Genetically Engineered Flagella Form Collagen-like Ordered Structures for Inducing Stem Cell Differentiation

Summary

Bacteria use flagella, the protein nanofibers on their surface, as a molecular machine to swim. Flagella are polymerized from monomers, flagellins, which can display a peptide by genetic means. However, flagella as genetically modifiable nanofibers have not been used in building bone extracellular matrix-like structures for inducing stem cell differentiation in non-osteogenic medium. Here we discovered that interactions between Ca²⁺ ions and flagella (displaying a collagen-like peptide (GPP)₈ on every flagellin) resulted in ordered bundle-like structures, which were further mineralized with hydroxyapatite to form ordered fibrous matrix. The resultant matrix significantly induced the osteogenic differentiation of stem cells, much more efficiently than wild-type flagella and type I collagen. This work shows that flagella can be used as protein building blocks in generating biomimetic materials.

Graphical Abstract

Highlights

• •Flagella as genetically modifiable protein nanofibers for bacterial swimming
• •Genetically engineering the flagella to surface display collagen-like peptides
• •Assembly and mineralization of flagella into mineralized collagen-like structures
• •Osteogenic differentiation of stem cells induced by mineralized collagen-like matrix

Genetic Engineering; Stem Cells Research; Materials Science; Biomaterials

Introduction

Nature is a school for material design by creating various sophisticated materials with delicate structures and spectacular properties (Siefert et al., 2019, Chen et al., 2018, Mao et al., 2016). Many biological materials exhibit hierarchical self-assembly properties and bear well-defined patterns and organizations such as bone, tendon, seashells, and chitin (Sanchez et al., 2005, Brennan et al., 2017, Gantenbein et al., 2018, Rauner et al., 2017, Liebi et al., 2015). The extracellular matrix (ECM) of bone is a particularly hierarchical example because it is assembled from type I collagen (COL I) and hydroxyapatite (HAP). At the nanoscopic scale, the collagen molecules form fibrils by self-assembly, within which HAP nanocrystals grow with c-axis preferred orientation along the fibrils (Glimcher and Krane, 1968). The hierarchically organized structures of bone from the nano- to microscale are critical for its resilience, strength, stiffness, and toughness (Meyers et al., 2008). ECM mimics are known to accommodate the proliferation and differentiation of cells such as stem cells needed for tissue regeneration (Zhao et al., 2019, Chen et al., 2012, Geckil et al., 2010). Hence, various fabrication approaches have been developed to promote the self-assembly of bio-inspired materials into mimics of natural ECM in bone, including liquid crystalline assembly (Dierking, 2015, Lee et al., 2002, van t'Hag et al., 2017), electrospinning (Lee and Belcher, 2004, Alamein et al., 2013), capillary forces (Lin et al., 2010, Zhang et al., 2018), foreign molecule-induced assembly (Cao et al., 2011, Borowko et al., 2017), competitive electrostatic interactions (Yoo et al., 2006, Mendes et al., 2017), and external force-assisted assembly (Chung et al., 2011, Tkacz et al., 2014). However, it is still challenging to fabricate such materials, even mimicking the lowest level of hierarchical organization, such as the lateral parallel ordered assembly of HAP/protein nanofibers in bone.

One feasible approach to bone-inspired materials is to use bionanofibers morphologically mimicking collagen fibrils (Cao and Mao, 2007, Zhu et al., 2011, Kaur et al., 2010, Zhang et al., 2010, Wang et al., 2014, Lauria et al., 2017, Farokhi et al., 2018, Sunderland et al., 2017). Bacterial flagella are protein nanofibers naturally protruding from the bacterial cell body. As molecular machines, their movement facilitated the swimming of the bacteria in liquid. Owing to their linear ordered nanostructures and ability to be genetically engineered to display multiple peptide motifs, flagella can be potentially employed as units for the fabrication of bone-like ECM (Scheme 1). Moreover, flagellins (FliCs), the monomers of flagella, have been used in vivo as a nontoxic drug (Burdelya et al., 2008). Bacterial flagellum is a helical nanofiber self-assembled from FliCs with some other minor proteins at the distal ends. It contains 11 subunits (FliCs) per two turns. It is about 14 nm wide and 10–15 μm long (Scheme 1A). The FliC is a globular protein with an overall shape resembling a capitalized Greek gamma (Γ), with the upper right corner (D2 and D3 domains) pointing outward and genetically modifiable to display foreign peptides. The lower left corner of the Greek gamma is made of the N- and C-terminal domains, termed D0 and D1, of FliC, and binds together through non-covalent interactions and faces the center of the flagella. The solvent-exposed central region (D2 and D3 domains) of FliC can be modified through insertion of amino acid sequences, a technique called flagella display (Westerlund-Wikstrom, 2000). Each FliC can display a copy of foreign peptide, resulting in a distance (along the long axis) of about 2.6 nm between the neighboring FliC subunits (Namba et al., 1989) and generating a nanofiber with homogeneous peptide display under precise control (Scheme 1A).

Scheme 1.

Display of a Collagen-like Peptide (GPP)₈ on Flagella, Biomimetic Assembly and Mineralization of the Resultant GPP8 Flagella, and BMSCs' Differentiation on the GPP8 Flagella Film

(A) Schematic of a flagellum. It is mainly composed of about 30,000 flagellin (FliC) subunits. It contains 11 subunits per 2 turns. The N and C termini of flagellin (D0 and D1) are highly conserved and face the center of the filament. However, the central D2 and D3 regions are hypervariable and can be genetically engineered to display (GPP)₈.

(B) Schematic of the production and self-assembly of HAP-mineralized flagellar filaments into a matrix, a closed-packed monolayer of mineralized flagella, which can support the adhesion, proliferation, and early osteogenic differentiation of BMSCs. After the peptide is displayed on the flagella by genetically engineering the bacteria (1), the flagella are purified from the surface of bacteria by vortexing (2). Then the flagella are first allowed to interact with Ca²⁺ ions and then with an HAP precursor solution to initiate the self-assembly and mineralization of the flagella (3). The resultant mineralized flagella bundles are deposited onto a polylysine-coated glass slide to form a matrix (4) to induce osteogenic differentiation of BMSCs (5 and 6).

Inspired by these excellent properties of the flagella, a kind of natural protein nanofiber morphologically mimicking collagen fibrils, here we genetically displayed a collagen-like peptide on the surface of flagella by genetic engineering (Scheme 1). The flagella are naturally protruding from bacterial cell body to assist cell swimming. We first detached flagella from the bacteria by vortexing. We then demonstrated formation of an inorganic-organic supramolecular nanocomposite with long-range-ordered architectures mimicking some aspects of bone ECM, by the self-assembly and mineralization of flagella. Briefly, we first allowed Ca²⁺ ions to interact with flagella to form ordered bundles. We then allowed the bundles to be incubated in an HAP precursor solution to induce HAP mineralization within the flagella bundles. The Ca²⁺ ions serve as both an inducer for the lateral assembly of the flagella into bundles and a precursor for HAP. The resultant ordered flagella structures were found to significantly induce osteogenic differentiation of bone marrow-derived stem cells (BMSCs) in a non-osteogenic basal medium.

Results and Discussion

Collagens bear a repeating triplet sequence of Gly-X-Y with X position often being Pro and Y often being hydroxylated Pro. Such triplets are the key to the triple helical conformation of collagens. Therefore, (Gly-Pro-Pro)_n or simply GPP_n is often considered collagen-like peptide (Fan et al., 2008). Thus we first genetically displayed a collagen-like peptide, (Gly-Pro-Pro)₈ (termed GPP8), on the solvent-exposed domain of every copy of FliC monomer constituting the flagella, leading to the production of flagella nanofibers with GPP8 fully presented on the side walls (Scheme 1). We incubated the resultant GPP8 flagella in a CaCl₂ solution (1–4 mM) and found that such pretreatment favored the assembly of flagella into bundles, with 2 mM CaCl₂ producing the most lengthy bundles (Figures S1 and S2). When the concentration of CaCl₂ solution was higher (10 or 20 mM), the flagella were depolymerized into subunits. Thus we chose to use 2 mM CaCl₂ solution to pretreat the flagella to form Ca²⁺-decorated flagella bundles. Then the flagella bundles were purified and incubated in a 4 mM HAP-supersaturated solution for 6 days to form HAP-mineralized higher-ordered nanostructures (Figure 1). Without pretreatment and mineralization, GPP8 flagella formed a randomly stacked layer with the constituent nanofibers showing characteristic sinusoidal wave pattern in water (Figures 1A and 1C). During the pretreatment, spontaneous self-assembly of flagella nanofibers occurred (Figures S1A and S1B), promoting the formation of mineralized parallel bundles after incubation in HAP-supersaturated solution (Figures 1B and 1D). At the same time, a layer of inorganic mineral can be visualized on the flagella (Figures 1B and 1D). Selected area electron diffraction (SAED) analysis verified that the mineral was HAP owing to the presence of characteristic (211), (002), and (004) planes. It should be noted that the flagella are invisible under transmission electron microscopy without staining but their morphology can be visualized once coated with HAP. Indeed, without pretreatment with calcium ions in the absence of other ions in an HAP-supersaturated solution (Li et al., 2012), flagella displaying a collagen-like peptide cannot self-assemble into ECM mimics with collagen-like parallel-aligned matrix, indicating that interaction between the flagella and calcium ions is important for the lateral assembly of flagella into ECM mimics.

Figure 1.

Transmission Electron Microscopic Images of GPP8 Flagella with or without Mineralization

(A and C) Transmission electron microscopic (TEM) images (A, low magnification; C, high magnification) of non-mineralized GPP8 flagella (due to the lack of pretreatment), showing that in water, flagella were randomly stacked and aggregated with a characteristic curly morphology. The flagella were stained with uranyl acetate. Owing to the localization of stain on the periphery of the fibers and hollow structure of flagella, tube-like filaments were observed at high magnification.

(B and D) Low (B) and high (D) magnification TEM images of flagella after pretreatment and mineralization. SAED clearly showed the characteristic (211), (002), and (004) planes of HAP (inset in B). The HAP nanocrystals located on the surface of flagella and between flagellar filaments when mineralized flagella form bundles are seen in (D). It should be noted that the TEM samples in (B) and (D) were not stained before imaging.

Scale bars, 1 μm in (A and B) and 400 nm in (C and D).

We observed a concentration-dependent self-assembly and mineralization behavior by varying concentrations of flagella in the HAP-supersaturated solution after pretreatment (Figures 2 and S3). At a low concentration of flagella (∼10 μg/mL), some monodisperse flagella were coated with a layer of inorganic mineral and a few flagella were assembled into bundled nanostructures (Figure 2A). At a higher concentration of flagella (∼40 μg/mL), more ordered nanostructures assembled from flagella bundles were observed (Figure 2B). Finally, at an even higher concentration of flagella (∼80 μg/mL), the mineralized flagella were assembled into a close-packed monolayer (Figure 2C). Owing to the formation of parallel bundles, most nucleated minerals were located in between neighboring filaments. The diameter of flagella is about 14 nm, and the channels between flagella filled by minerals are 25 ± 10 nm in diameter. Interestingly, the SAED pattern showed the diffraction ring for (211) plane and two pairs of diffraction arcs for (002) and (004) planes at the relatively straight segment of flagella bundles, showing the preferred crystallographic c-axis orientation of HAP along the flagella (Figure 2D). Energy-dispersive X-ray spectroscopy (Figure 2E) identified a Ca/P molar ratio of 1.65, similar to the theoretical ratio (1.67) in HAP [Ca₁₀(PO₄)₆(OH)₂]. This kind of nanostructure mimicked some key properties of bone ECM and rebuilt the structural orientation and organization between HAP and collagen fibrils in bone. The supramolecular self-assembly of GPP8 flagella was monitored by a reported turbidity assay at 320 nm (McMichael and Ou, 1979). We have found an exponential increase of absorption with the concentrations of flagella, verifying mineralized bundle formation (Figure 2F).

Figure 2.

Biomimetic Self-Assembly and Mineralization of GPP8 Flagella under Different Flagella Concentrations in the HAP-Supersaturated Solution after Pretreatment with Calcium Ions

(A) Only a few bundles were formed at a low flagella concentration (∼10 μg/mL).

(B) At a higher flagella concentration (∼40 μg/mL), more parallel bundles were formed.

(C) A close-packed monolayer was observed when the concentration of flagella reached ∼80 μg/mL.

(D) SAED pattern at the relatively straight segment of flagella bundles, showing the c-axis preferred orientation of HAP along the flagella (inset).

(E) Energy-dispersive X-ray spectroscopic analysis confirming a Ca/P molar ratio of 1.65. Some residues of Na and Cl came from the HAP precursor solution.

(F) An optical density measurement of flagella under different concentrations after mixing with 4 mM HAP-supersaturated solution. The solid line represents the curve-fitting result.

(G and H) A series of CD spectra of flagella under different concentrations of CaCl₂ (G, wild-type flagella; H, GPP8 flagella) during pretreatment.

(I) Schematic illustration of self-assembly due to interactions with Ca²⁺ ions. Ca²⁺ ions first interacted with carbonyl groups of GPP8 on flagella by chelation and induced a side-by-side aggregation. With the incorporation of phosphate ions, HAP nuclei were formed, leading to the formation of HAP on flagella.

Scale bars, 400 nm in (A–C) and 250 nm in (D).

In a control experiment, pretreated wild-type (WT) flagella exhibited neither nucleation nor ordered assembly after mixing with 4 mM supersaturated HAP solution for 6 days. The control experiment indicated that the presence of collagen-like peptide as well as the initial interaction between Ca²⁺ ions and the peptide on the flagella drove the self-assembly and mineralization. In another control, no bundled structures were observed in the absence of Ca²⁺ ions, even though the concentration of GPP8 flagella reached up to 200 μg/mL in water without pretreatment or mineralization. These results indicated that the initial interaction between collagen-like peptides displayed on the flagellar surfaces and the Ca²⁺ ions participated in directing the organized self-assembly.

To understand the importance of the pretreatment of GPP8 flagella by Ca²⁺ ions, GPP8 flagella (80 μg/mL, 20 μL) were mixed with different concentrations of CaCl₂ (200 μL) (Figures S1 and S2). At a low CaCl₂ concentration (1 mM), some lateral aggregation of filaments was observed (Figure S1A). With the increase of CaCl₂ concentration to 4 mM, thicker parallel flagella bundles were assembled into ribbon-like structures. Moreover, the constituent nanofibers lost their characteristic curly morphology but became “straight.” At the same time, some filaments were broken into shorter fragments (Figure S1B). Indeed, a high concentration of Ca²⁺ (0.5 M) could entirely depolymerize flagella (Wakabayashi et al., 1969). As expected, when the concentration of CaCl₂ reached 10 mM, almost all flagella were depolymerized into fragments. Surprisingly, some thicker and much more compact bundles were newly formed (Figures S1C and S2). We believe the bundles should be reassembled from the flagellar fragments or monomeric flagellins. At 20 mM CaCl₂, only thick and compact bundles could be observed without any flagellar fragments, implying that the flagellar fragments were totally depolymerized into monomers and then reassembled into compact bundles (Figure S1D). The formation of flagella bundles upon the increase of CaCl₂ concentration was also confirmed by the turbidity assay at 320 nm (Figure S4). At different concentrations of flagella (65 and 130 μg/mL), the turbidity initially increased with the concentration of CaCl₂ due to the formation of bundles, and then started to decrease under a higher concentration of CaCl₂ due to the depolymerization of flagella. It should be noted that no lateral aggregation or bundle formation occurred when a phosphate solution was mixed with GPP8 flagella.

The positively charged ions or molecules could induce the lateral aggregation of linear protein molecules (Cao et al., 2011, Wang et al., 2010). For example, the lateral packing of phage nanofibers in CaCl₂ was formed by the electrostatic interaction between Ca²⁺ and the negatively charged phage (Wang et al., 2010). The lateral assembly of flagella might partially arise from a similar mechanism due to the presence of large anionic surface-exposed domains (D2 and D3 from 405 to 454 amino acid positions) (Yonekura et al., 2003). On the other hand, Ca²⁺ promoted fibril formation of COL I in vitro by chelating with carbonyl oxygen on the COL I (Zhang et al., 2003). Circular dichroism (CD) spectroscopy and turbidity assays in situ also suggested that the interaction between Ca²⁺ and COL I induced the conformational change of COL I during the early phase of biomineralization (Cui et al., 2008). Molecular modeling and theoretical studies of the collagen-like peptide (GPP)_n also revealed that cations were bound to carbonyl groups and the final organization of Ca²⁺ ions was similar to that in HAP crystal (Yang and Cui, 2007). The CD spectra of WT and GPP8 flagella also revealed the interactions between Ca²⁺ and GPP8 peptide on flagella (Figures 2G and 2H). The two negative peaks around 222 and 208 nm of flagella, ascribed to the perpendicular and parallel π→π* bands of α-helix structure, respectively, were consistent with the major composition of flagellin (Yonekura et al., 2003, Kelly et al., 2005). After flagella were mixed with different concentrations of CaCl₂, no obvious peak changes could be observed in WT flagella (Figure 2G). However, with the increased concentrations of CaCl₂, the negative peak intensity was decreased in GPP8 flagella, indicating the possible conformational changes due to the interactions between the flagella and Ca²⁺ (Figure 2H).

Therefore, we propose that biomineralization of the bioengineered flagella takes place along with their lateral aggregation and ordered alignment (Figure 2I). During the pretreatment with Ca²⁺ ions, carbonyl groups on the flagella chelate Ca²⁺ ions, and when this process occurs in the neighboring filaments, a side-by-side aggregation appeared. The other polar amino acids on the flagellar surfaces might also help the lateral aggregation by the electrostatic interactions with Ca²⁺ ions, and water molecule-mediated hydrogen bonds should also facilitate the ordered alignment of the flagella (Brodsky and Ramshaw, 1997). After Ca²⁺ ions were coordinated with GPP8 flagella, phosphate ions are then electrostatically attracted onto the nanofiber surfaces, resulting in local supersaturation with respect to HAP and the subsequent formation of HAP nuclei. More minerals are then deposited preferentially along the flagellar surface and the channels between neighboring bundled filaments (Wang et al., 2010). The proposed oriented HAP growth along flagellar surface might shed light into the HAP formation mechanism on the collagen surfaces in bone.

The biomineralized flagella bundles, which mimicked biomineralized collagen fibers, the building blocks of bone, were deposited on polylysine-functionalized substrates to form a flagella-based film (Scheme 1). Thus, BMSCs were then cultured on the film in non-osteogenic basal medium (Figures 3A and S5). The initial interactions between the cells and substrates decided cell behaviors (Kaur et al., 2010, Smith et al., 2010). BMSCs spread on the flagella-based film but not as efficiently as on the COL I or polylysine substrates. They exhibited a range of morphologies from spindle to polygonal on GPP8 flagella. Occasionally, some cells with a curly morphology were observed on the GPP8 flagella and such morphology might be induced by contact guidance response (Gerecht et al., 2007), leading to oriented growth and alignment (Figure S6). BMSCs spread much less on the WT flagella than on the other substrates, implying lack of bioactive signals on the WT flagellar surface. Scanning electron micrographs of BMSCs on collagen and flagella also revealed similar morphologies as those under light microscopy (Figure S7). Surface spreading area measurements of BMSCs further confirmed that the cells on the flagella spread significantly less than those in the control polylysine substrates (Figure 3B).

Figure 3.

Fluorescence Images, Surface Spreading Area, and Proliferation of BMSCs

(A) Morphologies of BMSCs on the different substrates after 24 h. Most BMSCs exhibited a spindle-like morphology on the collagen and flagella-bearing film (WT and GPP8) but presented a well-spread morphology on polylysine (control). Scale bar, 100 μm. Cell nuclei and F-actin were marked by DAPI (blue) and fluorescein isothiocyanate-labeled phalloidin (green), respectively.

(B) Surface spreading area analysis of BMSCs.

(C) Cell proliferation on the different substrates. Collagen promoted cell proliferation, but cell proliferation on the flagella-bearing films decreased.

Compared with control, *p < 0.05, **p < 0.01; compared with collagen, ^@p < 0.05, ^@@p < 0.01; compared with GPP8 flagella, ^# p < 0.05). Collagen, collagen-coated surface; WT, wild-type flagella-coated surface; GPP8, mineralized GPP8 flagella bundles; control, polylysine surface.

The proliferation of BMSCs was measured after 3 days (Figure 3C). The growth rate of BMSCs on the flagella substrate was lower than that on the collagen or polylysine substrate. An inverse relationship between proliferation and differentiation was usually observed in many cell types (Ettinger et al., 2011). Thus the decreased proliferation of BMSCs on the flagella-bearing substrates implied the enhancement of differentiation on the substrates.

The osteogenic differentiation of BMSCs on the biomineralized flagella films was monitored by different osteogenic markers at both protein and gene levels, and biomineralized GPP8 flagella outperformed other groups in inducing the osteogenic differentiation of BMSCs (Figure 4). Both immunofluorescence and qRT-PCR indicated higher expression levels of two osteogenesis-specific markers, osteopontin (OPN) and osteocalcin (OCN), on biomineralized GPP8 flagella than on WT flagella on day 14 (Figures 4A and 4B). Both flagella films, GPP8 and WT, showed higher expression levels of the OPN and OCN than two control substrates (polylysine and COL I films). It was known that polylysine promoted the osteogenic differentiation of BMSCs (Galli et al., 2011). COL I was reported to enhance osteogenesis of BMSCs via ERK and Akt pathways (Tsai et al., 2010). Namely, the flagella substrates were more effective in inducing the osteogenic differentiation of BMSCs than polylysine and COL I. As a key transcription factor and early specific marker for osteogenesis, Runx2 was also up-regulated in the cells on the flagella film compared with non-flagella films (Figure 4B). After being cultured for 14 days, some cells on the GPP8 flagella film aggregated and formed calcified nodule-like structures, indicating the maturation on the osteogenic pathway (Figure S8). Quantitative analysis of alizarin red staining revealed that extracellular calcium deposits (calcified nodules) were highly increased on the flagella films compared with non-flagella films (Figure 4C). It is well known that HAP highly enhances BMSCs' differentiation toward osteoblasts (Wang et al., 2007). The nanofibrous topography can also enhance the osteogenic differentiation of stem cells (Zhu et al., 2011, Kaur et al., 2010, Smith et al., 2010). The synergistic effects of the nanotopographies of the flagella in combination with nucleated minerals and collagen-like peptide on the constituent flagella generated a microenvironment that could enhance the osteogenic differentiation of BMSCs toward osteoblasts. Thus the mineralized GPP8 flagella are more effective than non-mineralized WT flagella in inducing osteogenic differentiation, but the latter are still more effective than COL I. It should be noted that when non-collagen-like peptide is displayed on the flagella, the flagella cannot be assembled into the abundant collagen-like ordered structures seen on GPP8-displaying flagella, and consequently, the resultant flagella-based matrix can only promote the osteogenic differentiation in the osteogenic medium but cannot induce the osteogenic differentiation in the non-osteogenic basal medium (Li et al., 2019).

Figure 4.

Differentiation of BMSCs on the Mineralized Flagella Films

(A) Immunofluorescence images showed strong expression of OPN and OCN in the cells on the mineralized self-assembled GPP8 flagella film. Collagen and polylysine also promoted the expression of OPN and OCN but at a lower level than mineralized GPP8 flagella. DAPI (blue) marked the cell nuclei, and fluorescein isothiocyanate-labeled phalloidin (green) labeled the F-actin. OPN and OCN were labeled through rhodamine-tagged antibody (red). Scale bar, 25 μm.

(B) RT-qPCR analysis showing that the flagella film could enhance mRNA expression of Runx2, OPN, and OCN. Owing to the mineralization on the GPP8 flagella film, OPN expression was significantly increased. However, gene expression between the cells on the collagen and polylysine films did not show significant difference (compared with control, *p < 0.05, **p < 0.01; compared with collagen, ^@p < 0.05, ^@@p < 0.01; compared with GPP8 flagella ^#p < 0.05).

(C) The calcium-containing minerals (calcified nodules) were stained by alizarin red S after 2 weeks. Scale bar, 400 μm. An increased number of calcified nodules indicated a higher degree of mineralization. The cells were more matured on the mineralized flagella film, indicating accelerated differentiation of BMSCs. (**p < 0.01, versus control; ^@@p < 0.01, versus collagen; ^#p < 0.05, versus GPP8 flagella). Collagen, Col I-coated surface; WT, wild-type flagella-coated surface; GPP8, mineralized GPP8 flagella bundles; control, polylysine surface.

Compared with other display techniques such as phage display, which generally allows the display of short peptides (e.g., shorter than 20-mer) on the side wall (Zhu et al., 2011), much longer peptides (as long as a few hundred amino acids) can be tolerated on flagella without losing self-assembly properties (Westerlund-Wikström et al., 1997), making flagella capable of bearing a variety of functional peptides. Moreover, in combination with double or triple peptide display technology (Westerlund-Wikstrom, 2000), multifunctional flagella could be developed as building blocks to construct organized biomimetic scaffolds in the future. Flagella can also be easily purified in large amount with low cost and thus they are an excellent candidate for biomaterials applications. These features further allow the flagella-based ECM mimics to possess a variety of desired biological functions in regenerative medicine by genetic display of functional peptides instead of by using chemical conjugation. In addition, flagella have a Young's modulus within the range of 1–100 GPa (Flynn and Ma, 2004) whereas HAP single crystals have a Young's modulus of ∼150 GPa (Zamiri and De, 2011). Thus biomineralized collagen-like flagella structures are unique soft-hard hybrid materials that may also modulate stem cell fate through their mechanical cues other than the topographical cues. Finally, flagellins, the subunits of flagella, are anti-tumor and radioprotective agents (Hajam et al., 2017). Thus flagella-based ECM mimics may hold promise for regenerating tissues while avoiding tumor growth.

In conclusion, bacterial flagella, the molecular machines for driving the bacteria to swim, were employed as natural biotemplates to form ordered structures as building blocks for fabricating bone-inspired, osteogenic materials. With collagen-like peptides displayed on the side walls, the bioengineered flagella could interact with Ca²⁺ ions to form bundles and then mineralized in an HAP precursor solution to form bone ECM-like matrix. The presence of collagen-like peptides on the flagellar surfaces as well as the Ca²⁺ ions were the two important factors that contributed to the coupled self-assembly and biomineralization. The matrix assembled from the flagella was biocompatible and could support the adhesion and growth of BMSCs. Moreover, the nanotopography and surface chemistry of the flagella-based matrix significantly induced the osteogenic differentiation of BMSCs. The biomimetic nucleation and self-assembly of the bioengineered flagella represents a novel approach to bioinspired ordered and hierarchical materials.

Limitations of the Study

Mechanical cues can also influence the fate of stem cells. Although this study shows that flagella-based mineralized collagen-like structures can induce the osteogenic differentiation of BMSCs, this study can be further improved by investigating the stiffness of the collagen-like structure. It is likely that the topographical and chemical cues presented by the collagen-like structure play a dominant role in directing the osteogenic differentiation; studying the effect of the stiffness of the collagen-like structures on osteogenic differentiation will give us a more complete picture of the flagella-directed osteogenic differentiation.

Methods

All methods can be found in the accompanying Transparent Methods supplemental file.

Published: July 26, 2019