Acinetobacter Baumannii Phages: Past, Present and Future

Acinetobacter baumannii (A. baumannii) is one of the most common clinical pathogens and a typical multi-drug resistant (MDR) bacterium. With the increase of drug-resistant A. baumannii infections, it is urgent to find some new treatment strategies, such as phage therapy. In this paper, we described the different drug resistances of A. baumannii and some basic properties of A. baumannii phages, analyzed the interaction between phages and their hosts, and focused on A. baumannii phage therapies. Finally, we discussed the chance and challenge of phage therapy. This paper aims to provide a more comprehensive understanding of A. baumannii phages and theoretical support for the clinical application of A. baumannii phages.


Introduction
Acinetobacter baumannii (A. baumannii) is an essential Gram-negative pathogenic bacterium, widespread in nature [1]. A. baumannii can adhere to surfaces easily, as it has pods and pili [2]. Furthermore, since A. baumannii has strong invasive virulence factors, such as outer membrane proteins, lipopolysaccharides and phospholipases, the treatment of A. baumannii infection has been regarded as a great threat to clinical practice [3]. Antibiotics such as carbapenems, β-lactam antibiotics and polymyxins, are commonly used clinically to treat A. baumannii infections [4][5][6]. However, the treatment of multi-drug resistant (MDR) A. baumannii is further aggravated by the abuse of antibiotics and the evolution of bacteria. Bacteriophages are bacterial viruses that specifically recognize, infect, and replicate within the host bacteria [7,8]. Phages have been considered as therapeutic agents since the early 1920s as a result of their unique antibacterial ability. In addition, phages have the advantages of strong antibacterial ability, high quantity (10 30 -10 32 in the earth), and low toxic side effects to humans, and are considered as one of the most promising drugs to replace traditional antibiotics [9,10].

Antibiotic Resistance in Acinetobacter baumannii
Drug-resistant bacteria continue to emerge, posing a huge challenge to human health and safety. Some studies have shown that antibiotic resistance genes (ARGs) of bacteria, such as trimethoprim resistance genes (dfrA), existed long before the clinical application of antibiotics [11]. A. baumannii is one of the most common opportunistic pathogens in nosocomial infections. Bacteria can develop antimicrobial resistance (AMR) to a variety of antibiotics, such as β-lactams [12], quinolones [13] and polymyxin [14], through intrinsic resistance mechanisms, such as increased efflux pumps [15], decreased outer membrane proteins (OMPs) [16] and acquired resistance mechanisms [14]. In addition, phages can also mediate antibiotic resistance in bacteria through transduction. special cases include filamentous phage [43] and barely non-tailed phage [44]. At present, there are no definite reports that the host of filamentous phage is A. baumannii.
The viral nature of phages was controversial until the early 1940s, when they were observed by electron microscopy, confirming their particulate nature and enabling them to be classified based on morphology [45]. Classic electron microscopy images are formed by atoms of heavy metals such as uranium, which evaporate in a vacuum and allow for the sample to be struck at an angle. The introduction of negative stains (heavy metal salts that dry in thin layers, do not form crystals and can embed small particles, such as phages) in electron microscopy has resulted in more detailed images than earlier methods and revealed the complexity and diversity of phage morphology [46]. Assigning phages to different taxonomies is a fundamental step in phage research. As more and more new phages are discovered, ICTV's classification criteria are constantly changing. The most recent standard is the August 2022 phage classification system, which removes several major families in the order Siphoviridae, Podoviridae and Myoviridae. However, the classical description of its morphology as belonging, such as "podovirus", "myovirus", or "siphovirus", remains. The order "Caudovirales" was also deleted and replaced by the class Caudoviricetes, and a binomial system of nomenclature for species was established [47].
Genome sequencing revealed the abundance of the prophage. Comparative genomics showed the co-evolutionary relationship between phages and their host bacteria as an essential tool to reveal phage diversity [48] and provide the strategy of therapy [49]. The framework of the A. baumannii phage genome is mainly composed of genes from other phages, with a relatively small proportion of genes from the host A. baumannii [50]. Morphogenesis of the tail occurs frequently, the mutant Ab105-2phi∆CI404ad, genomic rearrangement increases the host range of the phage nearly 3-fold [51]. There is no genome similarity of vB_AbaS_TCUP2199 to other known phages [52]. It has been documented that most of the genes of bacteriophages are unknown [53]. Their genome size range is wide, and the structure of their genome is linear or circular [54]. The most common type of genome is dsDNA among known phages. The genome size of phage KARL-1 was determined to be 166,560 bp. A total of 253 ORFs were identified, involved in the replication, maturation and release of phage progeny. Whereas some encoded structural proteins, most ORFs encoded hypothetical phage-like proteins, and the rest are hypothetical proteins [55]. The genome of A. baumannii phage Abp9 contains 80 ORFs, but lacks any known virulence genes or lysogen-forming genes [56]. The two phage strains, WCHABP1 and WCHABP12S, contain seven structural proteins. In addition, both encode a gene for a protein-containing lysozyme that is also possessed by other phages of the genus Ap22virus [57].

Life Cycle and Biological Properties of Phages
Phage growth is determined by many parameters, namely, the phage bacterial adsorption rate constant, burst size, latent period, bacterial growth rate, phage and bacterial elimination rates, and the effect of controlled release of the phage [58,59].
Phages can be divided into temperate phages and lytic phages, which have different life cycles (Figure 1). Infection begins with the adsorption of phage, which interacts with specific receptors on the surface of host cells, and then the phage injects its DNA into the cytoplasm of the host, next, transcription and replication occur. Subsequently, once the synthesis and assembly of viral proteins are complete, the phage DNA is packaged into the capsid. Phage-encoded depolymerase hydrolyses the peptidoglycan layer of bacterial cells, leading to cell lysis and the release of mature virions. The phage progenies are released into the environment and they can then infect the next phage-sensitive receptor [60]. Temperate phages can enter either lytic or lysogenic cycles [61], for example, λ phage, after the injection of DNA into the cytoplasm. The temperate phage could choose to initiate a lytic cycle, which would be consistent with the virulent phage. It integrates its DNA into the host bacterial chromosome. At the same time, the viral genome is named a prophage. This phage expresses a specific phage repressor that represses transcription factors, including lytic cycle genes, and thus the phage enters a dormant state [62]. The integrated phage dsDNA replicates with the bacterial chromosome during cell division, and is thus passed down through generations in bacteria. When the phage is excised from the host chromosome, it exits the lysogenic cycle [63].
cells, leading to cell lysis and the release of mature virions. The phage progenies are re-leased into the environment and they can then infect the next phage-sensitive receptor [60]. Temperate phages can enter either lytic or lysogenic cycles [61], for example, λ phage, after the injection of DNA into the cytoplasm. The temperate phage could choose to initiate a lytic cycle, which would be consistent with the virulent phage. It integrates its DNA into the host bacterial chromosome. At the same time, the viral genome is named a prophage. This phage expresses a specific phage repressor that represses transcription factors, including lytic cycle genes, and thus the phage enters a dormant state [62]. The integrated phage dsDNA replicates with the bacterial chromosome during cell division, and is thus passed down through generations in bacteria. When the phage is excised from the host chromosome, it exits the lysogenic cycle [63]. The biological properties of different phages differ. vB-AbaM-IME-AB2 was able to adsorb host cells within 9 min with >99% adsorption, a 20 min latency period and a small outbreak size (62 PFU/cell) [64]. Two phages (WCHABP1 and WCHABP12) were obtained from hospital sewage, with different lysis ranges and burst sizes of 136 and 175 PFU/cell, respectively, with 99% adsorption within 10 min, and the MOIs of both were 0.1 [57]. Phage PD-6A3 had better activity at temperatures of 4-50 °C and from pH 5 to 10, with 90% adsorption within 5 min [65]. The clinical isolate PlyF307 is the first highly active therapeutic-specific A. baumannii phage against a Gram-negative bacterium, and rescued 50% of mice in a mouse animal model [66]. For Gram-positive bacteria, the rescued rates are usually higher [67,68].

Phage Recognition
The phage life cycle consists of five stages: adsorption, infestation, replication, assembly and release, the most important of which is the adsorption process [69] (Figure 2). Phage tail recognition sites are abundant and diverse, such as lipopolysaccharides, lipoproteins and capsules of A. baumannii, which are potential phage recognition sites. Phages expressing depolymerase degrade bacterial outer polysaccharides and promote phage recognition and initial adsorption to the host [70][71][72]. The biological properties of different phages differ. vB-AbaM-IME-AB2 was able to adsorb host cells within 9 min with >99% adsorption, a 20 min latency period and a small outbreak size (62 PFU/cell) [64]. Two phages (WCHABP1 and WCHABP12) were obtained from hospital sewage, with different lysis ranges and burst sizes of 136 and 175 PFU/cell, respectively, with 99% adsorption within 10 min, and the MOIs of both were 0.1 [57]. Phage PD-6A3 had better activity at temperatures of 4-50 • C and from pH 5 to 10, with 90% adsorption within 5 min [65]. The clinical isolate PlyF307 is the first highly active therapeutic-specific A. baumannii phage against a Gram-negative bacterium, and rescued 50% of mice in a mouse animal model [66]. For Gram-positive bacteria, the rescued rates are usually higher [67,68].

Phage Recognition
The phage life cycle consists of five stages: adsorption, infestation, replication, assembly and release, the most important of which is the adsorption process [69] (Figure 2). Phage tail recognition sites are abundant and diverse, such as lipopolysaccharides, lipoproteins and capsules of A. baumannii, which are potential phage recognition sites. Phages expressing depolymerase degrade bacterial outer polysaccharides and promote phage recognition and initial adsorption to the host [70][71][72].
The capsule is an important recognition site for A. baumannii phage and consists of capsular polysaccharide that wraps around A. baumannii and provides protection at the periphery [73][74][75][76][77][78]. Fernando et al. found that 24 strains of A. baumannii developed resistance to A. baumannii phage through evolution, and 20 of the phage-resistant strains showed capsule defects. Genetic comparison of the two resistant strains (AB900 and A9844) with the wild type, that both phage-resistant strains had a base deletion at the K site, resulting in a code-shifting mutation. The results showed that the gtr29 gene (encoding glycosyl transferase) was affected in AB900, and that the gpi gene (glucose-6-phosphate isomerase) was affected in A9844, causing capsule deletion [79]. Popova  host cells and observed that specific phages could not adsorb to the cells. The results obtained confirm that the capsular polysaccharides are the primary receptors for the phages and that tails pike depolymerase plays a crucial role in the initial step of phage-bacterial cell interaction [80]. There is no doubt that phage infestation of A. baumannii inevitably disrupts the capsule, regardless of whether the phage receptors are capsules. In basic research data on the mechanism of phage-specific recognition and hydrolysis of capsules, studies on A. baumannii phages emphasize the type of capsule of A. baumannii that can be lysed [76,77,[80][81][82][83]. and SAR-endolysin (green ovals with SAR domains). In the phage synthesis stage, holin or pinholin proteins are continuously synthesized and accumulate in the cytoplasm and IM in the form of a dimer. At the same time, endolysin protein is folded to form peptidoglycan-lytic enzymes and accumulates in the cytoplasmic matrix. SAR endolysin accumulates in the cytoplasm and IM without activity. When those proteins accumulate to a critical value, endolysin is released to lyse the peptidoglycan layer, and SAR endolysin refolds its conformation to obtain the active enzyme.
The capsule is an important recognition site for A. baumannii phage and consists of capsular polysaccharide that wraps around A. baumannii and provides protection at the periphery [73][74][75][76][77][78]. Fernando et al. found that 24 strains of A. baumannii developed resistance to A. baumannii phage through evolution, and 20 of the phage-resistant strains showed capsule defects. Genetic comparison of the two resistant strains (AB900 and A9844) with the wild type, that both phage-resistant strains had a base deletion at the K site, resulting in a code-shifting mutation. The results showed that the gtr29 gene (encoding glycosyl transferase) was affected in AB900, and that the gpi gene (glucose-6-phosphate isomerase) was affected in A9844, causing capsule deletion [79]. Popova et al. used tail spike depolymerase to degrade capsular polysaccharide layers surrounding A. baumannii host cells and observed that specific phages could not adsorb to the cells. The results obtained confirm that the capsular polysaccharides are the primary receptors for the phages and that tails pike depolymerase plays a crucial role in the initial step of phagebacterial cell interaction [80]. There is no doubt that phage infestation of A. baumannii inevitably disrupts the capsule, regardless of whether the phage receptors are capsules. In basic research data on the mechanism of phage-specific recognition and hydrolysis of capsules, studies on A. baumannii phages emphasize the type of capsule of A. baumannii that can be lysed [76,77,[80][81][82][83]. and SAR-endolysin (green ovals with SAR domains). In the phage synthesis stage, holin or pinholin proteins are continuously synthesized and accumulate in the cytoplasm and IM in the form of a dimer. At the same time, endolysin protein is folded to form peptidoglycan-lytic enzymes and accumulates in the cytoplasmic matrix. SAR endolysin accumulates in the cytoplasm and IM without activity. When those proteins accumulate to a critical value, endolysin is released to lyse the peptidoglycan layer, and SAR endolysin refolds its conformation to obtain the active enzyme.
With the invention and development of protein expression technologies and modern detection techniques, studies on phage and host recognition have been continuously reported. The tail fiber proteins gp52 and gp53 of A. baumannii phage AbTJ, labeled with fluorescein isothiocyanate, were combined with magnetic beads and the BL method for A. baumannii detection. The tail fiber proteins were observed to be bound to the surface of A. baumannii under fluorescence microscopy, but did not have a lytic function [84].
Arms races between bacteria and their phages promote each other to evolve for survival. Bacteria are forced to pay a price for phage threat evolution, especially when phage recognition sites are important to them. Phage-resistant strains are shown to suffer from adverse conditions, such as weakened metabolism, reduced virulence (capsule, virulence factors) and restored susceptibility to antibacterial substances (antibiotics) [79].

Phage Lyse Bacteria
Another critical stage of phage growth and reproduction is release. Phage synthesis of lysis-related functional proteins causes bacterial cells to rupture, thereby releasing the offspring. Phage therapy is a promising alternative route against the growing backdrop of severe global antibiotic resistance, but rapid bacterial resistance to phages limits the development of phage treatment. In the search for a breakthrough, the associated phage releases phase proteins that destroy bacteria and are considered to be a highly promising AMR agents. Therefore, the study of phage lysis bacterial pathways is imperative. This section focuses on the phage lysis pathway of A. baumannii.
In 1992, the classic holin-endolysin theory proposed that bacteriophages mainly rely on the synergistic cleavage of holin and endolysin proteins [60]. Holin protein is a kind of small molecular membrane protein that is synthesized in large quantities in the process of bacteriophage synthesis, to control the time of cleavage. Endolysin protein is a protease that hydrolyses peptidoglycan, which destroys the membrane structure of bacteria. Because of the comprehensive characterization of the lambda phage, the cleavage model is based on the lambda phage [85]. In the phage synthesis stage, holin protein is continuously synthesized and accumulates in the cytoplasm and intima in the form of a dimer. At the same time, endolysin protein is folded to form peptidoglycan-lytic enzymes and accumulates in the cytoplasmic matrix. When the holin and endolysin proteins accumulate to a critical value, holin "triggers" are triggered (in the lambda system, 50 min is triggered after the synthesis of holin and endolysin begins) [86,87]. The holin protein inserted into the intima randomly forms a nonspecific channel or hole that allows for the endolysin protein to pass through to cleave the peptide polysaccharide layer (Figure 2A). A sufficient amount of holin protein destroys the electrochemical balance of the membrane and causes local depolarization to form channels or pores. Studies by Gründling [88] have shown that the holin system can be triggered in advance by dinitrophenol (DNP) or sudden hypoxia, but premature cleavage is accompanied by a significant decrease in phage titer [89]. Due to the local action of the holin protein, the protuberance of bacterial cells at the action site of the holin protein was observed under a video microscope, followed by rupture that released substances in the cytoplasmic matrix [90]. A great deal of evidence shows that the holin proteins P2 Y and T4 T are different from the lambda system [91], the time of the holin "trigger" is determined by the holin protein itself [92][93][94][95][96].
With the deepening of molecular-level research, a new cleavage pathway, namely, the SAR endolysin and pinholin pathways, has been found [97]. The infiltration of SAR endolysin into the intima is not dependent on holin protein: it permeates the intima in the form of a membrane plug. SAR endolysin accumulates in the intima, but is not released, and the membrane plug form does not have the enzymatic activity to avoid premature opening and cleavage, leading to a low phage titer. When the molecular dynamic potential of the intima changes, such as in depolarization, SAR endolysin can be released from the intima to the peptidoglycan layer, and the enzyme active center can be constructed by folding, to catalyze the hydrolysis of the peptidoglycan layer [98,99]. The role of changing the dynamic potential of membrane molecules and control of the timing of cleavage could occur via a kind of holin protein that is not directly involved in the output and is only used as a timer [100]. To distinguish it from the holin protein with an output function in the classic holin-endolysin pathway, this kind of protein is called pinholin. Pinholin accumulates in the intima, and when the time for cleavage comes, it polymerizes into regular heptamer channels, destroys the membrane dynamic molecular potential, and promotes SAR endolysin to obtain activity and cleave peptidoglycan [101] ( Figure 2B). In addition, studies have shown that pinholin cannot release classic endolysin, which further proves that holin-endolysin and SAR endolysin/pinholin are two different cleavage pathways [102]. Both pathways lead to bacterial lysis, but different rupture processes were observed under the video microscope, and the two can be clearly distinguished. As mentioned above, there are significant local protuberances in the holin-endolysin pathway, which is due to the massive release of local endolysin, due to the formation of channels or holes by holin. SAR endolysin is uniformly distributed on the intima. When pinholin triggers the cleavage cycle, SAR endolysin releases and cleaves peptidoglycan from all directions. Therefore, the gradual contraction and final cleavage of bacteria can be observed under a video microscope.
Before 2012, based on the study of the above two pathways, it was speculated that the outer membrane is not a barrier blocking the release of the progeny phage. However, Berry et al. described a class of cleavage proteins called spanins, indicating that for most Gram-negative bacteria, the peptidoglycan layer and outer membrane are barriers that bacteriophages need to break through [90]. In their study, a mutant without spanin function was constructed. The morphological protuberance of the bacteria was observed and spread around, finally forming a ball that did not release intracellular material. At present, the characterized spanin systems are divided into two categories: (i) i-spanin and o-spanin two-component systems and (ii) u-spanin single-component systems [103]. The action mechanism of spanins is still in the exploratory stage. Young R provides a reasonable mode of action [104]. He proposed that spanins lead to the final lysis of bacterial cells through membrane fusion. The most suggestive elements for a membrane fusion model, besides the possible fusogenic character of purified Rz1 reported previously, are the primary structure of the Rz periplasmic domain and the conformational dynamics attendant to spanin complex formation [105]. Based on the results of Bryl et al., the binding of the Rz1 protein to the Rz homodimer leads to a substantial increase of alpha helix as well as the formation of coiled-coil oligomers, which evokes the coiled-coil dynamics of the SNARE system that is integral to trans-Golgi vesicle fusion [106,107]. SNARE proteins undergo coiled-coil oligomerization to bring the vesicle membrane and cell membranes into contact [108]. In the hypothesis of Young, in the two-component system, i-spanin is connected to the intima, o-spanin is tied to the adventitia, and the peptide polysaccharide layer is clamped to prevent folding and contraction. When endolysin or SAR endolysin releases cleavage peptidoglycans, their conformations change, and folding contraction may bring the intima and adventitia closer and lead to membrane fusion. Membrane fusion could destroy the molecular potential of the inner and outer membranes, leading to the collapse of the inner and outer membranes. Based on the data by Bryl et al., it is speculated that the factor promoting membrane fusion may be o-spanin [105]. These hypotheses do not seem to be suitable for the one-component system of u-spanin, but membrane fusion is the preferred model for u-spanin at present (Figure 3). that holin-endolysin and SAR endolysin/pinholin are two different cleavage pathways [102]. Both pathways lead to bacterial lysis, but different rupture processes were observed under the video microscope, and the two can be clearly distinguished. As mentioned above, there are significant local protuberances in the holin-endolysin pathway, which is due to the massive release of local endolysin, due to the formation of channels or holes by holin. SAR endolysin is uniformly distributed on the intima. When pinholin triggers the cleavage cycle, SAR endolysin releases and cleaves peptidoglycan from all directions. Therefore, the gradual contraction and final cleavage of bacteria can be observed under a video microscope.
Before 2012, based on the study of the above two pathways, it was speculated that the outer membrane is not a barrier blocking the release of the progeny phage. However, Berry et al. described a class of cleavage proteins called spanins, indicating that for most Gram-negative bacteria, the peptidoglycan layer and outer membrane are barriers that bacteriophages need to break through [90]. In their study, a mutant without spanin function was constructed. The morphological protuberance of the bacteria was observed and spread around, finally forming a ball that did not release intracellular material. At present, the characterized spanin systems are divided into two categories: (i) i-spanin and o-spanin two-component systems and (ii) u-spanin single-component systems [103]. The action mechanism of spanins is still in the exploratory stage. Young R provides a reasonable mode of action [104]. He proposed that spanins lead to the final lysis of bacterial cells through membrane fusion. The most suggestive elements for a membrane fusion model, besides the possible fusogenic character of purified Rz1 reported previously, are the primary structure of the Rz periplasmic domain and the conformational dynamics attendant to spanin complex formation [105]. Based on the results of Bryl et al., the binding of the Rz1 protein to the Rz homodimer leads to a substantial increase of alpha helix as well as the formation of coiled-coil oligomers, which evokes the coiled-coil dynamics of the SNARE system that is integral to trans-Golgi vesicle fusion [106,107]. SNARE proteins undergo coiled-coil oligomerization to bring the vesicle membrane and cell membranes into contact [108]. In the hypothesis of Young, in the two-component system, i-spanin is connected to the intima, o-spanin is tied to the adventitia, and the peptide polysaccharide layer is clamped to prevent folding and contraction. When endolysin or SAR endolysin releases cleavage peptidoglycans, their conformations change, and folding contraction may bring the intima and adventitia closer and lead to membrane fusion. Membrane fusion could destroy the molecular potential of the inner and outer membranes, leading to the collapse of the inner and outer membranes. Based on the data by Bryl et al., it is speculated that the factor promoting membrane fusion may be o-spanin [105]. These hypotheses do not seem to be suitable for the one-component system of u-spanin, but membrane fusion is the preferred model for u-spanin at present (Figure 3). Generally, the phage cleavage process of Gram-negative bacteria needs to break through three barriers: the intima, peptidoglycan layer and outer membrane. The cleavage of the intima and peptidoglycan layer is accomplished by two groups of matching proteins, namely, classical holin-endolysin and SAR endolysin/pinholin [104]. On this basis, the cleavage of the outer membrane at the spanin protein is an essential step [105]. At present, the characterized spanin system is divided into a u-spanin one-component system, and an i-spanin and o-spanin two-component system. In theory, there are four different combinations of bacteriophages in the cleavage of Gram-negative bacteria.

Phages Affect Bacterial Genomes
Bacteriophages have been widely recognized as natural and efficient carriers of certain bacterial toxins or resistance genes, including classic type I membrane-acting superantigens, type II porogenic lysin and type III exotoxins, such as diphtheria and botulinum toxin. In addition, there is a kind of effector protein encoded by bacteriophage in Gram-negative bacteria, which is a new type of bacterial virulence factor [109]. The life cycle of temperate phage is lysogeny. Lysogenic bacteriophages integrate their own genome into the host genome through integrase, replicate with the host genome and form offspring. Through this pathway, lysogen phages participate in the variation related to bacterial virulence or drug resistance [110,111], carrying genes encoding virulence (strong extracellular toxins), proteins related to promoting bacterial invasion, and various enzymes (superoxide dismutase [112], grape kinase, phospholipase, DNA enzyme, proteins that affect serum resistance and change antigenicity, superantigens, adhesion molecules, proteases, and mitogenic factors) [37,109,113,114]. In addition, the prophage genome can also bring quorum sensing and motility to bacteria [115]. The above phenomenon, which gives virulence to bacteria by bacteriophages, is called lysogen transformation [116,117]. Under the influence of inducers or adverse external factors (such as antibiotics), lysogen bacteriophages will enter the lytic cycle and produce offspring to find the host again. The induction and mobility of prophages promote the widespread of virulence genes, ARGs and mobile genetic elements and promote the evolution of bacteria [79,118,119].
Based on the genomic analysis of the current pandemic carbapenem-resistant A. baumannii, the Oxa23 gene related to drug resistance was detected and analyzed by Abouelfetouh et al.
The results have shown that eight of the 13 strains carrying the Oxa23 gene were located on the genome of the phiOXA prophage [111]. Phages with the Oxa23 gene were successfully isolated by mitomycin C induction. Transposon or plasmid-mediated drug resistance is a common mechanism, but in recent years, it has been reported that prophages lead to drug resistance, which may be another main mechanism of the horizontal transfer of these genes [120][121][122][123][124]. Similarly, based on the analysis of the evolution of A. baumannii in the clinical ward over 10 years, it was found that lysophage provided A. baumannii with virus defense proteins and quorum sensing functions [115].
Phages interact with bacteria while using bacteria for reproduction and provide bacteria with genes that are conducive to bacterial survival (ARGs, virulence genes, quorum sensing, etc.). This is a very interesting phenomenon; bacteriophages seem to protect their "food" from being naturally eliminated. In addition, it has been reported that the release of ARGs by phage lysis can be introduced into nondrug-resistant E. coli and produce related drug resistance [125].

Phages Act on Biofilms
A. baumannii with biofilms are a collective community cluster formed by the accumulation of a large number of bacteria, surrounded by self-secreted fibrin isopolymers. A large number of studies on hospital clinical outbreaks, severe infections and antibiotic resistance of A. baumannii have identified biofilm formation ability as the main virulence factor of A. baumannii, while providing strong adaptability and resistance to antibiotics [126][127][128][129][130][131].
In clinical practice, A. baumannii forms biofilms on the surfaces of hospital equipment or on retention equipment in patients [132], especially in urinary tract infections [133,134].
Phages act on biofilms, destroying the biofilm matrix, exposing receptors on the bacterial surface, and initiating the phage replication cycle. In addition, phage lysis of biofilms also increased the permeability to antibiotics, restoring the eradication effect of antibiotics on bacteria [152] (Figure 4). By comparing the antibiofilm activity of a single phage and a phage cocktail, it was found that the antibiofilm activity of the phage cocktail is significantly higher than that of a single phage [51]. Studies have shown that the combination of two different types of antibacterial agents, bacteriophage and antibiotics, is more effective than the use of a single component, and can eradicate biofilms [134,153]. Changes in the biofilm structure, caused by one or both of the antimicrobial agents, may account for the enhanced disinfection effect. For example, bacteriophages can make bacteria more sensitive to bacteriophages and certain antibiotics by disrupting their external structures and improving their internal metabolic state. Antibiotics themselves may also cause changes in the biofilm structure, thereby increasing the ability of phages to invade biofilms [152].
large number of studies on hospital clinical outbreaks, severe infections and antibiotic resistance of A. baumannii have identified biofilm formation ability as the main virulence factor of A. baumannii, while providing strong adaptability and resistance to antibiotics [126][127][128][129][130][131]. In clinical practice, A. baumannii forms biofilms on the surfaces of hospital equipment or on retention equipment in patients [132], especially in urinary tract infections [133,134]. Biofilms can effectively inhibit antibiotics elimination of bacteria, which is one of the important reasons for the high mortality of A. baumannii infection [135][136][137][138][139]. Biofilm-producing A. baumannii has become a serious challenge for clinical infection, and many chemical-, physical-, and biological-based methods have been developed to prevent or destroy biofilms [140][141][142][143]. However, bacteriophages that directly degrade bacterial biofilms are the most promising therapeutic method to solve this problem, so the study of phages acting on biofilms is crucial [144][145][146][147][148][149][150][151].
Phages act on biofilms, destroying the biofilm matrix, exposing receptors on the bacterial surface, and initiating the phage replication cycle. In addition, phage lysis of biofilms also increased the permeability to antibiotics, restoring the eradication effect of antibiotics on bacteria [152] (Figure 4). By comparing the antibiofilm activity of a single phage and a phage cocktail, it was found that the antibiofilm activity of the phage cocktail is significantly higher than that of a single phage [51]. Studies have shown that the combination of two different types of antibacterial agents, bacteriophage and antibiotics, is more effective than the use of a single component, and can eradicate biofilms [134,153]. Changes in the biofilm structure, caused by one or both of the antimicrobial agents, may account for the enhanced disinfection effect. For example, bacteriophages can make bacteria more sensitive to bacteriophages and certain antibiotics by disrupting their external structures and improving their internal metabolic state. Antibiotics themselves may also cause changes in the biofilm structure, thereby increasing the ability of phages to invade biofilms [152]. Phage disruption of biofilms relies on hydrolases such as phage endolysin and depolymerase, and Lu et al. characterized that endolysin PlyF307 significantly reduced plankton and biofilms of A. baumannii both in vitro and in vivo. PlyF307, the first highly active lysin protein against Gram-negative bacteria, rescues mice from lethal A. baumannii bacteremia. The endolysin belonging to other families can also lyse A. baumannii [154,155]. By comparing the engineered phage expressing depolymerase with the phage without depolymerase, it was found that the engineered phage significantly destroyed the biofilm of E. coli [66]. After Zhang et al. treated A. baumannii with bacteriophage AB3 and its LysAB3 for 24 h, the A. baumannii biofilm was significantly degraded [156]. Phage disruption of biofilms relies on hydrolases such as phage endolysin and depolymerase, and Lu et al. characterized that endolysin PlyF307 significantly reduced plankton and biofilms of A. baumannii both in vitro and in vivo. PlyF307, the first highly active lysin protein against Gram-negative bacteria, rescues mice from lethal A. baumannii bacteremia. The endolysin belonging to other families can also lyse A. baumannii [154,155]. By comparing the engineered phage expressing depolymerase with the phage without depolymerase, it was found that the engineered phage significantly destroyed the biofilm of E. coli [66]. After Zhang et al. treated A. baumannii with bacteriophage AB3 and its LysAB3 for 24 h, the A. baumannii biofilm was significantly degraded [156].
As bacteriophages continue to be discovered and characterized to dissolve biofilmrelated proteins, numerous studies have reported the use of bacteriophage endolysins or depolymerase for the efficient elimination of A. baumannii [157][158][159]. These antibacterial proteins do not have the advantage of self-replication, but are more difficult to tolerate than phages and less likely to elicit the body's immune defense. Similar to bacteriophages, bacteriophage protein antimicrobials work better with antibiotics. In the clinical eradication strategy of bacteria, the use of phage-associated protein antibacterial strategies has great potential and application prospects.

Transcriptomic Analysis
Transcriptomics investigates gene transcription and its regulation at the overall level and detects gene expression at the RNA level. At present, most of the research on the action of A. baumannii phage on A. baumannii is limited for characterization and local protein mechanism research, and the global analysis of phage and host interactions is lacking. Transcriptomic alignment analysis can be used to better understand the effect of phages on bacteria and provide a reference for the development of phage therapy [160].
Yang et al. analyzed the differential expression of genes after the infection of host AB1 by A. baumannii phage ϕAbp1 using global transcriptome analysis. The alignment showed only 15.6% (600/3838) of the host genes were differentially expressed; that is, only a small part of the bacterial genome can be used to complete phage reproduction. Through gene co-expression networks, an intermediate protein, gp34, was found to be involved in RNA polymerase synthesis and negatively interacts with many host ribosomal protein genes, which suggests that gp34 may be a key gene that inhibits or shuts down the translation process of the host. This also means that the replication and translation process of the phage overtakes the host cell. In phage late proteins, downregulation of both phage assembly and biosynthesis resulted in a lack of the expression of phage structural proteins. In addition, upregulation of host proteolytic proteins was observed, suggesting that the host has a defense mechanism against phages. Based on GO and KEGGplus network analysis, despite the challenge of the host stress response, the regulation of phages remains precise and powerful. In bacteria, phage infection suppresses the expression of relevant virulence genes, but activates resistance genes. The underlying mechanism of this phenomenon is still unclear, but it presents new challenges for the clinical use of phage-antibiotic combination regimens [161].
The global analysis of transcriptomics provides a basic research direction for studying the interaction between phages and bacteria. At the same time, data analysis based on transcriptomics can judge the treatment potential of phages from a more holistic perspective, to analyze the changes in virulence and drug resistance after bacterial infection and select appropriate phages to formulate treatment plans. At present, transcriptomic analyses of A. baumannii phages and their hosts are very scarce. Although many reports have verified the antibacterial potential of phages in vitro or in vivo, there are no data to analyze the changes in phages and bacteria globally [162][163][164][165][166][167][168][169]. This brings a huge workload to the clinical formulation of phage therapy or phage-antibiotic combination therapy.

Bacteriophage Resistance Mechanism of A. baumannii
In the tens of millions of years of struggle between bacteria and phages, bacteria have evolved a variety of defense systems [170] to resist phage infection, and anti-phage mutant strains have been found constantly. Ambroa A et al. conducted whole-genome sequencing analysis on 18 clinical strains of A. baumannii and identified 118 to 171 genes related to phage resistance, including the abortion infection (Abi) system, CRISPR-Cas system, and restricted-modification (R-M) system [171]. Their research suggests that the number of genes associated with resistance to phages may be growing year by year ( Figure 5).
Phage therapy, as an alternative therapy for multi-resistant A. baumannii (MRAB), extensive drug-resistant A. baumannii (XDRAB) and pandrug-resistant A. baumannii (PDRAB) infections, has gradually become the focus of attention. A more comprehensive understanding of the resistance strategy of A. baumannii against phage attacks is helpful to evaluate the antimicrobial ability of phages. In addition, according to the molecular mechanism of bacterial and phage resistance to each other, it is helpful to develop new phage therapies to solve the increasingly complex problem of antibiotic resistance and avoid the influence of the bacteriological anti-phage strategy on phage therapy. Viruses 2023, 15, x FOR PEER REVIEW 12 of Phage therapy, as an alternative therapy for multi-resistant A. baumannii (MRAB extensive drug-resistant A. baumannii (XDRAB) and pandrug-resistant A. baumann (PDRAB) infections, has gradually become the focus of attention. A more comprehensiv understanding of the resistance strategy of A. baumannii against phage attacks is helpf to evaluate the antimicrobial ability of phages. In addition, according to the molecul mechanism of bacterial and phage resistance to each other, it is helpful to develop ne phage therapies to solve the increasingly complex problem of antibiotic resistance an avoid the influence of the bacteriological anti-phage strategy on phage therapy.

Adsorption Inhibition
Inhibition of phage adsorption by phage receptor deletion or structural change is th first step taken by A. baumannii to defend against phage infection. Phages are recognize with host surface receptors by receptor-binding proteins (RBPs) typically located at th end of phage tail fiber [73]. The capsule of A. baumannii is the most common receptor fo phage [79,166,172,173].
Phages ΦFG02 and ΦCO01 can infect A. baumannii strains AB900 and A9844, respe tively. Gordillo Altamirano F et al. found that both strains, AB900 and A9844, had sing nucleotide deletions at the K locus (capsule biosynthesis locus) [174] after coincubatio with phages, which affected the gene gtr29 encoding transferase and the gene gpi encod ing glucose-6-phosphate isomerase, respectively. This resulted in the production phage-resistant mutants, ΦFG02-RAB900 and ΦCO01-RA9844, with missing capsules. A a result, the adsorption of phages ΦFG02 and ΦCO01 to the strain was interrupted, whic hindered the infection of the phage with A. baumannii [79]. Similarly, single-base deletion in the gtr6 gene, which encodes glycosyltransferase in A. baumannii, AB5001, resulted in change in the structure of the K3-type capsular polysaccharide (CPS), thereby inhibitin the ability of the phage to infect the bacteria [172]. The adsorption test proved that th capsular membrane was the main receptor of phage Phab24, while the outer membran was the secondary receptor [166]. Wang X et al. obtained phage-resistant mutant strain

Adsorption Inhibition
Inhibition of phage adsorption by phage receptor deletion or structural change is the first step taken by A. baumannii to defend against phage infection. Phages are recognized with host surface receptors by receptor-binding proteins (RBPs) typically located at the end of phage tail fiber [73]. The capsule of A. baumannii is the most common receptor for phage [79,166,172,173].
Phages ΦFG02 and ΦCO01 can infect A. baumannii strains AB900 and A9844, respectively. Gordillo Altamirano F et al. found that both strains, AB900 and A9844, had single nucleotide deletions at the K locus (capsule biosynthesis locus) [174] after coincubation with phages, which affected the gene gtr29 encoding transferase and the gene gpi encoding glucose-6-phosphate isomerase, respectively. This resulted in the production of phageresistant mutants, ΦFG02-RAB900 and ΦCO01-RA9844, with missing capsules. As a result, the adsorption of phages ΦFG02 and ΦCO01 to the strain was interrupted, which hindered the infection of the phage with A. baumannii [79]. Similarly, single-base deletions in the gtr6 gene, which encodes glycosyltransferase in A. baumannii, AB5001, resulted in a change in the structure of the K3-type capsular polysaccharide (CPS), thereby inhibiting the ability of the phage to infect the bacteria [172]. The adsorption test proved that the capsular membrane was the main receptor of phage Phab24, while the outer membrane was the secondary receptor [166]. Wang X et al. obtained phage-resistant mutant strains with the changed structure of the capsular and extracellular membrane due to gene mutations, by coculturing phage Phab24 and A. baumannii. Importantly, both studies found that phage-resistant mutant strains were resistant to antibiotics due to the absence of a bacterial capsule. Among them, Gordillo Altamirano F et al. demonstrated that phage-resistant strains ΦFG02-RAB900 and ΦCo01-RA9844 were more sensitive to cephalosporins and β-lactams. They also found that strains resistant to Phab24 increased their sensitivity to colistin [166]. These findings support the combination of bacteriophages and antibiotics in the treatment of A. baumannii infections.

CRISPR-Cas in Acinetobacter baumannii
CRISPR-Cas consists of clustered regularly spaced short palindromic repeats (CRISPR) and CRISPR-associated (cas) genes, and is an acquired immune system [175]. It can be divided into the class 1 CRISPR-Cas systems (including three types, I, III and IV), where the effector complex consists of multiple Cas proteins, and the class 2 CRISPR-Cas systems (including three types, II, V and VI), where the effector complex consists of a single Cas protein [176]. The CRISPR array consists of the spacer from exogenous gene sequences and repeat sequences. CRISPR-Cas systems insert the protospacers of phage DNA into the CRISPR array. When the phages invade again, CRISPR-derived RNA (cr-RNA) induces the Cas proteins to cut the target genes, thus preventing the phage from infesting the bacteria [177].
There is an evolving and complex CRISPR-Cas system in the genome of A. baumannii [122]. At present, the I-F CRISPR-Cas system has been found in A. baumannii. Tyumentseva M et al. compared the strains carrying the I-F1 and I-F2 CRISPR-Cas systems, and found that the A. baumannii strain with the I-F2 system had more CRISPR arrays and stronger resistance to phage invasion than the A. baumannii strain with the I-F1 system [178]. In addition, they found that the CRISPR-Cas system appeared to be associated with virulence factors, with a higher proportion of virulence genes in strains that lacked the CRISPR-Cas systems. Karah N et al. investigated the CRISPR-Cas isoform I-Fb locus in 76 A. baumannii isolates from 14 countries and found that the locus was derived from a common ancestor. In addition, they demonstrated that CRISPR-based methods could be used for A. baumannii. These isolates can be divided into 40 CRISPR-based sequence types (CSTS), with the CST1 isolate originating from Iraq and three CST19 isolates originating from Thailand [179]. In addition, the CRISPR-Cas system belonging to the CST8 subtype was also detected in the bla OXA-23containing A. baumannii ST409 isolated from Greece by Galani V et al. [180]. Mangas EL et al. performed genomic analysis of nearly 2500 A. baumannii and demonstrated that a group of strains with fewer plasmids (717 strains) had more CRISPR array and cas genes. This may be because the CRISPR-Cas system prevents plasmids from entering the bacteria. Moreover, they also found that this group of strains had more genes associated with the biomembrane formation [181].
Through the CRISPR-Cas system, bacteria can not only resist phages, but also obtain more ARGs. Some studies have shown that A. baumannii isolates with the CRISPR-Cas system usually carry more ARGs and have a stronger drug resistance [182]. At the same time, with the specific recognition property of the CRISPR-Cas system, it can be used as a tool to knock out, modify or silence drug-resistance genes in the bacterial genome, which is a new scheme for treating drug-resistant bacteria [183,184]. However, not all genes associated with CRISPR-Cas contribute to bacterial resistance. To explore the relationship between the csy1 gene in the type I-Fb CRISPR-Cas system and A. baumannii resistance, Guo T et al. knocked out the csy1 gene of A. baumannii AB43, and found that the AB43∆csy1 mutant showed antibiotic resistance. Transcriptome analysis proved that the csy1 gene could lead to the antibiotic sensitivity of A. baumannii by regulating related genes, such as the Cas protein and efflux pumps [15]. Although the CRISPR-Cas system is ubiquitous in A. baumannii, not all A.-baumannii-resistant strains have the CRISPR-Cas system. For example, A. baumannii NCIMB8209, which contains bla OXA-51-like genes on its chromosomes, has a low ability to form biofilms and membranes and a slow movement rate. Wholegenome sequencing analysis showed that the NCIMB8209 genome does not contain the CRISPR-Cas system [185].

Quorum Sensing
Quorum sensing (QS) is an intercellular communication system that can coordinate population density and regulate gene expression through QS signals, which helps A. baumannii respond quickly to environmental changes [186]. QS can promote the production of A. baumannii biofilms [187], which can protect bacteria from phage infection by shading phage receptors and other ways. A. baumannii possesses an abaI/abaR QS system, with n-acyl-homoserine lactone (AHL) as a signaling molecule, which consists of an AHL synthase (encoded by abaI) and a transcriptional regulator (encoded by abaR). The abaI and abaR genes are thought to have been obtained by A. baumannii from Thiobacillus neapolitanus by horizontal gene transfer [188]. The abaI/abaR QS system is widely present in A. baumannii. Tang J et al. collected 80 clinical isolates of A. baumannii from Jilin Province, China, from 2012 to 2017; 61 strains carried abaI and abaR genes, 24 of the 61 strains could produce AHL, and 76 of the 80 strains showed the ability to form biofilms [189]. In addition, they found that the abaI and abaR genes were positively correlated with bacterial drug resistance rates and surface-related motility. AbaM is a gene located between abaR and abaI; Lopez-Martin M et al. confirmed that ∆abaM mutant strains formed three times more biofilms than wild-type strains, and increased AHL production, indicating that abaM is a negative regulator of AHL production and biofilm formation [190]. QS also plays a role in the regulation of bacterial virulence factors. Sun X et al. evaluated the virulence of A. baumannii ATCC17978 with the abaI/abaR QS system. They found that ∆abaI and ∆abaIR mutant strains were completely nonvirulent, while the ∆abaR mutant remained virulent [191].

Other Mechanisms of Bacterial Resistance to Phages
Restriction-modification (R-M) systems that protect bacteria from bacteriophage infection can be classified into four categories: I−IV. The R-M system consists of genes encoding methyltransferase (MTase) and restriction enzyme (REase). Among them, MTase methylates host genome recognition sites, and REase recognizes unmethylated foreign genes and cleaves foreign DNA [192]. The R-M systems are the common defense system in bacteria, and Ambroa et al. identified genes associated with the R-M system in A. baumannii strains Ab2000 and Ab2010 [171]. When the phage gene injects into the host cell, the REase of the R-M system cleaves the phage gene by recognizing specific sites, while the host gene is not recognized by the Rease due to methylation by Mtase, thus resisting the phage infection [193].
Another bacteriophage-resistant strategy is abortive infection (Abi), in which bacteriophageinfected bacteria lyse their cells before phage maturation to protect other bacteria from infection [194].
The toxin-antitoxin (TA) system is widely present in A. baumannii [195][196][197] and is composed of toxin proteins and antitoxin proteins or antitoxin noncoding RNA [198]. Among them, toxins can regulate many important cellular processes in cells, which limits the growth of bacteria [199], and antitoxins can inhibit the activity of toxins. The antitoxin is unstable and easily degraded by protease. TA is believed to be associated with the resistance of A. baumannii [200] and is considered a promising drug target [201]. TA has also been shown to inhibit the biosynthesis of A. baumannii cell walls [202]. In addition, type IV TA has been found to mediate Abi, and thus play an anti-phage role [203]. Compared with A. baumannii, the anti-phage effect of TA has been studied more deeply in Escherichia coli [204,205]. Song S et al. found that the type I TA system Hok/Sok could inhibit phage T4 infection, and the possible mechanism of action is that phage T4 blocks the host transcription process, thereby blocking the production of Hok and Sok (antitoxin Sok is the single-stranded RNA). Because the antitoxin Sok is unstable and degraded first, the toxin Hok can play an anti-phage role [206].
At present, the CRISPR-Cas system is the most deeply studied anti-phage strategy in A. baumannii. However, R-M systems [207], Abi [208] and TA [206] systems, which have been extensively studied in other bacteria, have been studied relatively little in A. baumannii. However, it is believed that with increasing attention to phage therapy, research on the defense mechanism of bacteria against phages will continue to deepen, and the more specific action mechanism of A. baumannii against phage systems will be gradually clarified.

Phage Therapy for Drug-Resistant A. baumannii
With the abuse of antibiotics, MDR bacterial infections have emerged as a potential danger. Phages have been widely reported as a natural antibacterial agent, being used to deter the spread of MDR A. baumannii and to treat infections caused by A. baumannii [10,209]. Depending on the antibacterial mechanism, phage therapy can be classified as phage single therapy, phage cocktail therapy [59], the combination of phages with antibiotics, and application of phage products (lysozyme, depolymerase) ( Figure 6).
A. baumannii. However, R-M systems [207], Abi [208] and TA [206] systems, which have been extensively studied in other bacteria, have been studied relatively little in A. baumannii. However, it is believed that with increasing attention to phage therapy, research on the defense mechanism of bacteria against phages will continue to deepen, and the more specific action mechanism of A. baumannii against phage systems will be gradually clarified.

Phage Therapy for Drug-Resistant A. baumannii
With the abuse of antibiotics, MDR bacterial infections have emerged as a potential danger. Phages have been widely reported as a natural antibacterial agent, being used to deter the spread of MDR A. baumannii and to treat infections caused by A. baumannii [10,209]. Depending on the antibacterial mechanism, phage therapy can be classified as phage single therapy, phage cocktail therapy [59], the combination of phages with antibiotics, and application of phage products (lysozyme, depolymerase) ( Figure 6).

Phage Single Therapy
Phages used as antimicrobial agents need to have the characteristics of a broad host range, short latent period, high burst size, and limited frequency of resistant bacteria [56,210]. Single-phage treatment is primarily used as a proof-of-concept for phage agents during design and testing. Although single-phage treatment has been successfully tested in animal models, data on clinical treatment are still lacking [211,212]. Phage vB_ABAP_PD-6a3, reported by Minle Wu et al., showed a lysis rate of approximately 32.4% against clinically resistant A. baumannii, which was approximately three times higher than the usual phage lysis rate [213]. However, the therapeutic effectiveness of single phages in the clinic may be reduced due to the existence of an arms race between

Phage Single Therapy
Phages used as antimicrobial agents need to have the characteristics of a broad host range, short latent period, high burst size, and limited frequency of resistant bacteria [56,210]. Singlephage treatment is primarily used as a proof-of-concept for phage agents during design and testing. Although single-phage treatment has been successfully tested in animal models, data on clinical treatment are still lacking [211,212]. Phage vB_ABAP_PD-6a3, reported by Minle Wu et al., showed a lysis rate of approximately 32.4% against clinically resistant A. baumannii, which was approximately three times higher than the usual phage lysis rate [213]. However, the therapeutic effectiveness of single phages in the clinic may be reduced due to the existence of an arms race between resistant bacteria and phages [214]. Cocktail therapy is one of the most common methods in clinical treatment, as a single phage cannot always lyse various bacteria in the clinic.

Phage Cocktail Therapy
Cocktail therapy is a personalized treatment method that is often designed based on the specific conditions of patients. It targets a single bacterium or bacteria. Cocktail therapy is effective in preventing the emergence of resistant bacteria through a train of agents of different phages compared to simultaneous agents of phage mixtures [215]. Even if resistant bacteria appear during phage therapy, bacteria will be targeted by new phages that are added in the course of treatment. This strategy keeps the population density of bacteria at a low level for a long time and curbs the emergence of resistant bacteria [216].

Synergistic Effect of Bacteriophages and Antibiotics in A. baumannii
The main obstacle of bacteriophage therapy is the tolerance of bacterial evolution, which is faster than bacteriophage evolution. However, the evolution of bacteria to avoid phage lysis has resulted in certain sacrifices, including restoring sensitivity to antibiotics [217]. Based on this characteristic, the dual survival pressure exerted by the combination of bacteriophages and antibiotics can greatly slow the evolution of bacteria and provide sufficient clinical treatment time [168,169]. The ϕAB182 combined with antibiotics can eliminate the biofilm of MDR A. baumannii [218]. Therefore, the synergistic effect of bacteriophage and antibiotics on MDR bacteria is the most used and effective treatment strategy in clinical practice. A 42-year-old man's left tibia was infected with MDR K. pneumoniae and MDR A. baumannii in a car accident. After treatment with the combination of antibiotics (colistin, Romepenem) and bacteriophage (bacteriophage AbKT21phi3, bacteriophage KpKT21phi1), the drug-resistant bacteria were completely eliminated, and the patient's wound cured quickly [219]. The synergistic effect of bacteriophage and antibiotics has been proven to improve therapeutic efficiency in vivo and in vitro [51]. Pharmacological synergy occurs when the combined effect of multiple drugs is greater than the effect of a single drug, and it is typified by certain adjuvants that can block resistance mechanisms or improve drug pharmacokinetics. Antibiotics with appropriate concentrations can significantly improve the amount of phage lysis of bacteria. The reason is that lower concentrations of antibiotics inhibit the division of bacteria; therefore, the amount of biosynthesis of bacterial cells can be increased at this concentration [220,221]. In addition, antibiotics can promote phage lysis of bacteria, as morphological changes in bacteria cause changes in the peptidoglycan layer, which may lead to a higher sensitivity of bacteria to phage lyases (lysozymes, holins) [222]. A new approach was developed for the evaluation of A. baumannii phage-antibiotic synergic function [223]. The synergistic effect of these two effects (an increase in the number of phages and an improvement in the rate of bacterial lysis by phages) leads to superior results when phages and antibiotics work synergistically against drug-resistant bacteria.

Applications of Phage Enzymes
The bactericidal mechanism of phage enzymes is different from the mechanism of other agents. Compared to phages and antibiotics, phage enzymes are able to lyse a broader range of pathogenic bacteria without causing drug resistance. It is speculated that in the future, agents related to phage enzymes will be widely used for biological control, a strategy that will not only effectively contain the growth and spread of pathogenic bacteria, but also reduce the economic burden for patients [224]. The phage Petty, isolated by A.C. Hernandez-Morales et al., resulted in a clear halo around the center of the plaques, which suggests that phage Petty encodes a depolymerase that is capable of degrading the extracellular polysaccharides (EPS) of bacteria [152,225]. The enzyme was demonstrated to degrade the EPS of its host strain, AU0783, and reduce its viscosity. When the EPS of the bacteria is degraded, it will increase the accessibility of phage or other antimicrobial agents to the material inside the biofilm, and improve the efficiency of the bactericide. Hugo Oliveira et al. isolated a lytic phage, vB_AbaM_B9, of which ORF69 encodes a depolymerase (B9gp69) that degrades the EPS of A. baumannii ST-K30 and ST-k45, and does not cause drug resistance for the related bacteria [82]. Lysin packaged in hydrogel dressing is a safe and effective cure [226]. Furthermore, phage enzymes have some commercial value, as they can be protected by patents (which is different from phages). This advantage may facilitate the development and application of phage products in medicine, animal husbandry and food safety in the future [77,78,227].

Discussion
Phages are abundant in nature, and they can be obtained without complicated experimental manipulation and equipment, for experimental research or clinical practice. The technology of bacteriophage isolation is simple, and the experimental cost is low, which provides a basis for the industrial or clinical rapid formulation of personalized treatment.
In vivo, bacteriophages can replicate and reproduce in their hosts, maintain high titers and reduce the adverse effects of multiple administrations.
Currently, phage therapy offers significant advantages in the context of a pandemic of MDR A. baumannii. However, there are also some limitations. The strong host specificity of phages makes it difficult to obtain pre-prepared phage cocktails, which need to be personalized for different patients [228], with a current high cost of the treatment [229]. First, in the absence of widespread theoretical knowledge of phage therapy, many patients disagree with the use of phages for treatment. This is one of the reasons why phage therapy is not approved by the FDA and EMA as a clinical drug treatment [230]. Furthermore, the uncertainties in the mode of administration, route, efficiency, timing, pharmacokinetics and immunomodulatory mechanisms of in vivo treatment of phages partially deter the clinical application of phage therapy [217]. In addition, the treatment of Gram-negative drug-resistant bacteria also requires consideration of endotoxin release, and the response of each person to endotoxin is very variable [231][232][233]. The survey data of Lynn El Haddad et al. showed that bacteriophages were effective (87%) and safe (67%), but only 35% of the researchers focused on the specific reasons why bacteriophages were resistant to host bacteria [234].
With the rapid spread and epidemic of MDR pathogens, it has become apparent that the development of antibiotics has failed to keep pace with the evolution of drugresistant bacteria, and attention is being paid to phage therapy worldwide. Due to the diversity of phages, there are few studies of universal mechanisms based on phage and bacterial resistance, which has troubled in vivo studies of phage therapy. In the meantime, researchers are working individually and there is no effective organization to form a public phage bank to share resources and help build a system for the rapid preparation of personalized phages.
The maturation of next-generation sequencing technology and the development of genome editing technology provide the basis for phage modification and phage therapy. Based on second-generation sequencing technology, an abundant source of phages that exist in nature will be further exploited and utilized to provide a large number of candidates for clinical treatment. The development of gene editing technology enables further optimization of phages to be constructed, with engineered phages that meet the desired targets. Computer technology or artificial intelligence (AI) technology can also advance phage therapy based on phage genomic information, physiological properties, receptor binding and other information. In addition, the application of phage therapy in complex infections is promising. It has been shown to be effective in the treatment of patients with COVID-19 and secondary CRAB infection [235]. Alongside that, phage formulations are effective against challenging intracellular infections [236].
In this paper, we reviewed the current epidemiology and resistance mechanisms of drug-resistant A. baumannii, and described and discussed the current development of phages against drug-resistant A. baumannii. With the rapid spread of clinical MDR A. baumannii, phage therapy is a promising treatment for MDR A. baumannii infections. The combination of next-generation sequencing, gene editing and AI technologies makes it feasible to customize and optimize personalized phages, and there is no doubt that this will facilitate advances in phage therapy.