In this study, we have for the first time employed an experimental zebrafish infection- treatment model to mimic the conditions under which antibiotic resistance (mediated by a naturally occurring R-plasmid) transfer takes place in the intestinal microbiota during an infection caused by a resistant pathogen treated with effective or ineffective antibiotic treatments.
We were able to establish an infection with A. hydrophila resulting in disease symptoms similar to those previously described [10, 11] but with no mortality 3 days post- infection, as intended in our study design. Rodriguez et al.  and Pullium et al.  observed per-acute cases of A. hydrophila infection with high mortality rates within a few hours possibly related to intraperitoneal injection of bacterial extracellular toxins and/or enzymes. In our study, re-isolation of the challenge organism from the kidneys of exposed fish supported the clinical findings and confirmed systemic infection. It has also been shown that A. hydrophila produces an array of virulence factors that induce strong inflammatory responses [34–36]. The induction kinetics of some of the zebrafish intestinal immune system genes revealed an Acute Phase Response (APR), that is the immediate host inflammatory reaction which counteract challenges such as tissue injury and infection . In the current study A. hydrophila infection resulted in a clear increase in expression of the genes encoding the pro-inflammatory cytokines TNF α, IL-1β and IL-8. These cytokines are important inducers of APR resulting in increased production of Acute Phase Proteins (APPs) , such as C3. C3 is central in elimination of bacterial threats . A systematic study of APR in zebrafish has shown striking similarities with mammals in function and induction of involved genes . The fact that 1 IL-1β and IL-8 are highly induced while C3 remains moderately expressed is consistent with the expected expression profile at the early stages of infection (3 days in our case).
The composition of the zebrafish intestinal bacterial microbiota and its interaction with the host and the environment has previously been studied by cultivation and culture-independent methods [28, 40]. In the present study this microflora and the experimentally introduced pRAS1 harboring A. hydrophila were impacted by various antibiotic treatments. Recent studies have shown that Real-Time PCR with species-specific or universal probes is an accurate and sensitive method for quantification of total bacterial populations as well as individual species from the intestinal contents [41–45].
In our study a broad spectrum of 16S rDNA primers were used since bacteria can have different genome sizes and different rrn operon copy numbers. There are different concepts for considering the rrn operon numbers in quantitative 16S rDNA-based experimental systems [43, 44, 46]. Ott et al. , have provided accurate and stable figures of similar bacterial concentrations in clinical samples with application of universal primers and specific probes. In the present study, 16S rDNA gene copy numbers were significantly decreased after effective flumequine treatment, whereas sub-lethal flumequine or the clinically relevant ineffective tetracycline, trimethoprim and sulphonamide treatments caused minimal change. The reduction in 16S rDNA gene copy number following treatment with flumequine might be the result of killing of pathogenic A. hydrophila and a disturbed and reduced commensal flora. In mammals and humans, it is well known that antibiotics can change the composition of the bacterial populations in the intestines [48–50]. Studies concerning the distribution of antibiotic resistant bacterial isolates in zebrafish facilities are, however, limited. Previous studies performed in our laboratory Cantas et al. , have shown a relatively low level of tetracycline (12-20%), trimethoprim (25-32%), sulphonamide (28-36%) and quinolone (0.5-4.8%) antibiotic resistant bacteria in the Gram negative cultivable gut flora in four different zebrafish facilities, one of which supplied the zebrafish for the present study. This would leave potential recipient flora for plasmid transfer in all treatment groups.
The minimal change in total 16S rDNA copy number following treatment with clinically relevant levels of tetracycline, trimethoprim and sulphonamide may be explained by multiplication of the resistant A. hydrophila pathogen due to the decreased competition following killing of the susceptible part of the normal intestinal microbiota.
The active involvement of the selected tra-genes in the DNA conjugation process is described . The traD gene encodes an inner membrane protein with putative ATPase activity for DNA transport during bacterial conjugation. This protein forms a ring-shaped structure in the inner membrane through which DNA is passed to the transferosome [18, 51]. However, it has been shown that the virB4 and virD11 genes may, in addition, mediate conjugative transfer via a C-terminal ATPase function during pili assembly which is more efficient on surfaces than in liquids [52, 53]. pRAS1 is transferred approximately 1000× faster on solid surfaces compared to the frequency in liquid media [Kruse and Sørum 1994, unpublished data]
The genes of the conjugative transfer system studied i.e. traD, virB11 and virD4, were found to be differently expressed between the treatment groups. The expression of transfer genes was found to be low following sulphonamide and flumequine treatment, whereas treatment with a sub-inhibitory level of flumequine, clinical relevant levels of tetracycline and trimethoprim resulted in increased expression. Several factors have been proposed that could explain these differences; i) the susceptible gut microbiota was reduced in number leaving behind a variable number of potential conjugation recipients , ii) the donor potential and the genetic advantages/disadvantages of the specific plasmid in conjugating to the available recipient population , iii) the antibiotic itself might regulate the higher or lower expression levels of pRAS1 mobility genes resulting in possible different transfer frequencies. An increased transfer frequency induced by antibiotic exposures (tetracycline and trimethoprim) has been demonstrated for conjugal transfer of pRAS1 plasmid in sediment microcosm experiments .
A most remarkable result of the current study was the strongly increased expression levels of the selected plasmid transfer genes in the intestinal microbiota following treatment with tetracycline, trimethoprim (plasmid encoded resistance) and ineffective concentrations of flumequine. The low concentration of the quinolone flumequine was chosen to mimic the low concentration in the intestinal lumen when administering the drug intramuscularly or intravenously for treatment purposes, in in-appetent animals offered in-feed antibiotics, or by exposure to environmental residues from the water [55, 57, 58]. It has been shown that administration of sub-therapeutic levels can interfere with DNA replication (e.g. quinolones) [59, 60], folic acid synthesis (e.g. trimethoprim) , protein synthesis (e.g. tetracycline)  as well as cell wall synthesis (e.g. β-lactams)  and may induce the so-called SOS response  which can promote acquisition and dissemination of antibiotic resistance genes [57, 65]. Thus, our results reinforce the need for great caution in the use of SOS-inducing antibiotics to avoid induction of resistance transfer following antibiotic therapy. It is known that the LexA protein as part of the SOS response binds to the LexA box preceding the intI gene and thereby increasing the transcription rate of the intI gene resulting in an increased gene cassette exchange rate in the integron . There is no recognized LexA box found close to the promoters of the traD, virB11 and virD4 genes of the pRAS1 plasmid sequence (data not shown). However, the occurrence of LexA targets in promoter sequence areas in vivo without the existence of a putative LexA box in the DNA sequence has been demonstrated. This indicates the assistance by an additional unknown factor in regulation of LexA gene expression in vivo .
An equally remarkable finding was the impact of antibiotic treatments on the expression of innate immunity genes. The decreased TNF α and C3 expression in the zebrafish's intestine after non-effective tetracycline treatment is in accordance with earlier reports [68, 69] relating tetracyclines to posttranscriptional blockage of cytokine production . Whereas, sulphonamide and trimethoprim treatments that have no impact on the growth of pathogenic A. hydrophila had little impact on IL-1β and IL-8, as expected. In contrast, the sub-inhibitory level of flumequine caused 40 and 20 fold increases in the expressions of IL-1β and IL-8, respectively. In addition effective flumequine treatment caused 200 and 100 times higher expressions of those genes, respectively. Hypothetically, this may be related to the immunomodulatory properties of those drugs [71, 72] and in the diminished number (killed) of pathogenic A. hydrophila that can no longer depress the immune system by its virulence factors when the effective flumequine treatment was employed [73, 74].
We have for the first time termed this clear, aggressive, immunological activity at the molecular level as 'Charged Immune Attack, (CIA)', which describes the inevitably strong revenge of the innate immune response against the weakened bacterial infection, as mediated by a short period with an effective antimicrobial treatment. The reason for this bias is not known, but both human and veterinary medical practitioners have observed that a single dose of antibiotics, sometimes surprisingly, may cure an infection. We think that the current results provide a glance into subtle and immediate effects of chemotherapy on the host's innate immune system that may be responsible for such outcomes. Further studies are needed to shed new light on the current findings and to clarify the underlying mechanisms.
For methodological reasons, most studies of in vivo conjugal plasmid transfer have been performed by adding donors and limited numbers of recipients in germ free animals [75, 76] or by challenging conventional fish with genetically tagged bacteria . To the best of our knowledge, this is the first report on the effect of antibiotic treatment of an infection on the expression of the tra genes of an R-plasmid harbored by the infecting pathogen and the early immune signals in a host model. Real-Time PCR technology offers a fast and reliable quantification of the mRNA production of any target sequence in a sample . The results add information to our knowledge about development of antibiotic resistance in infected hosts including the clinical infection treatment and control scenario.