Background

The peptidoglycan (PG) cell wall is a crosslinked polymer network that surrounds most bacterial cells, defines their morphology, and protects them against osmotic lysis. This polymer is made of muropeptide units, which consist of two amino sugars, N-acetylglucosamine (NAG) and N-acetylmuramic acid (NAM), with a peptide side chain that contains unique D-amino acids attached to the latter. The sugar subunits are polymerized into glycan strands by glycosyltransferases (GTases) via β-1,4 glycosidic bonds, and adjacent glycan strands are crosslinked by transpeptidases (TPases) through their peptide moieties [5]. While class A penicillin-binding proteins (aPBPs) contain both GTase and TPase domains, SEDS (shape, elongation, division, and sporulation) proteins are monofunctional GTases that work in concert with monofunctional TPases called class B PBPs (bPBPs) [6]. Additionally, many bacteria utilize LD-transpeptidases (LDTs), which are structurally distinct from PBP transpeptidase domains and facilitate different crosslinking reactions [7].

Because the entire PG layer of each bacterial cell is a covalently closed single molecule, PG expansion, e.g., the incorporation of new glycan strands, requires hydrolases, also known as autolysins, to generate openings in the existing PG network. Endopeptidases (EPs) are particularly important as “space-making” enzymes that cleave peptide crosslinks and thus allow such insertion [8–15]. Bacteria constantly remodel their PG using large repertoires of autolysins. Each bond in the PG matrix has a specific type of enzyme with the ability to break it: besides EPs, lytic transglycosylases (LTGs) break glycan strands, while amidases and carboxypeptidases (CPs) remove amino acids from different termini of peptides [10,15–19].

Quantitative analysis of the activities of PG-modification enzymes has been a major challenge in the field. PG labeling using fluorescence-conjugated D-amino acids [20,21], carbohydrates [22,23], and antibiotics [24,25] allows in situ probing of newly synthesized PG in live cells. In addition, high-performance liquid chromatography (HPLC), ultrahigh-performance liquid chromatography (UPLC), and mass spectrometry on purified, muramidase-digested PG sacculi produce high-resolution muropeptide profiles that provide quantitative architectural information, including muropeptide composition, glycan chain length, and the crosslinks between them [3,26,27]. However, rather than monitoring individual enzymatic activities, both methods reflect the combined activities of many PG-modification enzymes.

Many in vitro assays have been developed to study individual enzymes. For instance, purified lipid II, a PG precursor, after being labeled with fluorescent dyes or radioactive isotopes, can be used to quantify the activities of purified GTases [6,28]. Zymogram and Remazol Brilliant Blue (RBB) assays are widely used on autolysins. Zymogram runs denatured autolysins in SDS-polyacrylamide gels supplemented with PG as a substrate. After electrophoresis, the gel is transferred to a renaturation solution. The refolded autolysins then degrade PG, and their activities can be reflected by the clearing zones on gels [29]. In contrast, the RBB assay quantifies the release of RBB, a dye that covalently attaches to PG on its sugar moieties. Autolysins can degrade the insoluble, RBB-modified PG and thus release the dye into solution, whose absorbance can be readily measured using a spectrophotometer [2,3,30,31]. However, as most PG-modification enzymes reside in the membrane and periplasm, these techniques face significant hurdles in protein expression, solubilization, and refolding. Moreover, the techniques for lipid II purification and labeling are not available for most microbiology laboratories.

Single-particle microscopy is a powerful imaging technique in bacterial cell biology. Due to the diffraction of light, each fluorophore, no matter how small, generates a signal that covers an area of 300–500 nm in diameter [32]. Because conventional fluorescence microscopy detects all fluorophores in the sample, hundreds to thousands of such signals inside a small bacterial cell (typically ~1 μm wide) create a blurry picture, where the dynamics of single protein particles cannot be resolved [33]. Using single-particle tracking photo-activated localization microscopy (sptPALM), only a few fluorophores are randomly switched on in each cell, enabling accurate determination of their locations. Capturing time-lapse images, we can track the movement of individual particles with high spatial and temporal resolution (~160 nm and up to 10-ms, e.g., 100 frames/s) and quantify their distribution in subcellular regions [1–4,34–36]. Here, we refer to isolated, single fluorescent spots under sptPALM as “single particles” rather than “single molecules” because they could represent protein complexes [37,38].

Here, we describe a sptPALM-based method to quantify the binding of enzymes to PG in vivo, which serves as a proxy for their enzymatic activities. Biochemical reactions on the PG scaffold are unique due to the stark size difference between the enzymes and their substrates. While enzyme particles are in nanometer scales, their substrates, the PG sacculi, span several micrometers. Under the microscope, PG remains stationary, but PG-related enzymes are free to move. When diffusive enzymes bind to PG, their mobility decreases. Consequently, increased enzyme binding to PG under specific growth or genetic conditions leads to a decrease in its diffusion coefficient. Using this method, the diffusion of many PG-modification enzyme particles, including both synthases and autolysins, has been documented [3,4,39,40]. In these studies, the diffusion coefficient (D) was used to quantify particle mobility, where high D values indicate high particle mobility and thus reflect low PG binding. However, the negative correlation between D and PG binding is not necessarily linear.

Much of our knowledge on PG assembly and modification is based on mutant phenotypes. However, many SEDS and their partner bPBPs are essential and thus undeletable. In addition, aPBPs and PG hydrolases are notoriously redundant in most bacteria, where strains lacking single or several of these enzymes rarely display significant phenotypes (B. subtilis remains viable when 41 of the 42 PG hydrolases are absent) [10,17,41–44]. Phenotypes, if they exist, are unreliable for assigning functions to PG-related enzymes, because even when two enzymes have distinct functions, their null mutants could produce similar phenotypes [4,16]. This method captures the diffusion of hundreds to thousands of protein particles, enabling us to elucidate how PG-modification enzymes regulate each other in live cells, even in the absence of pronounced “macro-phenotypes” in morphology, growth, or division. For this reason, this protocol is especially useful for studying the PG-modification enzymes that are either redundant (such as some aPBPs and most autolysins) or essential (such as the Rod components).