Roca‑Pinilla et al. Microb Cell Fact (2020) 19:175 https://doi.org/10.1186/s12934-020-01425-x Microbial Cell Factories RESEARCH Open Access Exploring the use of leucine zippers for the generation of a new class of inclusion bodies for pharma and biotechnological applications Ramon Roca‑Pinilla1, Sara Fortuna2, Antonino Natalello3, Alejandro Sánchez‑Chardi4,5, Diletta Ami3, Anna Arís1* and Elena Garcia‑Fruitós1* Abstract Background: Inclusion bodies (IBs) are biologically active protein aggregates forming natural nanoparticles with a high stability and a slow-release behavior. Because of their nature, IBs have been explored to be used as biocatalysts, in tissue engineering, and also for human and animal therapies. To improve the production and biological efficiency of this nanomaterial, a wide range of aggregation tags have been evaluated. However, so far, the presence in the IBs of bacterial impurities such as lipids and other proteins coexisting with the recombinant product has been poorly stud‑ ied. These impurities could strongly limit the potential of IB applications, being necessary to control the composition of these bacterial nanoparticles. Thus, we have explored the use of leucine zippers as alternative tags to promote not only aggregation but also the generation of a new type of IB-like protein nanoparticles with improved physicochemi‑ cal properties. Results: Three different protein constructs, named GFP, J-GFP-F and J/F-GFP were engineered. J-GFP-F corresponded to a GFP flanked by two leucine zippers (Jun and Fos); J/F-GFP was formed coexpressing a GFP fused to Jun leucine zipper (J-GFP) and a GFP fused to a Fos leucine zipper (F-GFP); and, finally, GFP was used as a control without any tag. All of them were expressed in Escherichia coli and formed IBs, where the aggregation tendency was especially high for J/F-GFP. Moreover, those IBs formed by J-GFP-F and J/F-GFP constructs were smaller, rougher, and more amorphous than GFP ones, increasing surface/mass ratio and, therefore, surface for protein release. Although the lipid and carbo‑ hydrate content were not reduced with the addition of leucine zippers, interesting differences were observed in the protein specific activity and conformation with the addition of Jun and Fos. Moreover, J-GFP-F and J/F-GFP nanoparti‑ cles were purer than GFP IBs in terms of protein content. Conclusions: This study proved that the use of leucine zippers strategy allows the formation of IBs with an increased aggregation ratio and protein purity, as we observed with the J/F-GFP approach, and the formation of IBs with a higher specific activity, in the case of J-GFP-F IBs. Thus, overall, the use of leucine zippers seems to be a good system for the production of IBs with more promising characteristics useful for pharma or biotech applications. Keywords: Inclusion bodies, Aggregation, Recombinant protein, Leucine zippers, Jun, Fos, Purity *Correspondence: anna.aris@irta.cat; elena.garcia@irta.cat 1 Department of Ruminant Production, Institute of Agriculture and Food Research and Technology (IRTA), 08140 Caldes de Montbui, Spain Full list of author information is available at the end of the article © 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. Roca‑Pinilla et al. Microb Cell Fact (2020) 19:175 Page 2 of 13 Background reporter protein fused to Jun and Fos LZ [23] at different Inclusion bodies (IBs) are protein nanoparticles rang- positions. The purity and formation of GFP IBs have been ing from 50 to 800 nm formed during the production studied to determine the possible role of LZ in obtaining of recombinant proteins and often localized in the cyto- purer and better controlled IBs. plasmic space of bacterial cells [1], although their forma- tion in the periplasm has also been described [2]. In the Results past, IBs have been regarded as an undesired byproduct Construct design and modelling of recombinant protein production processes [3]. How- In this study three different protein constructs, named ever, over the last 15 years it has been broadly proven GFP, J-GFP-F and J/F-GFP, have been engineered to eval- that proteins forming IBs are biologically active [4–7]. uate the effect of Jun and Fos LZ on the protein aggrega- The presence of native proteins forming such aggregates tion process (Fig. 1a). J-GFP-F is a single fusion protein has prompted to assess their potential as biomaterials for consisting on a GFP flanked by Jun and Fos at N- and a wide range of applications, including biocatalysis, tis- C-terminus, respectively, whereas J/F-GFP is consti- sue engineering, and antimicrobial and cancer therapies tuted by two proteins (a GFP with Jun LZ at N-terminus [7–10]. They offer important advantages over their solu- (J-GFP) and a GFP with Fos at N-terminus (F-GFP)), ble counterparts such as high stability [11], slow-release simultaneously coexpressed (Fig. 1). GFP protein has behavior [12, 13], and production through cost-effective been used as a control without any LZ tag. The three- processes [14]. Aiming to improve the production and dimensional structural arrangements of the constructs biological efficiency of this nanomaterial, the increase of were built by iterative threading, taking advantage of the aggregation tendency of proteins of interest is a key the I-TASSER webserver [26]. The generated models aspect and has been evaluated through the use of dif- were visually inspected and possible three-dimensional ferent aggregation tags, being VP1 [6], GFIL8 [15] and arrangements of the J/F units with respect to the GFP ELK16 [16] three representative examples. Besides, the were selected as representative conformational arrange- scale-up of IB production and purification protocols have ments (Fig. 1b). In all constructs, the J/F subunits tended also been improved during the last decade [14, 17]. It has to assume a helical structure, as expected, but it was only been described that the recombinant protein forming this possible to generate highly ordered starting domains biomaterial coexists with other proteins such as chaper- for J/F-GFP. All models underwent 250 ns of atomis- ones, and also with lipids [18, 19]. However, the presence tic molecular dynamics simulations in full water solvent of these impurities has been poorly studied, and conse- (Additional file 1: Figure S1) showing the J/F elongated quently there is room to optimize IB composition. This is domains to be structurally unstable leading to partially an important challenge to be solved in terms of IB appli- disordered arrangements similar to those initially mod- cability, since for specific applications it is important to elled for J-GFP-F. Among the generated models for control the exact composition of these bacterial nanopar- J-GFP-F, no highly ordered structure was present, and ticles. Thus, exploring ways to control the IB formation the pool differed by the local arrangement of the J/F sub- process emerges as a central strategy to improve IB purity domains (as exemplified by two structures in Fig. 1b). and better control their physicochemical properties. In the present study, we have used an alternative approach Protein production and aggregation for the generation of a new type of IB-like protein nano- Cell growth was determined under the overexpres- particles using leucine zippers (LZ) as aggregation-seed- sion of the three different constructs used in this study ing domains with the aim to drive protein aggregation (Fig. 2). Interestingly, a reduction in the bacterial growth and improve IB properties for industrial applications. was observed when J-GFP-F and J/F-GFP were pro- In marked contrast to aggregation tags used till present duced, especially at 5 h (Fig. 2). Moreover, those pro- to increase protein aggregation, LZ are protein–protein teins that impaired the growth (J-GFP-F and J/F-GFP) interaction domains consisting of amphipathic α helices were produced at lower levels than their GFP counter- that dimerize in parallel, either as homodimers or heter- part (Table 1). Analyzing the protein yields at different odimers, to form a coiled-coil [20–22]. LZ dimerization production time points, it can be observed that there is motifs have already been explored as protein–protein a time-dependent production of both GFP and J/F-GFP, interaction drivers both in recombinant mammalian [23] while for J-GFP-F production values kept similar at dif- and bacterial [24, 25] cells. Thus, LZ specific properties ferent times post-induction (Table 1). make these peptides promising players to improve, con- Despite of the reduction in protein levels and in cell trol, and stabilize IB quality, thus boosting their potential viability for those proteins carrying Jun and Fos (Fig. 2 for pharma and biotech industrial applications. In this and Table 1), all constructs aggregated (Fig. 3). Spe- study, we have explored the aggregation profile of a GFP cifically, the analysis of the aggregation ratio showed Roca‑Pinilla et al. Microb Cell Fact (2020) 19:175 Page 3 of 13 Fig. 1 Recombinant construct. a Diagram of the protein-based constructs and their possible aggregation process as IBs depending on the presence or not of LZ. b Representative three-dimensional models of the J-GFP-F (two models: a and b) and J/F-GFP constructs (one model each), as generated by iterative threading. Construct domains are color coded as follow: GFP (green), Jun (blue), Fos (red) differences among the constructs used, being the aggre- Characteristics of purified IBs gation ratio higher for J/F-GFP than for GFP and J-GFP-F Aiming to explore the specific nanoarchitectural char- (Fig. 3 and Additional file 1: Table S1). acteristics of the protein nanoparticles formed using the three different tag combinations, IBs of each construct were purified. In all samples, qualitative and quantita- tive approaches with high resolution electron microscopy Roca‑Pinilla et al. Microb Cell Fact (2020) 19:175 Page 4 of 13 when forming IBs, the difference between soluble and IB-forming proteins was especially greater in the case of GFP (Fig. 5b). The conformational properties of the proteins embed- ded in the IBs were investigated by Fourier transform infrared (FTIR) spectroscopy [27]. The spectra were collected before and after hydrogen/deuterium (H/D) exchange to allow a better assignment of the Amide I band components to the protein secondary structures [28, 29]. GFP and J/F-GFP IBs displayed comparable IR response, while J-GFP-F IBs showed distinct peak posi- tions and relative intensities of the spectral components assigned to β-sheets (Fig. 6 and Additional file 1: Figure S3). In particular, the absorption spectra (both before and after H/D exchange) of J-GFP-F IBs were more similar to those of the soluble GFP (Additional file 1: Figure S3) and Fig. 2 Optical density values of recombinant bacteria cultures in LB the main β-sheet peak showed a higher downshift after medium after protein expression induction. Circles represent E. coli H/D exchange compared to GFP and J/F-GFP IBs (Fig. 6). pET22b/GFP (GFP), triangles E. coli pET22b/Jun-GFP-Fos (J-GFP-F), In the deuterated J-GFP-F, this component occurred at and squares represent E. coli pETDuet-1-Jun-GFP/Fos-GFP (J/F-GFP). around 1620 cm−1, a peak position near to that observed Different letters depict significant differences between the growth curves (p ≤ 0.0001) for the main β-sheet band of soluble GFP, which indeed was observed around 1622 cm−1 (Fig. 6) [30, 31]. Overall, these data suggested that the Jun and Fos motifs modu- lated the conformational features of the formed IBs. imaging of the ultrastructural morphometric of IBs showed a high number of well-formed nanoparticles Purity of IBs with round shape and nanoscale size in the three con- Interestingly, J-GFP-F and J/F-GFP nanoparticles have structs (Fig. 4). However, slight and important shape and a degree of protein purity more than twice than that size differences between GFP IBs and those nanopar- observed for parental GFP nanoparticles (Fig. 7). Besides, ticles formed with Jun or Fos LZ were detected. First of the analysis of lipid and carbohydrate content in IBs all, control GFP IBs showed a very homogeneous round indicated that the presence of Jun and Fos in J-GFP-F shape and smooth surface, suggesting a highly compact increased the presence of these co-contaminants while structure (Fig. 4). In contrast, Jun-Fos based IBs showed the strategy for the production of IBs using J/F-GFP a high variability in shape, significantly more amor- showed no significant differences with GFP in lipid and phous than GFP ones and with a high percentage of IBs carbohydrate content (Fig. 8). showing rough and porous surface. About the size, GFP nanoparticles showed the highest size with a diameter of Discussion 400–500 nm and a large mean area (Fig. 4), whereas Jun- IBs are protein aggregates ranging at nanoscale that have GFP-Fos and Jun-GFP/Fos-GFP IBs were significantly been widely studied from different perspectives. Since smaller with a lower area and a diameter around 250 nm most of the recombinant proteins form IBs when over- in both cases (Fig. 4). expressed in bacteria [3], and some of them are only pro- The analysis of the protein quality in terms of protein duced in this insoluble format, these nanoparticles have activity of GFP, J-GFP-F, and J/F-GFP IBs indicated that been broadly used as a source to obtain the soluble form the addition of Jun and Fos improved the functional of a wide range of protein of interest [32]. On the other protein content when flanking the GFP, whereas no dif- hand, by being protein nanoparticles rich in functional ferences were observed for J/F-GFP compared to GFP recombinant protein, IBs have also been explored as a (Fig. 5a). This indicates that the GFP forming J-GFP-F new class of biomaterial with promising applications in nanoparticles has better quality and, in consequence, biocatalysis, tissue engineering, and human and animal higher fluorescence per μg of protein than the two other therapies [7–9, 11]. However, although different aggre- constructs used. Interestingly, the comparison of the spe- gation tags have been used to promote their formation, cific fluorescence of GFP forming the IBs with the soluble the impact of these aggregation domains in the quality version of this protein indicated that although the activ- of protein aggregates has not been addressed so far. For ity of the soluble form of all the proteins was higher than that, in this work, we have explored if the use of Jun and Roca‑Pinilla et al. Microb Cell Fact (2020) 19:175 Table 1 Total protein yield (μg) of each recombinant construct at different times post-induction (1, 3, and 5 h) GFP J-GFP-F J/F-GFP p value Time (h) 1 3 5 1 3 5 1 3 5 Treat Time Treat × time Protein yield (μg) 52.39 ± 0.79 99.43 ± 1.03 223.82 ± 1.48 1.44 ± 0.12 1.23 ± 0.08 0.93 ± 0.07 5.12 ± 0.23 9.98±0.2 20.66 ± 1.96 < 0.0001 0.029 0.014 Page 5 of 13 Roca‑Pinilla et al. Microb Cell Fact (2020) 19:175 Page 6 of 13 Fos LZ dimerization motifs could drive the formation surface of J-GFP-F and J/F-GFP IBs could be indicative of IBs in a more controlled way in terms of quality. The of significant differences in protein production and nano- results obtained proved that the constructs containing particle formation. In fact, the nanoarchitectural aspect Jun and Fos (Fig. 1a), although they showed some toxic- of GFP IBs shows nanostructures with higher size, com- ity to the producer cells (Fig. 2), they were produced as pactness and smooth surface than Jun and Fos IBs, which IBs with aggregation ratios higher than GFP (Fig. 3), appear as more soft particles with a rough surface, more proving that the aggregation propensity can be improved amorphous shape, and lower mean size. These nanoscale by using this strategy. The enhancement of the aggrega- differences can play a great role in differential function- tion propensity can be attributed to the formation of Jun alities of proteins forming IBs in potential therapeuti- and Fos interactions among constructs, confirming that cal applications. Then, low size, amorphous shape, and while the Jun and Fos domains seem to explore a number rough surfaces increase surface/mass ratio and, there- of partially disordered conformations when free in solu- fore, potential protein release at in vitro and in vivo tion, this does not hinder the formation of stable aggre- conditions, becoming suitable and desired morphomet- gates. Interestingly, the aggregation ratio was observed ric characteristics for more efficiently releasing nano or to be higher for J/F-GFP than for J-GFP-F (Fig. 3). This micro-platforms of drug delivery systems. Moreover, can be rationalized in terms of competition between although smaller, GFP proteins forming J-GFP-F nano- intramolecular and intermolecular Jun/Fos interactions. particles had a specific activity significantly higher than In J-GFP-F the Jun and Fos domains are entangled due the other constructs that we tested (Fig. 5a), which is in to intramolecular interactions between them (Additional accordance with FTIR spectra (Fig. 6 and Additional file 1: Figure S1). Their association within the same con- file 1: Figure S3). In particular, after H/D exchange, the struct competes with the formation of dimers, or multim- main β-sheet peak of J-GFP-F IBs was observed to be ers, where Jun and Fos fragments belonging to different very close to that observed for the soluble GFP, indicating molecules interact and bond with each other. In both the maintenance of native-like conformational features. cases the two interacting fragments are the same, thus Although J/F-GFP IBs showed no differences in the qual- their energetic is expected to be close. At room tempera- ity of the protein forming the IBs when compared to GFP ture we might expect the competition and coexistence of (Fig. 5a), both J-GFP-F and J/F-GFP proteins forming IBs monomers, dimers, and multimers in solution. However, presented activities closer to the soluble form than GFP it is also worth noting that in J-GFP-F, each Jun unit will produced without tags (Fig. 5b). This indicates that the be close to a Fos domain; this corresponds to a high local presence of Jun and Fos had a positive impact on the IB concentration of the partner, thus favoring the intramo- protein quality. This result agrees with previous studies lecular entangled J-GFP-F. On the other hand, when the that have described that the recombinant protein pro- Jun/Fos domains are uncoupled, independently bound duced or the strain used can have an impact on the con- to different GFPs (as in J/F-GFP), there is obviously no formational quality of the proteins forming IBs [36–38]. competition between intermolecular and intramolecular On the other hand, the analysis of the elements form- interactions. ing such aggregates showed that J-GFP-F and J/F-GFP The study of the specific physicochemical character- IBs had less protein impurities (Fig. 7), which suggests istics of purified GFP, J-GFP-F and J/F-GFP IBs showed that Jun and Fos sequences drive a more controlled for- that, depending on the LZ strategy used, it’s possible to mation of the protein nanoparticles in terms of protein modulate IBs features such as specific activity, size, pro- composition. By contrast, the presence of lipids and car- tein purity and presence of contaminants such as lipids bohydrates could not be decreased by the addition of LZ and carbohydrates. The presence of LZ has a significant (Fig. 8). The amount of both lipids and carbohydrates in impact on both the size and shape of the aggregates, J/F-GFP IBs was comparable to the levels in GFP IBs, with IB diameters around 250 nm and more amorphous whereas in the case of J-GFP-F was even higher (Fig. 8). forms, while GFP IBs have a size of 400–500 nm and a This could be correlated with the LZ interactions in each higher average surface and higher roughness (Fig. 4). construct. As previously discussed, in J-GFP-F the Jun However, despite the differences in size, which can be and Fos domains of different molecules compete with correlated with the lower production yields achieved intramolecular interactions between the two domains, for those constructs containing LZ (Table 1), surface and this could drive the formation of less compact aggre- rugosity and roundness, ultrastructural morphometry gates which could contain more lipids and carbohydrates. of J-GFP-F and J-/F-GFP IBs (Fig. 4) agree with other Overall, these results demonstrate that aggregation- conventional IBs produced in E. coli [33] and other cell seeding domains based on LZ peptide–peptide inter- factories such as Lactococcus lactis [34] or Pichia pasto- action can drive the formation of a specific type of IBs, ris [35]. Interestingly, more amorphous shape and rough improving their quality in terms of protein content Roca‑Pinilla et al. Microb Cell Fact (2020) 19:175 Page 7 of 13 cloned into pET22b ­(AmpR) vector (pET22b-Jun-GFP- Fos), while Jun-GFP and Fos-GFP constructs were cloned in pETDuet-1 ­ (AmpR) to co-express them (pETDuet- 1-Jun-GFP/Fos-GFP). As a control, residues 2–238 of EGFP were cloned into pET22b. The DNA sequences corresponding to each gene sequence were codon opti- mized for its expression in Escherichia coli (GeneArt, Germany). Preparation of E. coli competent cells Escherichia coli BL21 (DE3) cultures were grown over- Fig. 3 Protein aggregation ratio (%) for each construct over time. night (ON) in LB medium at 37 °C with shaking at Black, dark grey, and light grey bars represent GFP, J-GFP-F, and 250 rpm. A 1/100 inoculum was done in 50 ml of LB and J/F-GFP, respectively. Significant differences for construct (p ≤ 0.05) and for time (p ≤ 0.1; Additional file 1: Table S1) the culture was grown until the optical density (­ OD600nm) reached a value between 0.2 and 0.4. After that, cultures were centrifuged (4000×g) at 4 °C for 15 min. Pellets were resuspended in 12.5 ml of cold and sterile 50 mM (Fig. 7), and in one of the approaches increasing spe- ­CaCl2 and incubated for 45 min in an ice bath. Cells were cific activity (Figs. 5a and 6). However, those IBs with centrifuged again as described above and resuspended in higher specific activity (J-GFP-F) are produced through 1.25 ml of cold and sterile 50 mM C ­ aCl2 in glycerol (15% a strategy that has a negative impact in the presence of v/v) to prepare aliquots of 200 µl, which were stored at contaminants such as lipids and carbohydrates (Fig. 8). − 80 °C. To transform the cells, 40 ng of plasmid DNA On the contrary, the strategy based on the coexpression were added to 200 µl of competent cells. The mixtures of Jun-GFP and Fos-GFP (J/F-GFP) to form IBs, even were incubated on ice for 30–60 min, warmed up to though it did not display any increase in the protein qual- 42 °C for 45 s and placed on ice for 30 s. After incubation, ity (Fig. 5a), had no negative impact in the content of 800 µl of LB media were added, and transformed cells lipids and carbohydrates (Fig. 8), therefore showing that were incubated at 37 °C for 1 h. Finally, the cells were it could be a promising approach for the production of plated on LB-agar plates containing the corresponding IBs with higher recombinant protein content and less antibiotic. protein impurities. Protein production Conclusions Escherichia coli BL21 (DE3)/pET22b-GFP, E. coli Altogether this study proved that the use of Jun and Fos BL21(DE3)/pET22b-Jun-GFP-Fos (J-GFP-F) and E. coli LZ is a good strategy for the production of IBs with more BL21(DE3)/pETDuet-1-Jun-GFP/Fos-GFP (J/F-GFP) appealing characteristics that might be useful for pharma ON cultures were inoculated in 50 ml of LB media or biotech applications. This is especially relevant for J/F- with 100 µg/ml ampicillin in 200-ml flasks at an initial GFP approach, which allowed producing hybrid IBs with ­OD600nm = 0.05. Each culture was grown at 37 °C and increased aggregation ratio and protein purity without 250 rpm until the ­OD600nm was 0.5 and 1 mM isopropyl- negatively affecting their activity and lipid and carbohy- β-d-thiogalactoside (IPTG) was added to induce recom- drate content. binant protein expression. At times 0, 1, 3, and 5 h after IPTG induction, 1 ml samples were collected for protein Methods fractioning. These cultures were performed by triplicate. Construction of expression plasmids The sequence encoding amino acid residues 2–238 of Protein fractionation the enhanced GFP (EGFP) was fused downstream of Samples of 1 ml were harvested by centrifugation at sequence encoding Fos 118–210 (bFos) or Jun 257–318 6000×g at 4 °C for 15 min and the pellet was resuspended (bJun) using the linker sequence encoding SGGGSGGS in 0.5 ml phosphate buffered saline (PBS) supplemented to construct Fos-GFP and Jun-GFP, respectively. For with protease inhibitor (Complete EDTA-free, Roche, the Jun-GFP-Fos construct, the sequence encoding Jun Switzerland) to prevent protein proteolysis. Then, ice- 257–318 (bJun) was fused at the N-terminus, while the jacketed samples were disrupted by sonication (2 cycles sequence encoding Fos 118–210 (bFos) was fused at of 1.5 min at 10% amplitude under 0.5 s cycles) (Branson the C-terminus, using in both cases the linker sequence Ultrasonic SA, Switzerland). These samples were centri- encoding SGGGSGGS. The Jun-GFP-Fos construct was fuged at 15,000×g and 4 °C for 15 min to separate soluble Roca‑Pinilla et al. Microb Cell Fact (2020) 19:175 Page 8 of 13 Fig. 4 Representative FESEM images of the isolated IBs for each construct: GFP IBs, J-GFP-F IBs and J/F-GFP IBs. Bars size: 200 nm. IB mean area ­(nm2), mean diameter (nm) and roundness (%) was calculated for each construct IB (****p ≤ 0.0001) and insoluble fractions protein fractions. Samples were 45 min, respectively. At that time, samples were loaded stored at − 80 °C. onto the gel. SDS-PAGE protein bands were transferred onto PVDF membranes and identified using a com- Protein determination mercial anti-GFP antibody (1:1000, sc-9996, Santa Cruz Soluble and insoluble protein fractions were analyzed Biotechnology, USA), followed by an incubation with a by denaturing SDS-PAGE (15% acrylamide). Denaturing secondary ALP-conjugated anti-mouse IgG (whole mol- buffer (Laemli 4×: Tris base 1.28 g, glycerol 8 ml, SDS ecule) antibody (1:20,000, A4313, Sigma-Aldrich, USA). 1.6 g, β-mercaptoethanol 4 ml, urea 9.6 g in 20 ml) was The amounts of recombinant protein were estimated by added to the insoluble and soluble fractions to a final comparison with known amounts (usually ranging from concentration of 1× (see protein fractioning). Soluble 125 to 1000 ng) of T22-GFP protein [36]. Protein bands and insoluble protein fractions were boiled for 10 and were visualized with a solution of NBT/BCIP (B6404, Roca‑Pinilla et al. Microb Cell Fact (2020) 19:175 Page 9 of 13 Fig. 5 Specific fluorescence of the protein constructs. IB specific fluorescence for the three constructs (a). Specific fluorescence ratio of the insoluble fraction compared to the soluble fraction specific fluorescence (b). Black, dark grey, and light grey bars represent GFP, J-GFP-F, and J/F-GFP, respectively. * shows statistically significant differences compared to the GFP construct (p ≤ 0.05), and different letters also show statistically significant differences (p ≤ 0.05) Fig. 7 Protein purity of IBs. For each construct, protein purity was determined by densiometric image-analysis, comparing the amount of the specific recombinant construct to the whole protein amount of the IB. Different letters depict statistically significant differences (p ≤0.05) Sigma-Aldrich, USA), and images were obtained using a Color Image Scanner. ImageJ software was used to per- form densitometric analyses of the bands. Determination of specific fluorescence Fluorescence intensity of the three constructs in 1 ml samples was determined in a Varian Cary Eclipse fluores- cence spectrometer (Agilent Technologies, Australia) at excitation and emission wavelengths of 480 and 510 nm, respectively. Since protein amounts in 1 ml samples for each construct, fraction and time are known (see protein determination), specific fluorescence was calculated by dividing the fluorescence intensity by the amount of pro- tein for each construct, time and fraction. Fig. 6 Second derivatives of the FTIR absorption spectra of GFP, J-GFP-F and J/F-GFP IBs and of soluble GFP. Samples were measured in form of protein films, obtained by solvent evaporation, and after Protein purity assessment re-hydration by D ­ 2O in order to allow H/D exchange. Selected peaks After separation, the gels were carefully transferred to a and the typical spectral region of the different protein secondary plastic tray filled with 200 ml of distilled water and agi- structures after H/D exchange are indicated. α, α-helices; β, β-sheets; tated at 50 rpm to remove SDS traces. The staining solu- βt, β-turns; RC, random coils tion (Coomassie Brilliant BlueR-250 Staining Solution, Roca‑Pinilla et al. Microb Cell Fact (2020) 19:175 Page 10 of 13 Fig. 8 Lipid (a) and carbohydrate (b) content of for each IB construct. Different letters indicate statistically significant differences (p ≤0.01 (a) and p ≤0.05, (b)) Bio-Rad, USA) was added into the gels (and incubated for Then, DNA was removed with DNAse at a final concen- 1 h at room temperature (RT) and revealed with destain- tration of 0.6 µg/ml and 0.6 mM M ­ gSO4 for 1 h at 37 °C. ing solution (50% ­H2O, 40% methanol, 10% acetic acid Samples were centrifuged at 15,000×g for 15 min at 4 °C. (v/v)) until bands were clearly visible. The gel images Pellets containing IBs were washed with 25 ml lysis buffer were acquired by a Color Image Scanner and analyzed (50 mM Tris–HCl pH 8, 100 mM NaCl, 1 mM EDTA and with the ImageJ software. Triton X-100 0.5% (v/v)). Finally, pellets were centrifuged at 4 °C for 15 min and 15,000×g and stored at − 80 °C Protein aggregation ratio until analysis. The IBs were quantified by western blot The amount of recombinant protein of the soluble and using a monoclonal anti-GFP antibody (1:1000, sc-9996; insoluble fraction for each construct was determined as Santa Cruz Biotechnology, USA). All incubations were explained above. After that, the aggregation ratio was cal- done under agitation. culated by dividing the quantity of the insoluble fraction for each time, replicate and construct by the quantity of Electron microscopy protein in the respective soluble fraction. Field Emission Scanning Electron Microscope (FESEM) was used to visualize the ultrastructural morphology Purification of protein nanoparticles (size and shape) of protein nanoparticles in a nearly Bacterial cultures were processed 3 h post-induction native state. For that, protein samples were directly through a combination of mechanical and enzymatic deposited over silicon wafers (Ted Pella, USA), air dried, disruption methods. Protease inhibitors (Complete and observed with a high resolution standard second- EDTA-free, Roche, Switzerland), phenylmethanesulpho- ary electron detector through a FESEM Merlin (Zeiss, nylfluoride (PMSF) and lysozyme were added to the cul- Germany) operating at 2 kV. As a quantitative morpho- ture at a final concentration of 0.4 mM (Sigma-Aldrich, metric measurement, the mean area of IBs for each con- USA) and 1 µg/ml (Sigma-Aldrich, USA), respectively. struct was analyzed as estimator of size with the Image J After 2 h of incubation at 37 °C and 250 rpm the culture software. The number of particles was 30, 9, 11, for GFP, was centrifuged at 6000×g and resuspended in 30 ml of J-GFP-F and J/F-GFP, respectively. Using mean area val- PBS supplemented with protease inhibitors (Complete ues, the mean diameter of each particles was calculated, EDTA-free, Roche, Switzerland). Then, the mixture was and the diameter was also used as an estimator of size. ice-jacketed and sonicated for 4 cycles of 1.5 min at The roundness of IB particles for each construct was also 10% amplitude under 0.5 s cycles (Branson Ultrasonic evaluated (n = 50/construct), as an assessment of particle SA, Switzerland). After sonication, the mixture was fro- shape. zen ON at − 80 °C. The mixture was thawed and Triton X-100 was added (0.4% (v/v)) and incubated for 1 h at RT. Total carbohydrate and lipid analysis After this treatment, the mixture was frozen at − 80 °C The total lipid amount in IBs was determined following for 2 h and then thawed for several cycles until no viable a sulfo-phospho-vanillin colorimetric assay. Briefly, 500– bacterial growth was detected. After that, 125 µl of Noni- 1000 µl of each sample were centrifuged at 15,000×g and det P40 (NP-40) was added and incubated for 1 h at 4 °C. 4 °C for 15 min. The supernatants were removed, and Roca‑Pinilla et al. Microb Cell Fact (2020) 19:175 Page 11 of 13 (SpeedVac™, ThermoFisher) and dissolved in 200 µl of each pellet was dried by means of a vacuum lyophilizer simulations and their analysis were run as implemented in the GROMACS package [40]. chloroform in a capped glass tube. The chloroform was evaporated at 63 °C in a fume hood. To each tube, 2 ml FTIR analysis of 18 M sulfuric acid were added. The samples were then IB samples were resuspended in a few hundred micro- incubated for 10 min in a boiling water bath. After that, liters of 20 mM of sodium phosphate buffer pH 7.4 each tube was cooled down in ice for 5 min. Five milli- and 2–3 µl of the suspension at 1 μg/μl were deposited liters of phosphoric acid-vanillin reagent (100 ml of 85% in several drops on a B ­ aF2 window and dried at RT in (v/v) phosphoric acid and 0.12 g of vanillin, (Sigma- order to obtain a protein film. In particular, different Aldrich, Germany)) were added to the tubes and incu- sample concentrations were analyzed in order to obtain bated for 15 min at 37 °C. The tubes were cooled for thin (to avoid excessive absorption) and uniform films 15 min in ice and absorbance was measured at 530 nm. without cracks [28]. Soluble GFP at 2 mg/ml in 20 mM For the standard curve, a range of 10-100 μg triolein (dis- sodium phosphate buffer pH 7.4, was analyzed in the solved in chloroform) was used. To determine the total form of a protein film as for IB samples. The transmis- content of carbohydrates present in the IBs samples, a sion FTIR absorption spectra were then acquired by phenol–sulfuric assay was performed. A glucose stand- the Varian 670-IR FTIR spectrometer coupled to the ard was prepared ranging from 0 to 150 nanomols and Varian 610-IR infrared microscope (both from Varian, 150-225 µl of IBs were centrifuged at 15,000×g and 4 °C Australia) equipped with a mercury cadmium telluride for 15 min. The supernatant was removed, and each pel- nitrogen-cooled detector. The following conditions were let was dried with a vacuum lyophilizer. A total of 150 μl employed: 2 cm−1 spectral resolution, 25 kHz scan speed, of 18 M sulfuric acid was added, followed immediately by 512 scan co-additions, and triangular apodization. Sev- 30 μl of 5% phenol. Samples were incubated at 90 °C for eral areas for each sample were measured to verify the 5 min and then cooled to RT, and absorbance was meas- reproducibility of the spectral results. Only absorption ured at 490 nm. All assays were performed in triplicate. spectra with the Amide I band intensity below 0.8 were considered reliable. For H/D exchange, the protein film Structure modelling on the B­ aF2 window was rehydrated by the deposition of J-GFP-F and J/F-GFP were modelled by iterative thread- 8 μl of D ­ 2O around the dried film. The chamber was then ing, as implemented the I-TASSER server [26] without tightly closed by a second window using a flat O-ring and applying any additional restrains. Given the primary incubated for 5 h at RT to allow H/D exchange [28, 34]. sequence of the three constructs the server allows gen- The FTIR spectra of the ­D2O-rehydrated samples were erating possible three-dimensional models by mul- collected as described above. tiple threading alignment. The quality of the models Protein spectra were obtained after subtraction of the was assessed both by their C-score, which was nega- proper reference spectra strictly collected under the same tive in all cases, and visually. For J-GFP-F, both the first conditions. The second derivatives [41] were calculated (C-score = − 1.86) and the second (C-score = − 2.36) after spectral smoothing by the Savitsky-Golay method. modes were chosen as these models showed different Data collection and analysis were performed using the arrangements of the Jun/Fos domains. For both J-GFP Resolutions-Pro software (Varian, Australia). (C-score = − 1.68) and F-GFP (C-score = − 2.87) the first model was chosen. Statistical analysis All quantitative data are presented as mean val- Molecular dynamics protocol ues ± standard error of the mean (x̅ ± SEM). Normality For all the systems, the free construct was minimized, of the data was determined by a Shapiro–Wilk test. For placed in a cubic box with a water layer of 0.7 nm and IB roundness, the data were normalized with the follow- ­Na+ ­Cl- ions to neutralize the system, and a second ing formula: sqrt(max(x + 1) - x), where x is IB round- minimization was performed. We used AMBER99SB- ness and max the maximal roundness value in the data. ILDN [37] force field and Simple Point Charge water. A one-way ANOVA was conducted for all quantitative NVT and NPT equilibrations were run for 100 ps, fol- results except for IB surface and specific fluorescence lowed by 250 ns NPT production run at 300 °K. The measures. For the latter, values were compared with an temperature was controlled with a modified Berend- independent sample t-test. Finally, we did Post hoc com- sen thermostat [38], the pressure with an isotropic Par- parisons using the Tukey HSD test for all data analyzed rinello-Rahman at 1 bar. The iteration time step was set by the ANOVA method. The level of significance was set to 2 fs with the Verlet integrator and LINCS [39] con- at p < 0.05. Measures were done in triplicate, except for IB straint. Periodic boundary conditions were used. All mean area, diameter and roundness, as indicated above. Roca‑Pinilla et al. Microb Cell Fact (2020) 19:175 Page 12 of 13 All statistical analyses were performed using the RStudio Ethics approval and consent to participate Not applicable. Statistical Software (RStudio, Inc., USA). Consent for publication Supplementary information Not applicable. Supplementary information accompanies this paper at https​://doi. Competing interests org/10.1186/s1293​4-020-01425​-x. The authors declare that they have no competing interests. Additional file 1. Figure S1. Overlap between starting model (lighter Author details 1 shades) and final configuration (darker shades) of the J-GFP-F (two Department of Ruminant Production, Institute of Agriculture and Food models: a and b) and J/F-GFP constructs (one model each), after 250 ns of Research and Technology (IRTA), 08140 Caldes de Montbui, Spain. 2 Depart‑ molecular dynamics simulation. The generated models were minimized, ment of Chemical and Pharmaceutical Sciences, University of Trieste, Via placed in a cubic water box, minimized again, equilibrated and, for each L. Giorgieri 1, 34127 Trieste, Italy. 3 Department of Biotechnology and Bio‑ construct, 250 ns of molecular dynamics simulation were performed. sciences, University of Milano-Bicocca, 20126 Milan, Italy. 4 Department of Evo‑ Large rearragements of the Jun/Fos domains were observed. Construct lutionary Biology, Ecology and Environmental Sciences, Faculty of Biology, domains are color coded as follow: GFP (green), Jun (blue), Fos (red). Fig- University of Barcelona (UB), 08028 Barcelona, Spain. 5 Microscopy Service, ure S2. (a) Representative FESEM images of the isolated IBs for each con‑ Autonomous University of Barcelona (UAB), 08193 Cerdanyola del Vallès, Spain. struct: GFP IBs, J-GFP-F IBs and J/F-GFP IBs. Bars size represent 200 nm. (b) Frequency distribution of IBs ultrastructural morphometry quantification Received: 23 June 2020 Accepted: 11 August 2020 for each construct: size (area ­(nm2) and diameter (nm)) and shape (round‑ ness (%)). Figure S3. A) FTIR absorption spectra of the protein films. B) FTIR absorption spectra collected after re-hydration of the protein films with ­D2O for 5 h. GFP and J/F-GFP IBs displayed similar absorption spectra both as film and after re-hydration, while J-GFP-F IBs showed distinct spectral References features. As a control, the absorption spectra of the soluble GFP are also 1. García-Fruitós E, Sabate R, de Groot NS, Villaverde A, Ventura S. Biological shown. Supplementary Table 1. Statistics for the protein aggregation role of bacterial inclusion bodies: a model for amyloid aggregation. FEBS ratio (%) for each construct over time. (a) Aggregation ratio (%) differences J. 2011;278(14):2419–27. https​://doi.org/10.1111/j.1742-4658.2011.08165​ between the three constructs and (b) aggregation ratio (%) differences .x. for each construct over time. Different letters mean statistically significant 2. Choi JH, Lee SY. Secretory and extracellular production of recombinant difference (Post-hoc Tukey HSD (THSD) comparisons). proteins using Escherichia coli. Appl Microbiol Biotechnol. 2004;64(5):625– 35. https​://doi.org/10.1007/s0025​3-004-1559-9. 3. Villaverde A, Corchero JL, Seras-Fanzoso J, Garcia-Fruitós E. Func‑ Abbreviations tional protein aggregates: just the tip of the iceberg. Nanomedicine. LZ: Leucine zipper; IB: Inclusion body; GFP: Green fluorescent protein; IPTG: 2015;10(18):2881–91. Isopropyl-β-d-thiogalactoside; FESEM: Field Emission Scanning Electron Micro‑ 4. García-Fruitós E, Vázquez E, Díez-Gil C, Corchero JL, Seras-Franzoso J, scope; H/D: Hydrogen/deuterium. Ratera I, et al. Bacterial inclusion bodies: making gold from waste. Trends Biotechnol. 2012;30(2):65–70. https​://linki​nghub​.elsev​ier.com/retri​eve/ Acknowledgements pii/S0167​77991​10016​85. The authors are indebted to ICTS ‘‘NANBIOSIS”, more specifically the Protein 5. Ventura S, Villaverde A. Protein quality in bacterial inclusion bodies. Production Platform of CIBER in Bioengineering, Biomaterials & Nanomedicine Trends Biotechnol. 2006;24(4):179–85. https​://linki​nghub​.elsev​ier.com/ (CIBER- BBN)/IBB, at the UAB sePBioEs scientific-technical service (http://www. retri​eve/pii/S0167​77990​60005​27. nanbi​osis.es/unit/u1-prote​in-produ​ction​-platf​orm-ppp/) and to ‘‘Servei de 6. García-Fruitós E, González-Montalbán N, Morell M, Vera A, Ferraz RM, Arís Microscopia” (UAB). The authors also acknowledge CINECA for the availability A, et al. Aggregation as bacterial inclusion bodies does not imply inacti‑ of high performance computing resources and support. vation of enzymes and fluorescent proteins. Microb Cell Fact. 2005;4:1–6. 7. Rinas U, Garcia-Fruitós E, Corchero JL, Vázquez E, Seras-Franzoso J, Vil‑ Authors’ contributions laverde A. Bacterial Inclusion Bodies: Discovering Their Better Half. Trends RRP carried out most of the experiments, performed the data analysis and Biochem Sci. 2017;42(9):726–37. https​://linki​nghub​.elsev​ier.com/retri​eve/ prepared a draft of the manuscript. SF performed the structural modeling pii/S0968​00041​73002​69. and computational study. AN and DA carried out the FTIR analysis and ASC 8. de Marco A, Ferrer-Miralles N, Garcia-Fruitós E, Mitraki A, Peternel S, Rinas was in charge of electron microscopy analysis. EGF and AA directed the study U, et al. Bacterial inclusion bodies are industrially exploitable amyloids. and prepared the final manuscript. All authors read and approved the final FEMS Microbiol Rev. 2019;43(1):53–72. https​://acade​mic.oup.com/femsr​ manuscript. e/artic​le/43/1/53/51442​14. 9. Pesarrodona M, Jauset T, Díaz-Riascos ZV, Sánchez-Chardi A, Beaulieu Funding M, Seras-Franzoso J, et al. Targeting antitumoral proteins to breast This work was supported by an IRTA Cutting Edge Research project. We are cancer by local administration of functional inclusion bodies. Adv Sci. indebted to CERCA Programme (Generalitat de Catalunya) and the European 2019;6(18):1900849. https​://doi.org/10.1002/advs.20190​0849. Social Fund for supporting our research. RRP received a PhD fellowship from 10. Roca-Pinilla R, López-Cano A, Saubi C, Garcia-Fruitós E, Arís A. A new gen‑ Secretaria d’Universitats I Recerca del Departament d’Economia i Coneixe‑ eration of recombinant polypeptides combines multiple protein domains ment de la Generalitat de Catalunya (AGAUR), EGF a post-doctoral fellowship for effective antimicrobial activity. Microb Cell Fact. 2020;19(1):122. https​ from INIA (DOC-INIA)), and DA a post-doc research fellow (Assegno di Ricerca) ://doi.org/10.1186/s1293​4-020-01380​-7. from the University of Milano-Bicocca. We acknowledge the CINECA Award 11. Gifre-Renom L, Seras-Franzoso J, Rafael D, Andrade F, Cano-Garrido O, N. HP10BTJPER, 2017, for the availability of high performance computing Martinez-Trucharte F, et al. The biological potential hidden in inclusion resources and support. bodies. Pharm. 2020;12(2):157. https​://doi.org/10.3390/pharm​aceut​ics12​ 02015​7. Availability of data and materials 12. Unzueta U, Cespedes MV, Sala R, Alamo P, Sánchez-Chardi A, Pesarrodona The datasets used and/or analysed during the current study are available from M, et al. Release of targeted protein nanoparticles from functional bacte‑ the corresponding author on reasonable request. rial amyloids: A death star-like approach. J Control Release. 2018;279:29– 39. https​://linki​nghub​.elsev​ier.com/retri​eve/pii/S0168​36591​83017​80. Roca‑Pinilla et al. Microb Cell Fact (2020) 19:175 Page 13 of 13 13. Unzueta U, Seras-Franzoso J, Céspedes MV, Saccardo P, Cortés F, Rueda 29. Barth A. Infrared spectroscopy of proteins. Biochim Biophys Acta - Bioen‑ F, Garcia-Fruitós E, Ferrer-Miralles N, Mangues R, Vázquez E, Villaverde erg. 2007;1767(9):1073–101. A. Engineering tumor cell targeting in nanoscale amyloidal materials. 30. Scheyhing CH, Meersman F, Ehrmann MA, Heremans K, Vogel RF. Temper‑ Nanotechnology. 2017;28(1):15102. https​://doi.org/10.1088/0957- ature-pressure stability of green fluorescent protein: a Fourier transform 4484/28/1/01510​2. infrared spectroscopy study. Biopolymers. 2002;65(4):244–53. 14. Seras-Franzoso J, Peternel S, Cano-Garrido O, Villaverde A, García-Fruitós 31. Tedeschi G, Mangiagalli M, Chmielewska S, Lotti M, Natalello A, Brocca S. E. Bacterial inclusion body purification. New York: Humana Press; 2015. p. Aggregation properties of a disordered protein are tunable by pH and 293–305. https​://doi.org/10.1007/978-1-4939-2205-5_16. depend on its net charge per residue. Biochim Biophys Acta - Gen Subj. 15. Wang X, Zhou B, Hu W, Zhao Q, Lin Z. Formation of active inclusion 2017;1861(11):2543–50. bodies induced by hydrophobic self-assembling peptide GFIL8. Microb 32. Singhvi P, Saneja A, Srichandan S, Panda AK. Bacterial inclusion bodies: a Cell Fact. 2015 \;14(1):88. http://www.micro​bialc​ellfa​ctori​es.com/conte​ treasure trove of bioactive proteins. Trends Biotechnol. 2020;38(5):474–86. nt/14/1/88. https​://linki​nghub​.elsev​ier.com/retri​eve/pii/S0167​77991​93030​63. 16. Wu W, Xing L, Zhou B, Lin Z. Active protein aggregates induced by termi‑ 33. Cano-Garrido O, Rodríguez-Carmona E, Díez-Gil C, Vázquez E, Elizondo nally attached self-assembling peptide ELK16 in Escherichia coli. Microb E, Cubarsi R, et al. Supramolecular organization of protein-releasing Cell Fact. 2011;10(1):9. https​://doi.org/10.1186/1475-2859-10-9. functional amyloids solved in bacterial inclusion bodies. Acta Biomater. 17. Slouka C, Kopp J, Spadiut O, Herwig C. Perspectives of inclusion bodies 2013;9(4):6134–42. https​://linki​nghub​.elsev​ier.com/retri​eve/pii/S1742​ for bio-based products: curse or blessing? Appl Microbiol Biotechnol. 70611​20058​31. 2019;103(3):1143–53. https​://doi.org/10.1007/s0025​3-018-9569-1. 34. Cano-Garrido O, Sánchez-Chardi A, Parés S, Giró I, Tatkiewicz WI, Ferrer- 18. Rinas U, Bailey J. Protein compositional analysis of inclusion bodies pro‑ Miralles N, et al. Functional protein-based nanomaterial produced in duced in recombinant Escherichia coli. Appl Microbiol Biotechnol. 1992. microorganisms recognized as safe: a new platform for biotechnol‑ https​://doi.org/10.1007/BF002​40735​. ogy. Acta Biomater. 2016;43:230–9. https​://doi.org/10.1016/j.actbi​ 19. Fahnert B, Lilie H, Neubauer P. Inclusion bodies: formation and utilisation. o.2016.07.038. Berlin: Springer; 2004. p. 93–142. https​://doi.org/10.1007/b9399​5. 35. Rueda F, Gasser B, Sánchez-Chardi A, Roldán M, Villegas S, Puxbaum 20. Hakoshima T. Leucine Zippers. Encyclopedia of Life Sciences. Chichester: V, et al. Functional inclusion bodies produced in the yeast Pichia Wiley; 2005. https​://doi.org/10.1038/npg.els.00050​49. pastoris. Microb Cell Fact. 2016;15(1):166. https​://doi.org/10.1186/s1293​ 21. Porte D, Oertel-Buchheit P, Granger-Schnarr M, Schnarr M. Fos Leu‑ 4-016-0565-9. cine Zipper variants with increased association capacity. J Biol Chem. 36. Unzueta U, Céspedes MV, Ferrer-Miralles N, Casanova I, Cedano J, 1995;270(39):22721–30. https​://doi.org/10.1074/jbc.270.39.22721​. Corchero JL, et al. Intracellular CXCR4 + cell targeting with T22-empow‑ 22. Vinson C, Myakishev M, Acharya A, Mir AA, Moll JR, Bonovich M. Classifica‑ ered protein-only nanoparticles. Int J Nanomed. 2012;7:4533–44. tion of human B-ZIP proteins based on dimerization properties. Mol Cell 37. Lindorff-Larsen K, Piana S, Palmo K, Maragakis P, Klepeis JL, Dror RO, Biol. 2002;22(18):6321–35. https​://mcb.asm.org/conte​nt/22/18/6321. et al. Improved side-chain torsion potentials for the Amber ff99SB 23. Hu C-D, Kerppola TK. Simultaneous visualization of multiple protein protein force field. Proteins Struct Funct Bioinforma. 2010. https​://doi. interactions in living cells using multicolor fluorescence complementa‑ org/10.1002/prot.22711​. tion analysis. Nat Biotechnol. 2003;21(5):539–45. http://www.natur​e.com/ 38. Bussi G, Donadio D, Parrinello M. Canonical sampling through artic​les/nbt81​6. velocity rescaling. J Chem Phys. 2007;126(1):014101. https​://doi. 24. Choi S, Lee SJ, Yeom S, Kim HJ, Rhee YH, Jung H, et al. Controlled localiza‑ org/10.1063/1.24084​20. tion of functionally active proteins to inclusion bodies using Leucine 39. Hess B, Bekker H, Berendsen HJC, Fraaije JGEM. LINCS: a linear constraint Zippers. PLoS ONE. 2014;9(6):e97093. https​://doi.org/10.1371/journ​ solver for molecular simulations. J Comput Chem. 1997;18(12):1463– al.pone.00970​93. 72. https​://doi.org/10.1002/%28SIC​I%29109​6-987X%28199​ 25. Han GH, Seong W, Fu Y, Yoon PK, Kim SK, Yeom SJ, et al. Leucine zipper- 709%2918%3A12%3C146​3%3A%3AAID​-JCC4%3E3.0.CO%3B2-H. mediated targeting of multi-enzyme cascade reactions to inclusion 40. Pronk S, Páll S, Schulz R, Larsson P, Bjelkmar P, Apostolov R, et al. bodies in Escherichia coli for enhanced production of 1-butanol. Metab GROMACS 45: a high-throughput and highly parallel open source Eng. 2017;40:41–9. https​://doi.org/10.1016/j.ymben​.2016.12.012. molecular simulation toolkit. Bioinformatics. 2013;29(7):845–54. https​:// 26. Roy A, Kucukural A, Zhang Y. I-TASSER: a unified platform for automated doi.org/10.1093/bioin​forma​tics/btt05​5. protein structure and function prediction. Nat Protoc. 2010;5(4):725–38. 41. Susi H, Byler DM. Resolution-enhanced Fourier transform infrared spec‑ http://www.natur​e.com/artic​les/nprot​.2010.5. troscopy of enzymes. Methods in enzymology. Cambridge: Academic 27. Ami D, Natalello A, Taylor G, Tonon G, Maria Doglia S. Structural analysis of Press; 1986. p. 290–311. protein inclusion bodies by Fourier transform infrared microspectroscopy. Biochim Biophys Acta. 2006;1764(4):793–9. https​://linki​nghub​.elsev​ier. com/retri​eve/pii/S1570​96390​50044​01. Publisher’s Note 28. Natalello A, Doglia SM. Insoluble protein assemblies characterized by Springer Nature remains neutral with regard to jurisdictional claims in pub‑ Fourier transform infrared spectroscopy. Methods in molecular biology. lished maps and institutional affiliations. Clifton: Humana Press Inc; 2015. p. 347–69. 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