DNA or amino acid sequences of BIT33_RS14560 gene or protein (corresponding Accession Number: CP058289.1 or WP_002047564.1) and its homologs were searched using BLAST at the (NCBI) website¹ and downloaded in a separate FASTA format. These sequences were analyzed to construct phylogenetic trees to determine the phylogenetic relationships between the selected strains (Supplementary Tables S1, S2). The best DNA or protein model (GTR + G + I) with the highest parameter was determined and chosen before the construction of phylogenetic trees. Dendrograms were generated by Maximum Likelihood (ML) with bootstrap values corresponding to 1,000 replications using the (MEGA) 11.0 software (Tamura et al., 2021).

Bacterial strains and plasmids used in this study are listed in Table 1. Of 62 A. baumannii isolates, 43 strains (69.4%) including two XDR-Ab isolates, XAb53 (Strain No. A53; Stored on June 2, 2021) and XAb50 (Strain No. A50; Stored on May 19th, 2020), were isolated from the First Affiliated Hospital of Sun Yat-sen University, the other 19 strains (30.6%) were obtained from the Microbiology Laboratory of State Key Laboratory of Respiratory Diseases, Guangzhou Medical University. Sources and proportion of 62 A. baumanii isolates used in this study were presented in Supplementary Table S3. Escherichia coli DH5α competent cells (Cat. No. 9057; TaKaRa) were purchased from Guangzhou Ruizhen Biotechnology Co., Ltd. The suicide plasmid pMo130-telR (Cat. No. P4951-ea; Wonder Biotech) was purchased from Guangzhou Qunlan Biotechnology Co., Ltd. Bacterial culturing was performed using Luria-Bertani (LB) agar medium (Cat. No. CM0337B; OXOID) or LB broth medium (Cat. No. CM0405; OXOID).

Overnight culture of A. baumannii strain was inoculated into fresh LB broth until the mid-log phase (OD = 0.6–0.8) was reached. Total RNA was extracted using the E.Z.N.A.® Bacterial RNA Kit (Cat. No. R6950-01; Omega). The quality and concentration of RNA were determined via spectrophotometry (NanoDrop 2000c; Fisher Scientific, United States). For reverse transcription PCR (RT-PCR), reverse transcription was performed using ProFlex PCR System cycler (Applied Biosystems, United States). The purified RNA (1 μg/ml) was reverse-transcribed into cDNA by the Evo M-MLV RT Premix for qPCR kit (Cat. No. AG11706; Accurate Biology), in a 20 μl reaction mixture containing 4 μl of Evo M-MLVRT Master Mix and 15 μl of RNAase free water, which was incubated for 15 min at 37°C and 5 s at 85°C. Then, the qRT-PCR assay was performed on a Roche LightCycler® 960 real-time PCR system. Here, SYBR Green Premix Pro Taq HS qPCR kit (Cat. No. 78 AG11702; Accurate Biology) was employed. Around 2 μl of the resulting cDNA products were put into a 20 μl reaction mixture containing 10 μl of SYBR® Green Pro Taq HS Premix II, 0.8 μl (or 10 μM) of each primer and 6.4 μl RNase free water. The parameters of PCR were set as followed: initial heat activation at 95°C for 30 s, followed by 35 cycles of the two-step cycling condition of denaturation at 95°C for 10 s and then annealing at 60°C for 30 s. The melting data of PCR products were measured and collected following the completion of PCR cycling. The rpoB gene was used as an internal reference for normalization. △CT value, defined as the CT value of the target gene minus that of the reference rpoB, was used to calculate the BIT33_RS14560 mRNA levels of A. baumannii. For each sample, three technical replicates were performed and the gene transcript levels were determined using the 2^-△△Ct method. Primers used in this part are listed in Table 2.

All primers used in this study are listed in Table 2. PCR amplification was achieved using DNA polymerase (Cat. No. AG12202; Accurate 79 Biology). A plasmid DNA extraction kit (Cat. No. D2156-01; Omega), a DNA fragment purification kit (Cat. No. 9761; 77 TaKaRa), and a DNA Ligation kit (Cat. No. 6023; Accurate Biology) were also employed. The novel BIT33_RS14560 gene was PCR-amplified from the genomes of A. baumannii ATCC 19606 using specific primers RS14560-EcoRI-F and RS14560-6his-BamHI-R with the following conditions—initial hold at 94°C for 4 min, followed by 32 cycles of 98°C for 10 s, 55°C for 5 s, 72°C for 8 s, and final extension at 72°C for 7 min. Purified PCR products and pWH1266 plasmid were digested with BamHI and EcoRI, then, according to the manufacturer’s instructions, ligated to make a recombinant plasmid of overexpression, named as p-RS14560. After digesting with EcoRI and BamHI, blunt ends to the large fragments of pWH1266 plasmid were created with high-fidelity DNA polymerase and then self-ligated to make into an empty plasmid, known as p. Each plasmid was transformed into E. coli DH5α competent cells and then introduced into the A. baumannii ATCC 19606 by electroporation (1.75 kV). After overnight culture (16–18 h) at 37°C, monoclonal colony PCR was performed with Pwh1266-F and Pwh1266-R primers to select positive transformants on LB agar supplemented with 50 μg/ml Carbenicillin.

For the overexpression of XDR-Ab, based on the above work, we reconstructed a specific recombinant plasmid by replacing the Ampicillin resistance (AmpR) cassette in the p-RS plasmid with the Tellurite resistance (TelR) cassette. Briefly, the TelR cassette was amplified from the plasmid pMo130-telR using telR-PstI-F/telR-PvuI-R primers. After purification, these PCR products and p-RS plasmid were digested with PstI and PvuI then ligated to make a new plasmid p-telR-RS14560. Similarly, a corresponding empty plasmid p-telR was constructed according to the above-mentioned method. Each plasmid was transformed into E. coli DH5α competent cells and then electroporated into A. baumannii XAb53 and XAb50 isolates at the same voltage of 1.75 kV. Of particular note, when selecting positive transformants with telR-F/telR-R primers, the concentration of tellurite added into the LB agar should not be completely fixed and rigid, that is, 25 μg/ml for E. coli DH5α and 80 μg/ml for XAb53 or XAb50 isolates. Of note, these two XDRAB clinical isolates were randomly selected according to their results of multilocus sequence typing, that was, XAb50 was assigned to ST1145 (gltA-1, gyrB-3, gdhB-3, recA-102, cpn60-2, gpi-97, and rpoD-3), an ST recently prevalent at the First Affiliated Hospital of Sun Yat-sen University, while XAb53 was considered a new ST (gltA-1, gyrB-3, gdhB-3, recA-102, cpn60-1, gpi-103, and rpoD-26). The media should be cultured overnight for 24–30 h at 37°C.

We performed a microtiter-plate test to evaluate the biofilm formation ability of A. baumannii according to the method described previously (Rodríguez-Baño et al., 2008), but with certain modifications. Prior to inoculation, all strains were transferred from the stock cultures onto LB agar plates and incubated aerobically at 37°C for 18 h. Colonies from these plates were suspended in phosphate-buffered saline (PBS). After a 0.5 McFarland turbidity standard suspension (corresponding to 1.5 × 10⁸ CFU/ml) of each isolate was determined, a 100-fold dilution with sterile LB broth was prepared. Then, 200 μl of each bacterial suspension were filled into no less than six wells of a sterile 96-well flat-bottomed plastic tissue culture plate with a lid. Negative and positive control wells contained sterile LB broth and standardized suspension of P. aeruginosa strain ATCC 27853, respectively. The plates were covered and incubated aerobically at 37°C for 0, 6, 12, 18, 24, 30, 36, 42, 48, 60, and 72 h, respectively. Then, the content of each well was gently aspirated, and each well was washed three times with 250 ml of ultrapure deionized water and left to dry. Each well was stained for 20 min at room temperature with 200 μl of 10 g/L crystal violet. Excess stain was rinsed off by sterile water. After the plates were air-dried, each well was added with 200 μl of 33.3% glacial acetic acid and resolubilized for 20 min. Absorbance at 570 nm (OD570) of each well was measured by Epoch 2 microplate reader (BioTek). A cut-off OD (ODc) was defined as three standard deviations (SDs) above the mean OD of the negative control. All strains were classified into the following categories (Stepanović et al., 2007): non-adherent (OD ≤ ODc, 0), weakly adherent (ODc < OD ≤ 2 × ODc, +), moderately adherent (2 × ODc < OD ≤ 4 × ODc, ++), or strongly adherent (4 × ODc < OD, +++). The results were averaged after all tests were carried out three times.

Bacterial cell suspensions at a concentration equivalent to a 0.5 McFarland Standard were prepared and followed by a 100-fold dilution with sterile LB broth. To make an infection model, the last instar healthy larvae with 2–3 cm long and 180–250 mg in weight were selected and randomly divided into groups with 10 individuals in each group. Each insect was carefully given an injection with 20 μl of bacterial solution (1.5 × 10⁶ CFU/ml) and put in a Petri dish of 8 cm in diameter. Infect at least 30 larvae for each strain. Then place the dishes at 37°C in an incubator and check larval mortality regularly over a 6 h period for 3 days. A. baumannii 5075, a strain with high virulence, was served as the positive control while PBS served as the negative. Mortality data were analyzed using the Kaplan–Meier method.

Four single colonies, two derived from the overexpressed strain constructed by ATCC 19606, and two from the corresponding empty plasmid strain, were picked into 3 ml LB medium supplemented with 50 μg/ml Carbenicillin and inoculated, respectively. These cultures were grown for 18 h at 37°C and cells were then collected by centrifugation at 6,000 rpm for 10 min and washed three times with RNase free water. Total RNA extraction and the subsequent transcriptome sequencing were performed by Shanghai Majorbio Bio-pharm Technology Co., Ltd. After the quality analysis, clean data were obtained and analyzed on the online platform of Majorbio Cloud Platform.² Basic functional annotation analysis was performed on the provided reference genome (Organism: A. baumannii; Genome ID: GCF_008632635.1). The sequences were then annotated with the following databases including NCBI Nr, NCBI Nt, Pfam,³ eggNOG⁴, SwissProt⁵, GO⁶, and KEGG.⁷ Differentially expressed genes (DEGs) were identified by DESeq2 software, Version 1.24.0.⁸ Those mRNAs with absolute value log2[Fold Change (FC)] ≥ 1 and p-adjust < 0.05 were considered as significant.

The IBM SPSS 26.0 statistical software was used for statistical analysis. The t test was applied to compare the difference of measurement data, which were presented as mean ± SD. Spearman’s rank correlation was conducted to analyze the correlations between biofilm formation and gene expression levels. The Kaplan–Meier method was used to construct a survival curve, with a calculation of p value by the log-rank test. p < 0.05 was considered to be statistically significant.

In the case of the DNA sequences, BIT33_RS14560 gene widely existed in all 14 A. baumannii strains with different identities and two clades were identified in the phylogenetic tree (Supplementary Figure S1). In the monophyletic clade I, the BIT33_RS14560 gene was found to most closely related to A. baumannii ATCC 17961 and they both clustered into a subgroup with other five A. baumannii isolates. Interestingly, the pattern strain A. baumannii ATCC 17978 was subgrouped with another six isolates into clade II, indicating a distant relationship with A. baumannii ATCC 19606.

When it came to the amino acid sequences, the phylogenetic tree showed that the BIT33_RS14560 membrane transporter was most closely related to A. baumannii WP_0657191181 (Supplementary Figure S2). In addition, this protein of interest was clustered into a subgroup with its homologous proteins A. baumannii HAV5588984.1, A. baumannii HAV6132876.1, A. baumannii HAV4460275.1, and Klebsiella pneumoniae SSW87290.1. What is more, the target protein was constituted into an evolutionary clade together with Acinetobacter nosocomialis WP_004709884.1, A. nosocomialis HAB71326.1, A. nosocomialis WP_207694075.1 as well as A. baumannii HAV5002142.1.

In this study, when biofilm biomass was measured at 72 h by crystal violet staining, 62 A. baumannii isolates showed a various range of OD570 from 0.09 to 2.29 nm, indicating varying abilities of biofilm formation. Among all the isolates, 9 (14.5%), 28 (45.2%), and 17 (27.4%) were strong, moderate, and weak biofilm producers, respectively, while no biofilm was observed in eight (12.9%) isolates. Interestingly, we found that these A. baumannii isolates also differed considerably in their mRNA levels of BIT33_RS14560, with the △CT ranging from 0.47 to 19.51 (Table 3). Spearman’s rank correlation analysis indicated that the △CT value of BIT33_RS14560 was negatively correlated with the biofilm rank of those A. baumannii isolates (R = − 0.569, p = 0.000), which suggested that a stronger biofilm-forming ability could be followed by a relative higher the expression level of BIT33_RS14560 (Supplementary Table S4).

Fragment of BIT33_RS14560 coding sequence (CDS) plus its preceding promoter was cloned from the ATCC 19606 genome and 1% agarose gel electrophoresis of the PCR products yielded an expected band, 1,632 bp in size. The purified PCR products were inserted into pWH1266 to construct a vector p-RS14560 for the overexpression of BIT33_RS14560 in ATCC 19606. PCR analysis of the cloned plasmid followed by gene sequencing showed that the target fragment was successfully and correctly cloned into pWH1266 plasmid and the cloned gene was the same as BIT33_RS14560 (NCBI Reference Sequence: NZ_MJHA01000008.1) in the GenBank. These findings indicated that the expression vector p-RS14560 was successfully constructed (Supplementary Figures S3, S4). A 3,075 bp TelR cassette was successfully cloned from the suicide vector pMo130telR and inserted into the p-RS14560 plasmid to construct a new vector p-telR-RS14560. These were confirmed by obtaining two expected bands, 3,075 and 3,526 bp in size, respectively (Figure 1). Surprisingly, we managed to overexpress the target gene BIT33_RS14560 in two XDR-Ab isolates with this recombinant plasmid (see this article below).

As mentioned above, those clinical A. baumannii isolates with higher BIT33_RS14560 mRNA levels tended to possess a stronger capacity of biofilm formation. To further investigate this relationship, a standard A. baumannii strain ATCC 19606, together with two XDR-Ab isolates, were transformed with the overexpression plasmids, p-RS14560 and p-telR-RS14560, respectively. To confirm if over-expression of BIT33_RS14560 occurred in the ATCC 19606 and these two XDR-Ab isolates, relative quantification of the BIT33_RS14560 transcript levels were determined by qRT-PCR with the corresponding strains transformed with empty plasmids as the control. In ATCC 19606, there was no significant difference between empty plasmid group and wild-type group, indicating that the transformation of two recombinant plasmids had little effect on the expression of BIT33_RS14560 gene (p > 0.05). For ATCC 19606 strain with the plasmid overexpressing the gene of interest, when compared with its control group, the expression level of the BIT33_RS14560 was significantly increased at 311.5 folds (p < 0.0001). For XAb53 strain, the relative expression levels of BIT33_RS14560 in the overexpression group was surprisingly higher at about 586 1000-fold than that of its control group (p < 0.0001). Similarly, for the other XDR-Ab strain, XAb50, the overexpression group significantly displayed approximately 619 1000-fold higher BIT33_RS14560 mRNA levels in comparison with its empty plasmid group (p < 0.0001). These results indicated the success of overexpression of BIT33_RS14560 in the three A. baumannii strains (Figure 2). The breathtaking multiples of overexpression in two XDR-Ab isolates may probably come from their very low transcript levels (or relative high △CT values) of this target gene, as shown in Table 3.

To verify the effect of BIT33_RS14560 overexpression on the biofilm formation of three A. baumannii strains, we evaluated the biofilm biomass with the microtiter-plate test at different points of time. We found that these three wild-type A. baumannii strains, including ATCC 19606 and other two XDR-Ab isolates, differed in their peak time of biofilm formation, but normally displayed relatively high levels of production within 12–48 h. When compared with each control group transformed with empty plasmids, we observed significantly higher biofilm biomass in the experimental groups of BIT33_RS14560-overexpression, whether in the standard strain ATCC 19606 or the two isolates XAb53 and XAb50 (Figure 3).

We used the G. mellonella infection model to investigate the effect on the virulence of the standard strain ATCC 19606 when BIT33_RS14560 was overexpressed. None of the G. mellonella died within 72 h when given with the equivalent PBS. By contrast, all G. mellonella died within 12 h following infection with the high-virulent strain A. baumannii 5075, which served as the positive control. When infected with the ATCC 19606 wild-type strain, G. mellonella mortality was 40.0% within 12 h and then rose to 70.0% at 24 h. In contrast, G. mellonella mortality following infection with the BIT33_RS14560 overexpression strain was 60.0% at 12 h and reached 90% within 24 h. No significant difference was observed in the survival rates between the wild-type strain ATCC 19606 and its empty plasmid strain 19,606-p (p > 0.05). As shown in Figure 4, Kaplan–Meier survival curve analysis showed that the survival rate of G. mellonella was significantly lower when infected with the BIT33_RS14560 overexpression strain (χ² = 8.462, p = 0.004).

To better understand the mechanisms of the relationship between BIT33_RS14560 overexpression and A. baumannii biofilm formation, in the current study, we conducted a comparative transcriptomic analysis of four samples, including the treated group (two samples derived from ATCC 19606 with target gene overexpression) and the control group (two samples derived from ATCC 19606 with empty plasmid). A total of over 7.0 Gigabyte raw data were obtained, and the percentage of Q30 (an important index for the assessment of RNA-Seq quality, a higher value usually indicates a lower probability of base error) reached above 94.65%. Totally, 3,232 expressed genes were detected in this analysis, including 3,208 mRNAs and 24 small RNAs (sRNAs). Under the threshold of |log2(FC)| ≥ 1 and p-adjust < 0.05, a total of 12 DEGs (seven upregulated and five downregulated) were screened for subsequent analysis. The volcano plot and heatmap of these DEGs were drawn in Figures 5, 6, respectively. To further explore the biological functions of these DEGs, we performed GO and KEGG enrichment analysis. As presented in Figure 7, none of these DEGs were involved in GO biological process or cellular component category. However, in molecular function(MF)category, they significantly enriched in “magnesium transmembrane transporter activity, phosphorylative mechanism” (GO:0015444), “3-oxoacid CoA-transferase activity” (GO:0008260), “acetate CoA-transferase activity” (GO:0008775), “ATPase-coupled cation transmembrane transporter activity” (GO:0019829), and “transferase activity, transferring glycosyl groups” (GO:0016757; all FDR<0.05). In terms of the KEGG pathway, “aminobenzoate degradation” (map00627) and “valine, leucine, and isoleucine degradation” (map00280) were involved (FDR < 0.05; Figure 8).

To investigate the distribution of the BIT33_RS134560 gene and its encoding protein in A. baumannii strains as well as their relationship, we conducted phylogenetic tree analysis. Our results showed that this gene existed in all the 14 A. baumannii strains of different genomic homology (Supplementary Figure S1). Similarly, the BIT33_RS14560 protein and its homologous ones could be found in many other strains of A. baumannii, even in other Acinetobacter spp. and K. pneumoniae (Supplementary Figure S2). These results indicated a wide distribution of the BIT33_RS134560 gene and its protein in A. baumannii strains. It was interesting to note that this target gene in ATCC19606 was very closely related to ATCC17961, but not to ATCC17978, although they were all pattern strains of A. baumannii. We have not figured out why and how this gene evolved the way it did. Further study of the difference in the expression of the BIT33_RS14560 gene between these two clades may shed light on the causes of this phenomenon.

In addition, that the BIT33_RS14560 protein was closely related to the MFS homologs of A. nosocomiae and K. pneumoniae indicated potential pathogenicity of this protein due to the fact that the latter two bacteria were also important pathogens of nosocomial infections (Nho et al., 2015; Wang et al., 2020). Therefore, we believed this gene of interest needed to be noticed.

The E. coli/Acinetobacter shuttle vector pWH1266 containing AmpR and TetR was initially generated for the cloning experiments of Acinetobacter calcoaceticus and E. coli (Hunger et al., 1990) and has been successfully applied in the overexpression construction of A. baumannii (Zander et al., 2013; Liou et al., 2014). However, for XDR-Ab, it should be mentioned that due to its extreme drug resistance, the number of antibiotic selection markers for transformants was greatly limited. One solution to this problem was that the antibiotic concentration should be increased as much as possible within the tolerant range of the vector’s resistance. In this case, we made our attempt to transform the recombinant plasmid p-RS14560, which contained only AmpR, into an XDR-Ab isolate by increasing the concentration of Carbenicillin from 50 to 1,700 μg/ml. However, we failed because the XDR-Ab was still able to grow at such a high concentration of Carbenicillin. To achieve this goal of overexpression in XDR-Ab, a replacement of AmpR by other resistant genes was the key to the deal.

When we made our attempts to knock out this target gene using a marker-less method (Amin et al., 2013), the TelR cassette of pMo130telR vector caught our attention. This suicide vector was derived from the modification of the pMo130 vector (Hamad et al., 2009) by carrying a TelR cassette (Sanchez-Romero et al., 1998). Researchers had tested A. baumannii isolates, including MDR-Ab isolates, and found that they were susceptible to tellurite (Amin et al., 2013). We also tested several XDR-Ab isolates to confirm their susceptibilities to tellurite and found that they could be counter-selected in LB agar containing tellurite of 30–100 μg/ml (data not shown). Fortunately, we managed to overexpress the target gene BIT33_RS14560 in two XDR-Ab isolates by replacing the AmpR with the TelR. The transformation of two recombinant vectors, p-telR and p-telR-RS14560, might to some extent make an impact on the transcript expression of BIT33_RS14560. This could be explained by the fitness cost of bacteria (San Millan and MacLean, 2017; Nang et al., 2018). Large plasmids could impose a metabolic burden for host bacterial strains (Ma et al., 2018). However, to our knowledge, this was the first time to achieve stable gene overexpression in XDR-Ab by the replacement of AmpR with telR, which would provide guidance for further investigations on XDR-Ab strains.

Like many other bacteria, A. baumannii displays a robust biofilm formation on many abiotic surfaces in hospital environments, raising the risk of chronic infections. Previous studies have shown that the degree of biofilm formation varied considerably depending on the A. baumannii isolates (Rodríguez-Baño et al., 2008). Our study also showed different abilities of biofilm formation among 62 A. baumannii isolates when measured at 72 h. Those with robust abilities of biofilm formation were more likely to survive under desiccation and nutrition-limiting conditions, which explained why some clinical isolates possessed more tenacious vitality in hospitals than others. As shown in Supplementary Table S4, Spearman’s rank correlation analysis suggested that a stronger biofilm-forming ability could be followed by a relatively higher expression level of BIT33_RS14560 (the △CT value vs. biofilm rank, R = −0.569, p = 0.000). This was validated by overexpression of the BIT33_RS14560 gene in the standard strain ATCC 19606 and the two isolates XAb53 and XAb50 (Figure 3).

Here, three wild-type A. baumannii strains used for the construction of BIT33-RS14560 overexpression normally displayed relatively high levels of production within 12–48 h, which was in accordance with the previous study (Rodríguez-Baño et al., 2008). However, the enhancing effects of BIT33_RS14560 on biofilm formation appeared to be variable in these three A. baumannii strains (Figure 3). Within 48 h, for the ATCC 19606 strain, overexpression of BIT33_RS14560 exhibited a gradual upward trend of facilitating biofilm formation, while in the XAb53 isolate; this effect seemed to be more stable, keeping it in fairly high levels of biofilm biomass. However, when it came to XAb50 isolate, the amount of biofilm remained relatively unchanged over the past 30 h, then, rose sharply into about 3.24-fold (data not shown) at 48 h. Further investigation is warranted to elucidate the mechanisms for this phenomenon.

Currently, A. baumannii biofilm formation, of which the mechanisms are still not fully elucidated, is still a tough issue in the treatment of refractory and chronic infections. As one of the largest superfamily of secondary carriers known to date, MFS family membrane transporters exist widely in the whole biological world (Reddy et al., 2012) and they are considered to be closely related to a wide variety of life phenomena due to a basic function of assisting in transmembrane transport of certain substances, including monosaccharides, oligosaccharides, amino acids, enzyme cofactors, drugs, and so on (Lorca et al., 2007; Chen et al., 2008; Yen et al., 2010). In bacteria, MFS superfamily proteins have been found not only to play an important role in the transport of many substances but also to be a member of the efflux pumps large families (Pasqua et al., 2019), which are involved in biofilm formation and drug resistance (Bay et al., 2017; Saranathan et al., 2017; Poudyal and Sauer, 2018; Tang et al., 2020).

BIT33_RS14560, although considered to be categorized into the MFS family, is still lacking reports about its effects on the biofilm formation of A. baumannii. In this study, we conducted a comparative transcriptomic analysis to investigate the possible mechanisms by which BIT33_RS14560 overexpression had an impact on the biofilm formation and virulence of A. baumannii ATCC19606. As shown in Table 4, we found that three DEGs, including F3P16_RS12800, F3P16_RS04575, and F3P16_RS04585, although displayed different up or down multiples, were hypothesized to be related more or less to bacterial biofilm formation or virulence.

F3P16_RS12800 was annotated as an OprD family outer membrane porin. The OprD family was described first for P. aeruginosa, but ever since that time, it has been found in many metabolically versatile soil bacteria and comprises over 100 members (Tamber et al., 2006). In the current study, we found that the overexpression of BIT33_RS14560 led to a significant upregulation of F3P16_RS12800, suggesting that in A. baumannii, BIT33_RS14560 may enhance biofilm formation by upregulating the expression of this outer membrane porin. This was supported by a previous comparative proteomics study (Cabral et al., 2011) revealing that some membrane proteins including OprD-like protein were involved in the adhesion and biofilm formation of A. baumannii.

Most importantly, we got F3P16_RS04575 and F3P16_RS04585, both encoded proteins belonging to the glycosyltransferase (GT) family and were upregulated when BIT33_RS14560 was overexpressed. GTs mainly participate in the transfer process of glycosyl from an activated donor to a receptor, such as sugar, lipid, protein, and nucleic acid and play an important role in pathogenic adhesion, biofilm formation, and virulence (Breton et al., 2006; Ribet and Cossart, 2010a). Structural analysis of three-dimensional (3D) protein fold has shown that the catalytic domains of GTs are categorized into GT-A, GT-B, and GT-C three types, the most common of which are GT-A and GT-B (Moremen and Haltiwanger, 2019). GT-B or GT-C is believed to play a role in adhesion, biofilm formation, and virulence of pathogenic bacteria while GT-A enters the host cell and interferes it with post-translational modifications thus affecting its signal transduction, protein translation, and immune response. It was also important to note that the protein glycosylation of GT was closely related to the virulence of pathogenic bacteria (Ribet and Cossart, 2010b; Yakovlieva and Walvoort, 2020). Many bacterial pathogens were found to bear toxic GTs that interfere with host post-translational modifications to promote their own survival and replication (Lu et al., 2015). Interestingly, when we further analyzed with the Carbohydrate-Active Enzymes (CAZy) database,⁹ we found that the proteins encoded by F3P16_RS04575 and F3P16_RS04585 belonged to GT-C and GT-A, respectively. This might explain why ATCC 19606 became more virulent when BIT33_RS14560 was overexpressed (Figure 4).

It was reported that proteins and carbohydrates were the main components of biofilm formation (Costerton et al., 1995). Amino acid metabolism, glycerolipid, and tricarboxylic acid cycle were also observed to be mostly affected during bacterial biofilm formation (Lu et al., 2019). Our results of GO and KEGG analysis (Figures 7, 8) also indicated that most DEGs participated in complicated metabolic processes including amino acid, carbohydrate metabolism, which also suggested the complexity of bacterial biofilm formation. However, the specific effects of these DEGs on A. baumannii biofilm formation and what metabolic pathways these DEGs may be involved in remained further investigations.