the oligomerization behavior of IbpA_Ec-R93A is different from that of the WT, the oligomeric states of the other γ-proteobacterial IbpA-R93 mutants were almost identical to those of the WT IbpAs (Fig. 4, C and D). In addition, even when IbpA_Ec-R93A forms smaller oligomers in the presence of IbpB_Ec, it is still suppression-inactive (Fig. 6, C and D). Collectively, these results strongly suggest that the defect of R93 mutants in the translation suppression activity through impaired mRNA binding is not caused by size variation in oligomers, implying a more complex mechanism behind the suppression activity.

The incorporation of IbpA-R93A into IbpA-WT impaired the WT-mRNA interaction (Fig. 5, C–E). If we assume that the hetero-oligomer formation is not compromised in the R93A mutant, the incorporation of the mutants into the WT oligomer perturbs the WT-mediated mRNA binding. In contrast to R93A, IbpB_Ec can strengthen the interaction between IbpA_Ec-WT and mRNA, possibly due to the reduced fibril formation of IbpA_Ec-WT (Fig. 6B). Furthermore, the impaired suppression activity of IbpA_Ec-R93A cannot be activated in the presence of IbpB_Ec (Fig. 6, C and D), even though R93A interacts with IbpB_Ec and forms more nonfibrous structures (Fig. 6A). All of these collectively elucidated why the expression levels of the reporter GFP, in E. coli WT and ΔAB cells overexpressing IbpA_Ec-R93A, were comparable (Fig. 1D, lanes 13 and 14). This would be attributed to the impaired suppression activity of endogenous IbpA_Ec-WT by overexpressed IbpA_Ec-R93A, coupled with the inability of the endogenous IbpB_Ec to activate the suppression function of IbpA_Ec-R93A, despite the potential formation of heterocomplexes of IbpA_Ec-R93A with endogenous IbpA_Ec-WT and IbpB_Ec in vivo.

The AlphaFold2-predicted structures indicate that the monomeric structure of the R93A mutant is nearly identical to the WT (Fig. S10), as expected for dimeric structures (Fig. S9). Previous analyses have already revealed that sHsp dimerization is promoted by the β6 loop of one ACD interacting with a β2 strand of the partner’s ACD, while oligomerization is induced by the interplay between the IXI/V motif in CTD and a β4/β8 groove of the neighboring dimer (4). The formation of heterodimers and hetero-oligomers of IbpA-IbpB has been found to be essentially the same as that of a single sHsp (16). All of these interacting regions are independent of the β6/β7-bridging loop region containing R93 (Fig. 3D) which ensures the interplays of IbpA-R93A with IbpA-WT and IbpB.

Overall, this study demonstrates that IbpA-mediated self-suppression function is conserved in γ-proteobacteria, and that the cationic residues in ACD, particularly R93, are essential for this function. Since the Arg93 mutations in IbpA ACD regions have little effect on chaperone activity, the conservation of Arg93 in γ-proteobacterial IbpAs implies that the translation suppression function would be critical for the cellular function of IbpAs, distinct from their chaperone role as a sequestrase.

Experimental procedures

Plasmid construction and cloning

Constructions of pBAD30-ibpA_Ec 5′ UTR-gfp and pCA24N-ibpA_Ec were reported previously (18). The ibpAB_Ec chimeras were produced by standard cloning procedures and Gibson assembly. Single mutations in pCA24N-ibpA_Ec were achieved by site-directed mutagenesis. The pCA24N-ibpA_Cn and pCA24N-ibpA_Vh were individually constructed by subcloning of ibpA_Cn and ibpA_Vh DNA fragments into pCA24N using Gibson assembly. The ibpA_Cn and ibpA_Vh DNA fragments were amplified from pET3a-ibpA_Cn and pET3a-ibpA_Vh plasmids, which were kindly provided by Dr Krzysztof Liberek (15). The IbpA_Cn-R93A and IbpA_Vh-R94A mutants were also constructed individually by site-direct mutagenesis from pCA24N-ibpA_Cn and pCA24N-ibpA_Vh. The pBAD30-ibpA_Cn 5′ UTR-gfp and the pBAD30-ibpA_Vh 5′ UTR-gfp were generated by subcloning of the ibpA_Cn 5′ UTR and the ibpA_Vh 5′ UTR DNA fragments into pBAD30-gfp, respectively. The DNA fragments of the ibpA_Cn and the ibpA_Vh 5′ UTRs were amplified from DNA oligos commercially produced by Integrated DNA Technologies. The sequence information of the ibpA_Cn and the ibpA_Vh 5′ UTRs was obtained from the National Library of Medicine NBRC 105707 and NBRC 15634. All plasmids were amplified in E. coli DH5α. Primers used for cloning and mutagenesis were shown in Table S2.

Reporter assay

The pBAD30-ibpA 5′ UTR-gfp plasmids, as reporter plasmids carrying gfp gene with ibpA 5′ UTR sequence in the upstream, were transformed alone or cotransformed with IbpA-expression plasmids, harboring ibpAB chimera/ibpA mutant sequences in pCA24N vector, into E. coli BW25113 WT and a ΔibpAB-deleted strain (18). After preculture in LB medium at 37 °C with 180 rpm overnight, the cells were diluted to fresh LB medium with the induction of 2 × 10⁻⁴% arabinose until growing to A₆₆₀ of 0.2 ∼ 0.3. The cells containing only pBAD30-ibpA 5′ UTR-gfp plasmids were incubated for more 30 min, while the cells possessing both pBAD30-ibpA 5′ UTR-gfp and pCA24N-ibpA plasmids were further induced with 0.1 mM IPTG for 30 min to achieve the overexpression of exogenous IbpAs. Next, the cells were harvested and treated with the same volume of 10% trichloroacetic acid to precipitate proteins. After incubation on ice for 15 min and centrifugation at 20,000g for 3 min at 4 °C, the pellets were washed by ice-cold acetone and then centrifugated again. After two times of acetone washing, 1× SDS sample buffer (0.25 M Tris–HCl pH 6.8, 5% SDS, 5% (w/v) sucrose, 0.005% (w/v) bromophenol blue, and 5% (w/v) 2-mercaptoethanol) was applied to dissolve the pellets, and then the samples were incubated at 37 °C for 15 min. The SDS-treated samples were applied to 12% polyacrylamide SDS gels and then transferred from the gels to polyvinylidene fluoride membranes based on standard immunoblotting procedures. The membranes were blocked with 1% skim milk in TBS-T buffer (20 mM Tris–HCl pH 7.5, 137 mM NaCl, and 0.2% (w/v) Tween 20). Mouse anti-sera against GFP (mFx75, Wako) or rabbit anti-sera against FtsZ (kindly provided by Dr Shinya Sugimoto from Jikei Medical University) was used as the primary antibody with the dilution of 1:10,000. The secondary antibody was horseradish peroxidase (HRP)-conjugated anti-mouse or anti-rabbit IgG (Sigma-Aldrich) used in the same dilution factor. Subsequently, the samples were visualized by Dual Chemiluminescent substrates (Millipore) and detected by a LAS 4000 mini imager (Fujifilm).

Protein purification

To purify IbpA_Ec and R93A mutants, E. coli BW25113ΔAB cells at A₆₆₀ of 0.5 were used to overexpress IbpA_Ec-WT and -R93A with the induction of 0.5 mM IPTG at 37 °C for 4 h. The harvested cells were lysed by sonication (Branson Ultrasonics) in buffer A (50 mM Hepes–KOH pH 7.6, 10% glycerol, 1 mM DTT, and 0.1 M KCl). The following anion exchange chromatography was performed as described (18). His-IbpB_Ec was purified with Ni-NTA agarose upon the denaturation by 6 M urea. Purification of IbpA_Cn WT and R93A expressed in E. coli was the same as that of IbpA_Ec. To purify IbpA_Vh WT and R94A, pET3a-ibpA_Vh or ibpA_Vh R94A was overexpressed in E. coli BL21(DE3) cells at A₆₆₀ of 0.5 with the induction of 10 μM IPTG at 28 °C for 20 h. The harvested cells were lysed by sonication in buffer B (50 mM Tris–HCl pH 7.5, 10% glycerol, and 1 mM DTT). IbpA_Vh was soluble in E. coli cells, so supernatants of the lysates were collected by low-speed centrifugation (10,000g, 10 min, 4 °C) to remove E. coli inclusion bodies including endogenous IbpA_Ec. Then, one more centrifugation with high speed (30,000 rpm, 30 min, 4 °C) was carried out to remove the pellet of membrane vesicles and ribosomal particles. Next, the supernatants were applied to QAE resin (Toyopearl, Tosoh), and the flowthrough fractions were collected since most of the native IbpA_Vh-WT and IbpA_Vh-R94A were found in these fractions. Then the flow-throughs were applied onto fresh QAE resin after being dialyzed against buffer C (50 mM Tris–HCl pH 7.5, 10% glycerol, 1 mM DTT, and 6 M urea). The denatured IbpA_Vh was eluted by increasing salt concentration to 200 mM NaCl in buffer C. The IbpA_Vh-containing fractions were then dialyzed against buffer C and applied to another round of QAE chromatography with the elution of decreasing pH from buffer C to buffer D (50 mM citric acid-NaOH pH 5, 10% glycerol, 1 mM DTT, and 6 M urea). Finally, urea was gradually removed by dialysis against buffer E (50 mM Tris–HCl pH 7.5, 10% glycerol, 1 mM DTT, and 0.1 M KCl) from buffer E containing 4 M urea, 2 M urea, 1 M urea to buffer E without urea. Purification of firefly luciferase was performed as described previously (33). Protein concentrations were determined by the Bradford method with standard bovine serum albumin. All concentrations of IbpA in this study were in dimeric units.

Reconstituted cell-free translation

PUREfrex (GeneFrontier) with Cy5-labeled tRNA^fMet was used to express RNA templates produced by CUGA7 in vitro transcription kit (Nippon Gene) in the presence or the absence of purified IbpAs or the mutants (1 μM). The reaction mixtures were incubated at 37 °C for 2 h, then mixed with the same volume of 2× SDS sample buffer (0.5 M Tris–HCl pH 6.8, 10% SDS, 10% (w/v) sucrose, 0.01% (w/v) bromophenol blue, and 10% (w/v) 2-mercaptoethanol), and boiled at 95 °C for 5 min. The mixtures were separated by SDS-PAGE, and visualized by an Amersham Typhoon scanner (Cytiva), and finally quantified by Multi gauge software (Fujifilm).

SDG centrifugation

To investigate the oligomeric states of the purified IbpAs and the mutants, the proteins (3 μM) in buffer F (50 mM Hepes–KOH pH 7.6, 0.1 M KCl, 5 mM DTT, and 20 mM Mg-acetate) were applied onto a 11 ml 10% to 30% (w/v) sucrose gradient in buffer F and then ultracentrifuged with a Beckman SW 41 Ti rotor (35,000 rpm, 4 °C, 80 min). The centrifuged samples were collected from the top to the bottom by a fractionator (BioComp). Then, the top six fractions, the bottom six fractions, and the aggregates attached to the tube bottom were analyzed by SDS-PAGE, and detected by standard Western blotting procedures using rabbit anti-sera against IbpA (Eurofin) as primary antibody and HRP-conjugated anti-rabbit IgG as secondary antibody (Sigma-Aldrich). For IbpA_Cn and IbpA_Vh detection, Coomassie brilliant blue (CBB) was used for visualization.

To investigate the interaction of IbpAs with substrate proteins, the purified IbpAs (12 μM) and the purified luciferase (3 μM) in buffer F were mixed and incubated at 50 °C for 30 min. After that, the mixtures were applied onto a 10% to 50% (w/v) sucrose gradient and then centrifuged as described above. The protein distributions were verified by SDS-PAGE and visualized by CBB staining.

To investigate interaction with IbpB_Ec, the mixtures of IbpA_Ec (3 μM) and IbpB_Ec (7 μM) in buffer F were incubated at room temperature for 30 min and then applied onto a 10% to 30% (w/v) sucrose gradient. Centrifugation and fraction collection were performed as above. The fractions were analyzed by SDS-PAGE in the presence of 6 M urea followed by CBB staining.

Transmission electron microscopy

Purified IbpA_Ec-WT or IbpA_Ec-R93A (2 μM) in the absence or presence of IbpB_Ec (4.8 μM) was applied on carbon-coated copper grids. The samples were allowed to absorb for 1 min before negatively stained with 1% methylamine tungstate at pH 7 for 1 min. The staining was repeated twice. The observation was performed with a JEOL 1400 Plus electron microscopy.

Filter binding assay

The ibpA 5′ UTR-gfp mRNA produced from CUGA7 in vitro transcription kit (Nippon Gene) was attached to a 3′-terminal biotinylated nucleotide using Pierce RNA 3′ End Biotinylation kit (Thermo Fisher Scientific). The biotin-labeled mRNA (0.1 μM) was incubated with different ratio of IbpA in buffer G (100 mM sodium phosphate buffer pH 7.5, 0.1 M NaCl, 5 mM EDTA, 5 mM DTT, and 10% glycerol) at room temperature for 30 min after which the mixtures were fixed with 1% formaldehyde for 10 min followed by an addition of 0.25 M glycine to stop the cross-linking reactions in 5 min. A nitrocellulose membrane (Amersham Protran 0.2 μM NC, GE HealthCare, Life Sciences) was presoaked in buffer G and then overlaid on a positively charged nylon membrane (BrightStar-Plus, Invitrogen). The protein-mRNA mixtures were applied onto a 96-well slot-blot apparatus and then filtered through the double membranes by vacuum. The protein-mRNA complexes were trapped in the top nitrocellulose membrane, while the free mRNA samples passed through the nitrocellulose membrane and were caught by the bottom nylon membrane. Finally, biotin-labeled mRNAs were detected with streptavidin-HRP (Thermo Fisher Scientific) according to the protocol prepared for Chemiluminescent Nucleic Acid Detection Module kit (Thermo Fisher Scientific). To examine the interaction of IbpA_Ec-WT with the mRNA in the presence of IbpA_Ec-R93A, the mixture containing the mRNA (0.1 μM), IbpA_Ec-WT (1 μM) and different ratios of IbpA_Ec-R93A was applied to the filter assay, and the biotin-labeled mRNA was detected by streptavidin Alexa Fluor 647 conjugate (Invitrogen). To investigate the effect of IbpB_Ec on IbpA_Ec-mRNA interaction, the mixture containing the mRNA (0.1 μM), IbpA_Ec (1 μM) and IbpB_Ec (2.4 μM) was applied to the filter assay, and the biotin-labeled mRNA was detected by streptavidin Alexa Fluor 647 conjugate (Invitrogen).

Statistical analysis

One-way ANOVA was used for calculating statistical significance. All experiments were conducted at least three times independently, and the mean values ± SD were represented in the figures.

Data availability

Data in this manuscript have been uploaded to the Mendeley Dataset public repository (https://doi.org/10.17632/gj92kb2wdd.1).

Supporting information

This article contains supporting information.

Conflict of interest

The authors declare that they have no conflicts of interest with the contents of this article.

Reviewed by members of the JBC Editorial Board. Edited by Ursula Jakob