Engineering Nanoreactors Using Bacterial Microcompartment Architectures

Abstract

Bacterial microcompartments (BMCs) are organelles that encapsulate enzymes involved in CO₂ fixation or carbon catabolism in a selectively permeable protein shell. Here, we highlight recent advances in the bioengineering of these protein-based nanoreactors in heterologous systems, including transfer and expression of BMC gene clusters, the production of template empty shells, and the encapsulation of non-native enzymes.

Introduction

A key goal of synthetic biology is to engineer metabolic pathways to produce bulk chemicals for medical, agricultural, and industrial purposes using microbial cell factories. Factors that reduce the efficiency of engineered pathways include crosstalk of metabolites, toxic intermediates, and inhibitory products. Eukaryotes have evolved compartmentalizing organelles to overcome these obstacles. Bacteria also have organelles, known as bacterial microcompartments (BMCs) [1–3]. BMCs contain enzymes that catalyze sequential reactions and a private pool of cofactors (e.g., NAD⁺/NADH, coenzyme A, and ATP) within a protein shell. The BMC shell serves as a selectively permeable interface between the encapsulated pathway and the cellular environment. Because they self-assemble entirely from proteins, BMCs are becoming a viable platform for engineering novel nanoreactors.

Functionally diverse BMCs are bioinformatically predicted to be present in at least 23 different bacterial phyla [2]. Cyanobacteria and some chemoautotrophs produce BMCs (α- or β-carboxysomes) that encapsulate carbonic anhydrase (CA) and ribulose-1,5-bisphosphate carboxylase/oxygenase (RuBisCO) to enhance CO₂ fixation. However, the majority of functionally diverse BMCs are catabolic (metabolosomes), utilized by heterotrophs to degrade a range of carbon compounds in niche environments. Most metabolosomes contain a signature enzyme, such as a propanediol dehydratase (PDH) [4], an ethanolamine-ammonia lyase (EAL) [5], or a glycyl-radical enzyme [6, 7], that defines the function of the BMC [e.g., propanediol utilization (PDU) or ethanolamine utilization (EUT) BMCs, or glycyl-radical enzyme microcompartment (GRM)]. Metabolosome cores also include four conserved [1, 2] enzymes: an aldehyde dehydrogenase (AldDH) [8], an alcohol dehydrogenase (AlcDH) [9], and a phosphotransacylase (PTAC) [10, 11]. In addition, many BMC loci encode ancillary proteins that support organelle function, such as the transport of substrates and recycling co-factor (e.g., ATP and vitamin B₁₂) [1, 2].

The core enzymes of carboxysomes and metabolosomes are encapsulated by a shell comprised of proteins that form hexamers (BMC-H) [12], pseudohexamers/trimers (BMC-T) [13, 14], and pentamers (BMC-P) [15, 16] (Figure 1a). The hexameric shell proteins typically contain a pore at the symmetry axis, with a diameter of 4 – 10 Å [17, 18••] and electrostatic properties [1, 6, 12, 14, 19] suited to the passage of small charged metabolites across the shell [1]. Confinement of sequential enzymatic reactions by the BMC shell facilitates substrate channeling, thereby enhancing catalytic efficiency. [1, 20–24]. The shell also acts as a barrier, preventing potentially toxic/volatile intermediates from diffusing into the cytoplasm [21–23•]. Most BMCs are predicted to form from the inside out, the core proteins coalesce into a bolus around which a shell assembles [1, 25] (Figure 1b). A short helical extension on a subset of core proteins, the encapsulation peptide (EP), facilitates the aggregation of the core enzymes [10, 26, 27] and their subsequent encapsulation by the shell [1, 25, 28–32]. The structure and native functions of BMCs have been reviewed elsewhere [1, 3, 20, 24, 33, 34]. The aim of this review is to highlight the recent efforts that adapt the BMC architectures for the development of novel nanoreactors in heterologous systems.

Figure 1.

Shell proteins (a) and assembly of BMCs (b–d). (a) Representatives of a BMC-H protein (PDB 5DJB) (left), a BMC-T protein (PDB 5DIH) (middle), and a BMC-P protein (PDB 2QW7) (right). Pf indicates Pfam identification. Individual polypeptide chains are colored differently. (b) Cartoon representation of BMC assembly from the inside out, where the primary role of the EP is in shell recruitment [25, 35••] (c) of empty BMC shell assembly, and (d) of targeting enzymes to the lumen of BMC shells using EPs [36••].

Overview of BMC engineering

Efforts to engineer BMCs have involved both the transfer and expression of BMC operons in heterologous systems, the production of empty shells (Figure 1c), and the encapsulation of heterologous cargo using encapsulation peptides (Figure 1d). Subsequent efforts have focused on building a core based on protein domain interactions and tuning shell permeability to support encapsulated metabolism.

Heterologous expression of BMC gene clusters in E. coli

BMCs are encoded by gene clusters, providing a ready genetic module for heterologous expression. The 21-gene PDU operon of Citrobacter freundii was the first demonstration of the potential for “transplanting” a metabolic module in E. coli [37]. Electron microscopy of thin sections of strains expressing the operon revealed polyhedral bodies (Figure 2a) and recombinant metabolosome demonstrated diol dehydratase activity. Follow-up work showed that the recombinant PDU BMCs have similar morphology and mechanical properties as wildtype (WT) PDU BMCs [38]. Similarly, the α-carboxysome operon of a chemoautotroph was expressed in E. coli (Figure 2b), generating carboxysome-like particles and active RuBisCO [39].

Figure 2.

Physical characterization of engineered BMCs. (a) Thin section electron micrograph of E. coli expressing the complete PDU BMC operon (arrows) (scale bar 300 nm). Asterisks mark unknown granular dense matter. Inset is an enlarged view of the polyhedral bodies (scale bar 96 nm) that show regular substructures (arrows) [37]. (b) Thin sections electron micrograph of E. coli expressing the carboxysome genes of Halothiobacillus neapolitanus viewed by TEM (scale bar 500 nm) showing polyhedral bodies (magnified in inset) [39]. (c) Negatively stained electron micrograph of purified HO BMC shells (scale bar 50 nm). Reproduced with permission from [40]. Copyright Elsevier 2014. (d) Surface representation of the HO shell crystal structure. Shell proteins are colored blue (BMC-H), green (BMC-T), and yellow (BMC-P). Reproduced with permission from [41••]. Copyright The Association for the Advancement of Science 2017.

Heterologous expression of BMC shells

A variety of BMC shells from both carboxysomes [42] and metabolosomes have been shown to assemble in E. coli [37, 40, 43–45]. Compared to fully packaged native counterparts, recombinant empty shells tend to be smaller; only recombinant EUT shells, formed from all of the or a single EUT shell protein, were observed to be similar to WT EUT BMCs in size [45]. The number of these recombinant EUT shells were shown to increase when co-expressed with a putative cupin domain [46]. When the shell protein genes of a metabolosome of unknown function (from Haliangium ochraceum ) were expressed in E. coli, homogeneous, robust shells were formed and readily purified (Figure 2c) [40], enabling crystallization [41••]. The atomic resolutions structure of the 6.5 MDa empty shell is estimated to be able to accommodate threehundred 30 kDa proteins (Figure 2d). Interestingly, the structure revealed that the interactions between the shell proteins are largely governed by shape complementarity rather than salt bridges and hydrogen bonds between conserved residues. The structure provides scalable shell assembly principles that likely apply to all BMCs, and provides a blueprint for shell engineering.

Encapsulation of non-native proteins using EPs

EPs on BMC core enzymes have been shown to interact with shell proteins [28, 29, 47•, 48•], facilitating their encapsulation. EPs are typically short amphipathic helical extensions connected by a poorly conserved linker to a subset of core proteins [1, 30, 31]. They have been identified on the essential β-carboxysome protein CcmN [28], and on the signature (e.g., PDH [31, 32], EAL [30, 31, 45], and aldolase [2, 31, 49]) and conserved core (e.g., AldDH [29, 30, 40, 50], AlcDH [30, 31], and PTAC [10, 11, 31]) enzymes of metabolosomes. Recently, designed EPs were developed for the PDU shell using both rational and library-based strategies [47•].

EPs have been employed to target non-native proteins to the interior of BMC shells. For example, fluorescent proteins (FPs) can be encapsulated inside native and recombinant carboxysome and metabolosome shells by full length core enzymes or by EPs alone [40, 42–45, 48•, 51, 52]. Green fluorescent protein fused to the EP of AldDH was used to develop an assay for the rapid quantification of encapsulated protein (Figure 3a) [51]. Likewise, various enzymes have been targeted to the interior of recombinant PDU shells, producing prototypes for the engineering of nanoreactors based on BMC architectures. For example, the EPs from the native enzymes of the PDU BMC were used to encapsulate exogenous pyruvate decarboxylase and AlcDH (Figure 3b) into synthetic shells made from PDU shell proteins to generate ethanolproducing BMCs in E. coli [50]. Recently, the EP of AldDH was used to separately encapsulate three different enzymes, one of which remained active in varying pH conditions in vitro [53•]. The EP of AldDH was also used to generate a polyphosphate-synthesizing BMC in E. coli (Figure 3c) [54••]. Because ATP is the substrate of the encapsulated enzyme, this study also shows that the shell of the engineered BMC is permeable to ATP (as is native PDU BMCs [1]), but the specific shell protein that allows ATP transit is unknown. Beyond the encapsulation of enzymes for biosynthesis, the EP of the AldDH was used to sequester a cytotoxic enzyme inside recombinant PDU shells, to restrict its toxic effects [55••].

Figure 3.

Heterologous expression of BMC shell proteins and cargo. (a) Fluorescent microscopy image of S. enterica expressing PDU BMC shell proteins and EP_AldH-GFP. Reproduced with permission from [51]. Copyright John Wiley and Sons 2017. (b) Negatively stained electron micrograph of purified PDU BMC ethanol bioreactors from E. coli [50] (scale bar 100 nm). (c) Light microscopy of Neisser stained fixed E. coli cells expressing a polyphosphate kinase and PDU shell proteins, polyphosphate appears purple-black. Reproduced with permission from [54••]. Copyright John Wiley and Sons 2017.

Developing new methods for core assembly

In addition to targeting proteins for encapsulation, EPs cause their cognate enzyme to oligomerize [8, 10, 26, 27]. Taking advantage of this property, four enzymes tagged with the EP of PDH or AldDH resulted in a shell-free enzyme aggregate that was able to convert glycerol to 1,2-propanediol [36••]. Disadvantages of the EP-based approach include the loss of precise control of internal organization and stoichiometry. Control over core assembly may be obtained by visualizing the core components as an array of interacting protein domains. For example, utilizing the knowledge of core protein domain structures and their interactions in β-carboxysome formation [25], a chimeric protein (CcmC) was designed that could replace four gene products required for carboxysome assembly [35••]. CcmC is a synthetic protein consisting of domains to aggregate RuBisCO, a CA, and an EP for adherance to the carboxysome shell.

Domain-based engineering can be expanded to build chimeric cores that may improve the organization of the encapsulated enzymes. In addition, it has recently been observed that some core protein domains interact with the shell [42]. For instance, a chimeric core could be formed by the fusion of enzymatic domains from a specific metabolic reaction sequence; association with the shell could be mediated by one of these domains, or via an EP extension (Figure 4a, upper). The EP would facilitate the oligomerization of the fused domains and their subsequent encapsulation within the BMC shells. Alternatively, the stoichiometry of the enzymes may be controlled by utilizing a linear protein scaffold [56] fused to an EP (Figure 4a, lower). Enzymes are recruited to the core via specific protein-protein interactions with the scaffold domains.

Figure 4.

Cartoon representation of domain-based cores (a) and of shells with selective permeability (b). (a) Top panel: a chimeric protein tagged with an EP comprised of protein domains for a metabolic reaction. Lower panel: a linear scaffold tagged with an EP that is able to form protein-protein interaction with enzymes of a metabolic pathway. (b). Top panel: introduction of a foreign or a modified shell protein to tune permeability for specific metabolites. Lower panel: engineering of catalytic sites in the pore of shell proteins that will convert metabolites to the form required by the encapsulated pathway.

Modifying the BMC shell to tune permeability

Fundamental to the engineering of pathways within BMCs is the ability to tailor the permeability of the shell as the interface with metabolism to conduct the requisite substrates and products. The pores in BMC shells selectively control the passage of metabolites; accordingly, shell permeability can be tuned by modifying residues that surround the pores. Mutation of residues surrounding the pores have been shown to cause permeability defects [57], or enhance BMC function, due to alteration of small molecule diffusion rates [58•]. Shell permeability has also been modified by creating shell chimeras (Figure 4b). Chimeric β-carboxysome shells were generated by incorporating an α-carboxysome BMC-H protein, which structurally rescued a carboxysome-minus strain [59]. PDU BMCs incorporating shell proteins from the EUT BMC or β-carboxysome also have been produced, albeit with impaired function [60]. Recently, chimeric PDU shells were formed by substituting a EUT shell protein that has a smaller and more charged pore than its PDU counterpart [58•]. Strains expressing the chimeric BMCs showed improved growth over WT, demonstrating enhanced permeability to the substrate.

The prevailing model of the BMC shell as merely a passive semipermeable barrier is being reconsidered through engineering. PduT, a BMC-T protein from the PDU BMC, is the sole natural example of shell protein that binds a redox-active [4Fe-4S] cluster [61, 62]. The pore of a BMC-T protein, naturally devoid of a cofactor, was engineered to conduct electrons by incorporating a stable and redox-active [4Fe-4S] cluster, providing an example of generating a shell protein with a non-native function [18••]. This work provides a major step forward in the development of a synthetic shell permeable to electrons. This approach could be extended to the engineering of catalytic metal centers in the pores of shells, generating metabolites consumed by the encapsulated enzymes (Figure 4b).

Summary and Outlook

BMCs are proteinaceous organelles that encapsulate enzymes involved in anabolic or catabolic reactions. BMCs are genetic, structural, and functional modules. Future efforts in bioengineering of BMCs could lead to nanoreactors applicable for use in agriculture and energy sectors. For instance, the first steps toward installing carboxysomes into chloroplasts has been taken, including the heterologous expression of cyanobacterial RuBisCO and shell proteins [63, 64]. Likewise, the development of microbial cell factories that harbor engineered BMCs encapsulating a pathway that produces renewable alternatives to petroleum-based commodities is now in reach. The capacity to engineer BMCs as nanoreactors will be of significant value for synthetic biology and beyond.

• BMCs are natural metabolic modules found in diverse bacterial phyla.

• BMCs are proteinaceous organelles that encapsulate enzymes for CO₂ fixation or carbon catabolism.

• Bioengineering of BMCs has led to the production of empty synthetic shells.

• Encapsulation of enzymes led to BMC-based nanoreactors.

• BMC-based nanoreactors will be of significant value for synthetic biology.