Towards revealing the structure of bacterial inclusion bodies

Abstract

Protein aggregation is a widely observed phenomenon in human diseases, biopharmaceutical production, and biological research. Protein aggregates are generally classified as highly ordered, such as amyloid fibrils, or amorphous, such as bacterial inclusion bodies. Amyloid fibrils are elongated filaments with diameters of 6-12 nm, they are comprised of residue-specific cross-β structure, and display characteristic properties, such as binding with amyloid-specific dyes. Amyloid fibrils are associated with dozens of human pathological conditions, including Alzheimer disease and prion diseases. Distinguished from amyloid fibrils, bacterial inclusion bodies display apparent amorphous morphology. Inclusion bodies are formed during high-level recombinant protein production, and formation of inclusion bodies is a major concern in biotechnology. Despite of the distinctive morphological difference, bacterial inclusion bodies have been found to have some amyloid-like properties, suggesting that they might contain structures similar to amyloid-like fibrils. Recent structural data further support this hypothesis, and this review summarizes the latest progress towards revealing the structural details of bacterial inclusion bodies.

Introduction

Protein aggregation is a widely observed phenomenon in human diseases, biopharmaceutical production and biological research. Based on their morphology, protein aggregates are generally classified as either highly ordered, as in the case of amyloid fibrils, or amorphous, as in the case of bacterial inclusion bodies.

The term “amyloid” was first introduced by Rudolf Virchow to describe the starch-like pale waxy tissue abnormality,1 and amyloid fibrils are associated with dozens of human pathological conditions, including Alzheimer disease, Parkinson disease, diabetes type II and prion diseases.2–5 It has also been reported that a variety of bacteria can make functional amyloids.6–10 Under the electron microscope (EM), amyloid fibrils are elongated filaments with diameters of 6–12 nm.11–13 X-ray diffraction of aligned amyloid fibrils shows a characteristic pattern with a meridional reflection at 4.7 Å and an equatorial reflection at ∼10 Å, indicating a cross-β structure.14,15 High-resolution structural studies have shown that these filaments are comprised of sequence-specific cross-β structure, with intermolecular and in-register β-sheets parallel to the filament axis.15–24 Amyloid fibrils can bind with amyloid-specific dyes, such as Congo red25,26 and thioflavin T,27 and can be infectious and toxic as represented by the HET-s prion system.18,28

Distinguished from amyloid fibrils, bacterial inclusion bodies are classified as amorphous aggregates. They are protein aggregates generated during recombinant protein production in bacteria, and are a major concern in biotechnology.29,30 Formation of inclusion bodies may be caused by the high local concentration of nascent polypeptides emerging from ribosomes during overexpression, and insufficient chaperones presenting around to protect these nascent polypeptides from aggregation.29,31–33 Bacterial inclusion bodies are not just unstructured aggregates that are clusters of misfolded proteins sticking to each other through non-specific hydrophobic interactions.34,35 Rather, studies have shown that inclusion bodies have amyloid-like properties,30,32,36–43 i.e., binding with Congo red and showing birefringence (Fig. 1), seeding the aggregation of their soluble counterpart39 and inducing cytotoxicity in eukaryotic cells.44 These properties are indicative that inclusion bodies might contain structure reminiscent of amyloid fibrils. Indeed, recent data that further support this hypothesis have been presented, and it is the aim of this review to summarize the current progress towards revealing the structure of bacterial inclusion bodies.

Morphology of Inclusion Bodies

When inclusion bodies are formed, they are normally observed under EM as large, dark aggregates inside the host cells45–47 (Fig. 2A). After purified from cell lysate, inclusion bodies are amorphous (Fig. 2B), approximating sphere-like or rod-like shapes with diameters ranging from 0.2 µm to 1.2 µm.39,48–52 The size of inclusion bodies is probably related to the dimensions of the host cells in which they were produced, the protein sequences, and the physical conditions during protein production. Inclusion bodies of some proteins also release amyloid-like protofibrils or fibrils under certain conditions, examples are: (1) BMP2(13–74) incubated at 37°C for 12 hours52 (Fig. 2C); (2) BMP2(13–74) produced in host cells for 12 hours52 (Fig. 2D); (3) BMP2(13–74) partially disaggregated by 4 M urea solution (Fig. 2E); (4) ESAT-6 incubated at room temperature for 14 days52 (Fig. 2F); (5) Aβ42-GFP and Aβ42 after proteinase K proteolytic action53 (Fig. 2G); (6) HET-s(218–289) after three hours of expression46 (Fig. 2H). Since inclusion bodies of BMP2(13–74) and ESAT-6 do not display fibrils right after three hours of expression, it is possible that these inclusion bodies mainly contain immature and flexible protofibrils that mature into fibrils given time and proper temperature. On the other hand, the inclusion bodies of highly aggregation-prone, prion-forming domain, HET-s(218–289), may contain mature fibrils.

Inclusion Bodies are Often Enriched in β-Sheet Secondary Structure

The apparent amorphous inclusion bodies of different proteins were examined by Fourier transform infrared red (FTIR) spectroscopy, which is a method to analyze the secondary structure content of proteins in soluble as well as in aggregated form. For the FTIR spectrum of soluble VP1LAC, a β-galactosidase derivative with N-terminally fused VP1 capsid protein, peaks at ∼1,634 cm⁻¹, ∼1,644 cm⁻¹ and ∼1,654 cm⁻¹ are usually assigned to the β-sheet, random coil and α-helix conformations of the protein, respectively39 (Fig. 3A). In the FTIR spectrum of inclusion bodies of VP1LAC, additional sharp peaks at ∼1,621 cm⁻¹ and ∼1,691 cm⁻¹ emerge compared to the spectrum of soluble protein (Fig. 3A), which are indicative of newly formed β-sheet structures in inclusion bodies.39,53–57 For some proteins, the FTIR spectra of their inclusion bodies also show peaks at ∼1,634 cm⁻¹ and ∼1,651 cm⁻¹ (Fig. 3A), which suggests that these inclusion bodies also contain residual native-like β-sheet and a-helix structures of their soluble form.39,54–62

Inclusion Bodies Contain Cross-β Structure

Although no 3D structure of inclusion bodies is available, the tertiary structural content of inclusion bodies is, at least partially, determined by X-ray diffraction.52,63 The X-ray diffraction spectra of inclusion bodies shows a two-ring diffraction pattern (Fig. 3B), typical for the cross-β structure in amyloid fibrils, with a major reflection at 4.7 Å resolution interpreted as the spacing between strands in a β-sheet and a diffused reflection at ∼10 Å interpreted as the spacing between β-sheets. The circular profiles of the two reflections, rather than the typical orthogonal positions for the cross-β structure in amyloid fibrils, show that the cross-β structural entities in inclusion bodies are not strongly aligned as in amyloid fibrils.

The Cross-β Structure of Inclusion Bodies is Residue-Specific and Amyloid-Like

(a) Structural studies of inclusion bodies of ESAT-6, BMP2 (13–74) and MOG(ECD).52

To elucidate the residue-specific structural information, quenched hydrogen/deuterium exchange (H/D-exchange) experiments with solution nuclear magnetic resonance (NMR) were measured for three inclusion body-forming proteins that have distinctive native soluble folds that cover the folding spectrum: (1) The α-helical early secreted antigen 6-kDa protein (ESAT-6) (ESAT-6 folds only in complex with its protein partner CFP-10);64 (2) The mixed a-helical and β-sheet protein, residues 13–74 of the secretory human bone morphogenetic protein-2 [BMP2(13–74)];65,66 (3) The β-sheet extracellular domain of the human membrane protein myelin oligodendrocyte glycoprotein [MOG (ECD)] [MOG(ECD) contains one disulfide bridge].67,68 Inclusion bodies of all three proteins bind Congo red and thioflavin T, suggesting that they contain amyloid-like structures.

In the case of ESAT-6 inclusion bodies, residues 7–23 form hydrogen bonds as identified by NMR H/D-exchange experiment (Fig. 4). Figure 4A (left) is the [¹⁵N,¹H]-correlation NMR spectrum of dissolved monomeric ESAT-6 inclusion bodies, which contains cross-peaks corresponding to its backbone amides. Upon exchange of the inclusion bodies in D₂O buffer for 311 hours, only cross-peaks of residues 8–25 and 36–43 are still present (Fig. 4A, right), which is indicative of slow exchange. The H/D-exchange was followed over time, and it was found that all residues in the inclusion bodies display a heterogeneous biphasic behavior, with a very fast and a slow exchanging component (Fig. 4C). The detailed analysis of the H/D-exchange data shows that the major population (Fig. 4D and p > 1/2) of residues 7–23 in ESAT-6 inclusion bodies display slow exchange rates of 10^∓3 to 10⁻⁴ h⁻¹ and are therefore considered to be involved in hydrogen bonds (Fig. 4B and D). In contrast, the majority of residues 2–6 and 24–95 (Fig. 4D and p > 1/2) display fast exchange rates larger than 10¹ h⁻¹ and are therefore considered to be unprotected in H/D-exchange and conformationally disordered. Because soluble ESAT-6 is a α-helical protein, but the circular dichroism spectrum of ESAT-6 inclusion bodies is indicative of β-sheet conformation, and the x-ray diffraction shows a two-ring pattern that is typical for a cross-β structure,52 it is likely that the hydrogen bond-forming residues 7–23 of ESAT-6 contain a mainly amyloid-like, cross-β structure in its inclusion bodies.

To verify that residues 7–23 comprise the dominant component in the formation of cross-β structure in ESAT-6 inclusion bodies, aggregation-prone residues in ESAT-6 were mutated to the aggregation-interfering residue Arg.69 It was found that only the mutations F8R, I11R, I18R or V22R within the residue 7–23 segment abolished the formation of inclusion bodies, but not the mutations L36R, V54R or I76R. To confirm that residues 7–23 can form an amyloid-like cross-β structure, a peptide E20 corresponding to residues 6–25 of ESAT-6 was synthesized, and it can form amyloid fibrils under physiological conditions. In summary, residues 7–23 of ESAT-6 in bacterial inclusion bodies form a cross-β structure characteristic of amyloid-like fibrils with the remainder of sequence disordered.

In the case of BMP2(13–74) and MOG(ECD), similar results were also found: residues 62–67 of BMP2(13–74) and residues 85–95, 101–108, 111–118 of MOG(ECD) form a cross-β structure characteristic of amyloid-like fibrils with the remainder of the amino acid sequence disordered.

(b) Structural study of inclusion bodies of HET-s(218–289).46

[Het-s] is a prion protein involved in the self-recognition of the filamentous fungus P. anserina.70 The C-terminal region containing residues 218–289 of [Het-s] [HET-s(218–289)] is the prion-forming domain.18,71 HET-s(218–289) can form amyloid fibrils, which contain a β-solenoid with two layers of β-strands per monomer and is characterized by the formation of a triangular hydrophobic core.22 During protein production, HET-s(218–289) forms inclusion bodies that display [Het-s] prion infectivity.46 The ¹³C-¹³C proton-driven spin-diffusion (PDSD) spectra with solid-state NMR was measured for the inclusion bodies of HET-s(218–289), and the spectra reproduces all the cross-peaks visible for the amyloid fibrils of HET-s(218–289) (Fig. 5). Since the NMR chemical shifts are strongly dependent on the conformation of a polypeptide, the same chemical shifts of inclusion bodies and amyloid fibrils of HET-s(218–289) suggest that their molecular structures have to be virtually the same. This conclusion is also supported by NMR H/D-exchange data, which shows that the exchange pattern of the purified inclusion bodies closely resembles the exchange pattern of the HET-s(218–289) fibrils.

Heterogeneity of Inclusion Bodies

In addition to amyloid-like structure, native-like structure could be retained in inclusion bodies of some proteins,39,54–62,72 and inclusion bodies may contain phospholipids from the E. coli membrane as well as other proteins and possibly RNA.46 Also the H/D-exchange of solution NMR shows that a small population (usually less than 1/3) of protein inside inclusion bodies have different exchange rates than the major population that forms amyloid-like structure,52 indicating structural heterogeneity in inclusion bodies. So it is possible that, besides contaminants, inclusion bodies are comprised of mixtures of amyloid-like protofibrils/fibrils with unfolded, partially folded or even natively folded proteins.73 The ratio of amyloid-like structure versus other heterogeneous structure could be affected by several factors, such as the stability of the protein in its native fold, or the physical parameters used during protein production.54

The structural study of inclusion bodies of FHA2 provides a residue-specific analysis to show that inclusion bodies retain at least part of their native-like structure.72 FHA2 is the N-terminal 185-residue functional domain of the 221-residue HA2 subunit of the influenza virus hemagglutinin protein.74 Its sequence contains several “sequential pairs.” By selectively [¹³C,¹⁵N]-labeling these “sequential pairs” and measuring the rotational-echo double-resonance (REDOR) with solid-state NMR75 for the inclusion bodies of FHA2, REDOR can detect the signal of ¹³C carbonyl (¹³CO) nuclei which are directly bonded to ¹⁵N nuclei in the protein sequence. By comparing the backbone ¹³CO chemical shifts of these residues to the chemical shift of α-helix and β-sheet, the local secondary structure of FHA2 inclusion bodies can be determined. It was found that the backbone ¹³CO chemical shifts of residues Gly-1, Gly-4, Ala-7 and Leu-98 of FHA2 in inclusion bodies indicate an α-helix conformation. Considering that in the native soluble fold of FHA2, residues Gly-1, Gly-4 and Ala-7 lie in a N-terminal α-helix, and Leu-98 lies in an α-helix spanning residues 38–105, it suggests that some native-like structure is retained in inclusion bodies of FHA2.

Summary

Current structural studies have revealed that beneath the amorphous appearance, bacterial inclusion bodies are actually structured aggregates that contain residue-specific cross-β structure reminiscent of amyloid-like protofibrils or fibrils. Inclusion bodies may also contain a portion of heterogeneously structured proteins that may be native-like, partially folded or unstructured, and could retain native-like biological activities.76–79 High-resolution 3D structures of inclusion bodies need to be solved to understand their architecture in depth. Structural comparison of inclusion bodies and amyloid fibrils of HET-s(218–289) suggests that they share the same structure, and it would be interesting to make this comparison on proteins that are less aggregation-prone. Since polymorphism plays an important role in the formation of amyloid fibrils,80,81 it might also induce different cross-β structure in inclusion bodies and amyloid fibrils of the same protein.

By assuming that the observed amyloid-like nature of inclusion bodies holds for most of the other documented bacterial inclusion bodies,82 amyloid-like aggregation is probably a common intrinsic property of protein segments and consequently is observed in both eukaryotes and prokaryotes.3 These structural studies of bacterial inclusion bodies thus extend the possible structural landscape of proteins: in addition to an unfolded or folded state, each protein may also contain one or more segments that are capable of forming a sequence-specific, cross-β-sheet aggregated state. “The process of protein aggregation can thus be viewed as a primitive folding mechanism, resulting in a defined aggregated conformation with each aggregated protein having its own distinctive properties.”52

Abbreviations

EM	=	electron microscope
NMR	=	nuclear magnetic resonance
H/D-exchange	=	hydrogen/deuterium exchange
FTIR	=	fourier transform infrared red spectroscopy
ESAT-6	=	early secreted antigen 6-kDa protein
BMP2(13–74)	=	residues 13–74 of the secretory human bone morphogenetic protein-2

Figures and Tables

Figure 1 Congo red binding with amyloid fibrils and bacterial inclusion bodies. Congo red staining under bright field (left) and showing birefringence under cross-polarized light (right) when binding with amyloid fibrils of HET-s (upper) and inclusion bodies of mouse prion protein PrP(23–231) (lower). (Partial reproduction of Figs. 2,26 and S8,52).

Figure 2 Morphology of bacterial inclusion bodies observed under electron microscope. (A) An E. coli cell containing inclusion body of ESAT-6, (B) Inclusion bodies of BMP2(13–74) after purification, (C) Inclusion body of BMP2(13–74) after incubation at 37°C for 12 hours, (D) Inclusion body of BMP2(13–74) after grown in host cells for 12 hours, (E) Inclusion body of BMP2(13–74) after disaggregated in 4 M of urea solution, (F) Inclusion body of ESAT-6 after incubation at room temperature for 14 days, (G) Inclusion body of Aβ42 after proteinase K proteolytic action, (H) Inclusion body of HET-s(218–289) after three hours expression. Scale bars indicate 1 µm. (Partial reproduction of Figs. 5, S7,52 5,53 and S4,46).

Figure 3 (A) The FTIR spectrum of inclusion bodies of VP1LAC (continuous line) and soluble VP1LAC (broken line). VP1LAC stays in the soluble cell fraction as well as in inclusion bodies when produced in E. coli. (B) The X-ray diffraction of inclusion bodies of ESAT-6. (Partial reproduction of Figs. 8,39 and 2,52).

Figure 4 Residues 7–23 of ESAT-6 form cross-β structure in inclusion bodies (A) [¹⁵N,¹H]-HMQC-spectra of dissolved monomeric ESAT-6 inclusion bodies before (left) and after (right) exchanged in D₂O buffer for 311 hours. (B) Ribbon representation of the 3D structure of soluble ESAT-6, while in complex with CFP-10 (not shown).64 The green-colored segment corresponds to residues 7–23. (C) NMR H/D-exchange curves for residues A9, T23, K57 and S77 of ESAT-6. The peak intensities for each residue were plotted versus the H/D exchange time. (D) H/D-exchange rates k_ex/h, and the relative population P of the two exchange components against the amino acid sequence of ESAT-6. The exchange rates of the major population (p > 1/2) are colored green. If the minor population (p < 1/2) is present more than 1/3, the corresponding exchange rates are shown in grey. The secondary structures of the soluble conformation shown in (B) are highlighted in red. (Partial reproduction of Figs. 2 and S2,52).

Figure 5 ¹³C-¹³C solid-state NMR correlation spectrum of purified HET-s(218–289) inclusion bodies (blue) compared to a spectrum of in vitro fibrillized HET-s(218–289) (red). Additional cross-peaks, not belonging to HET-s(218–289), at ∼34 ppm, ∼60–100 ppm and ∼176–180 ppm are assigned to phospholipids, other proteins and RNA from E. coli. (Reproduction of Fig. 3,46).

Acknowledgements

I want to thank Prof. Roland Riek, Dr. Christos Tzitzilonis, Dr. Jason Greenwald and Carolin Buhtz for providing valuable comments and suggestions for this review.