Encapsulins in Terpene Biosynthesis: Enzyme Nanoreactors in Bacterial Secondary Metabolism

Abstract

Encapsulins are self-assembling protein nanocompartments widely distributed across prokaryotes that sequester diverse enzymes. While most encapsulin systems studied thus far are involved in nutrient storage or oxidative stress response, recent bioinformatic and experimental studies have also demonstrated their involvement in secondary metabolism, particularly terpenoid biosynthesis. In this perspective, we first present a comprehensive analysis of Family 2B encapsulin gene clusters likely involved in terpene or terpenoid biosynthetic pathways. We then highlight the structural features of Family 2B encapsulin shells, with a focus on their pore properties and putative ligand-binding domains. We review the mechanisms of enzyme cargo loading in Family 2B systems and examine known examples of terpenoid synthesis compartmentalized within Family 2B encapsulin shells. This is followed by a discussion of the molecular logic and potential functional advantages of enzyme encapsulation. Finally, we consider outstanding questions and future research directions aimed at elucidating the molecular details and physiological implications of encapsulin-mediated bacterial terpene biosynthesis.

Graphical Abstract

INTRODUCTION

Encapsulins are a large and diverse class of prokaryotic protein compartments formed from protomers exhibiting the HK97 phage-like fold and are likely of viral origin.^1–4 A recent comprehensive structure-based phylogenetic analysis of HK97-fold proteins suggests multiple bidirectional evolutionary transitions between encapsulins and HK97-fold viruses.⁵ This study highlights the similarity between HK97-fold viral procapsids and encapsulins as well as the conceptual similarity of viral scaffold proteins and encapsulin cargo proteins suggesting that viral procapsids represent the most likely evolutionary nexus for HK97-fold viruses and encapsulins. Encapsulins self-assemble into icosahedral shells with triangulation numbers of T = 1 (60 subunits), T = 3 (180 subunits), or T = 4 (240 subunits) and diameters ranging from approximately 20 to 45 nm in diameter.^3,6–11 The defining feature of encapsulins is their ability to internalize and sequester cargo proteins containing a short N- or C-terminal targeting peptide (TP) or longer N-terminal cargo-loading domain (CLD).¹²

Comparative sequence and genome neighborhood analyses have enabled the classification of encapsulins into four families, with each family having distinct structural and functional features (Figure 1A).^1,2,7 Family 1 encapsulins are the most extensively characterized family and play roles in oxidative stress resistance, iron storage, and detoxification.^6,11,13–17 Family 2 encapsulins are divided into subfamilies 2A and 2B, defined by the absence (2A) or presence (2B) of cyclic-nucleotide-like binding domain (CBD) and metal-binding domain (MBD) insertions in the encapsulin protomer. Family 2A encapsulins are associated with sulfur metabolism and have been shown to sequester large amounts of elemental sulfur via internalized cysteine desulfurase (CD) enzymes.^7,8 Genome neighborhood analysis of Family 2B encapsulin gene clusters initially suggested their now confirmed role in polyprenyl and terpenoid biosynthesis among other putative functions.^2,7,9,18,19 Family 3 encapsulins remain experimentally uncharacterized, but are associated with large peptide and polyketide biosynthetic gene clusters.^1,2 Family 4 encapsulins have also not been characterized in the context of encapsulin biology thus far. However, as part of a structural genomics project, a Family 4 structure was determined and shown to form a dimer consisting of heavily truncated protomers.²⁰ Family 4 systems are found mainly in archaeal thermophiles, where they are hypothesized to function to stabilize enzymes rather than form canonical icosahedral shells.^1,2

Figure 1.

Distribution and diversity of Family 2B terpenoid encapsulins. (A) Encapsulin systems can be classified into four families. Enc: encapsulin shell protein. Adapted with changes with open access permission via a creative common license (https://creativecommons.org/licenses/by/4.0/).⁴ (B) Schematic of the known and putative roles of Family 2B encapsulins in terpene biosynthesis. The C5 prenyldiphosphate precursors DMAPP and IPP may be converted by encapsulated PTs into acyclic terpenoid precursors such as GPP, FPP, and GGPP. Acyclic terpenoid precursors can then be used by encapsulated TCs to produce cyclic terpenoid products. Encapsulin systems predicted to contain both TC and PT cargos may produce cyclic terpenoids directly using C5 prenyl diphosphate precursors. (C) Operon types and phylogenetic distribution of Family 2B encapsulins likely involved in terpene biosynthesis. A phylogenetic tree of 2416 terpene-associated Family 2B encapsulin shell proteins is shown. Family 2B accessions were obtained from the UniProt database by searching for accessions containing both PFAM annotations PF19307 and PF00027.³² Operon types were assigned using the EFI-GNT server to filter operons according to cargo type.³⁵ The phylogenetic tree was assembled via a custom workflow using the NGPhylogeny server where sequences were aligned using MAFFT, trimmed using BMGE, and trees assembled using FastTree with 200 bootstrap iterations.^36–40 The phylogenetic tree was then visualized and annotated using the ITOL server.⁴¹

Family 2B encapsulin systems are the only encapsulins confirmed to be involved in anabolic pathways and secondary metabolism. As mentioned above, bioinformatic analyses have revealed that Family 2B encapsulins are abundant in operons encoding polyprenyl transferases (PT) and terpene cyclases (TC), key enzymes in terpenoid biosynthesis (Figure 1B,C).^2,7

Terpenoids represent the largest and most structurally diverse class of natural products.²¹ They play essential roles in primary and secondary metabolism and are found in all domains of life.^22,23 Vitamins, steroids, fragrances, hormones, pigments, and many pharmaceutical drugs are based on terpenoids.^22,24 The fundamental building blocks of all terpenes are the C5 isoprene units isopentenyl diphosphate (IPP) and dimethylallyl diphosphate (DMAPP). Through successive prenyl transferase-catalyzed additions of IPP to allylic diphosphates, starting with DMAPP, the acyclic polyprenyl diphosphate precursors geranyl diphosphate (GPP, C₁₀), farnesyl diphosphate (FPP, C₁₅), and geranylgeranyl diphosphate (GGPP, C₂₀) are assembled.^21–23,25 While polyprenyl diphosphates with chain lengths larger than C₂₀ are rare in prokaryotic secondary metabolism, many eukaryotes are able to routinely synthesize substantially longer polyprenyl chains.^26,27 These linear precursors can then undergo a wide range of enzyme-catalyzed modifications, most commonly carbocation-driven cyclization via TCs, to produce diverse and structurally complex terpenoids.^21,22 Terpenoids are categorized as mono- (C₁₀), sesqui- (C₁₅), or diterpenoids (C₂₀) based on the number of five-carbon isoprene units present in their carbon skeletons.

A recent study demonstrated that a Family 2B encapsulin from Streptomyces griseus (Sg Enc) plays a direct role in terpenoid biosynthesis.⁹ Previously misannotated as a transcription factor (EshA), Sg Enc was shown to sequester the terpene cyclase 2-methylisoborneol synthase (2-MIBS) and play a role in the biosynthesis of the volatile terpenoid 2-methylisoborneol (2-MIB).^28–31 This finding provides the first example of an encapsulin directly involved in secondary metabolism, specifically terpenoid biosynthesis. As of September 2025, the Uniprot database contained over 2400 sequences of Family 2B encapsulin shell proteins associated with PTs and TCs, primarily found in the phyla Actinomycetota, Myxococcota, and Cyanobacteriota.³² Many of the identified TCs are similar to 2-MIBS or geosmin synthase (GS). The organization of Family 2B encapsulin operons was found to be variable, containing PTs, TCs, both PTs and TCs, and often additional putative tailoring enzymes (Figure 1C).^2,7,9,18 Many operons encode two separate encapsulin genes, suggesting the formation of hetero-oligomeric two-component shells as recently reported for a nonterpenoid Family 2B encapsulin system.^2,7,18 Further adding to Family 2B encapsulin system complexity is the potential for cosequestration of cargo proteins encoded in satellite loci outside the Family 2B encapsulin biosynthetic gene cluster, a feature previously observed in some Family 1 encapsulin systems.^33,34 Approximately 20% of 2-MIBS- and 80% of GS-like genes in Actinomycetota are encoded in satellite loci not containing encapsulin genes, yet most still contain characteristic CLDs found in all directly encapsulin-associated TCs and may therefore be encapsulated as secondary cargos.⁹ Collectively, these data establish Family 2B encapsulins as widespread terpene nanoreactors in bacterial secondary metabolism.

This manuscript serves to provide current perspectives on the roles Family 2B encapsulins play in bacterial terpenoid biosynthesis. Initially, the structures of Family 1 and Family 2 shells will be discussed and compared, with an emphasis on the unique features observed in Family 2B encapsulins. Next, we specifically highlight the structure and putative functions of CBDs, the defining feature of Family 2B shells. We then summarize the characterized and predicted mechanisms of cargo loading in Family 2 encapsulins. Lastly, we provide a discussion on the known and predicted terpenoid gene clusters associated with Family 2B encapsulins, followed by an analysis of the biochemical logic of enzyme encapsulation. In closing, we suggest future research directions and potential synthetic biology applications of Family 2B encapsulins as nanoreactors for diversified combinatorial terpenoid biosynthesis.

ENCAPSULIN SHELL STRUCTURE

Family 1 and 2 encapsulin shell proteins share strong structural homology with HK97-type viral capsid proteins and are composed of three canonical primary domains: the axial domain (A-domain), peripheral domain (P-domain), and the extended loop (E-loop) (Figure 2A,B).^1,2,4 Family 1 encapsulins possess an N-terminal helix (N-helix) that interacts with the P-domain to form an interior TP-binding pocket important for cargo loading (Figure 2A). In contrast, Family 2 encapsulins exhibit an extended N-terminus containing an N-arm, N-helix, and a disordered N-extension that can be up to 80 residues in length in Family 2B encapsulins and likely protrudes from the encapsulin shell (Figure 2A). While the HK97 phage-like fold of Family 2 encapsulin shell proteins is highly conserved, only Family 2B encapsulins contain a CBD and MBD insertion within the E-loop.

Figure 2.

Encapsulin structure and assembly. (A) Structure of the Family 1 encapsulin protomer from Thermotoga maritima (PDB ID: 3DKT).³ Core HK97 phage-like domains (A-domain, P-domain, E-loop) are highlighted. The bar below the structure represents the locations of the annotated domains according to amino acid sequence. (B) Family 2 encapsulin protomers from Synechococcus elongatus (2A, PDB ID: 6X8T) and Streptomyces griseus (2B, PDB ID: 9BHV).^7,9 Dashed lines represent disordered or unresolved structural features. Variable lengths of the Family 2 N-extensions are shown in the inserted plot (adapted with permission from ref 2).² (C) Left: Family 1 encapsulin shells of T. maritima, Myxococcus xanthus (T = 3, PDB ID: 7S20), and Quasibacillus thermotolerans (T = 4, PDB ID:6NJ8).^3,11,45 Right: Family 2 encapsulin shells from S. elongatus, S. griseus, and a model of a two-component shell from Streptomyces lydicus (2B, PDB IDs: 9BJE, 9BIX).^7,9,18 Pentameric facets are shown in blue, CBDs are shown in green, and Enc 1 of the two-component shell is shown in light blue and light green for the encapsulin backbone and the CBD, respectively. (D) Interior and exterior views of the closed 2-fold pore of the S. elongatus Family 2A encapsulin shell.⁷ The N-arm and N-helix are shown in yellow to highlight their position at the 2-fold pore. E-loops were removed to better highlight the pore region. (E) The open 2-fold pore of the Sg Enc Family 2B encapsulin shell from S. griseus.⁹ E-loops and CBDs are not shown for visual clarity. (F) The open and closed pore states observed in the S. lydicus two-component shell.¹⁸

Family 1 encapsulins form icosahedral shells with T = 1 (60 protomers), T = 3 (180 protomers), or T = 4 (240 protomers) symmetry, ca. 20 to 45 nm in diameter (Figure 2C).^3,6,10,11,13 Family 2 encapsulins by contrast form exclusively T = 1 icosahedral shells with a diameter of ca. 25 nm for 2A systems and a diameter of ca. 29 nm for 2B shells (Figure 2C).^7–9,18,42 The externally facing CBDs and MBDs contribute to the larger diameter of 2B shells and are located around all 2-fold symmetry axes of the shell (Figure 2C). While the assembly of 2A and 2B shells is highly similar, only Family 2B encapsulin gene clusters commonly encode two distinct shell proteins. A nonterpenoid-associated Family 2B shell has recently been shown to assemble into pseudoicosahedral mixed two-component shells (Figure 2C).¹⁸

Encapsulin shells likely regulate the molecular flux of substrates and products of encapsulated enzymes through pores located at the 5-fold, 3-fold, and 2-fold symmetry axes. The size and properties of encapsulin pores can vary considerably based on the encapsulin family, symmetry axis, triangulation number, and identity of amino acids lining the pore.⁴ A major difference in the organization of Family 1, Family 2A, and Family 2B encapsulins lies in the arrangement of their 2-fold pores.⁴ 2-fold pores of Family 1 encapsulins are formed by interactions of the E-loops of two neighboring protomers, resulting in closed or nearly nonexistent 2-fold pores. In Family 2 encapsulins, the 2-fold pores are instead formed by interactions between the N-arms and N-helices of two protomers surrounding the 2-fold symmetry axis. In Family 2A encapsulins, this arrangement results in small, nearly closed 2-fold pores (Figure 2D).^7,8 In contrast, Family 2B encapsulin shells have been shown to form both large, and open elongated 2-fold pores as well as closed 2-fold pores. The Family 2B encapsulin Sg Enc from S. griseus has unresolved N-termini at the 2-fold pore, resulting in a large, elongated pore opening, approximately 15 × 45 Å in size. Two external CBDs sit above the 2-fold pore without occluding it (Figure 2C,E).⁹ A recently structurally characterized two-component Family 2B shell from Streptomyces lydicus shows two conformationally distinct 2-fold pore states, with an open state (unresolved N-termini) resembling the Family 2B Sg Enc 2-fold pore, and a closed state (resolved N-termini) resembling that of 2-fold pores in Family 2A shells (Figure 2F).¹⁸

The existence of open and closed 2-fold pore states in Family 2B encapsulins suggests that these pores may function as dynamic gates to regulate flux of substrates or products across the encapsulin shell. Dynamic pores have been observed in multiple Family 1 encapsulin systems, such as the 5-fold pore of a ferritin-like protein-associated encapsulin from Haliangium ochraceum and the pH-responsive 5-fold pore from a dye-decolorizing peroxidase-associated encapsulin from the acidophilic bacterium Acidipropionibacterium acidipropionici.^43,44 The physiological triggers and functional implications of the dynamic 5-fold pores observed in Family 1 systems are presently unknown. It is currently also unclear whether the state of the 2-fold pores in Family 2B encapsulins is strictly a consequence of interactions between N-termini and pore-lining residues, or if the pore state can be modulated by external factors such as pH, salt, or even small molecules. One possible mechanism of pore control is ligand binding to the externally accessible CBDs or MBDs. We will elaborate on this hypothesis in more detail below.

STRUCTURE AND POSSIBLE FUNCTIONS OF FAMILY 2B CBDS

CBD insertions are the defining feature of Family 2B encapsulins and are positioned above all 2-fold axes of symmetry within assembled Family 2B shells (Figure 3A).^9,18 CBDs do not directly contact 2-fold pore residues and do not occlude the pore in any thus far determined 2B shell structure. A small MBD subdomain of unknown function with poor sequence or structural similarity to any other characterized domain is located at the N-terminal end of the CBD (Figure 3B).^4,9 Four conserved residues within the MBD coordinate a single Ca²⁺ ion.⁹ Because the MBD seems to mediate an interaction between the CBD and the Family 2B HK97-like part of the shell protein, it has been suggested to play a role in controlling the conformation or orientation of the CBD.⁹ Family 2B CBDs share significant structural homology with cyclic adenosine monophosphate (cAMP)-binding domains (Figure 3B–E). However, multiple orthogonal experiments indicate that they do not bind cAMP.^4,9,18 Canonical cAMP-binding domains, such as the catabolite activator protein (CAP) of Escherichia coli, contain a strictly conserved arginine in the cAMP binding pocket that coordinates the cyclic phosphate group of cAMP to facilitate binding (Figure 3E).^46,47 In contrast, most Family 2B CBDs lack this arginine residue and contain a tryptophan at the corresponding site instead (Figure 3D). This has been proposed to prevent cAMP binding by CBDs and may suggest that Family 2B CBDs prefer alternative ligands.⁹

Figure 3.

CBD structure and putative ligand binding. (A) Top and side views of a locally refined cryo-EM density map (EMD-44557) of the 2-fold pore of the Sg Enc shell from S. griseus in the presence of 20 mM cAMP.⁹ HK97-fold shell density is shown in blue, CBD density is shown in green, and MBD density is shown in purple. (B) Structure of the Sg Enc CBD (PDB: 9BHU). A Ca²⁺ ion is shown as a green sphere within the MBD. (C) Structure of the CBD from E. coli CAP in complex with cAMP (PDB: 1G6N).⁷² (D) A close-up of the Sg Enc CBD binding pocket. Sequence conservation between residues of other Family 2B CBDs is highlighted in color according to ConSurf conservation score.⁷³ Residues labeled with asterisks are conserved cAMP-interacting residues between the Sg Enc CBD and the E. coli CAP CBD. W176 is shown in bold. Adapted with open access permission via a creative common license (https://creativecommons.org/licenses/by/4.0/).⁹ (E) A close-up of the CAP binding pocket in complex with cAMP (yellow). The cAMP-coordinating R82 residue is shown in bold. Adapted with open access permission via a creative common license (https://creativecommons.org/licenses/by/4.0/).⁹ (F) Ligand binding to external CBDs may control the state of the 2-fold pore (open or closed). Upon ligand binding to the open-state shell (left), the 2-fold pores and CBDs may change conformation and assume a closed state (right) in which the encapsulated TC cargo is made inactive by restricting substrate access to the shell interior. (G) Different putative CBD-binding ligands.

In many CBD-containing proteins, such as protein kinase A, ion-gated channels, and transcription factors, ligand binding by a CBD triggers a conformational change that regulates the activity of distal domains.^46,47 Such changes are typically mediated through a C-terminal α-helical domain connected to the CBD via a flexible hinge region. Family 2B CBDs instead display shorter C-terminal α-helical domains that transition into a flexible linker connected to the HK97-fold E-loop, which is unresolved in currently available structures. Structural analysis of Sg Enc in the presence of high concentrations of cAMP revealed no change in the orientation of the CBD (Figure 4A), but an alternative ligand may induce a conformational change of the CBD and/or the directly connected E-loop, resulting in the opening or closing of the 2-fold pores.^4,9 Functionally, this would allow Family 2B encapsulins to regulate the flux of substrates across the shell in a ligand-dependent manner. Controlling substrate access would in turn allow the activity of encapsulated cargo enzymes to be modulated (Figure 3F).

Figure 4.

Cargo loading mechanisms in Family 2B encapsulins. (A) Disorder plots generated using DISOPRED3 for four different Family 2B cargos.^2,74 Yellow boxes in the disorder plots represent the approximate locations of the CLDs. Adapted with open access permission via a creative common license (https://creativecommons.org/licenses/by/4.0/).² (B) An SDS-PAGE gel of in vivo cargo-loading experiments demonstrating that the CLD is necessary for loading 2-MIBS into the Sg Enc Family 2B encapsulin from S. griseus. Adapted with changes with open access permission via a creative common license.⁹ (C) Cryo-EM map of the Sg Enc encapsulin loaded with 2-MIBS cargo.¹⁷ The Sg Enc shell is colored in blue, the CBDs are green, and the CLD density is yellow. Representative shell-subtracted 2D classes show internal densities corresponding to 2-MIBS. (D) A close-up of the interactions between the conserved interface of the Sg Enc shell and the CLD density of 2-MIBS at the 3-fold symmetry axis. Individual protomers are represented by different background surface shading. The cartoon and stick representations are color coded by amino acid conservation as calculated by ConSurf.⁷³ CLD density is shown in yellow. Adapted with open access permission via a creative common license.⁹ (E) The amino acid sequence of the 2-MIBS CLD from S. griseus is poorly conserved outside of the consensus motif. Amino acids are color coded according to ConSurf conservation scores.⁷³ Sections of the CLD are labeled and color-coded accordingly. The consensus motif is shown as a sequence logo below the CLD sequence. Adapted with open access permission via a creative common license.⁹ (F) In vivo cargo-loading experiments using various truncations of the CLD, with each section of the CLD corresponding to the colors in panel (E). The top gel depicts cargo-loaded Sg Enc complexes stained by Coomassie. The bottom gel depicts the observed mNeonGreen fluorescence of cargo-loaded Sg Enc complexes from the same gel. Adapted with open access permission via a creative common license.⁹ (G) GS-like terpene cyclases have conserved sequence motifs similar to 2-MIBS-like terpene cyclases. Adapted with changes with open access permission via a creative common license.²

Proteins containing CBD-like folds are known to bind to diverse ligands such as heme, 2-oxoglutarate, chlorophenylacetic acid, and even peptides, making it difficult to directly predict a specific ligand for a given CBD.^48–53 The prevalence of Family 2B encapsulins associated with terpenoid biosynthetic gene clusters, such as those responsible for 2-MIB and geosmin production, may offer a clue to potential ligands. Expression of these gene clusters and the production of their respective terpenoids are closely linked to the complex developmental lifecycles of Actinomycetota, Myxococcota, and Cyanobacteriota, with changes in expression coinciding with transitions to stationary growth, nutrient limitation, and sporulation.^30,54–56

Potentially relevant classes of CBD ligands may include: second messengers, terpenoid-related precursors, central metabolites, and other molecules associated with quorum sensing and nutrient sensing (Figure 3G). Many of these molecules act as signals, triggering physiological and morphological changes. Second messenger molecules include nucleotide derivatives such as cyclic mono- and dinucleotides, guanosine 5′-diphosphate 3′diphosphate (ppGpp), and diadenosine tetra- and pentaphosphate (Ap₄A/Ap₅A).^28,57–63 Terpenoid precursors may include 1-deoxyxylulose 5-phosphate (DXP), methylerythritol phosphate (MEP), 2-C-methyl-D-erythritol-2,4-cyclodiphosphate (cMEPP), IPP, DMAPP, GPP, and FPP. DXP, MEP, and cMEPP are key regulatory molecules involved in the MEP pathway for terpenoid precursor biosynthesis, whereas IPP, DMAPP, GPP, and FPP are potential substrates for Family 2B encapsulin-associated TCs and PTs.⁶⁴ Central metabolites may include molecules such as nucleoside tri-, di-, and monophosphates, nicotinamide adenine dinucleotide (NADH) and nicotinamide adenine dinucleotide phosphate (NADPH), coenzyme A (CoA) and acetyl coenzyme A (acetyl-CoA), and 2-oxoglutarate.^57,65–67 Intracellular concentrations of these molecules generally report on cellular energy states and nutrient availability and their fluctuations are often accompanied by physiological or morphological changes. For instance, intracellular GTP levels decrease immediately prior to sporulation in Streptomyces and Bacillus species.^65,66 Notably, intracellular 2-oxoglutarate accumulates under nitrogen-limited conditions and it is known to bind to the CBD of the global transcription factor NtcA in Cyanobacteriota.⁵⁰ Other potential ligands include bacterial quorum sensing molecules like N-acyl homoserine lactones, γ-butyrolactones, and autoinducer-2 as well as molecules related to nutrient availability like N-acetylglucosamine (GlcNAc) and its derivatives, all of which also are regulators of secondary metabolism and cell morphology.^57,68–71

CARGO LOADING MECHANISMS FOR FAMILY 2 ENCAPSULINS

Cargo loading in encapsulins is mediated via TPs or CLDs located at the N- or C-termini of all native cargo proteins. In Family 1 encapsulins, the TP motif is typically a short, conserved sequence, found at the C-terminus of the cargo protein. Cargo proteins of Family 2A and Family 2B encapsulins instead rely on CLDs for cargo loading, which typically range in length from 50 to 300 residues and are most often located at the N-terminus of the cargo protein, but can also be located internally or at the C-terminus (Figure 4A). All so far identified CLDs are annotated as intrinsically disordered.

CLDs were first identified through the bioinformatic and structural characterization of the SrpI Family 2A shell from Synechococcus elongatus, which revealed a 255 amino acid-long N-terminal disordered domain fused to a CD cargo, necessary for cargo encapsulation (Figure 4A).⁷ Additional bioinformatic analyses revealed similar CLDs in Family 2B encapsulin cargo proteins, including CDs, TCs, and PTs (Figure 4A).^2,7 For TC cargos, 2-MIBS-like single domain TCs and geosmin synthase-like two domain TCs with distinct CLD arrangements could be identified. For the Sg Enc system, it was shown that the 2-MIBS CLD is required for cargo loading (Figure 4B).^9,19 Cryo-EM analysis of the 2-MIBS-loaded Sg Enc shell revealed poorly resolved nonshell density located at an internal pocket formed around the 3-fold axis of symmetry, likely representing CLD density (Figure 4C).⁹ Comparable poorly resolved CLD-shell interactions were observed in CD-loaded Family 2A shells from Acinetobacter baumannii.^7,8 The CLD binding pocked located at the 3-fold symmetry axis is conserved across Family 2 encapsulins and lined with hydrophobic residues (Figure 4D).^7–9,18

Sequence analysis of 2-MIBS-like TCs indicates that their N-terminal CLDs are rich in proline, glycine, and alanine with a typical length of around 120 residues. Their sequence conservation is poor aside from an often repeated GPxGxGTxxL consensus motif (Figure 4E).^2,7,9,12 In vivo cargo loading experiments with the Sg Enc 2-MIBS CLD showed that the consensus motif is not required for cargo loading, but did improve loading yield when present (Figure 4F).⁹ Further truncation experiments demonstrated that partial CLDs can still mediate cargo loading, suggesting that encapsulation likely relies on redundant, cooperative hydrophobic interactions rather than a single high-affinity binding mode as seen in Family 1 encapsulins. This cooperative binding may explain why GS-like TCs have two CLDs, one within a flexible linker between the two TC domains, and another at the C-terminus, each containing consensus motifs highly similar to the ones found in 2-MIBS-like TCs (Figure 4F).^2,8 Presently, no information on GS-loaded Family 2B shells has been reported in the literature.

No PT-loaded Family 2B encapsulin has been structurally or biochemically characterized to date. PTs lack clear consensus motifs and typically contain much shorter CLDs than TCs or CDs.² As with other CLDs, the CLDs of PTs are predicted to be disordered and rich in flexible hydrophobic residues including glycine, alanine, and proline. Given the similar biochemical properties of PT CLDs compared to TC and CD CLDs and the conserved hydrophobic 3-fold binding pocket present in all Family 2 shells, it seems likely that PT loading will follow similar rules.

FAMILY 2B ENCAPSULINS AS NATURAL TERPENOID NANOREACTORS

In this section, we discuss the known and putative roles of Family 2B encapsulins as terpenoid nanoreactors in bacterial terpenoid biosynthesis and the potential advantages of enzyme encapsulation. Currently, the only characterized terpenoid-associated Family 2B encapsulin is the 2-MIBS-loaded shell from S. griseus.⁹ As outlined above, many other putative Family 2B systems with PT and/or TC cargos have been computationally predicted.

PTs are responsible for the production of linear polyprenyl diphosphates including GPP (C₁₀), FPP (C₁₅), and GGPP (C₂₀) by the successive addition of the C5 precursor IPP to acyclic allylic diphosphate substrates starting with DMAPP (Figure 5A).^21–23,25 As the linear polyprenyl chain grows, the allylic chain is pushed into an elongation cavity within the PT, which is thought to dictate chain length.⁷⁵ Subtle differences in amino acid composition of the elongation cavity can lead to drastically different product length distributions, making it challenging to accurately predict the product profiles of PTs. Phylogenetic analysis suggests that Family 2B encapsulin-associated PTs are likely capable of producing acyclic polyprenyl diphosphates of different lengths, potentially including less common longer-chain polyprenyl compounds (C₂₅ to C₅₅), depending on the systems.^9,26

Figure 5.

Family 2B encapsulin-associated terpene biosynthetic pathways. (A) A representative biosynthetic gene cluster of a PT-associated Family 2B encapsulin from Streptomyces coelicolor. A potential reaction scheme is shown for the production of FPP. Additional putative tailoring enzymes are numbered. (B) The biosynthetic gene cluster for the 2-MIBS-associated Family 2B encapsulin from S. griseus with the reaction scheme for 2-MIB synthesis. (C) A biosynthetic gene cluster from M. xanthus associated with a GS-like TC and a reaction scheme for geosmin biosynthesis. In all panels, genes are color-coded by function. Gray genes are confirmed and or predicted tailoring enzymes.

2-MIB, a methylated C₁₁ monoterpenoid, and geosmin, a noncanonical C₁₂ sesquiterpenoid, are among the most common volatile odiferous terpenoids and are responsible for the earthy smell of soil and foul taste of contaminated or stale water.^22,54,76,77 The primary biological functions of these volatile terpenoids are currently unclear, but roles as attractants for bacterial spore dispersal and repellants against bacterial predators have been suggested.^25,55,78 2-MIB biosynthesis is well characterized.^79–82 Briefly, an S-adenosyl methionine-dependent methyl transferase (MT) methylates the acyclic C10 precursor GPP, producing the intermediate 2-methylgeranyl pyrophosphate (2-MeGPP). 2-MIBS then cyclizes 2-MeGPP to form 2-MIB (Figure 5B).

Geosmin biosynthesis has also been previously characterized.^21,22 GS is a bifunctional TC containing two distinct TC domains, both of which are required for geosmin biosynthesis.^21,83–85 In the first step, the substrate FPP is cyclized in a Mg²⁺-dependent reaction to form the cyclic sesquiterpene intermediate germacradienol which is then released into solution from the active site of the N-terminal TC domain (GS-A). Germacradienol then binds to the C-terminal TC domain (GS-B) where it undergoes a Mg²⁺-dependent cyclization and fragmentation reaction to form geosmin and acetone (Figure 5C).

Many terpenoids undergo extensive structural modification by diverse tailoring enzymes such as oxidases, reductases, and transferases, which dramatically expands the structural and functional diversity of terpenoid scaffolds.^22,25,86 Family 2B encapsulin operons associated with PTs and TCs often also contain additional putative tailoring enzymes.² A well-characterized tailoring enzyme is the MT that modifies GPP to form 2-MeGPP during 2-MIB biosynthesis.⁸² Other tailoring enzymes common in TC and PT systems include isomerases, dehydrogenases, epimerases, and deaminases.⁸⁶ Enrichment of these enzymes in TC and PT operons indicates that these systems are capable of producing modified and likely structurally diverse terpenoids.

Why are many terpenoid-associated enzymes sequestered within Family 2B encapsulin shells? We will subsequently discuss the molecular logic and potential advantages of enzyme encapsulation in these systems. Approximately 18% of Family 2B encapsulin operons encode for both a PT and TC, with some operons even encoding both 2-MIBS-like and GS-like genes. CLD-containing TCs are also frequently encoded in satellite loci genomically distant from Family 2B encapsulin genes. This may suggest that multiple coregulated PTs and/or TCs can be coencapsulated. Co-encapsulation and therefore colocalization of PT and TC cargos may lead to improved pathway flux via intermediate channeling, which has been observed in some bifunctional PT-TC fusion proteins.⁸⁷ In systems encapsulating a single cargo type, shell-induced substrate- or intermediate-concentrating effects caused by the selectively permeable protein shell may also improve product yield. For example, during geosmin biosynthesis, the intermediate germacradienol must leave the active site of GS-A and rebind to the active site of GS-B. The encapsulin shell may act as a diffusion barrier for released germacradienol, resulting in an increased local substrate concentration for GS-B, thereby improving overall pathway flux.

Enzyme encapsulation in specifically Family 2B shells, uniquely containing external CBDs, may allow for the regulation of cargo enzyme activity by controlling substrate access via ligand-induced conformational changes to the CBDs and 2-fold pores. By altering the substrate and product flux across the shell, the activity of internalized enzymes may be reduced or increased depending on an external signal, probably in the form of a small molecule ligand, sensed by CBDs. Such a mechanism would represent a novel regulatory paradigm for enzyme regulation, based on the physical sequestration of enzymes in dynamic protein shells. Compared to other types of regulation, this sequestration-based strategy would be highly responsive to concentration changes of the cognate ligand resulting in fast and flexible enzyme control. Aside from regulating the activity of internalized enzymes, enzyme encapsulation may also alter the product profiles of internalized enzymes. For example, PTs have been shown to produce longer or shorter acyclic polyprenyl diphosphate products in response to the concentration and relative ratio of DMAPP and IPP, and controlling the flux of the C5 precursors into the shell may lead to a preferred PT product profile.⁷⁵

Lastly, encapsulation may additionally function to stabilize and protect cargo enzymes from harsh conditions, especially during cellular transitions like sporulation where spores become desiccated and may be exposed to harsh conditions for extended time periods.⁸⁸

PERSPECTIVES AND FUTURE DIRECTIONS

The recent discovery and characterization of Family 2B encapsulins has established their widespread role as terpenoid nanoreactors in bacterial secondary metabolism. Nevertheless, the field is still nascent and many functional aspects of these terpenoid nanoreactors are still unclear. Current structural data is limited, with only a single characterized Family 2B shell associated with terpenoid biosynthesis. Further structural investigations of other terpenoid-related Family 2B shells will provide additional insight into their biological function and dynamics, particularly with respect to the organization of their 2-fold pores and the roles of CBD and MBD insertions. Additional structures may reveal whether variable pore states are only observed in two-component shells, or if other single-component shells also share this feature. The biochemical and functional characterization of CBDs may reveal a preferred ligand and highlight a chemical switch to control pore gating. CBDs could also be engineered to bind user-selected ligands to control 2-fold pore states for engineered nanoreactor applications.

Additional questions remain related to CLDs, encapsulation mechanisms, and the rationale for having mixed cargo shells. Do the predicted CLDs in GS-like and PT cargo proteins facilitate encapsulation in a similar manner as TC CLDs? Can a minimal CLD be designed and engineered to optimize loading of non-native cargos? Do shells coencapsulating both PTs and TCs exist naturally, and what is the functional advantage of coloading? Answering questions such as these will clarify the role of Family 2B encapsulins in bacterial terpenoid biosynthesis and secondary metabolism and will lay the foundation for rationally engineering Family 2B shells for synthetic biology applications.

ABBREVIATIONS

TP

targeting peptide

CLD

cargo-loading domain

CBD

cyclic-nucleotide-like binding domain

MBD

metal-binding domain

CD

cysteine desulfurase

PT

polyprenyl transferase

TC

terpene cyclase

IPP

isopentenyl diphosphate

DMAPP

dimethylallyl diphosphate

GPP

geranyl diphosphate

FPP

farnesyl diphosphate

GGPP

geranylgeranyl diphosphate

2-MIBS

2-methylisoborneol synthase

2-MIB

2-methylisoborneol

GS

geosmin synthase

cAMP

cyclic adenosine monophosphate

CAP

catabolite activator protein

ppGpp

guanosine 5′-diphosphate 3′diphosphate

Ap₄A

diadenosine tetraphosphate

Ap₅A

diadenosine pentaphosphate

DXP

deoxyxylulose 5-phosphate

MEP

methylerythritol phosphate

cMEPP

2-C-methyl-D-erythritol-2,4-cyclodiphosphate

NADH

nicotinamide adenine dinucleotide

NADPH

nicotinamide adenine dinucleotide phosphate

CoA

coenzyme A

Acetyl-CoA

acetyl coenzyme A

GlcNAc

N-acetylglucosamine