Qian
Tian
,
Yousef
Bagheri
,
Puspam
Keshri
,
Rigumula
Wu
,
Kewei
Ren
,
Qikun
Yu
,
Bin
Zhao
and
Mingxu
You
*
Department of Chemistry, University of Massachusetts, Amherst, Massachusetts 01003, USA. E-mail: mingxuyou@chem.umass.edu
First published on 28th December 2020
With highly precise self-assembly and programmability, DNA has been widely used as a versatile material in nanotechnology and synthetic biology. Recently, DNA-based nanostructures and devices have been engineered onto eukaryotic cell membranes for various exciting applications in the detection and regulation of cell functions. While in contrast, the potential of applying DNA nanotechnology for bacterial membrane studies is still largely underexplored, which is mainly due to the lack of tools to modify DNA on bacterial membranes. Herein, using lipid–DNA conjugates, we have developed a simple, fast, and highly efficient system to engineer bacterial membranes with designer DNA molecules. We have constructed a small library of synthetic lipids, conjugated with DNA oligonucleotides, and characterized their membrane insertion properties on various Gram-negative and Gram-positive bacteria. Simply after incubation, these lipid–DNA conjugates can be rapidly and efficiently inserted onto target bacterial membranes. Based on the membrane selectivity of these conjugates, we have further demonstrated their applications in differentiating bacterial strains and potentially in pathogen detection. These lipid–DNA conjugates are promising tools to facilitate the possibly broad usage of DNA nanotechnology for bacterial membrane analysis, functionalization, and therapy.
As another promising functional unit, surprisingly, DNA has rarely been used for bacterial membrane engineering.8 DNA can form highly precise and programmable self-assembly and well-defined nanostructures. A large variety of DNA nanodevices have been developed for nanomedicine, diagnostics, computing, biophysical characterization, and structural biology.9–11 More recently, these DNA nanodevices have been modified onto mammalian cell membranes to generate artificial membrane channels, detect membrane signaling, monitor membrane biophysical phenomena, regulate cell-surface and intercellular interactions, and deliver genes and various cargos.12–23
Our goal in this study is to provide a simple and efficient approach to functionalize DNAs onto bacterial cell membranes, with the hope of achieving similar level of versatile decoration, biosensing, and therapy as that on the mammalian cell membranes. Different methods have been developed to engineer bacterial membranes, including metabolic labeling, chemical cross-linking, and hydrophobic insertion.3,24 For example, a number of DNA aptamer molecules have been identified to selectively recognize specific bacterial membrane targets and further used for the pathogen detection.25–28 However, the membrane modification efficiency and affinity of these aptamers are often limited. In addition, the identification of aptamers for many bacterial strains is still challenging with the current time-consuming and labor-intensive screening process. Considering the straightforward procedure of hydrophobic insertion, i.e., simply by incubation, we wondered if it is possible to insert DNA oligonucleotides onto bacterial membranes just by adding a hydrophobic moiety, such as lipids.
Lipid–DNA conjugates have emerged as a potent tool for the modification of mammalian cell membranes.16,17,29–31 These amphiphilic conjugates have attracted great interest due to their simple procedure, fast insertion, and high efficiency.32,33 Nevertheless, considering the inherent differences in the membrane composition of bacterial and mammalian cells, lipid–DNA conjugates that have been tested on mammalian cell membranes may not function similarly on bacterial membranes. We hope to demonstrate here that by fine-tuning the hydrophobic lipid moieties, these lipid–DNA conjugates could be also potentially used for bacterial membrane engineering.
In this study, we constructed a library of lipids, with different structure and hydrophobicity, to conjugate with DNA oligonucleotides. Interestingly, the membrane insertion efficiency of these lipid–DNA conjugates is highly dependent on the bacterial species, and even the strains. Selective bacterial membrane modification can be achieved based on the choice of lipid moieties. We have further demonstrated that these lipid–DNA conjugates can be used for the detection of various target bacteria, including methicillin-resistant Staphylococcus aureus (MRSA). With selective, rapid, and efficient modification, these lipid–DNA moieties will pave the way for the potential versatile applications of DNA nanostructures and devices for bacterial membrane analysis and regulation.
After purification and validation, we first asked if these lipid–DNA conjugates can be inserted onto the membranes of Gram-negative Escherichia coli (E. coli) cells. We chose two commonly used E. coli strains: a K-12 strain, TOP10, and a B strain, BL21. After incubating 1 μM of each lipid–DNA conjugate with the cells for 1 h at 37 °C and washing away free conjugates, cell membrane fluorescence signal was imaged with a confocal microscope. As shown in Fig. 1b, among these conjugates, the 18:1-DNA exhibited obvious fluorescence signals on most (74%) TOP10 cell membranes (Fig. 1c and Table S2†). While the 18:0–18:0 and 18:1–18:1-based conjugates can modify 25% and 28% of the TOP10 cells, respectively. There are some clear cell-to-cell variations in the membrane modification efficiency. These variations among individual cells can be due to their differences in the membrane compositions, phases of growth, aggregation status, etc. All other lipid–DNA conjugates have minimal modification (3–12%) on both TOP10 and BL21 cell membranes (Fig. 1c and S1†). As a control, the DNA oligonucleotide itself will not be inserted onto the bacterial membranes (Fig. S1†).
We also used super-resolution structured illumination microscopy to further confirm that the observed fluorescence signal was indeed from the bacterial cell surfaces (Fig. S2a†). To study if these lipid–DNA conjugates are located in the outer or inner membranes, we added 30% sucrose-containing M9 medium to induce plasmolysis of these DNA-modified E. coli cells. Plasmolysis results in the shrinkage of bacterial cytoplasm, which further leads to the separation of the inner and outer membranes. Indeed, using DiI-C12 to specifically label the inner cytoplasmic membrane of E. coli,34 we can clearly visualize the cellular shrinkage and internalized Dil-C12 fluorescence signal after treating with 30% sucrose (Fig. S2b†). In contrast, under the same condition, the fluorescence of 18:1-DNA conjugate stayed on the membranes, indicating these lipid–DNA conjugates are located on the outer membranes of E. coli cells (Fig. S2b†).
We also studied the membrane insertion of these lipid–DNA conjugates onto another type of Gram-negative bacteria, Pseudomonas aeruginosa (P. aeruginosa). However, none of these conjugates can be modified on the P. aeruginosa membranes (<1%, Table S2†). Our data indicated that the membrane insertion of lipid–DNA conjugates has specific preference on the bacterial species.
Next, we also asked if these lipid–DNA conjugates can be inserted onto the membranes of Gram-positive bacteria, such as Corynebacterium glutamicum (C. glutamicum), Staphylococcus aureus (S. aureus), and Micrococcus luteus (M. luteus). After 1 h incubation, three types of lipid–DNA conjugates, 18:0–18:0, 18:1–18:1, and 16:0–16:0, can efficiently modify the membranes (80–94%, Table S2†) of both C. glutamicum and S. aureus (Fig. 2). In contrast, the cholesterol–DNA conjugate can be selectively inserted onto C. glutamicum membranes, while the 18:1-DNA conjugate prefers S. aureus. There is no lipid–DNA conjugate in the library that can modify M. luteus cells though (Table S2†). Indeed, different Gram-positive bacterial species can also be modified distinctly with these lipid–DNA conjugates (Fig. 2b and c).
We next asked if the membrane insertion kinetics is influenced by the incubation temperature. To test this, we repeated the above-mentioned kinetic measurement at 4 °C. Similar as that of 37 °C, very fast membrane insertion was observed, with half maximum fluorescence shown at 7–20 min and 90% of maximum signal exhibited at 25–60 min. Interestingly, for the E. coli TOP10 cells, a faster membrane insertion and higher maximum fluorescence signal was shown at 4 °C (Fig. 3a). While in contrast, for S. aureus and C. glutamicum, a lower insertion kinetics and efficiency was observed at 4 °C than 37 °C (Fig. 3b and c). As shown in the following section, we believe this difference in the bacteria-specific temperature effect is likely due to different hydrophobicities of these membranes. Temperature can affect the membrane insertion of the lipid–DNA conjugates.
We also wondered if these lipid–DNA conjugates can stay on these bacterial membranes for a long time. By elongating the incubation time, on the TOP10 cell membranes, the fluorescence signal of 18:1-DNA was shown to be decreased by ∼50% after ∼90 min incubation at 37 °C. While at 4 °C, a much higher membrane probe density and persistence was observed, with only a ∼30% reduction in fluorescence after 2 h incubation (Fig. 3d). Interestingly, the 16:0–16:0-DNA conjugate was highly stable on both S. aureus and C. glutamicum membranes. Even after 24 h incubation at 37 °C, ∼40–50% of the conjugate was still on these cell membranes (Fig. 3e and f). These highly stable modifications can be potentially useful for long-term membrane analysis and regulations.
To further study the correlation between the membrane density of lipid–DNA and the observed fluorescence intensity, we prepared 1,2-dilauroyl-sn-glycero-3-phosphocholine (DLPC)-based supported lipid bilayers containing different concentrations of lipid–DNA conjugates. Lipid–DNA conjugates can be homogenously distributed on these supported lipid bilayers. Under the same imaging condition as that for the bacterial membrane studies, a linear correlation between the membrane fluorescence intensity and lipid–DNA density was observed (Fig. S7b†). Based on this standard calibration curve, we can now quantify the bacterial membrane insertion efficiency of each lipid–DNA conjugate.
On C. glutamicum membranes, up to 0.7 DNA per nm2 area can be inserted at 37 °C with the help of the 16:0–16:0-DNA conjugate. Similarly, the highest DNA density on S. aureus membranes was 0.6 nm−2 when 1 μM of 18:0–18:0-DNA was added for 1 h at 37 °C (Table S3†). In comparison, the maximum DNA modification on E. coli TOP10 membranes was achieved at 4 °C (0.4 nm−2), rather than 37 °C (0.2 nm−2), with the addition of 1 μM 18:1-DNA conjugate (Table S3†). Indeed, the lipid–DNA conjugates can be efficiently modified onto bacterial membranes. Meanwhile, on each type of these bacterial membranes, there are some clear variations in the modification efficiency among different lipid–DNA conjugates. After 1 h incubation with 1 μM of each conjugate, up to 13.5-fold difference in the membrane DNA density was shown. For the same lipid–DNA conjugate, a 1.8–7.4-fold variation in the maximum modification efficiency was observed on these E. coli TOP10, S. aureus, and C. glutamicum cells.
Fig. 4 The relationships between the hydrophobicity of each lipid–DNA conjugate and their corresponding membrane densities. These membrane densities were measured after 1 h incubation of each lipid–DNA conjugate with (a) E. coli TOP10, (b) S. aureus, and (c) C. glutamicum cells at 37 °C. The relative hydrophobicity was determined from an HPLC assay.28 Error bars represent the standard error of the mean values as analyzed from at least 50 cells in each case from different imaging regions. |
While in contrast, for the E. coli TOP10 cells, both the least hydrophobic (18:1-DNA) and most hydrophobic (18:1–18:1-DNA and 18:0–18:0-DNA) conjugates exhibited high membrane modification efficiency (Fig. 4a). There seems to be no clear correlation between the lipid–DNA hydrophobicity and the TOP10 membrane modification efficiency. Compared to S. aureus and C. glutamicum, these E. coli K12 cells are known to be more hydrophilic due to the existence of lipopolysaccharide chains on the outer membranes.38–40 The membrane hydrophobicity of these E. coli cells can be further reduced at low temperatures due to the increased content of unsaturated fatty acids.41 This fact may also explain the above-mentioned higher TOP10 membrane modification of 18:1-DNA at 4 °C compared to 37 °C (Fig. 3a). Bacterial membranes may indeed prefer the insertion of lipid–DNA conjugates of similar hydrophobicity.
We also wanted to study the effect of DNA length on the bacterial membrane modification. For this purpose, we synthesized a cholesterol–DNA conjugate based on an 80 nt-long DNA oligonucleotide. After incubating this conjugate with E. coli TOP10 and S. aureus cells, respectively, at 37 °C for 1 h, the observed cell membrane fluorescence signal was quite similar as that of the 20 nt cholesterol–DNA conjugate (Fig. S8†). These data indicated that the effect of DNA length on the membrane insertion efficiency may not be as dramatic as that of the lipid moiety. Efficient membrane insertion of both short and long oligonucleotides can be achieved with the help of these lipid–DNA conjugates.
We would like to also mention that in these Gram-positive C. glutamicum and S. aureus bacteria, there are some cell walls located outside of the plasma membranes. The thickness and composition of these cell walls will likely affect the membrane insertion efficiency of these lipid–DNA conjugates. Some of these lipid–DNA may actually anchor directly into the cell walls. Indeed, compared to cell-wall-embedded Gram-negative cells, these Gram-positive bacteria exhibit overall higher lipid–DNA modification efficiency (Fig. 4).
To further demonstrate other potential applications of these membrane-anchored lipid–DNA conjugates, we asked if these lipid–DNA conjugates can be used for the selective detection of bacteria from a mixture. We chose to study two bacterial mixtures, E. coli BL21 + S. aureus, and E. coli BL21 + TOP10. These BL21 cells have been transformed with a red fluorescent protein, RFP670, for easy distinction. Considering the bacterial membrane selectivity of these lipid–DNA conjugates (Tables S2 and S3†), a FAM-labelled 18:1–18:1-DNA and a 18:1-DNA conjugate was respectively used to target S. aureus and TOP10 cells in the mixture. Indeed as expected, these conjugates can selectively modify S. aureus and TOP10 cell membranes. 96% and 92% of the labeled cells were correct target bacteria in the presence of E. coli BL21 (Fig. 5a).
We finally asked if it is possible to use these lipid–DNA conjugates to distinguish target bacteria in a more complex cell system. We realized that it is still difficult in using just a single lipid–DNA conjugate for this purpose, instead, a pattern-based bacterial detection using a simple array of lipid–DNA conjugates may be feasible. To test this idea, we wondered if a pair of two lipid–DNA conjugates can be enough to distinguish E. coli TOP10, E. coli BL21, C. glutamicum, and S. aureus. Indeed, based on the specific recognition pattern of 18:1-DNA and 16:0–16:0-DNA, all these strains can be categorized into separate clusters in a linear discriminant analysis (Fig. 5b). Not only this pair of 18:1-DNA/16:0–16:0-DNA conjugates, other pairs of lipid–DNA conjugates, such as cholesterol–DNA/18:1–18:1-DNA, can also be used to differentiate each of these bacterial strains (Fig. S11†). More interestingly, these simple lipid–DNA arrays are able to not only distinguish bacteria from mammalian cells, but also bacteria of minor differences, for example, S. aureus vs. methicillin-resistant S. aureus (MRSA), a clinically important health-threating bacterial pathogen (Fig. S11†).42 Indeed, these lipid–DNA conjugates can be potentially used for the selective detection of various bacterial species, including antibiotic-resistant superbugs.
This study can potentially largely extend the applications of DNA nanotechnology in the field of microbiology. With the help of these lipid–DNA conjugates, versatile DNA scaffolds, structures, and devices can now be functionalized onto various types of bacterial membranes. These membrane-anchored functional DNA nanodevices can be likely used for the generation of artificial signaling pathways, analytical and biophysical characterization of bacterial membranes, structural regulation, and therapy.
Footnote |
† Electronic supplementary information (ESI) available: Materials and methods, Fig. S1–S11, Table S1–S6. See DOI: 10.1039/d0sc06630c |
This journal is © The Royal Society of Chemistry 2021 |