Gil‑Garcia et al. Microb Cell Fact (2020) 19:117 https://doi.org/10.1186/s12934-020-01375-4 Microbial Cell Factories RESEARCH Open Access Coiled‑coil inspired functional inclusion bodies Marcos Gil‑Garcia, Susanna Navarro and Salvador Ventura* Abstract Background: Recombinant protein expression in bacteria often leads to the formation of intracellular insoluble protein deposits, a major bottleneck for the production of soluble and active products. However, in recent years, these bacterial protein aggregates, commonly known as inclusion bodies (IBs), have been shown to be a source of stable and active protein for biotechnological and biomedical applications. The formation of these functional IBs is usually facilitated by the fusion of aggregation-prone peptides or proteins to the protein of interest, leading to the formation of amyloid-like nanostructures, where the functional protein is embedded. Results: In order to offer an alternative to the classical amyloid-like IBs, here we develop functional IBs exploiting the coiled-coil fold. An in silico analysis of coiled-coil and aggregation propensities, net charge, and hydropathicity of different potential tags identified the natural homo-dimeric and anti-parallel coiled-coil ZapB bacterial protein as an optimal candidate to form assemblies in which the native state of the fused protein is preserved. The protein itself forms supramolecular fibrillar networks exhibiting only α-helix secondary structure. This non-amyloid self-assembly propensity allows generating innocuous IBs in which the recombinant protein of interest remains folded and func‑ tional, as demonstrated using two different fluorescent proteins. Conclusions: Here, we present a proof of concept for the use of a natural coiled-coil domain as a versatile tool for the production of functional IBs in bacteria. This α-helix-based strategy excludes any potential toxicity drawback that might arise from the amyloid nature of β-sheet-based IBs and renders highly active and homogeneous submicromet‑ ric particles. Keywords: Coiled-coil protein, Fusion tag, Functional inclusion bodies, Fluorescent proteins, Protein engineering Background conditions, according to the “living on the edge” hypoth- Biotechnological and pharmaceutical industries exploit esis [5]. The heterologous expression of these molecules microorganisms as cell factories in order to produce their in bacteria leads to intracellular concentrations that are biological products, including therapeutic proteins, such several times above their natural solubility limits. As a as hormones, enzymes for replacement therapies, or anti- result, these proteins might establish non-native inter- bodies [1, 2]. However, the production of these molecules molecular interactions, which would facilitate their in their soluble and functional states faces significant bar- aggregation into inclusion bodies (IBs) in the bacterial riers [3, 4]. Proteins have been shaped by natural selec- cytosol [6]. tion to remain soluble and functional under physiological Traditionally IBs were thought to be formed by mis- folded conformations and thus devoid of any function- ality. However, data is accumulating to indicate that, at *Correspondence: salvador.ventura@uab.cat least for specific proteins and production conditions, Institut de Biotecnologia i de Biomedicina and Departament de Bioquímica i Biologia Molecular, Universitat Autònoma de Barcelona, IBs might exhibit significant activity [7, 8]. IBs are easy 08193 Bellaterra, Barcelona, Spain to purify and can be stored for long periods, thanks © The Author(s) 2020. This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creat​iveco​ mmons​.org/licen​ses/by/4.0/. The Creative Commons Public Domain Dedication waiver (http://creat​iveco​mmons​.org/publi​cdoma​in/ zero/1.0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data. Gil‑Garcia et al. Microb Cell Fact (2020) 19:117 Page 2 of 16 to their inherent stability [9]. They have a nanometric involved in Z-ring formation during the bacterial cell size (50–1000 nm) [9–12], and ~ 90% of them are com- division process [42, 43]. posed of the target protein [13–15]. These properties, We first show that the ZapB self-assembles to form combined, make them active nanoparticles, which are α-helix-based fibrillar networks, and afterward, we dem- finding increasing applications in biotechnology and onstrate how this property allows its use as a tag to form biomedicine. In this way, the ability to immobilize non-amyloid and non-toxic IBs which preserve the activ- enzymes in IBs has been exploited to build up reusable ity of the attached polypeptides. catalysts [16, 17], and IBs have been used as nanocar- riers or/and nanopills to deliver antitumoral polypep- Results and discussion tides in the body [18, 19]. Selection of a polar and non‑aggregating coiled‑coil We have shown that, generically, a significant pro- protein for the production of functional IBs portion of the protein contacts that sustain IBs have an A wide range of fusion tags has been used to induce IBs amyloid-like nature [20–22]. When embedded in IBs, formation. They comprise small artificial peptides [44], proteins exhibit a significant increase in β-sheet content, and aggregation-prone natural proteins or domains [22, relative to their soluble counterparts, and, often, become 37, 45], which are fused to functional globular proteins. A able to bind typical amyloid dyes. For specific proteins characteristic property of most of these tags is that they or protein fusions, this amyloid scaffold coexists with promote the formation of aggregates sustained by collec- functional conformations [23, 24]. However, because the tive intermolecular β-sheet interactions. Alternatively, process of protein aggregation into non-native intermo- two different coiled-coil domains have been used for the lecular β-sheet structures necessarily involves the popu- production of fluorescent and/or catalytically IBs [37]. lation of misfolded species, a fraction of the recombinant Nevertheless, the biophysical properties of these active protein is necessarily inactivated to build up the amyloid aggregates were not assessed, and thus it is not known if structure that sustains the IB [22]. the coiled-coil encoding sequences keep their native heli- IBs have been assimilated to natural functional amy- cal structure in the IBs or they had just transitioned to loids [25], which are non-toxic for their host cells or a conventional amyloid-like assembly. Indeed, the forma- organisms [26–28]. However, different studies indicate tion of β-sheet-rich amyloid fibrils by aggregation-prone that this is not because functional amyloids are intrin- coiled-coil sequences is behind the onset of several neu- sically non-cytotoxic [29], but instead because, in these rodegenerative disorders [46]. A similar transition in specific cases, nature has evolved dedicated mecha- coiled-coil-tagged proteins during IBs formation might nisms to prevent amyloid-associated toxicity [27, 30, 31]. turn these aggregates potentially toxic. The artificial formation of IBs lack these natural control We selected the E. coli protein ZapB as a scaffold to mechanisms and, although bacterial aggregates have obtain functional IBs. ZapB is an 81 residues-long pro- been assumed to be innocuous, it cannot be entirely dis- tein whose 3D-structure (PDB: 2JEE) consists of two carded that toxic β-sheet conformations can be incorpo- α-helical polypeptide chains arranged in anti-parallel rated or released from these amyloid-like inclusions. orientation to form a dimeric coiled-coil of 116 Å (PDB: In order to overcome the two above-described limita- 2JEE) [42]. In the crystal structure, individual coiled-coils tions, we introduce here an IBs production strategy that interact close to their termini, which already suggested exploits a natural coiled-coil protein to promote non- that, under appropriate conditions, these helical modules amyloid supramolecular interactions. In the last years, might self-assemble into supramolecular structures [42]. the use and design of coiled-coil domains as building The propensity to form a stable coiled-coil assembly in blocks in protein assemblies have attracted significant solution is encoded in the protein sequence. The higher attention [32, 33]. The self-organization capacity of these the coiled-coil propensity, the lowest the probability to structures has been exploited to create different nano- transition into an aggregated β-sheet structure since sta- structures, such as nanofibers [34] and nanocages [35, ble α-helices protect against aggregation [47, 48]. We cal- 36]. Furthermore, two different coiled-coil domains have culated the coiled-coil propensity of ZapB and compared been previously used to generate active IBs: the tetramer- it with that of the two coiled-coil domains used as IBs ization domain of the tetrabrachion protein (TDoT) from formation tags in previous studies (3HAMP and TDoT) Staphylothermus marinus [16, 37–40] and the 3HAMP using four different algorithms: COILS [49], PCoils [50], coiled-coil, which was derived from the oxygen sensor MARCOIL [51] and DeepCoil [52]. Additional file 1: protein Aer2 from Pseudomonas aeruginosa [37, 41]. Figures S1–S3 show the coiled-coil probability profiles In this work, we apply this strategy to build up func- for ZapB, 3HAMP and TDoT. The four algorithms coin- tional IBs using ZapB, a non-essential Escherichia coli cide to predict a very high coiled-coil propensity along (E. coli) protein consisting of two anti-parallel α-helices, the complete ZapB sequence. In the case of 3HAMP, the Gil‑Garcia et al. Microb Cell Fact (2020) 19:117 Page 3 of 16 programs identify a region of high propensity close to Polarity and hydrophobicity are critical negative and the N-terminus and two additional stretches with low to positive contributors to protein sequence aggregation pro- moderate propensity. This is consistent with the homo- pensity, respectively. Their role is crucial at the nucleation dimeric 3HAMP structure, in which parallel monomers step of the aggregation reaction [60, 61]. These biophysical exhibit three successive domains (HAMP1, 2, and 3), properties were analyzed by calculating the Grand Average each about 50 residues long and bridged by flexible link- of Hydropathicity (GRAVY score) provided by the Prot- ers. For TDoT, only DeepCoil is able to identify a sig- Param server, according to the Kyte-Doolittle scale [62]. nificant coiled-coil propensity in the central part of the Positive GRAVY scores indicate higher hydrophobicity, sequence. This makes sense, since TDoT is a parallel and whereas negative values correspond to polar sequences. right-handed coiled-coil tetramer, which is based on the As it is shown in Fig. 2c, ZapB has the more negative 11-residue repeat, and COILS, PCoils and MARCOIL GRAVY score (− 0.980), being this value three times were trained to identify canonical heptad repeats, where higher than the secondly ranked sequence (VP1, GRAVY DeepCoil was aimed to identify both kinds of periodici- score = − 0.307), thus indicating that ZapB is a highly polar ties. Therefore, we used this last algorithm to compare protein, relative to compared sequences. the average coiled-coil probabilities of ZapB, TDoT and We also checked the secondary structure propensity of 3HAMP primary sequences. As it can be seen in Fig. 1 the six tags. Cryptic regions of significant β-sheet pro- ZapB seems to be a better coiled-coil former than the pensity might exist even in sequences that usually fold proteins it has been compared to. into α-helices [63]. These stretches might favor aggre- Not only the coiled-coil propensity accounts for the gation into amyloid-like structures upon coiled-coil ability to maintain the native state in the multimeric unfolding or once the polypeptide chain emerges from state, but also the protein’s intrinsic aggregation pro- the ribosome [64]. To this aim, we used the PSIPRED pensity, which facilitates the conformational shift to [65] and GOR [66] servers. ZapB is predicted to be com- aggregated β-sheet states. We analyzed this property pletely α-helix, with an extremely high propensity for this for ZapB, TDoT, 3HAMP, and other three non-coiled- secondary structure; meanwhile, 3HAMP and specially coil sequences used previously as IB-tags, namely the TDoT are predicted to have a significantly lower α-helical amyloid β-peptide (Aβ42) [53, 54], the viral capsid propensity and predicted β-strand segments are identi- peptide VP1 [45, 55], and the signal sequence of E. coli fied at their sequences (Additional file 1: Figures S4, S5). TorA (ssTorA) [56, 57]. In order to do that, we used The above-described analyses converge to indicate two of the most popular sequence-based aggregation that the sequence of ZapB is more polar, less aggrega- prediction servers, Aggrescan [58] and TANGO [59]. tion-prone, devoid of cryptic β-sheet regions, and with As it can be seen in Fig. 2a, b, ZapB was predicted as highest α-helical and coiled-coil propensities than any the least aggregation-prone sequence in this polypep- previously used IBs tag. Each of these individual prop- tide set. Aggrescan (Fig. 2a), predicts ZapB to be the erties disfavors the potential aggregation of the ZapB most soluble sequence with a significant difference, sequence into β-sheet-rich aggregates. However, aggrega- relative to the other proteins or domains. In the case tion can still occur from the folded state of the proteins of TANGO (Fig. 2b), ZapB remains as the less aggrega- [67]. We have recently developed AGGRESCAN3D, an tion-prone sequence, in this case, together with ssTorA. algorithm that allows predicting a protein’s aggregation- propensity taking into account the structural context [68–70]. When we analyzed the published 3D structures of the three coiled-coil domains (ZapB, PDB:2JEE; TDoT, PDB:1FE6; and 3HAMP, PDB:3LNR), ZapB turned to be the less aggregation-prone structure, displaying a highly soluble surface (Additional file 1: Figure S6). Finally, we assessed the net charge of the different IBs tagging sequences. This value is essential because if the high predicted solubility of ZapB comes together with a high positive charge, this will be a significant drawback for its use in nanomedical applications. Cationic sequences bind to negatively charged nucleic acids and cell membranes, which then become incor- porated into IBs during their formation or purification, Fig. 1 Coiled-coil probability for ZapB, TDoT, and 3HAMP. The making them compositionally heterogeneous. The cal- coiled-coil propensity is represented according to DeepCoil predictor culated net charge for the six tags is represented in Gil‑Garcia et al. Microb Cell Fact (2020) 19:117 Page 4 of 16 Fig. 2 Aggregation propensity predictions, hydrophaticity scores, and net charge values. Aggregation propensity values calculated with a Aggrescan (Na4vSS value) and b TANGO (AGG​ value) for the six selected tags. For both predictors, the more positive the value, the higher the aggregation propensity. c GRAVY hydrophaticity score of the different tags according to the Kyte-Doolittle scale. d The calculated net charge of the sequences at physiological pH Fig. 2d. ZapB is the most acidic of the sequences (net the C-terminus of the full-length protein, and it was charge − 8), the net charge being two times higher recombinantly produced in E.coli at 30 °C. The protein than that of the second most anionic tag (TDoT, net was expressed at high yield (> 1 g/L culture) (Additional charge − 4). These tags differ in their lengths. In order file 1: Figure S8) and purified from the soluble cell frac- to obtain a value independent of the protein size, we tion by IMAC. This soluble fraction was significantly calculated the net charge per residue (NCPR) (Addi- viscous, and an imidazole gradient should be used for tional file 1: Figure S7) [71]. According to the NCPR ZapB homogeneous purification (Additional file 1: values, ZapB is again the most acidic tag (NCPR: Figure S9). This unusual viscosity already suggested a − 0.099), followed by TDoT (NCPR: − 0.077). There- certain degree of ZapB self-assembly, which was fur- fore, no interaction is expected between DNA, RNA, ther corroborated by transmission electron microscopy or membranes and ZapB. (TEM) analysis of the purified protein. Abundant long Overall, we can conclude that, theoretically, ZapB protein nanofibers 20 ± 5 nm in width and displaying fulfills all the requirements to work as a tag to promote a regular striated pattern were observed (Fig. 3a), in the formation of non-amyloid-like functional IBs; excellent agreement with the nanostructures visualized still, to act as such, the domain should be able to self- in a previous study [42]. assemble, despite its high predicted solubility in both We analyzed the secondary structure content of the folded and unfolded states. self-assembled ZapB by circular dichroism (CD) spectroscopy. The spectrum is characteristic of an ZapB self‑assembles into coiled‑coil, non‑amyloid, α-helical structure, with two minima at 208 nm and nanofibers 222 nm (Fig. 3b). Some studies have revealed that the In order to test the ability of ZapB to form pro- 222/208 nm ratio allows discriminating individual tein assemblies driven by interactions between α-helices from coiled-coils, owing to the different natively folded coiled-coils, a His-tag was added at periodicity of the two folds. A 222/208 nm ratio ≥ 1 is Gil‑Garcia et al. Microb Cell Fact (2020) 19:117 Page 5 of 16 Fig. 3 ZapB coiled-coil assembly biophysical characterization. a TEM image of ZapB fibers upon negative staining. b Far-UV circular dichroism spectrum. c Congo-Red absorbance spectra and d Th-T fluorescence emission spectra indicative of coiled-coil structures, and ratios ≤ 0.86 spectroscopy and Fourier Transform Infrared Spectros- can be attributed to individual α-helices in a protein copy (FTIR). structure [72, 73]. For ZapB, the 222/208 nm ratio The Far-UV CD spectrum of ZapB IBs resembles the is 0.99, indicating that, as anticipated by the protein one obtained for the nanofibrillar solution (Fig. 4a). The crystal structure (PDB: 2JEE), ZapB keeps the coiled- packing of ZapB into IBs seems to favor the preservation coil fold in the macromolecular fibrillar assembly. of the coiled-coil conformation since the 222/208 nm The absence of a β-signature in the CD spectrum of ratio of these aggregates is 1.14. We recorded the infrared self-assembled ZapB is consistent with a non-amyloid spectra of ZapB IBs in the amide I region of the spectrum nature. This trait was confirmed using the Congo Red (1700–1600 cm−1), corresponding to the absorption of (CR) and Thioflavin-T (Th-T) amyloid dyes. Both analy- the main chain carbonyl group and sensitive to protein ses indicated that the ZapB nanofibers do not have an conformation (Fig. 4b). It is important to note here, that amyloid-like nature, since the spectra of both dyes in in coiled-coils, the supercoil bending of the α-helices the presence and absence of the protein (buffer alone) results in a spectral splitting of the α-helical IR amide are fairly similar (Fig. 3c, d). I band and in a shift to lower wavenumbers [74]. FTIR measurements on the model coiled-coil GCN4 dem- ZapB IBs exhibit a coiled‑coil conformation onstrated the assignment of the solvated portion of the In order to explore whether, like the nanofilaments puri- coiled-coil to a low helix frequency (1631 cm−1); in con- fied from the soluble cell fraction, ZapB IBs are sustained trast, the buried helix frequency (1651 cm−1) is observed by coiled-coil interactions, they were purified from the for residues in the interior of the coiled-coil. These two insoluble cell fraction (Additional file 1: Figure S10) signals alone accounted for the 78% of the GCN4 spec- and their secondary structure content analyzed by CD tral area [75, 76]. Similarly, the major contributors to ZapB IBs IR spectra are two signals at 1632 cm−1 and Gil‑Garcia et al. Microb Cell Fact (2020) 19:117 Page 6 of 16 Fig. 4 Conformational properties of ZapB IBs. a Far-UV circular dichroism spectrum. b FTIR absorbance spectrum in the amide I region (dashed line). The different component bands (solid lines) were obtained by deconvolution of the absorbance spectrum. c SEM image of a ZapB IB. d Congo-Red absorbance spectra. e Th-T fluorescence emission spectra 1651 cm−1 bands, accounting for the 28% and the 35% of emission in E. coli intact cells. As expected, the GFP sig- the absorbance spectrum area, respectively. Additional nal was confined mainly in IBs placed at the poles of cells bands at 1620 cm−1 and 1672 cm−1 were detected, the (Fig. 5b). latest one being also present in the IR spectra of GCN4 GFP is considered the default fluorescent protein for and other coiled-coil proteins [75]. When observed using most applications, but when dealing with deep in vivo Scanning Electron Microscopy (SEM) (Fig. 4c), these imaging, red-shifted fluorescent proteins are preferred IBs displayed a spherical shape. The non-amyloid nature since, at these wavelengths, light absorption by tissues is of the IBs was confirmed using both CR and Th-T amy- significantly lower [77], among them mCherry is one of loid dyes, observing similar absorbance and fluorescence the most used variants [78, 79]. We N-terminally fused spectra for the IBs and the buffer alone, respectively ZapB to mCherry, in order to asses if we can obtain red (Fig. 4d, e). fluorescent IBs with potential in vivo applications. Overall, the CD and FTIR data converge to indicate ZapB-mCherry was expressed in E. coli, and the solu- that ZapB IBs consist mostly of coiled-coil molecules. ble and the insoluble fraction were separated as above. In this case, ~ 55% of the fusion protein is present in ZapB‑GFP and ZapB‑mCherry proteins are produced the insoluble fraction (Fig. 6a), whereas in non-tagged as fluorescent IBs in E. coli mCherry only ~ 5% of the protein is insoluble (Additional In order to test the ability of ZapB to assist the formation file 1: Figure S12). The difference between the fraction of of functional IBs inside bacteria, ZapB was N-terminally ZapB-GFP and ZapB-mCherry located in the respective fused to GFP and expressed in E. coli. After induction of insoluble fractions likely owes to the highest solubility protein expression, the soluble and the insoluble cellular of the mCherry structure when compared with GFP, as fractions were separated by centrifugation and analyzed assessed using the AGGRESCAN3D algorithm (Addi- by SDS-PAGE. As shown in Fig. 5a, the ZapB-GFP fusion tional file 1: Figure S13). When the location of the red protein was mainly located in the insoluble fraction fluorescence was monitored using confocal microscopy, (~ 95%), whereas, non-tagged insoluble GFP accounts highly fluorescent IBs become evident at the poles. How- for ~ 30% of the recombinant protein (Additional file 1: ever, their discretization was more difficult than in the Figure S11). Thus, the fusion of the two polypeptides case of ZapB-GFP, due to the soluble fusion protein fluo- facilitates GFP deposition. We used fluorescence con- rescent background (Fig. 6b). focal microscopy to localize the GFP fluorescence Gil‑Garcia et al. Microb Cell Fact (2020) 19:117 Page 7 of 16 Fig. 5 Cellular distribution of ZapB-GFP IBs. a SDS-PAGE of intact cells, soluble, and insoluble cellular fractions. The arrow indicates the ZapB-GFP. b Visualization of ZapB-GFP IBs in intact E. coli cells by confocal microscopy. Some IBs located at the poles of cells are indicated GFP and mCherry maintain native spectral properties or mobility between soluble and assembled fluorescent in ZapB‑based IBs proteins. In order to evaluate the impact of the coiled-coil struc- Overall, these results allow us to confirm that GFP and ture in the functionality of the attached fluorescent pro- mCherry keep their activity and, likely, their native con- teins when embedded in the IBs, we purified both IBs formation inside ZapB-induced IBs. from the insoluble fraction (Additional file 1: Figures S14, S15). The analysis of the purified protein aggregates ZapB‑GFP inclusion bodies contain a coiled‑coil using an epifluorescence microscope and adequate filters conformation allowed us to observe the presence of abundant green We selected the ZapB-GFP fusion as a model system to and red fluorescent particles for ZapB-GFP and ZapB- further study the properties of ZapB promoted IBs. We mCherry, respectively (Fig. 7a,b). We compared the analyzed the secondary structure content of these IBs spectral properties of the fluorescent proteins trapped using CD spectroscopy and FTIR. The far-UV CD spec- in the IBs with those of their soluble and non-tagged trum of ZapB-GFP is of course influenced by the all counterparts. As it is shown in Fig. 7c, soluble GFP and β-sheet structure of GFP (Fig. 8a); still, the two main sig- ZapB-GFP IBs presented identical excitation and emis- natures of α-helices could be detected in the IBs, with the sion maxima at 495–496 and 512 nm, respectively. The global minimum placed at 222 nm and inflection of the same behavior was observed when comparing soluble spectrum at 208 nm (Fig. 8a). mCherry with ZapB-mCherry IBs, both displaying exci- When ZapB-GFP IBs were analyzed using FTIR in the tation and emission maxima at 589 and 604 nm, respec- amide I region of the spectrum, we could detect, again, tively (Fig. 7d). The emission spectra of ZapB-GFP and the characteristic helical band at 1651 cm−1, accounting ZapB-mCherry overlap perfectly with that of the respec- for 39% of the absorbance area. However, now the major tive soluble untagged fluorescent proteins, whereas, signal is located at 1626 cm−1 (49% of the area). This in both cases, the left side of the excitation spectrum is band, likely results from the sum of the low-frequency slightly red-shifted when the protein is located within the coiled-coil signal and the β-sheet signal of the fused GFP IBs, which likely respond to differences in crowding and β-barrel (Fig. 8b). An additional band at 1678 cm−1, likely Gil‑Garcia et al. Microb Cell Fact (2020) 19:117 Page 8 of 16 Fig. 6 Cellular distribution of ZapB-mCherry IBs. a SDS-PAGE of intact cells, soluble, and insoluble cellular fractions. The arrow indicates the ZapB-mCherry. b Visualization of ZapB-mCherry IBs in intact E. coli cells by confocal microscopy. Some IBs located at the poles of cells are indicated arising from the mixed contribution of the two structur- compared the spectral properties of GFP in these aggre- ally different domains in the fusion at high frequencies gates with those of ZapB-GFP. Figure 9a demonstrates was also evident. Finally, the non-amyloid character of that Aβ42-GFP and ZapB-GFP IBs share the same excita- ZapB-GFP IBs was corroborated using the CR amyloid tion and emission spectra, indicating that the active and dye, observing similar absorbance spectra in the presence properly folded GFP they contain is in a similar environ- of ZapB-GFP IBs, soluble GFP or only buffer (Fig. 8c). mental context. The size of the IBs was analyzed using Overall, the secondary structure analyses of ZapB-GFP dynamic light scattering (DLS). Both IBs exhibited simi- IBs suggest that they contain a significant proportion of lar sizes, with moderately polydisperse distributions and coiled-coil conformations and do not have an amyloid- calculated average diameters of 462.2 ± 69.51 nm and like nature. This is likely also the case for ZapB-mCherry 463.3 ± 99.9 nm, for ZapB-GFP and Aβ42-GFP, respec- IBs, since their binding to Th-T and CR are negligible tively. The DLS data suggested that ZapB-GFP IBs are (Additional file 1: Figure S16). quite homogeneous in size (Additional file 1: Figure S19). Effectively, when these aggregates were imaged by Scan- Comparison of ZapB‑based and amyloid‑like IBs ning Electron Microscopy (SEM), it was observed that We wanted to compare the properties of the above they correspond to submicrometric spherical assemblies described coiled-coil-based IBs with those of model (Fig. 9b), and accordingly, they can be assimilated to pro- amyloid-like IBs. To this aim, we selected fusion of the tein nanoparticles. Alzheimer’s related β-amyloid peptide (Aβ42) and GFP. Once confirmed that ZapB-GFP and Aβ42-GFP IBs In previous studies, we have characterized in detail the share spectral properties and dimensions, we wondered properties of Aβ42-GFP IBs [22, 80, 81]. The high amy- if the GFP activity in both nanostructures was similar. To loid propensity of the Aβ42 peptide drives the incor- this aim, we analyzed the GFP fluorescence intensity of poration of the GFP moiety into β-sheet enriched and both IBs by fluorescence microscopy. The mean fluores- fluorescent IBs. cence intensity values as obtained from images (Addi- We expressed Aβ42-GFP (Additional file 1: Figure S17) tional file 1: Figure S20) quantification of 50 isolated and purified its IBs (Additional file 1: Figure S18) and fluorescent dots in each sample, using ImageJ, revealed Gil‑Garcia et al. Microb Cell Fact (2020) 19:117 Page 9 of 16 Fig. 7 Epifluorescence microscopy images and spectral properties of ZapB-GFP and ZapB-mCherry IBs. a Fluorescence microscopy image of purified ZapB-GFP IBs. b Fluorescence microscopy image of purified ZapB-mCherry IBs. c Excitation and emission spectra of ZapB-GFP IBs and soluble GFP. d Excitation and emission spectra of ZapB-mCherry IBs and soluble mCherry that ZapB-GFP IBs exhibited two times higher activity are expected to interfere less with the folding and struc- (3713 ± 91.25 a.u.) than Aβ42-GFP IBs (1839 ± 25.23 a.u.) ture of the GFP moiety, and indeed, many coiled–coil (Fig. 10). This observation is not surprising since the domains naturally exist and function appended to globu- assembly of Aβ42-GFP IBs depends on an aberrant inter- lar domains [83]. action between hydrophobic Aβ42 regions, which leads to a relatively rapid aggregation into amyloid-like struc- tures, with the subsequent inactivation of a least a frac- ZapB‑GFP IBs are innocuous for human cells tion of the attached globular domains, with their most One of the potential advantages of coiled-coil inspired aggregation-prone sequence stretches contributing to IBs, relative to amyloid-like IBs, is that in the absence stabilize the aggregate through amyloid-like contacts of intermolecular β-sheet assemblies that might elicit [22]. Indeed, we have demonstrated that the activities cytotoxicity, these α-helical-based assemblies should of the IBs formed by 20 Aβ42-GFP variants, bearing dif- be non-toxic for human cells. To confirm this extent, ferent mutations in the Aβ42 moiety, inversely correlate we incubated HeLa cells with increasing concentrations with the aggregation propensities of the peptides [82]. tored their viability using ­PrestoBlue® fluorescent assay. of ZapB-GFP IBs (from 0 to 12 µM) for 72 h and moni- Aβ42-GFP IBs inactivation was favored by increased β-sheet propensity and hydrophobicity and counter- As it can be observed in Fig. 11a, the IBs turned to be acted by increased net charge [54]. ZapB has a negligi- innocuous at any of the assayed concentrations, which ble β-sheet propensity, is polar, highly charged and has should make them suitable for in vivo applications. This a very low aggregation propensity, compared with Aβ42, is contrast with Aβ42-GFP IBs, which exhibit a mod- all these factors likely contributing to the higher activity erate and concentration-dependent toxicity for HeLa of ZapB-GFP IBs. In addition, the inter- and supramo- cells (Fig. 11b), in good agreement with the toxicity we lecular assembly of ZapB is directed by native interac- described previously for Aβ42 IBs [84]. tions, and not by non-native contacts, as in Aβ42, which Gil‑Garcia et al. Microb Cell Fact (2020) 19:117 Page 10 of 16 finding amazing applications in biotechnology [86] and biomedicine [18, 19]. Conventionally, IBs are enriched in intermolecular β-sheet structure, and contain both amyloid-like and native protein conformations [24]. Unavoidably, a frac- tion of the target protein becomes inactivated to form the amyloid skeleton that provides mechanical stability to IBs. In order to endorse IBs with tailored functionali- ties, the particular protein of interest is usually fused to an aggregation-prone polypeptide that act as IB-forming tag [44, 87]. This allows to recruit otherwise soluble pro- teins into IBs. However, the high-aggregation potential of these tags compromises the folding of the target protein, a fraction of which establishes non-native contacts and becomes an integral part of the amyloid-like architecture, and thus inactive [22]. Because of their amyloidogenic nature, active IBs are assimilated to natural functional amyloids [88]. It is important to note here that, as pathogenic amyloids, functional amyloids found in nature are also inher- ently cytotoxic [29]; the difference being that nature has evolved dedicated mechanisms to prevent the toxicity of natural assemblies. In the absence of these control mechanisms it cannot can be completely discarded that amyloid-like IBs might contain or release toxic protein conformers. A way to avoid protein inactivation by non-native interactions and potential toxicity is to run away from non-native-β-sheet based IBs and develop native-α-helix based IBs. We take here a step towards this direction by implementing ZapB based IBs. Catalytic coiled-coil based IBs have been described previously [16, 17, 37–41]. However, our computational analysis suggests that these sequences display lower α-helical and coiled-coil propen- sities and higher intrinsic aggregation propensities than ZapB, which suggests that a partial transition towards amyloid-like structures upon overexpression cannot be fully discarded, especially because the IBs they promoted Fig. 8 Conformational properties of ZapB-GFP IBs. a Far-UV circular were not conformationally characterized. dichroism spectrum of ZapB-GFP IBs and soluble GFP. b FTIR To avoid the above-mentioned α-helix to β-sheet tran- absorbance spectrum of ZapB-GFP IBs in the amide I region (dashed line). The different component bands (solid lines) were obtained by sition, we have selected ZapB. A natural sequence with deconvolution of the spectrum using PeakFit software. c Congo-Red extremely high α-helical and coiled-coil propensities, absorbance spectra a neglectable intrinsic aggregation propensity and a high polar and anionic character. This coiled-coil pro- tein exhibits the ability to spontaneously assemble into Conclusions α-helical macromolecular fibrils devoid of any amyloid IBs have been traditionally regarded as waste reser- character. We show how this property can be exploited to voirs containing only misfolded and thus non-active and obtain innocuous, spherical, relatively homogeneous and useless proteins. However, it is now clear that at least highly active sub-micrometric coiled-coil inspired IBs. some proteins can retain certain activity when embed- ZapB expands our existing toolbox to generate immobi- ded in these aggregates, turning them into functional lized enzymes or biomedical nanocarriers, among other sub-micron particles [23, 85]. These nanostructures are applications. However, the present work constitutes a proof-of-concept, and the ability of ZapB to facilitate the Gil‑Garcia et al. Microb Cell Fact (2020) 19:117 Page 11 of 16 Fig. 9 Spectral properties and morphology of ZapB-GFP IBs. a GFP-Excitation and emission spectra of ZapB-GFP IBs, Aβ42-GFP IBs, and soluble GFP. b SEM image of ZapB-GFP IBs formation of functional IBs when fused to other protein folds should still be demonstrated. Methods In silico analysis For the analysis of the coiled-coil tendency, four differ- ent predictors were used: Coils, DeepCoil, PCoils and MARCOIL. For the different predictions, the Coiled-coil probability (%) was calculated in order to obtain a quan- titative value of the coiled-coil tendency of the analyzed sequences. Aggregation propensity predictions were performed using Aggrescan and TANGO webservers. In the case of Fig. 10 GFP fluorescence of ZapB-GFP and Aβ42-GFP inclusion bodies. Quantification of GFP fluorescence intensity from Aggrescan, the Normalized a4v Sequence Sum for 100 fluorescence microscopy images using ImageJ. The corresponding residues ­(Na4vSS) was employed as the aggregation pro- intensity of 50 individual fluorescent dots for each sample was pensity value. In the case of TANGO, predictions were analyzed and averaged. The statistical values were derived using the made using the default parameters and the AGG​ value Mann–Whitney Test. A p value < 0.001 is indicated as ***. Error bars was selected as the aggregation propensity value. correspond to SEM Fig. 11 Citotoxicity of ZapB-GFP and Aβ42-GFP IBs. HeLa cells viability was assessed after adding different concentrations (from 2 to 12 µM) of ZapB-GFP a and Aβ42-GFP b IBs for 72 h. PBS buffer was used as a control Gil‑Garcia et al. Microb Cell Fact (2020) 19:117 Page 12 of 16 For the aggregation propensity predictions consider- was cooled in ice and sonicated 3 min at 15% amplitude ing the 3D structure, AGGRESCAN3D webserver was under 1 s cycles. After that, 4 µL of Nonidet P40 (NP- used with the default parameters and using the follow- 40) were added and the mixture incubated at 4 °C for ing PDBs as an input: PDB: 2JEE for ZapB, PDB: 1FE6 for 1 h. Then, 10 µL of 1 mg/mL DNase I and 12 µL of 1 M TDoT and PDB: 3LNR for 3HAMP, PDB: 2Y0G for GFP ­MgSO4 were added and the mixture was further incu- and PDB: 2H5Q for mCherry. In the case of ZapB and bated at 37 °C for 45 min. IBs were collected by centrif- 3HAMP proteins, the dimeric structures were generated ugation at 15.000g for 15 min at 4 °C and washed with with PISA. Protein structures and their surfaces were buffer A (50 mM Tris, 100 mM NaCl, pH 7.4) containing represented with Pymol (DeLano Scientific, LLC). 0.5% Triton X-100. All incubations were done under gen- The Grand Average of Hydropathicity (GRAVY score) tle agitation. Finally, IBs were washed three times with and the net charge at physiological pH were calculated buffer A (50 mM Tris, 100 mM NaCl, pH 7.4) to remove using the ProtParam tool. The net charge per residue remaining detergent. The purity of the IBs was checked (NCPR) was calculated dividing the net charge by the by SDS-PAGE. Protein concentration was estimated total number of residues. measuring the absorbance at 280 nm in a Specord 200 Finally, secondary structure predictions were per- Plus spectrophotometer (Analytik Jena, Germany). formed using GOR and PSIPRED webservers. Circular dichroism (CD) spectroscopy Protein production and purification For the analysis of the secondary structure, ZapB protein, ZapB gene fragment (Additional file 1: DNA and amino ZapB and ZapB-GFP IBs were exhaustively resuspended acid sequences of ZapB protein) was cloned into a pET- in buffer A (50 mM Tris, 100 mM NaCl, pH 7.4). Solu- 21a vector between NdeI and BamHI restriction sites. ble GFP was diluted in buffer A (50 mM Tris, 100 mM The ZapB-GFP and ZapB-mCherry fusion proteins NaCl, pH 7.4) at 5 µM. Sedimentation problems were were generated inserting both target proteins after ZapB avoided with thorough resuspension of the sample before sequence, using a linker (SIPGA) and BamHI and EcoRI the measurements. The correct resuspension of the sam- restriction sites. ple was checked before and after of each measurement. For the production of soluble and His-tagged ZapB, Far-UV CD spectra of the different protein solutions GFP and mCherry proteins, transformed E. coli BL21 were recorded using a 1 nm bandwith, a response time of (Invitrogen, USA) cells were grown aerobically in Luria 1 s, and a scan speed of 100 nm/min in a Jasco-710 spec- Broth (LB) medium supplemented with 100 µg/ml of tropolarimeter (Jasco Corporation, Japan), thermostated ampicillin. Protein expression was induced with 1 mM at 25 °C. Ten accumulations were averaged for each Isopropyl β-d-1-thiogalactopyranoside (IPTG) at 30 °C spectrum. for 12 h. For protein purification, cells expressing the recombinant protein were harvested by centrifugation Transmission electron microscopy (TEM) (5000g for 20 min). After cell lysis by sonication and frac- For TEM sample preparation, 10 µL of 10 µM of ZapB tionation, the soluble fraction was collected and injected protein were deposited onto a carbon-coated copper grid in a Histrap FF 5 mL column (GE Healthcare, USA) using for 10 min and the excess of liquid was removed with fil- an ÄKTA (GE Healthcare, USA). After purification, pro- ter paper, followed by a negative stain with 10 µL of 2% teins were dialyzed in buffer A (50 mM Tris, 100 mM (w/v) uranyl acetate for 1 min. Grids were exhaustively NaCl, pH 7.4). The purity of these proteins was checked scanned using a JEM 1400 transmission electron micro- by SDS-PAGE. scope (JEOL Ltd, Japan) operating at 80 kV, and images For the production of ZapB, ZapB-GFP, ZapB-mCherry were acquired with a CCD GATAN ES1000W Erlangshen and Aβ42-GFP IBs, transformed E. coli BL21 (Invitro- camera (Gatan Inc., USA). The width of fibers was ana- gen, USA) cells were grown aerobically in LB medium lyzed using ImageJ, averaging the measures of 5 individ- supplemented with 100 µg/mL of ampicillin for ZapB, ual fibers. ZapB-GFP and ZapB-mCherry transformed cells, and with 50 µg/mL of kanamycin for Aβ42-GFP transformed Scanning electron microscopy (SEM) and dynamic light cells. Protein expression was induced with 1 mM IPTG scattering (DLS) at 30 °C for 12 h. For IBs purification, cells were centri- Scanning electron microscopy (SEM) was used in order fuged at 5000g for 20 min. After that, IBs were purified to analyze the morphology of ZapB and ZapB-GFP IBs. as described [89]. Briefly, pelleted cells were resuspended To do that, 10 µL of sample resuspended in water were in 400 µL of buffer A (50 mM Tris, 100 mM NaCl, pH deposited on silicon wafers (Ted Pella Inc., USA), air- 7.4) plus 4 µL of 100 mM PMSF and 6 µL of 10 mg/mL dried and observed using a SEM Merlin (Zeiss Merlin, lysozyme. After 1 h of incubation at 37 °C, the mixture Germany) operating at 2 kV. Gil‑Garcia et al. Microb Cell Fact (2020) 19:117 Page 13 of 16 Dynamic light scattering (DLS) was used for a quan- GFP and mCherry fluorescence spectra titative determination of ZapB-GFP and Aβ42-GFP IBs Excitation and emission spectra of soluble GFP, ZapB- size. The size of these nanoparticles was determined GFP IBs and Aβ42-GFP IBs in buffer A (50 mM Tris, using a Zetasizer Nano ZS (Malvern Instruments Lim- 100 mM NaCl, pH 7.4) were analyzed in a Jasco FP-8200 ited, UK) at 25 °C. Three different measures of ten runs fluorescence spectrofluorometer (Jasco Corporation, were recorded for each sample. Japan). Emission spectra were obtained recording the emitted fluorescence between 500 and 600 nm. Excita- tion spectra were obtained by exciting the samples in a Cell fractionation 400–500 nm range. The distribution of the expressed ZapB, GFP, mCherry Excitation and emission spectra of soluble mCherry and the fusion proteins (ZapB-GFP, ZapB-mCherry and and ZapB-mCherry IBs in buffer A (50 mM Tris, Aβ42-GFP) in E. coli cells was analyzed by SDS-PAGE. 100 mM NaCl, pH 7.4) were analyzed using Jasco After protein expression at 30 °C for 12 h, cells were har- FP-8200 fluorescence spectrofluorometer (Jasco Cor- vested by centrifugation (5000g for 20 min) and resus- poration, Japan). Emission spectra were obtained pended in buffer A (50 mM Tris, 100 mM NaCl, pH 7.4). recording the emitted fluorescence between 600 and After that, disrupted cells (total fraction) by sonication 700 nm. Excitation spectra were obtained by excit- were centrifuged at 15.000g for 15 min at 4 °C, and super- ing the samples in a 500–600 nm. Three spectra were natant (soluble fraction) was separated from pellet (insol- accumulated at 25 °C with slit widths of 5 nm, a 0.5 nm uble fraction). The insoluble fraction was resuspended in Interval, and a 1000 nm/min scan rate for each sample. the same volume of buffer A (50 mM Tris, 100 mM NaCl, pH 7.4) than the soluble fraction and the different frac- tions were heated at 98 °C for 10 min. After that, 10 µL Fourier transform infrared spectroscopy (FTIR) of each fraction were loaded separately into SDS-PAGE. Samples of ZapB and ZapB-GFP IBs were washed Band intensity quantification was performed using the with ­H2O to remove the presence of salts. Both sam- ImageJ software in order to estimate the percentage of ples were placed on the ATR crystal and dried out protein in each fraction. under ­N2 flow. The experiments were carried out in a Bruker Tensor 27 FTIR (Bruker Optics, USA) supplied Confocal microscopy with a Specac Golden Gate MKII ATR accessory. Each E. coli BL21 cells expressing ZapB-GFP and ZapB- spectrum consists of 32 acquisitions measured at a mCherry proteins at 30 °C for 12 h were centrifuged resolution of 1 cm−1. Data were acquired and normal- and resuspended in PBS pH 7.4 to an OD of 0.1. 10 µL ized using the OPUS MIR Tensor 27 software (Bruker of resuspended cells were deposited on top of micros- Optics, USA). IR spectrum was fitted employing a copy poly-l-lysine glass slides, covered with coverslips nonlinear peak-fitting equation using PeakFit package and observed in a Leica SP5 confocal fluorescence micro- v4.12 (Systat Software, USA). The area for each Gauss- scope (Leica Microsystems, Germany). ian curve was calculated in the amide I region from 1700 to 1600 cm−1 using second derivative deconvolu- Epifluorescence microscopy tion method in PeakFit package v4.12 (Systat Software, 10 µL of the ZapB-GFP, ZapB-mCherry and Aβ42-GFP USA). IB samples in buffer A (50 mM Tris, 100 mM NaCl, pH 7.4) were deposited on top of microscopy glass slides and covered with coverslips. The fluorescence was observed Toxicity assay using an Eclipse Ts2R-FL inverted microscope (Nikon, HeLa cells were cultured in Dulbecco’s Modified Eagle Japan) using a C-LED470 filter for GFP fluorescence and Medium (DMEM) supplemented with 10% Fetal Bovine a C-LED525 for mCherry fluorescence. Serum (FBS) and seeded into 96-well plates. ZapB-GFP To determine the GFP-fluorescence intensity of ZapB- and Aβ42-GFP IBs were resuspended in PBS pH 7.4 and GFP and Aβ42-GFP IBs, images were analyzed using the added at a range from 2 to 12 µM. For control, the same ImageJ software. 50 fluorescent dots were selected in each volume of PBS pH 7.4 was added. Treated and control of ­PrestoBlue® reagent (ThermoFisher Scientific, USA) image maintaining the same dimensions (height × width) cells were incubated for 72 h at 37 °C, and then 10 µL for the different selections. After that, the intensity of these different fluorescent dots was calculated and the was added and incubated for 10 min. To determine cell average and SEM values were estimated. viability, fluorescence signal was measured by exciting at 560 nm and collecting at 590 nm in a Victor3 fluorescent plate reader (Perkin Elmer, USA). Gil‑Garcia et al. Microb Cell Fact (2020) 19:117 Page 14 of 16 Thioflavin T (Th‑T) and congo red (CR) binding Availability of data and materials All data generated and analyzed during this study are shown in this article and For the Th-T binding assay, ZapB, mCherry. ZapB and it Additional file 1. ZapB-mCherry IBs, were diluted in buffer A (50 mM Tris, 100 mM NaCl, pH 7.4) and incubated with 25 µM Ethics approval and consent to participate Not applicable. Th-T. Emission fluorescence was recorded using a Jasco FP-8200 spectrofluorometer (Jasco Corporation, Japan) Consent for publication in the 460–600 nm range, using an excitation wave- We state that all the authors (Marcos Gil-Garcia, Susanna Navarro and Salvador Ventura) mutually agree that this research work should be submitted to length of 440 nm and an emission bandwith of 5 nm. Microbial Cell Factories; it is an original work of the above-mentioned authors; The same buffer with 25 μM Th-T and without protein and that the article has not been published or submitted in any other peer- was employed as a control. In the case of ZapB-GFP, reviewed journal. Th-T binding assay was not performed due to an over- Competing interests lap between the fluorescence spectra of GFP and Th-T. The authors declare that they have no competing interests. For the CR binding assay the different IBs and soluble Received: 26 March 2020 Accepted: 25 May 2020 proteins were diluted in buffer A (50 mM Tris, 100 mM NaCl, pH 7.4) and mixed with CR to a final concentra- tion of 10 µM CR. Optical absorption spectrum was recorded in the range from 375 to 700 nm in a Specord References 200 Plus spectrophotometer (Analytik Jena, Germany). 1. Ferrer-Miralles N, Domingo-Espín J, Corchero J, Vázquez E, Villaverde A. Spectrum of protein alone was acquired to subtract Microbial factories for recombinant pharmaceuticals. Microb Cell Fact. protein scattering. 2009;8:1–8. 2. Rosano GL, Ceccarelli EA. Recombinant protein expression in Escherichia coli: advances and challenges. Front Microbiol. 2014;5:1–17. Supplementary information 3. Schrödel A, De Marco A. Characterization of the aggregates formed Supplementary information accompanies this paper at https​://doi. during recombinant protein expression in bacteria. BMC Biochem. org/10.1186/s1293​4-020-01375​-4. 2005;6:1–11. 4. De Marco A. Recombinant polypeptide production in E. coli: towards a rational approach to improve the yields of functional proteins. Microb Additional file 1: Figure S1. Coiled-coil predictions for ZapB. Figure Cell Fact. 2013;12:1–8. S2. Coiled-coil predictions for 3HAMP. Figure S3. Coiled-coil predictions 5. Tartaglia GG, Pechmann S, Dobson CM, Vendruscolo M. Life ond the for TDoT. Figure S4. Secondary structure prediction by PSIPRED server. edge: a link between gene expression levels and aggregation rates of Figure S5. Secondary structure prediction by GOR server. Figure S6. human proteins. Trends Biochem Sci. 2007;32:204–6. AGGRESCAN3D structural aggregation propensity predictions for ZapB, 6. Villaverde A, Carrio MM. Protein aggregation in recombinant bacteria: TDoT and 3HAMP. Figure S7. Net charge per residue (NCPR) of the differ‑ biological role of inclusion bodies. Biotechnol Lett. 2003;25:1385–95. ent tags. Figure S8. SDS-PAGE of the cellular distribution of ZapB. Figure 7. Slouka C, Kopp J, Spadiut O, Herwig C. Perspectives of inclusion bodies S9. SDS-PAGE of ZapB purification by IMAC. Figure S10. SDS-PAGE of for bio-based products: curse or blessing? Appl microbiol biotechnol. purified ZapB IBs. Figure S11. SDS-PAGE of the cellular distribution of GFP. Appl Microbiol Biotechnol. 2019;103:1143–53. Figure S12. SDS-PAGE of the cellular distribution of mCherry. Figure S13. 8. Rinas U, Garcia-Fruitós E, Corchero JL, Vázquez E, Seras-Franzoso J, Vil‑ AGGRESCAN3D structural aggregation propensity predictions for GFP and laverde A. Bacterial inclusion bodies: discovering their better half. Trends mCherry. Figure S14. SDS-PAGE of purified ZapB-GFP IBs. Figure S15. Biochem Sci. 2017;42:726–37. SDS-PAGE of purified ZapB-mCherry IBs. Figure S16. Characterization of 9. García-Fruitós E, Rodríguez-Carmona E, Díez-Gil C, Ferraz RM, Vázquez E, the non-amyloid nature of ZapB-mCherry IBs. Figure S17. SDS-PAGE of Corchero JL, et al. Surface cell growth engineering assisted by a novel the cellular distribution of Aβ42-GFP. Figure S18. SDS-PAGE of purified bacterial nanomaterial. Adv Mater. 2009;21:4249–53. Aβ42-GFP IBs. Figure S19. DLS spectra of ZapB-GFP and Aβ42-GFP IBs. 10. García-Fruitós E, Seras-Franzoso J, Vazquez E, Villaverde A. Tunable Figure S20. Epifluorescence microscopy images of ZapB-GFP and Aβ42- geometry of bacterial inclusion bodies as substrate materials for tissue GFP IBs. DNA and amino acid sequences of ZapB protein. engineering. Nanotechnology. 2010;21:205101. 11. De Marco A, Ferrer-Miralles N, Garcia-Fruitós E, Mitraki A, Peternel S, Rinas U, et al. Bacterial inclusion bodies are industrially exploitable amyloids. Acknowledgements FEMS Microbiol Rev. 2019;43:53–72. We are grateful to UAB microscopy service of Barcelona for their technical 12. Margreiter G, Messner P, Caldwell KD, Bayer K. Size characterization of advice. inclusion bodies by sedimentation field- flow fractionation. J Biotechnol. 2015;138:67–73. Authors’ contributions 13. Rinas U, Bailey JE. Protein compositional analysis of inclusion bodies MGG contributed to acquisition, and analysis of data, preparation of the produced in recombinant Escherichia coli. Appl Microbiol Biotechnol. manuscript and carried out the experiments and analysis. SN carried out 1992;37:609–14. the cytotoxicity experiments. SV contributed to the concept and design of 14. Carrió MM, Villaverde A. Construction and deconstruction of bacterial the investigation in addition to data analysis, preparation and revision of the inclusion bodies. J Biotechnol. 2002;96:3–12. manuscript. All authors read and approved the final manuscript. 15. Fahnert B, Lilie H, Neubauer P. Inclusion bodies: formation and utilisation. Adv Biochem Eng Biotechnol. 2004;89:93–142. Funding 16. Jäger VD, Lamm R, Kloß R, Kaganovitch E, Grünberger A, Pohl M, et al. This work was funded by the Spanish Ministry of Economy and Competitive‑ A synthetic reaction cascade implemented by colocalization of two ness BIO2016-78310-R to S.V and by ICREA, ICREA-Academia 2015 to S.V. proteins within catalytically active inclusion bodies. ACS Synth Biol. M.G.G. was supported by the Spanish Ministry of Science and Innovation via a 2018;7:2282–95. doctoral grant (FPU16/02465). 17. Krauss U, Jäger VD, Diener M, Pohl M, Jaeger KE. Catalytically-active inclusion bodies—carrier-free protein immobilizates for application in biotechnology and biomedicine. J Biotechnol. 2017;258:136–47. Gil‑Garcia et al. Microb Cell Fact (2020) 19:117 Page 15 of 16 18. Pesarrodona M, Jauset T, Díaz-Riascos ZV, Sánchez-Chardi A, Beaulieu 41. Kloss R, Karmainski T, Jäger VD, Hahn D, Grünberger A, Baumgart M, et al. ME, Seras-Franzoso J, et al. Targeting antitumoral proteins to breast Tailor-made catalytically active inclusion bodies for different applications cancer by local administration of functional inclusion bodies. Adv Sci. in biocatalysis. Catal Sci Technol. 2018;8:5816–26. 2019;6:1900849. 42. Ebersbach G, Galli E, Møller-Jensen J, Löwe J, Gerdes K. Novel coiled-coil 19. Céspedes MV, Cano-Garrido O, Álamo P, Sala R, Gallardo A, Serna N, cell division factor ZapB stimulates Z ring assembly and cell division. Mol et al. Engineering secretory amyloids for remote and highly selective Microbiol. 2008;68:720–35. destruction of metastatic foci. Adv Mater. 2019;1907348:1–9. 43. Den Blaauwen T. Prokaryotic cell division: flexible and diverse. Curr Opin 20. Ventura S, Villaverde A. Protein quality in bacterial inclusion bodies. Microbiol. 2013;16:738–44. Trends Biotechnol. 2006;24:179–85. 44. Wang X, Zhou B, Hu W, Zhao Q, Lin Z. Formation of active inclusion bod‑ 21. Carrió M, González-Montalbán N, Vera A, Villaverde A, Ventura S. ies induced by hydrophobic self-assembling peptide GFIL8. Microb Cell Amyloid-like properties of bacterial inclusion bodies. J Mol Biol. Fact. 2015;14:1–8. 2005;347:1025–37. 45. Rueda F, Gasser B, Sánchez-Chardi A, Roldán M, Villegas S, Puxbaum V, 22. Morell M, Bravo R, Espargaró A, Sisquella X, Avilés FX, Fernàndez- et al. Functional inclusion bodies produced in the yeast Pichia pastoris. Busquets X, et al. Inclusion bodies: specificity in their aggregation Microb Cell Fact. 2016;15:166. process and amyloid-like structure. Biochim Biophys Acta Mol Cell Res. 46. Pelassa I, Corà D, Cesano F, Monje FJ, Montarolo PG, Fiumara F. Associa‑ 2008;1783:1815–25. tion of polyalanine and polyglutamine coiled coils mediates expansion 23. García-Fruitós E, González-Montalbán N, Morell M, Vera A, Ferraz RM, disease-related protein aggregation and dysfunction. Hum Mol Genet. Arís A, et al. Aggregation as bacterial inclusion bodies does not imply 2014;23:3402–20. inactivation of enzymes and fluorescent proteins. Microb Cell Fact 47. Päiviö A, Nordling E, Kallberg Y, Thyberg J, Johansson J. Stabilization BioMed Central. 2005;4:1–6. of discordant helices in amyloid fibril-forming proteins. Protein Sci. 24. García-Fruitós E, Sabate R, De Groot NS, Villaverde A, Ventura S. Biologi‑ 2004;13:1251–9. cal role of bacterial inclusion bodies: a model for amyloid aggregation. 48. Brandenburg E, Von Berlepsch H, Gerling UIM, Böttcher C, Koksch B. Inhi‑ FEBS J. 2011;278:2419–27. bition of amyloid aggregation by formation of helical assemblies. Chem A 25. Cano-Garrido O, Rodríguez-Carmona E, Díez-Gil C, Vázquez E, Elizondo Eur J. 2011;17:10651–61. E, Cubarsi R, et al. Supramolecular organization of protein-releasing 49. Lupas A, Van Dyke M, Stock J. Predicting coiled coils from protein functional amyloids solved in bacterial inclusion bodies. Acta Biomater. sequences. Science. 1991;252:1162–4. 2013;9:6134–42. 50. Gruber M, Söding J, Lupas AN. Comparative analysis of coiled-coil predic‑ 26. Fowler DM, Koulov AV, Balch WE, Kelly JW. Functional amyloid—from tion methods. J Struct Biol. 2006;155:140–5. bacteria to humans. Trends Biochem Sci. 2007;32:217–24. 51. Delorenzi M, Speed T. An HMM model for coiled-coil domains and a com‑ 27. Fowler DM, Koulov AV, Alory-Jost C, Marks MS, Balch WE, Kelly JW. parison with PSSM-based predictions. Bioinformatics. 2002;18:617–25. Functional amyloid formation within mammalian tissue. PLoS Biol. 52. Ludwiczak J, Winski A, Szczepaniak K, Alva V, Dunin-Horkawicz S. 2006;4:0100–7. DeepCoil-a fast and accurate prediction of coiled-coil domains in protein 28. Maji SK, Perrin MH, Sawaya MR, Jessberger S, Vadodaria K, Rissman RA, sequences. Bioinformatics. 2019;35:2790–5. et al. Functional amyloids as natural storage of peptide hormones in 53. Villar-Piqué A, Espargaró A, Sabaté R, de Groot NS, Ventura S. Using bacte‑ pituitary secretory granules. Science. 2009;325:328–32. rial inclusion bodies to screen for amyloid aggregation inhibitors. Microb 29. Jackson MP, Hewitt EW. Why are functional amyloids non-toxic in Cell Fact. 2012;11:1–11. humans? Biomolecules. 2017;7:1–13. 54. De Groot NS, Aviles FX, Vendrell J, Ventura S. Mutagenesis of the central 30. Berson JF, Theos AC, Harper DC, Tenza D, Raposo G, Marks MS. Pro‑ hydrophobic cluster in Aβ42 Alzheimer’s peptide: side-chain properties protein convertase cleavage liberates a fibrillogenic fragment of a correlate with aggregation propensities. FEBS J. 2006;273:658–68. resident glycoprotein to initiate melanosome biogenesis. J Cell Biol. 55. García-Fruitós E, Arís A, Villaverde A. Localization of functional 2003;161:521–33. polypeptides in bacterial inclusion bodies. Appl Environ Microbiol. 31. Guyonnet B, Egge N, Cornwall GA. Functional amyloids in the mouse 2007;73:289–94. sperm acrosome. Mol Cell Biol. 2014;34:2624–34. 56. Jong WSP, Vikström D, Houben D, Berg van Saparoea HB, Gier JW, Luirink J. 32. Huang PS, Oberdorfer G, Xu C, Pei XY, Nannenga BL, Rogers JM, et al. High Application of an E. coli signal sequence as a versatile inclusion body tag. thermodynamic stability of parametrically designed helical bundles. Sci‑ Microb Cell Fact. 2017;16:1–13. ence. 2014;346:481–5. 57. Jong WSP, ten Hagen-Jongman CM, Vikström D, Dontje W, Abdallah AM, 33. Fletcher JM, Harniman RL, Barnes FRH, Boyle AL, Collins A, Mantell J, de Gier JW, et al. Mutagenesis-based characterization and improvement et al. Self-assembling cages from coiled-coil peptide modules. Science. of a novel inclusion body tag. Front Bioeng Biotechnol. 2020;7:1–13. 2013;340:595–9. 58. Conchillo-Solé O, de Groot NS, Avilés FX, Vendrell J, Daura X, Ventura S. 34. Ing NL, Spencer RK, Luong SH, Nguyen HD, Hochbaum AI. Electronic AGGRESCAN: a server for the prediction and evaluation of “hot spots” of conductivity in biomimetic α-helical peptide nanofibers and gels. ACS aggregation in polypeptides. BMC Bioinform. 2007;8:65. Nano. 2018;12:2652–61. 59. Fernandez-Escamilla AM, Rousseau F, Schymkowitz J, Serrano L. Predic‑ 35. Gradišar H, Božič S, Doles T, Vengust D, Hafner-Bratkovič I, Mertelj A, tion of sequence-dependent and mutational effects on the aggregation et al. Design of a single-chain polypeptide tetrahedron assembled from of peptides and proteins. Nat Biotechnol. 2004;22:1302–6. coiled-coil segments. Nat Chem Biol. 2013;9:362–6. 60. Chiti F, Calamai M, Taddei N, Stefani M, Ramponi G, Dobson CM. Studies 36. Ljubetič A, Lapenta F, Gradišar H, Drobnak I, Aupič J, Strmšek Ž, et al. of the aggregation of mutant proteins in vitro provide insights into the Design of coiled-coil protein-origami cages that self-assemble in vitro genetics of amyloid diseases. Proc Natl Acad Sci USA. 2002;99:16419–26. and in vivo. Nat Biotechnol. 2017;35:1094–101. 61. Chiti F, Stefani M, Taddei N, Ramponi G, Dobson CM. Rationalization of the 37. Jäger VD, Kloss R, Grünberger A, Seide S, Hahn D, Karmainski T, et al. Tailor‑ effects of mutations on peptide and protein aggregation rates. Nature. ing the properties of (catalytically)-active inclusion bodies. Microb Cell 2003;424:805–8. Fact. BioMed Central. 2019;18:1–20. 62. Kyte J, Doolittle RF. A simple method for displaying the hydropathic 38. Diener M, Kopka B, Pohl M, Jaeger KE, Krauss U. Fusion of a coiled-coil character of a protein. J Mol Biol. 1982;157:105–32. domain facilitates the high-level production of catalytically active 63. Kallberg Y, Gustafsson M, Persson B, Thyberg J, Johansson J. Prediction of enzyme inclusion bodies. ChemCatChem. 2016;8:142–52. amyloid fibril-forming proteins. J Biol Chem. 2001;276:12945–50. 39. Lamm R, Jäger VD, Heyman B, Berg C, Cürten C, Krauss U, et al. Detailed 64. Gendoo DM, Harrison PM. Discordant and chameleon sequences: small-scale characterization and scale-up of active YFP inclusion body their distribution and implications for amyloidogenicity. Protein Sci. production with Escherichia coli induced by a tetrameric coiled coil 2011;20:567–79. domain. Elsevier Ltd: J Biosci Bioeng; 2020. 65. McGuffin LJ, Bryson K, Jones DT. The PSIPRED protein structure prediction 40. Kloss R, Limberg MH, Mackfeld U, Hahn D, Grünberger A, Jäger VD, et al. server. Bioinformatics. 2000;16:404–5. Catalytically active inclusion bodies of l-lysine decarboxylase from E. coli for 1,5-diaminopentane production. Sci Rep. 2018;8:1–11. Gil‑Garcia et al. Microb Cell Fact (2020) 19:117 Page 16 of 16 66. Garnier J, Osguthorpe DJ, Robson B. Analysis of the accuracy and 79. Shaner NC, Patterson GH, Davidson MW. Advances in fluorescent protein implications of simple methods for predicting the secondary structure of technology. J Cell Sci. 2007;120:4247–60. globular proteins. J Mol Biol. 1978;120:97–120. 80. De Groot NS, Ventura S. Effect of temperature on protein quality in bacte‑ 67. Gil-Garcia M, Banó-Polo M, Varejao N, Jamroz M, Kuriata A, Díaz-Caballero rial inclusion bodies. FEBS Lett. 2006;580:6471–6. M, et al. Combining structural aggregation propensity and stability 81. De Groot NS, Sabate R, Ventura S. Amyloids in bacterial inclusion bodies. predictions to redesign protein solubility. Mol Pharm. 2018;15:3846–59. Trends Biochem Sci. 2009;34:408–16. 68. Zambrano R, Jamroz M, Szczasiuk A, Pujols J, Kmiecik S, Ventura S. 82. de Groot NS, Ventura S. Protein activity in bacterial inclusion bodies cor‑ AGGRESCAN3D (A3D): server for prediction of aggregation properties of relates with predicted aggregation rates. J Biotechnol. 2006;125:110–3. protein structures. Nucleic Acids Res. 2015;43:W306–13. 83. Lupas AN, Gruber M. The structure of α-helical coiled coils. Adv Prot 69. Kuriata A, Iglesias V, Kurcinski M, Ventura S, Kmiecik S. Aggrescan3D stan‑ Chem. 2005;70:37–78. dalone package for structure-based prediction of protein aggregation 84. Dasari M, Espargaro A, Sabate R, Lopez Del Amo JM, Fink U, Grelle G, et al. properties. Bioinformatics. 2019;35:3834–5. Bacterial inclusion bodies of alzheimer’s disease β-amyloid peptides 70. Kuriata A, Iglesias V, Pujols J, Kurcinski M, Kmiecik S, Ventura S. Aggres‑ can be employed to study native-like aggregation intermediate states. can3D (A3D) 2.0: prediction and engineering of protein solubility. Nucleic ChemBioChem. 2011;12:407–23. Acids Res. 2019;47:W300–7. 85. Gifre-Renom L, Seras-Franzoso J, Rafael D, Andrade F, Cano-Garrido O, 71. Tedeschi G, Mangiagalli M, Chmielewska S, Lotti M, Natalello A, Brocca Martinez-Trucharte F, et al. The biological potential hidden in inclusion S. Aggregation properties of a disordered protein are tunable by pH bodies. Pharmaceutics. 2020;12:157. and depend on its net charge per residue. Biochim Biophys Acta. 86. Wang R, Li J, Dang D, Hu J, Hu Y, Fan J. Bacterial production of maize and 2017;1861:2543–50. human serine racemases as partially active inclusion bodies for d-serine 72. Zhou NE, Zhu B-Y, Kay CM, Hodges RS. The two-stranded α-helical coiled- synthesis. Enzyme Microb Technol. 2020;137:109547. coil is an ideal model for studying protein stability and subunit interac‑ 87. Carratalá JV, Cano-Garrido O, Sánchez J, Membrado C, Pérez E, Conchillo- tions. Biopolymers. 1992;32:419–26. Solé O, et al. Aggregation-prone peptides modulate activity of bovine 73. Zhou NE, Kay CM, Hodges RS. Synthetic model proteins: the relative interferon gamma released from naturally occurring protein nanoparti‑ contribution of leucine residues at the nonequivalent positions of the cles. N Biotechnol. 2020;57:11–9. 3–4 hydrophobic repeat to the stability of the two-stranded α-helical 88. Céspedes MV, Fernández Y, Unzueta U, Mendoza R, Seras-Franzoso J, coiled-coil. Biochemistry. 1992;31:5739–46. Sánchez-Chardi A, et al. Bacterial mimetics of endocrine secretory gran‑ 74. Reisdorf WC, Krimm S. Infrared amide I’ band of the coiled coil. Biochemis‑ ules as immobilized in vivo depots for functional protein drugs. Sci Rep. try. 1996;35:1383–6. 2016;6:1–10. 75. Heimburg T, Schünemann J, Weber K, Geisler N. FTIR-spectroscopy of 89. Rodríguez-Carmona E, Cano-Garrido O, Seras-Franzoso J, Villaverde A, multistranded coiled coil proteins. Biochemistry. 1999;38:12727–34. García-Fruitós E. Isolation of cell-free bacterial inclusion bodies. Microb 76. Manas ES, Getahun Z, Wright WW, DeGrado WF, Vanderkooi JM. Infrared Cell Fact. 2010;9:1–9. spectra of amide groups in α-helical proteins: evidence for hydrogen bonding between helices and water. J Am Chem Soc. 2000;122:9883–90. 77. Deliolanis NC, Kasmieh R, Wurdinger T, Tannous BA, Shah K, Ntziachristos Publisher’s Note V. Performance of the red-shifted fluorescent proteins in deep-tissue Springer Nature remains neutral with regard to jurisdictional claims in pub‑ molecular imaging applications. J Biomed Opt. 2008;13:044008. lished maps and institutional affiliations. 78. Shaner NC, Steinbach PA, Tsien RY. A guide to choosing fluorescent proteins. Nat Methods. 2005;2:905–9. Ready to submit your research ? Choose BMC and benefit from: • fast, convenient online submission • thorough peer review by experienced researchers in your field • rapid publication on acceptance • support for research data, including large and complex data types • gold Open Access which fosters wider collaboration and increased citations • maximum visibility for your research: over 100M website views per year At BMC, research is always in progress. Learn more biomedcentral.com/submissions