Fitness Cost and Compensation Mechanism of Sulfonamide Resistance Genes (Sul1, Sul2, and Sul3) in Escherichia Coli

Background: The tness cost of antibiotic resistance is a crucial factor to determine the evolutionary success of resistant bacteria. Even if the selection pressure in the environment is eliminated, drug-resistant bacteria can still compensate for drug-resistant genes' tness cost through some compensation mechanisms. The tness cost and compensatory evolution of antibiotic resistance are an essential part of bacterial evolution. Result: Engineered bacteria with the same genetic background that carry sulfonamide resistance gene were generated to explore the tness cost of sulfonamide resistance gene in Escherichia coli. There were signicant differences in the protein expression of the two-component system pathway (iZ, iA, iC and lrhA), folate biosynthesis pathway (sul1, sul2 and sul3), ABC transporter system (ugpC, rbsA and gsiA), and outer membrane pore protein OmpD through the comparative analysis of differential proteins compared to sensitive bacteria. Thus, we could speculate the possible tness compensation mechanism. Finally, qRT-PCR was used to verify the functions of some differential proteins at the transcriptional level. Conclusions: The study of tness cost assessment and compensatory evolution of bacterial resistance will help understand the development track of antibiotic resistance of bacterial pathogens and provide new ideas for solving antibiotic resistance issues. on total colonies patched. Three biological The growth kinetics of studied The initial optical bacterial monitored triplicate. NH 4 HCO 3 ) was added, mixed well at 600 ×g for 60 s, and placed at 37 ℃ for 16-18 h. The collecting tube was replaced and centrifuged at 14000 ×g for 15 min. Then, 40 μL 25 mM NH 4 HCO 3 was added and centrifuged at 14000 ×g for 15 min, and the ltrate was collected. The peptides were desalted by C 18 cartridge, lyophilized, and redissolved in 40 μL 0.1% formic acid solution. The peptides were quantied by photometry at OD 280 . of difference PET23a-sul1, BL21: PET23a-sul2


Introduction
The spread of MDR (multidrug-resistant) amongst Gram-negative bacteria has emerged as one of the most pressing global public health threats [1]. As synthetic broad-spectrum antimicrobial agents, sulfonamides have inhibitory activity against most G + and Gbacteria [2]. While sulfonamides play a substantial role in preventing and treating animal bacterial diseases, bacteria have developed extensive and robust resistance to them [3]. Bacteria usually produce resistance to sulfonamides through two different strategies: (i) gene mutation and (ii) gene substitution. In gene mutation, the gene folP of dihydrosphenoic acid synthase (DHPS), located on the chromosome, produces drug resistance [4]. In gene substitution, drug resistance is generated by obtaining DHPS replacement genes sul1, sul2 and sul3. The expression products of the latter mechanism have a lower a nity for sulfonamides [5]. The second mechanism of sulfonamides resistance is more widespread [6].
Several recent reports have shown that sul gene is found and prevalent in Escherichia coli from human and livestock origins, and it may have stable vertical and horizontal transmissions [6].The base sequences of sul1, sul2, and sul3 have about 50% homology with each other [7]. The primary mechanism of drug-resistant gene transmission is the horizontal or vertical transfer of mobile genetic elements carrying drug-resistant genes [8]. Gene sul has been identi ed on both chromosomes and plasmids. At present, plasmids, integrons, transposons and other mobile gene elements (MGEs) have been found to be related to the transmission of sulfonamides resistance [9].
The sul gene originated from animal-associated bacteria. It has been frequently isolated from livestock and is supposedly sustained by sulfonamide's heavy usage in the veterinary sectors [10,11]. The genes sul1 and sul2 were rst published by Swedbery and Radstrom, respectively, in 1983 and 1985. Sul2 was more common and widely distributed in clinical Escherichia coli than the gene sul1 [12]. The sul1 gene is mainly associated with other resistance genes on the integron retention fragment [13]. The sul2 gene is often associated with streptomycin resistance genes found in small, unbound plasmids with a broad host range [14]. In 1990, Martin CT found a gene similar to sul1 in Mycobacterium, but the gene had missed the promoter codon, and the codon had been inserted further upstream, so the gene was named sul3 gene [15]. China has a vast territory with numerous pig farms, making the detection range of sulfonamides in pig faeces samples di cult. Furthermore, the highest detection concentration can reach 50 mg/kg [16][17][18].
The development of antibiotic resistance often comes with a tness cost, de ned by reduced competitive ability in an antibioticfree environment. This phenomenon usually allows the tter, often susceptible strain to outcompete the resistant one [19]. Given the biological basis of transmission and epidemic, the tness cost and compensation mechanism of sul gene in E. coli have not been reported.
The present study aims to investigate the tness cost of sulfonamide resistance genes in E. coli. bacterial competition in vitro test was used to assess the tness cost of sul genes in E. coli. The label-free proteomics and real-time uorescent quantitative PCR were applied to explore adaptive changes of sul gene in E. coli strains at different protein expression levels. This study may help to explain the possible adaptive mechanism of these genes, which provides theoretical support to control their transfer and spread.

Methods
Bacterial strain and plasmids E. coli DH5α and E. coli BL21 were used as the clone and the expression strains respectively. The sulfonamide resistant genes sul1, sul2 and sul3 were inserted into polyclonal sites of the constitutive expression plasmid pET23a (Novagen, USA) using BamH I and Sac I enzymes through the gene recombination method. Recombinant plasmids pET23a-sul1, pET23a-sul2 and pET23a-sul3 carrying sulfonamide resistance genes were transformed into E. coli BL21 cells by heat shock transformation method. Single colonies of the transformed plasmids were selected and inoculated into LB liquid medium and incubated at 37°C and 200 rpm for 5 h. General primers of the vector T7 promoter and terminator were used for colony PCR identi cation.

Antimicrobial susceptibility test
MICs were determined by broth dilution method with an inoculum of 10 5 CFU ml −1 in LB medium. Each test was repeated three times at least. Plasmid stability and growth kinetics Transconjugants E. coli BL21: pET23a-sul1, E. coli BL21: pET23a-sul2 and E. coli BL21, pET23a-sul3 and E. coli BL21: pET23a were propagated by serial transfer for 14 days of passage. The culture broths were serially diluted in 0.9% saline and plated onto LB agar without sulfonamide. Approximately 100 colonies were randomly chosen and replica plated onto sulfonamidecontaining and antibiotic-free LB agar plates. The percentage of plasmid retention was calculated by dividing the number of colonies on sulfonamide-containing LB agar by the total number of colonies patched. Three biological replicates were included for each group. The growth kinetics of E. coli BL21 and its transformants carrying the plasmids pET23a-sul1, pET23a-sul2, pET23a-sul3 and pET23a were studied by inoculation in 150 mL of fresh LB broth. The initial optical density at 600 nm (OD 600 ) was 0.6, and the bacterial growth was monitored by hourly recording OD 600 for 24 h at 37 ℃. Experiments were performed in triplicate.

Motility test
Engineered bacteria and control strains were inoculated into MH liquid medium. When the initial OD 600 value was 0.3, 5 μL aliquot of bacterial suspension was taken and inoculated into the center of a 0.4% MH agar medium plate (Ø=9 cm) and incubated at 37℃ for 48 h.

In vitro competition experiments
To assess the tness effect of sul1, sul2, sul3 in the bacterial host, E. coli BL21: pET23a-sul1, E. coli BL21: pET23a-sul2 and E. coli BL21 pET23a-sul3 were used to compete against E. coli BL21: pET23a. E. coli BL21: pET23a was mixed with the control strain in the ratio of 1:1 by volume. The bacterial suspension was transferred to fresh LB medium at a volume ratio of 1:100 every 24 h for 5-6 times. Before each transfer, the mixed bacterial suspension was plated on the non-resistant LB medium to calculate the total number of colonies. 100 single colonies were picked from the non-resistant plate and inoculated on the LB resistant plate (4 mg/ mL sulfamethoxazole concentration). The number of sulfamethoxazole resistant colonies was calculated by Eg. 1.

Eq. 1. Selection coe cient = [ln(E/R)t -ln(E/R)0]/T
In which, E is experimental group cell number, R is control group cell number, and T is passage number.
Preparation and quantitative analysis of proteome samples The engineered and control strains were cultured for 7 h to the logarithmic growth phase. Bacterial suspensions were centrifuged at 5000 ×g and 4 °C for 10 min. After decanting the supernatant, cells were thoroughly washed twice by the addition of PBS and subsequent centrifugation under the same condition. Each sample added a moderate amount of SDT lysis buffer (SDS, dithiothreitol, Tris) for 60 s. The sample was then ultrasonized for 10 s at each interval of 15 s for 10 cycles. After 15 min in boiling water bath, the sample was centrifuged at 14000 ×g for 40 min. After centrifugation, the supernatant of each sample was quanti ed with BCA kit and analysed by SDS-PAGE.

Enzymatic hydrolysis of proteins
A 30-μL aliquot of protein solution was taken from each sample, DTT was added to the nal concentration of 100 mM, boiled in water for 5 min. Two hundred μL UA Buffer was added and mixed well, then transferred into a 10 kD ultra ltration tube and centrifuged at 14000 ×g for 15 min. One hundred μL IAA buffer (100 mM IAA in UA) was added, followed by 600 ×g oscillation for 60 s. The dark reaction was carried out for 30 min followed by centrifugation at 14000 ×g for 15 min. Subsequently, 100 μL UA buffer was added and centrifuged at 14000 ×g for 15 min, repeated twice. One hundred μL NH 4

Bioinformatics analysis
MaxQuant software was used for database identi cation and quantitative analysis. The relevant parameters and instructions were as follows: the maximum number of permissible leak-cut sites was 2; the mass tolerance of primary ion and secondary ion plasmid were 6 and 20 ppm, respectively. All databases of the library are uniprot_Escherichia_coli_1124415_20180910.fasta. The quantitative strength values of unique peptide and razor peptide were adopted to carry out protein quanti cation using the LFQ algorithm. After the proteomic data were analyzed and processed, the screened differential proteins were annotated and attributed to protein function.
Transcriptional level veri cation of differentially expressed proteins The total RNA extraction method was carried out according to the total RNA extraction kit for biological bacteria (Baitek). Realtime PCR primers were designed for 16 related genes and 16S DNA using Primer 5.0 software. The primers used in qRT-PCR are presented in supplementary materials.

Results
Contribution of sul to sulfonamide resistance MIC tests con rmed that the sulfonamide resistance gene was generated successfully.

Bioinformatics analysis of differentially expressed proteins
Gene Ontology (GO) functional annotation analysis In terms of cell composition, the protein changes were mainly located in the cell membrane and cell-matrix among all six groups. These changes are re ected primarily in catalytic activity and binding functions, which involved cells' metabolic process.

KEGG pathway analysis
In the paired comparison between the engineered strains carrying sul1, sul2 and control strains, proteins involved in the pathways were relatively concentrated. Apart from the glucose and nucleotide metabolisms, they were also involved in the ABC transporter and the two-component systems. However, in the paired comparison between E. coli BL21: PET23a-sul3 strains in the experimental and control groups, the metabolic pathways were relatively dispersed, mainly focused on glucose and nucleotide metabolism pathways.

Analysis of the variation trend of differentially expressed proteins
In E. coli BL21: pET23a-sul1 and E. coli BL21: pET23a-sul2 strains, proteins iA, iC and their upstream transcriptional activator iZ in the two-component system were signi cantly upregulated. However, lrhA, the transcriptional inhibitor expressed by type I pili, was signi cantly downregulated.
The protein SUL1, which is involved in folic acid metabolism, was signi cantly upregulated in E. coli BL21: pET23a-sul1 strain. Besides, molybtrexate adenosine transferase, molybtrexate synthase catalyzed subunit and cyclopyranxate monophosphate synthase were upregulated by 1.7 and 1.6 times, respectively. 2-amino-4-hydroxy-6-hydroxymethyl dihydroteropterine diphosphate kinase was changed from zero to zero, and 4-amino-4-deoxy branched-acid lythase was altered from existing to non-existing. The protein expression of QueE (7-carboxy-7-deazaguanine synthase), which is involved in the pathway of folic acid metabolism, was downregulated 1.8 times in E. coli BL21: pET23a-sul2 strain. E. coli BL21: pET23a-sul3 strain signi cantly reduced the number of differential proteins involved in the pathway of folic acid metabolism, and signi cantly upregulated the expression of dihydrospteric acid synthase SUL3 In this study, obvious differences were observed in the ABC transporter system pathway by KEGG pathway analysis. The expressions of ATP-binding proteins ugpC, rbsA and gsiA, which are closely related to energy supply, were signi cantly increased in E. coli BL21: pET23a-sul1 or E. coli BL21: pET23a-sul2, but there was no signi cant change in E. coli BL21: pET23a-sul3.
Analysis of individual differentially expressed proteins showed that the differential expression ratio of the outer membrane pore protein ompD was the highest among all the proteins. In E. coli BL21: pET23a-sul3, the quantitative strength of ompD LFQ protein was 1.7 × 10 9 , which was increased by 29.93 times compared with the control strain E. coli BL21: pET23a.
The comparative quantitative analysis results showed that the variation trend of target gene mRNA level expression obtained by real-time PCR analysis was the same as that obtained by Label-free proteomic analysis ( Figure 6).

Discussion
Sulfonamide resistance genes sul1, sul2, and sul3 can mediate high levels of drug resistance of Escherichia coli to sulfonamides. Several studies have demonstrated these three drug resistance genes are widespread in the pork industry [20][21][22][23]. This study was to investigate the biological basis of the prevalence and transmission of sulfonamide-resistant genes sul1, sul2 and sul3.
The sulfonamide resistance genetic engineering strains with the same genetic background were constructed by gene cloning and recombination techniques. In recent years, the construction of drug-resistant genes in engineered bacteria has been more and more used to research the tness cost of drug-resistant genes in strains. Nang et al recombined mobile colistin resistance (mcr-1) on a low-copy and broad-spectrum host range vector pBBR1MCS-5 and switched it to Klebsiella pneumonia B5055 to study the tness cost of mcr-1 gene [24]. As expected, drug-resistant genetically engineered bacteria constructed in this study showed high resistance to sulfonamide in MIC tests.
The evolution and persistence of antibiotic resistance in the bacterial population depend on a complex calculus rooted in biological tness cost associated with the resistance and the impact of the resistance pressure. In the presence of high levels of antibiotics, access to antibiotic resistance mechanisms offers an adaptive advantage over susceptible competitors [25]. However, there are often deleterious effects in the absence of antibiotic resistance mechanisms, typically observed as an increased generation time and reduced survival in a host [19]. These observations suggest that susceptible, higher tness populations should generally outcompete resistant bacteria after removing selection pressure. However, several factors can cause the stability of this resistance: 1) co-selection of resistance genes with other functions that confer a tness advantage [26,27]; 2) presence of resistances that impose a very low or no cost [28,29]; 3) compensatory evolution that reduces the tness cost, often without loss of the resistance [30,31] Studies have shown that the change of tness cost will affect normal physiological functions of bacteria to a certain extent, such as growth ability, movement ability, virulence and plasmid stability. In general, there was no signi cant difference between E. coli BL21: pET23a-sul1, E. coli BL21: pET23a-sul2 and control E. coli BL21: pET23a (P>0.05). E. coli BL21: pET23a-sul3 showed a signi cant (P<0.05) decrease in growth capacity, locomotion capacity, and loss of plasmid. E. coli BL21: pET23a-sul3 also showed a certain tness cost in the in vitro competition test. Sul1 and sul2 had lower tness cost, which illustrated one of the important reasons for the widespread prevalence of these two genes from the biological basis. Theoretically, suppose the adaptability of speci c genotypes of drug-resistant bacteria is lower than the average adaptability of the population. In that case, it will be a disadvantage in competition, and these genotypes may be eliminated gradually. However, it is noteworthy that E. coli BL21: pET23a-sul3 adaptability was restored in response to sulfonamides selection pressure. Therefore, if the selective pressure of sulfa drugs persists in nature, the prevalence of sul3 gene will gradually increase.
Bacterial proteome changes dynamically with growth environment and growth stage, and there are different protein expression pro les under various conditions. Label-free quantitative proteomics was used to explore the possible tness mechanisms further.
The two-component system (TCS) is the main pathway for bacteria to complete chemotactic movement [32,33]. It was found that in E. coli BL21: pET23a-sul1 and E. coli BL21: pET23a-sul2 strains, proteins iA, ic and their upstream transcriptional activator iZ were signi cantly upregulated in the TCS. However, lrhA, a transcriptional inhibitor expressed by type I pili, was signi cantly down-regulated. The increase of FliZ protein expression can promote the expression of FliA and FliC, increase the synthesis of agellin, and make the cells gain better motility and improve the cells survival [34,35]. As a transcriptional regulatory inhibitor of the agellum, motility and chemotactic gene lrhA can promote type I pili and bio lm formation by inhibiting lrhA [36,37]. However, there was no signi cant change in these genes in E. coli BL21: pET23a-sul3, which may be one reason for its lower adaptability than that of sul1 and sul2.
Most prokaryotes must synthesize folate themselves. DHPS is the target enzyme of sulfonamides, which binds to it and inhibits its activity, to reveal antibacterial effects [38]. Sul1, sul2 and sul3 are dhps substitution genes, and their expression products all act as dihydrosphenoate synthase. The function of dihydrosphenoate synthase substitution enzyme exerted by SUL3 may not be su cient to balance the in uence of 2-amino-4-hydroxy-6-hydroxymethyl dihydrosphenoidine diphosphate kinase overexpression. Silent expression of 4-amino-4-deoxybranch acid cleasthase on the folic acid metabolism pathway may affect the synthesis of folic acid and reducing adaptability.
Multidrug e ux transporters are prevalent in the antibiotic resistance mechanisms as they bestow an ability to bacteria to evade most current therapies [39,40]. ATP-binding cassette transporters are essential biomolecule transport systems in organisms that rely on ATP hydrolysis to obtain energy for transporting various biomolecules [39]. In this study, the energy supply by ATP binding protein UgpC, RbsA, GsiA in E. coli BL21: pET23a-sul1 and E. coli BL21: pET23a -sul2 increased signi cantly. After obtaining exogenous sul1 and sul2 genes, the strain was subjected to ne regulation by ATP energy supply system in the logarithmic phase. However, there was no noticeable change in E. coli BL21: pET23a -sul3 strains. The changes of the ABC transporter system are closely related to the strain's physiological state and growth conditions. Therefore, the association between the ABC transporter system and bacterial adaptability needs to be further veri ed.
Outer membrane pore protein OmpD is involved in discharging bacterial metabolic poisons, which are mainly caused by metabolic changes and insu cient bacterial carbon sources. OmpD is overexpressed when metabolism is inhibited, or carbon sources are not su cient [41]. In this study, the expression of OmpD protein in E. coli BL21: pET23a-sul3 was signi cantly increased. Thus, it was speculated that E. coli BL21: pET23a-sul3 metabolism was inhibited, leading to insu cient carbon source for bacterial metabolism. Accordingly, it was attempted to upregulate OmpD to participate in the excretion of toxin in bacterial metabolism, which consumed energy and affected its tness cost.

Conclusions
In this study, the compensation mechanism of sulfonamide resistance genes in E. coli was explored in constructed engineered bacteria carrying sul1, sul2 and sul3 genes. The results showed that the tness cost of sul3 was signi cantly higher than sul1 and sul2 (P<0.05). The sulfonamide resistant genes obtained by bacteria directly affect the folic acid metabolism pathway of E. coli. sul1 and sul2 genes make more differential proteins involved in folic acid synthesis. Meanwhile, ATP-binding proteins such as UgpC, RbsA and GsiA, which are closely related to energy supply, were upregulated, making folic acid synthesis in bacteria more accurately regulated. In addition, the agella-related proteins FliA and FliC and their upstream transcriptional activator FliZ in the two-component system were signi cantly upregulated, while the type I pili transcriptional inhibitor LrhA was downregulated so that the bacteria could obtain better motility and maintain their survival. The tness compensation mechanism of sul3 was relatively weak. The considerably upregulated OmpD of E. coli BL21: pET23a-sul3 indicated that the bacteria's metabolism was inhibited and energy was insu cient, showing a high adaptive cost, leading lower prevalence of sul3 than sul1 and sul2. Tables Table 1 The signi cant differentially expressed proteins in recombinant bacteria   SDS-PAGE electrophoresis of total protein. a1, a2 and a3 represent three biologically repeated total proteins of strain E. coli BL21: PET23a. S1-1, S1-2 and S1-3 represented total protein of three biologically repeated strains of E. coli BL21:pET23a-sul1; S2-1, S2-2 and S2-3 represented the total protein of three biologically repeated strains of E. coli BL21: pET23a-Sul2; S3-1, S3-2 and S3-3 represent three biologically repeated total proteins of strain E. coli BL21: pET23a-Sul3.