This column highlights recently published articles that are of interest to the readership of this publication.
This column highlights recently published articles that are of interest to the readership of this publication. We encourage ABRF members to forward information on articles they feel are important and useful to Clive Slaughter, AU-UGA Medical Partnership, 1425 Prince Avenue, Athens GA 30606. Tel; (706) 713-2216: Fax; (706) 713-2221: Email; [email protected] or to any member of the editorial board. Article summaries reflect the reviewer’s opinions and not necessarily those of the Association.
The release of glycans from glycoconjugates for structural analysis by mass spectrometry is conveniently accomplished in the case of N-glycans with the enzyme peptide-N4-(N-acetyl-α-glucosaminyl) asparagine amidase (PNGase). In the case of O-glycans, however, base-mediated β-elimination, the most commonly used method, yields a glycan with a terminal reducing sugar that is associated with a significant problem. The ring configuration of the terminal reducing sugar is in equilibrium with an open-chain configuration in which the hydrogen at the C-2 position is acidic. This can cause loss of the monosaccharide from the reducing end by β-elimination of C-3-linked substituents. In turn, the new reducing glycan can undergo loss in the same way. Vos et al. propose a new, mild, convenient method to prevent such ‘peeling reactions.’ The method is compatible with base-labile substituents. It employs oxidative release of glycans by hypochlorite, but conducts the reaction at neutral pH rather than the usual alkaline conditions. This results in selective formation of an anomeric glycoside of lactic acid from threonine-linked glycans, or glycolic acid from serine-linked glycans, that prevents ring opening. The method preserves information about the amino acid to which the glycan is linked, and the carboxylate group on the anomeric tag promotes ionization in negative ion mass spectrometry. The authors show that sialic acid, sulfates, and acetyl esters remain intact during the release. When combining analysis of N- and O-glycans, they suggest first cleaving the N-glycans with PNGase, then deploying oxidative release of the O-glycans.
Native mass spectrometry is a promising approach to investigation of physiologically important interactions among membrane proteins and effects of physiologic changes in the lipid environment in which these interactions occur. Encouraging results have been obtained in studies of integral membrane proteins embedded in nanodiscs, but tight control of the lipid milieu for successful mass spectrometry in these experiments precludes reproduction of physiologically relevant characteristics such as membrane curvature, fluidity and lateral protein mobility. Panda et al. here show how integral membrane proteins that are embedded in liposomes can be analyzed. Liposomes allow lipid composition to be adjusted to emulate physiological membranes. Their diameter can be changed to vary membrane curvature. The abundance of cholesterol and unsaturated lipids can be used to regulate membrane fluidity. And their protein-to-lipid ratio can be adjusted to emulate variation in physiologic abundance. The authors accomplish desorption of protein ions directly from liposomes using an Orbitrap Q-Exactive Ultra-High Mass Range mass spectrometer system from Thermo Fisher Scientific (Waltham, MA). In this instrument, conventional collision-induced dissociation of ions in the collision cell may be supplemented by pre-activation in the source region by collision with neutral gas molecules. For some proteo-liposomes (principally those in which an integral membrane protein is embedded in a lipid bilayer, not a uni-lamellar membrane), supercharging of analyte ions with glycerol 1,2-carbonate is additionally employed to reduce the energy required for lipid removal. The authors apply the methodology to study factors affecting the interaction between VAMP2 on synaptic vesicles and t-SNARE on neuronal synaptic membranes during vesicle docking for neurotransmitter release. They also investigate factors affecting the oligomeric state of the bacterial sugar transporter semisweet. The methodology is anticipated to be useful for study of interactions between the components of yet more challenging multi-protein complexes when reconstituted within the same liposome.
A very high-throughput method is described for screening the stability of proteins encoded by mutational cDNA libraries. Established methodology for the purpose involves yeast display of encoded proteins. Susceptibility to proteolysis is used as an index of polypeptide unfolding in proteins displayed clonally on yeast cells. Proteolytic cleavage releases an attached fluorescent tag from the cell surface, and remaining fluorescent cells are sorted for identification of their encoded protein by deep DNA sequencing. The present method uses cDNA display instead of yeast display. A DNA library is transcribed and translated in a cell-free system that produces proteins that remain covalently attached to their cDNA at the C-terminus. Their N-termini bear a PA epitope tag. Proteolytic digestion of unfolded proteins separates the cDNA from the PA tag, but folded (protease-resistant) proteins remain attached to their cDNA when immunoprecipitated with anti-PA antibodies. Their identity is then inferred, and the proteins quantified, by deep DNA sequencing. To control for the effects of protease specificity, two different proteases – trypsin and chymotrypsin – are employed. In a panel of 331 natural protein domains and 148 domains designed de novo, the authors assess the stability of all single amino acid variants, including single deletions and 2 insertions at each position. They also test comprehensive double mutations at 559 site-pairs spread across 190 of the domains. The resulting dataset shows how individual amino acids and pairs of amino acids contribute to folding stability. The data are anticipated to help build deep learning models for the prediction of folding stabilities and the effects of mutations.
This paper documents the capabilities of a new Orbitrap Tribrid instrument from Thermo Fisher, the Orbitrap Ascend. The general architecture of Tribrid mass spectrometers includes an ion inlet, followed in turn by a quadrupole mass filter, a C-trap/Orbitrap, an ion routing multipole (IRM) for ion accumulation and higher-energy collisional dissociation (HCD), and a quadrupole linear ion trap. The Orbitrap Ascend is updated in several ways. The largest change is the addition of a second IRM installed in front of the C-trap. In this ‘front IRM’ an ion packet isolated in the quadrupole can be accumulated at the same time as a previous ion packet is being routed into the HCD cell via the ‘rear IRM,’ then returned to the C-trap/Orbitrap for Fourier transform product-ion scanning. The shortened duty cycle enabled by this change increases the instrument’s scan speed. Other modifications include new ion funnel optics in the inlet region that help reduce in-source fragmentation. The effect of these and diverse additional updates is a substantial improvement in performance for shot-gun proteomic analysis. Compared to the Orbitrap Elipse, the authors demonstrate an increase of up to 140% in the number of tryptic peptides identified in a sample-limited setting, up to 50% increase in the number of phosphopeptides from which phosphorylation sites could be identified, and a 2-fold increase in the number of N-glycopeptides detected. They also show a 9-14% increase in the number of peptides quantified in TMT11-plex experiments.
The genomes of all species contain short open reading frames (sORFs) capable of encoding peptides or microproteins. Ribosome profiling (Ribo-seq) has revealed that thousands of sORFs are indeed translated. In studying the physiological significance of putative microproteins, it is generally expected that those with definable biological function will be evolutionarily conserved and will have some minimum size. Sandmann et al. argue that neither of these expectations is necessarily valid. They survey 7,264 sORFs in the human genome documented as translated in the GENCODE reference catalog on the basis of Ribo-seq data. They find that 90% of them lack structurally conserved homologs in non-primate mammals, indicating recent evolutionary origin; 4,101 emerged within the group of primates plus their close relatives, the colugos. The GENCODE Ribo-seq catalog imposes a 15-amino acid length cut-off for peptides encoded by sORFs listed, but the authors identify in prior datasets a further 221 translated sORFs in the size range of 3-15 amino acids, including 38 with direct evidence for the peptide’s existence. The authors investigate function by performing proteomic interaction screening. They synthesize on cellulose membranes tiled 15-amino acid peptides covering the sequences coded by 271 sORFs, and identify by liquid chromatography-mass spectrometry proteins from a cell lysate that bind to each peptide. Functions of the proteins identified include mRNA splicing, translational regulation, and endocytosis. The authors emphasize that while such interactome evidence is suggestive of function, each instance requires experimental verification in physiological context. The work presents a pathway to begin investigating the role of a remarkable group of peptides about which very little is presently known.
Expansion microscopy is a technique for acquiring information about cells or tissues below the diffraction limit using a diffraction-limited fluorescent microscope. It works by physically and isotropically magnifying the cells or tissues in a swellable polyelectrolyte hydrogel. Proteins, nucleic acids or lipids in the sample are covalently linked to the hydrogel before swelling. Methods that achieve 4x expansion are in routine use. Klimas et al. here describe methodology achieving ≥10x expansion that is compatible with a broad range of tissue types and fixation methods, conserves biomolecule classes comprehensively without requiring separate, molecule-specific anchoring steps, and permits biomolecules to be labeled (stained) for identification after expansion. The authors specify a hydrogel composed of 4% (w/v) N,N-dimelthylacrylamide acid, 34% (w/v) sodium acrylate, 10% (w/v) acrylamide, and 0.01% (w/v) N,N’-methylenebisacrylamide (bis) for support of distortion-free expansion up to 11x, and employ methacrolein, a small molecule used in classical fixation protocols to modify biomolecules in a similar way to formaldehyde, for anchoring biomolecules to the gel. The authors use their methodology to investigate synaptic proteins in mouse brain, podocyte foot processes in formalin-fixed, paraffin-embedded human kidney, and defects in cilia and basal bodies of drug-treated human lung organoids.
Reinhardt et al. present a super-resolution fluorescence microscopy technique which supplements recently developed implementations of the MINFLUX approach. MINFLUX achieves spatial precision better than 2 nm (20 Å). The new technique is an example of so-called localization microscopy, in which target molecules are labelled with fluorescent tags that undergo sparse emission events known as ‘blinking’. Blinking from any one target is recorded at low resolution, e.g. 10-20 nm, at positions spread around the true position of the fluorophore. When multiple such localizations from the same target are averaged, the true position may be defined with very high precision. Two target molecules remain unresolved if they produce spatially overlapping distributions of blinking. But if blinking from neighboring molecules could be distinguished by color, barcode or some other feature, they could be unambiguously grouped and the targets potentially resolved. The authors accomplish this in a modified implementation of an existing method in which blinking is achieved when target molecules labeled with a DNA barcode are allowed to bind transiently with a complementary ‘imager’ DNA strand conjugated to a fluorescent dye. When the imager strand associates with the barcode strand, the target fluoresces, but upon dissociation from the barcode strand, target fluorescence ceases. The transient nature of the binding produces blinking. In the present implementation of this process, orthogonal barcodes on neighboring target molecules are used to collect positional information sequentially in a series of imaging-washing-imaging cycles. The process is called resolution enhancement by sequential imaging (RESI). It discriminates fluorescent tags spaced less than 1 nm apart. An example of the application of this methodology is provided in the authors’ proof-of-principle studies of the arrangement of neighboring Nup96 proteins within the nuclear pore complex, and patterns of association between CD20 membrane receptors on B cells that are targeted for immunotherapy. The authors report that it takes 100 min to collect localization data for molecules in an area of 67 x 67 µm2, so the methodology is suitable only for fixed cells in which the dynamic motions amenable to study by MINFLUX are suppressed. Nevertheless, the entire super-resolution procedure can be accomplished with off-the-shelf labeling reagents and a simple inverted fluorescence microscope operated under standard conditions.
Much of our knowledge of the cell divisions and morphogenetic cellular changes that occur during pre-implantation development of the human embryo rest upon study of mouse embryos. Mouse embryos can be subjected to invasive manipulation, notably transfection with genes encoding fluorescent proteins, or microinjection, that are precluded in the study of human embryos on ethical grounds. But mouse and human development are significantly different in their trajectories. Consequently, substantial uncertainty still exists about processes that are of clinical relevance for in vitro fertilization, such as what kinds of errors may lead to aneuploidy during the early mitotic divisions following fertilization. Domingo-Mueles et al. here perform live imaging of human embryos at high spatiotemporal, subcellular resolution by the introduction of fluorescent dyes that do not interference with development: SPY555-DNA for labeling genomic DNA, and SPY555-actin for labeling F-actin. They track chromosome segregation, blastocyst formation and compaction, cellular polarization, and hatching (emergence from the zona pellucida). Mosaic aneuploidy in embryos is generally attributed to errors occurring during mitosis, but the authors interestingly observe an additional process in trophectoderm cells: nuclear budding associated with shedding of DNA into the cytoplasm. They associate this process with cellular mechanical stress, and, disturbingly, they also show that it is associated with trophectoderm biopsy for the purpose of genetic testing during assisted reproductive technology procedures. The functional consequences for the embryo are presently unknown, but are now amenable to further investigation by direct imaging.
Three groups describe ground-breaking developments in methods for non-invasive, single-cell imaging of neurons in large populations firing action potentials in vivo. The work of all three groups centers upon the development of new protein sensors termed genetically-encoded voltage indicators (GEVIs), which respond to cell membrane depolarization. Most current GEVIs decrease fluorescence emission when cells depolarize, hence requiring detection of decreased signal strength against an inherently high fluorescence background. GEVIs that instead increase fluorescence with depolarization would be strongly advantageous. Platisa et al. develop two new such GEVIs with increased fluorescence response for detecting depolarization spikes occurring at high frequency. They employ low flux, two-photon imaging to minimize photobleaching and photodamage with a two-photon microscope engineered for rapid imaging over a large field of view, and they develop a self-supervised, deep learning algorithm for denoising to enhance detection of firing events. Evans et al. re-engineer two existing GEVIs for positive, and greater magnitude, response to depolarization, that works in one-photon and two-photon regimes. They demonstrate temporal resolution capable of tracking 1-ms spikes at 1,000 Hz, sufficient for clearly recording the firing of neurons in vivo, and for prolonged monitoring without photobleaching in two-photon mode. Tian et al. evolve two new GEVIs based on rhodopsin with improved signal-to-noise and, in one case, much improved on/off kinetics. Using optogenetic stimulation in vivo, they are able to characterize in vivo synaptic coupling between pairs of notably fast-spiking neurons. By providing robust, sensitive, highly parallelized optical methods for the study of electrophysiological interactions between neurons, these contributions promise to expand research into the function of neural networks previously constrained by the inherent limitations of work with microelectrodes.
Single-cell, spatial mapping of the abundance of mRNAs on a genome-wide scale is providing extensive information about mechanisms of transcriptional control and their physiological impact. However, the abundance of the proteins encoded depends additionally on translational control mechanisms. These are not as readily amenable to spatial mapping. Single-cell ribosome profiling quantifies transcriptome-wide protein translation, but does not preserve spatial information, while imaging methods provide information about spatial distribution, but only for limited numbers of different transcripts. To meet the resulting need, Zeng et al. here provide methodology for highly multiplexed spatial mapping of translation at cellular and subcellular scales. Their method uses in situ hybridization with oligonucleotide probes barcoded to recognize particular target mRNAs. The probes mediate rolling circle amplification if, and only if, the target mRNA is associated with a ribosome. Once amplified, the barcodes are identified by sequencing in situ. The sequencing involves the use of fluorescently labeled nucleotides that are imaged to localize the target mRNA. Three probes are required to complete the translation-specific amplification: (1) an open “padlock” probe complementary to the target mRNA that bears a gene-specific barcode; (2) a primer for rolling circle amplification that hybridizes both to the mRNA and to the padlock probe; and (3) a “splint” probe which hybridizes both to ribosomal RNAs and to the padlock probe, and is required to circularize the padlock probe, allowing amplification to proceed. Zeng et al. use this scheme to map the translation of 981 genes in HeLa cells. They identify translation of different mRNAs at particular stages of the cell cycle. They also identify functionally linked mRNAs whose translation is co-regulated for which translation is spatially highly correlated to particular subcellular structures. In mouse brain slices, they map the translation of 5413 genes across 119,173 cells. This work intriguingly reveals translation-controlled remodeling during oligodendrocyte maturation, widespread differences in translation in neurons and glial cells between different anatomical regions, and subcellular differences in regulation, for example between cell body and periphery. The authors mention that similar proximity-based 3-probe methodology might be deployed for study of RNA-RNA interactions, RNA-protein interactions, and RNA modifications. They anticipate that combination with other imaging techniques will facilitate integrated spatial mapping of epigenome, transcriptome and translatome in the same samples.
In the mouse, the totipotent state – the capability to generate all differentiated cells in embryonic and extra-embryonic tissues, and to form an entire organism – is restricted to zygotes and 2-cell embryos. The totipotent state is distinct from the pluripotent state. Pluripotency has been induced in differentiated cells, and stem cells in a state of induced pluripotency can be maintained in culture. Hu et al. now identify a combination of small molecules that can induce a state closely resembling totipotency from pluripotent mouse stem cells. The required molecules are: the retinoic acid analog TTNPB; the glycogen synthase kinase 3β inhibitor 1-azakenpaullone; and WS6, a modulator of NF-κB signaling. The resulting cells resemble totipotent cells in their transcriptome, epigenome and metabolome. They can form embryonic and extraembryonic cell types in culture and in teratomas, and contribute to embryonic development when introduced into 8-cell embryos. Capability to generate a whole new organism remains to be demonstrated. Meanwhile, the ability to maintain induced totipotent stem cells in culture provides new opportunities to study early cell differentiation during embryonic development.
Coulter counters work on the principle that particles passing through a pore positioned between two reservoirs containing a conductive fluid will produce a transient perturbation in the pore’s conductance. This methodology is known as resistive pulse sensing. The size of the particle determines the magnitude of the perturbation, and its speed of travel determines the duration of the event. The concentration of particles determines the number of events per unit time. The diameter of the pore restricts the particle sizes that can be detected. Pore sizes can be chosen to detect particles ranging from micrometers (for cells) to nanometers (for single molecules). The challenge inherent in the methodology is that pore diameter affects fluid volume throughput, and hence sampling efficiency and concentration limit-of-detection. Vaidyanathan et al. fabricate a chip-based device that improves the limit-of-detection by incorporating 5 pores in parallel. Using pores of effective diameter 350 nm, they enumerate virus particles and extracellular vesicles. They achieve a limit-of-detection of 5.5 x 103 particles/mL. This is estimated to represent an improvement of 2-3 orders of magnitude when compared to existing resistive pulse sensors. Because the method detects any particle within a specified size range, the authors incorporate into their device provision for affinity purification of particles of interest from biological fluids to maintain specificity. On pillared microchips, they use anti-MUC16 antibodies to select extracellular vesicles from ovarian cancer cells and an aptamer to select COVID-19 viral particles from saliva. The device is expected to find application in a wide variety of clinical screening tests, in addition to its research utility.
Three consecutive papers in Nature represent first fruits of the Human BioMolecular Atlas Program (HuBMAP) initiative. The goal of this initiative is to compile cellular maps of normal human tissues, documenting spatial cellular organization in terms of gene function, protein function and metabolic status at the single-cell level. The consortium conducting this work has developed and deployed a multitude of analytical tools to fulfill this goal. Hickey et al. map distinct regions along the length of the intestine. For multiplexed mapping of protein expression, they employ the CODEX technology, in which tissue sections are stained with a panel of protein-specific antibodies conjugated to different DNA barcodes, serially visualized by hybridization to cognate, fluorophore-conjugated oligonucleotides for imaging by fluorescence microscopy. They also collect information about gene regulation by open chromatin mapping using single-nucleus assay for transposase-accessible chromatin with sequencing (snATAC-seq). Lake et al. map the kidney in both the normal state and pathologic states representing various stages of injury (acute and chronic) or recovery. To perform spatial transcriptome sequencing at the single-cell level, they employ Slide-seqV2, in which spatial information is acquired by transferring RNA from a tissue section to a dense array of barcoded beads fabricated by split-pool phosphoramidite synthesis and indexed by sequencing-by-ligation. Greenbaum et al. map development of the maternal-fetal interface by studying sections of placentae representing different gestational stages during the first 20 weeks of pregnancy. They perform multiplexed imaging of a panel of antibodies against 37 different cell-type specific markers using a mass spectral imaging technique. Each antibody is conjugated to a different metal isotope, and the distribution of the metal atoms is determined at high spatial resolution by matrix-assisted laser desorption-time-of-flight (MALDI-TOF) imaging. These, and other studies from the consortium even more recently published, represent a new era of histology in which cellular architecture is described in spatial terms in global molecular detail at the single-cell level.
Inclisiran, a chemically modified, double-stranded small interfering RNA (siRNA) that inhibits translation of the proprotein convertase subtilisin/kexin type 9 (PCSK9), increases expression of LDL receptor and has recently received FDA approval for the treatment of hypercholesterolemia. Inception of the use of therapeutic RNAs to treat conditions of such high prevalence provides an impetus for development of streamlined methods for the synthesis of nucleic acids in very large quantities. Classical solid-phase phosphoramidite chemistry is not ideally suited for this purpose because of its large consumption of monomer precursors and organic solvents, its limited percentage yield, and the necessity for product purification. Moody et al. here address these problems. They describe a biocatalytic methodology for the production of short, modified oligonucleotides from unprotected nucleoside triphosphates in a single operation. The sequence of the oligonucleotide product is templated by a catalytic self-priming hairpin oligonucleotide. Nucleotide triphosphate building blocks are added to the hairpin by a DNA polymerase to create a hairpin duplex. Once the product sequence is complete, and the polymerase dissociates, an endonuclease V is able to bind. This enzyme selectively cleaves a single strand of the duplex downstream of an inosine residue built into the hairpin sequence, thereby releasing the product and regenerating the template. The product accumulates with repeated cycles of extension and cleavage, without generating oligonucleotide by-products that would need to be purified away. This methodology is expected to contribute substantially to the manufacture of oligonucleotide therapeutics at scale.
For the experimental exploration of functional effects of mutations on encoded proteins, there is an ongoing need for the synthesis of families of kilobase-sized DNA sequences that bear pre-determined mutations dispersed across the length of a common background sequence. De novo synthesis of variant DNA sequences is inherently wasteful in this setting because it entails making the common core sequence repeatedly. Liu et al. describe a new alternative methodology for efficiently building large families of variants. The procedure makes use of a pair of DNA polymerases: Q5 and Q5U. Q5 stalls upon encountering a uracil nucleotide in the template sequence, whereas its modified counterpart Q5U can polymerize DNA on a template containing uracil and inosine bases. In a process termed template-guided amplicon assembly, Liu et al. first synthesize a template DNA sequence with uracil at chosen positions. They then allow this template to anneal to oligonucleotides having the desired variant sequences at the chosen positions. Q5U then extends these oligonucleotides on the template and Taq DNA ligase ligates the extended oligonucleotides together. In this way, kilobase DNA sequences bearing 6-8 nucleotide mismatches or insertions, or deletions of 6-20 nucleotides, can be synthesized. A desired variant is then selectively amplified within the pool by PCR. Q5 polymerase is used to perform the amplification without interference from the original uracil-containing template. Products are validated by nanopore sequencing. The authors use this methodology to synthesize numerous multi-site variants in diverse sequences up to 10 kb at low cost.
Hematopoietic stem cell gene therapy presently involves harvesting and ex vivo engineering of a large number of a patient’s stem cells, careful quality control of the engineered products, and reintroduction to the patient following conditioning chemotherapy to deplete endogenous hematopoietic stem cells. In vivo genetic engineering, by contrast, could simplify the procedure and decrease its toxicity and cost. Breda et al. conduct pre-clinical tests in mice using lipid nanoparticles to deliver mRNAs to accomplish conditioning. The mRNA cargo for this purpose encodes the protein p53 up-regulated modulator of apoptosis (PUMA), which stimulates apoptotic cell death of cell targeted by the nanoparticle. Delivery of nanoparticles is directed by conjugating them with anti-CD117 antibodies. These bind to the cell surface protein c-KIT, which is expressed at high levels on hematopoietic stem cells, multipotent progenitors, and common myeloid progenitors. Importantly, the cell killing they induce is nongenotoxic. Because lipoprotein uptake by hepatocytes mediated by lipoprotein receptors may additionally cause hepatotoxicity, the authors also incorporate liver-specific miRNA binding sites for mir-122 into the 3’-untranslated region of the mRNA payload. This suppress expression of the mRNA, hence minimizing hepatocyte killing. The results encourage optimism that lipid nanoparticles may become clinically useful in the treatment of diverse clinical conditions based on genetic engineering.
Biofilms are aggregates of microorganisms that become embedded in an extracellular matrix of polymeric substances of their own production. Biofilms are associated with many hard-to-treat clinical conditions, including chronic pulmonary diseases (e.g. cystic fibrosis, chronic obstructive pulmonary disease) and ventilator-associated diseases (e.g. ventilator-associated tracheobronchitis, ventilator-associated pneumonia). Among the reasons for the resistance of these conditions to antibiotics is that the extracellular matrices are hard for antibiotics to penetrate. Mazzolini et al. engineer in a human lung bacterium, Mycoplasma pneumoniae, the capability to degrade the extracellular matrix of the biofilm-forming human respiratory pathogen Pseudomonas aeruginosa. They demonstrate in pre-clinical studies the engineered bacterium’s capability to prevent or treat infection with P. aeruginosa. P. aeruginosa biofilms are comprised of DNA, proteins, and the polysaccharides Pel and Psl. The polysaccharide alginate is also a principal component of mucoid strains of P. aeruginosa. The authors begin with a genome-reduced, attenuated strain of M. pneumoniae lacking genes conferring pathogenicity. They introduce into this strain genes encoding the enzymes PelAh, PslGh and the alginate lyase Al-II for degradation of the three major targeted biofilm polymers elaborated by P. aeruginosa. They also incorporate the capability to express pyocin L1 and pyocin S5, two antimicrobials with bactericidal activity against the target P. aeruginosa cells. They show that the engineered strain of M. pneumoniae eliminates infection in treated mice, whose survival is enhanced. It also dissolves biofilms on endotracheal tubes in ICU patients on prolonged mechanical ventilation. These results encourage further exploration of engineered bacteria for the treatment of human respiratory diseases.