Phenotypical Comparison of Pseudomonas Aeruginosa Isolated from Human and Veterinary Samples; Impact of Host Adaptation on Infection Pathogenesis

Purpose: Determinants of virulence in Pseudomonas aeruginosa vary strongly depending on its habitat. In this study, we analyzed these alterations depending on the host organism in isolates cultured from canine ears and compared it to clinical extended-spectrum antibiotic-resistant Pseudomonas aeruginosa isolates (XDR), clinical antibiotic-sensitive (non-XDR) from humans and environmental isolates (EI) analyzed during our rst study in 2017. Methods: A total of 22 veterinary isolates cultured from canine ears (VET) were examined for spontaneous biolm formation, stress response in biolm formation induced by meropenem, in vitro tness, susceptibility to human serum and polymorphonuclear leukocytes and the genetically determined virulence factors toxA, exoS, exoT, exoU, exoY, nan1, cif, lasA and lasB. Results: We observed signicantly elevated spontaneous biolm formation and serum susceptibility in VET isolates compared to EI and non-XDR isolates as well as signicantly decreased in vitro tness compared to XDR isolates. The VET isolates resembled most the XDR subgroup of isolates previously cultured from blood. Within the environmental isolates, we observed an increase of spontaneous biolm formation and exoU presence in isolates cultured from community water samples over hospital water samples to pool samples. Conclusions: Considering the distinct differences in some features of the examined VET isolates, a higher degree of phenotypical adaption can be assumed. Increased biolm formation seems to be a common and characteristic event in isolates adapted to a specic habitat. Therefore amplication of potentially more virulent Pseudomonas aeruginosa strains in domestic animals may lead to elevated zoonotic risk for example for pet owners.


Introduction
Among the most relevant facultative pathogenic organisms, Pseudomonas aeruginosa (Pa) is a Gram negative bacterium that causes nosocomial infections and is one of the most important pathogens for patients with cystic brosis. Although Pa inhabits a wide spectrum of habitats, including moist environments like in-house water supplies, the genetic similarity between isolates of different origins, including human and veterinary clinical isolates, is surprisingly high [1]; and, in veterinary medicine, equally sporadic diseases and outbreaks occur [2][3][4][5][6][7]. Due to this wide spectrum of hosts, a zoonotic potential exists [8,9].
In a recent study, we revealed alterations in pheno-and genotypical properties between clinical extendedspectrum antibiotic-resistant isolates (XDR), non-multiresistant clinical isolates (non-XDR) and environmental isolates cultured from community-based water samples (EI), including signi cantly higher spontaneous and induced bio lm formation in XDR than in non-XDR clinical isolates. Furthermore, XDR isolates showed increased in vitro tness but were more susceptible to normal human serum than both non-XDR and EI. The endemic XDR strains ES-1 and ES-1 showed properties with values at the extremes of the distributions, often with opposite manifestations for particular properties [10].
The present study examines an additional group of 22 isolates cultured from canine ears (VET). This group was chosen because of its importance in veterinary medicine and the possibility of exposure and transmission to animal keepers. Furthermore, we split the already in 2017 evaluated XDR and EI isolates into subgroups, consisting of isolates cultured from similar clinical material to assess their properties in relation to infection site.
Pa is primarily considered a cause of nosocomial infections in humans. We wanted to extend the spectrum of our observations by adding a group of clinical isolates from a distinct material of animal origin to learn more about Pa in veterinary medicine and to elucidate the mechanisms of persistence and pathogenesis of Pa.

Methods
Samples consisted of a total of 22 Pa isolates of clinical material from canine ears (VET) sampled in various veterinary clinics or practices. These were cultivated and determined in 2015 and 2016 at the Institute for Hygiene and Infectious Diseases of Animals at the University of Giessen.
Of the 296 isolates cultured at Heidelberg University Hospital from 2009 to 2014, 211 clinical extendedspectrum antibiotic-resistant isolates (XDR), which are resistant to piperacillin, ceftazidime, cipro oxacin and meropenem [11], and 47 environmental isolates cultured from community-based water samples (EI) were subdivided further. The 45 non-multidrug resistant clinical isolates (non-XDR) resistant to 2, at most, of the aforementioned antibiotics have not been subdivided further.
The XDR isolates used in Kaiser et al. (2017) were subgrouped as follows: 15 XDR blood isolates cultured from blood samples, 20 XDR invasive isolates cultured from abscesses and intraoperative swabs, 28 XDR wound isolates cultured from wounds, 55 respiratory isolates cultured from samples of the upper and lower respiratory tract, 31 XDR urinary isolates cultured from urine and urinary tract swabs, 22 XDR colonizing isolates cultured from samples of rectal and vaginal swabs, 15 cfXDR isolates cultured from sputum of cystic brosis patients, the endemic strains ES-1 (13 isolates) and ES-2 (12 isolates) cultured from various clinical materials and assigned with RAPD PCR with primers 208 and 272 [12].
The EI isolates cultured from in-house water were subdivided into 10 EI community isolates from community water supply, 25 EI hospital isolates from drinking water samples from hospital facilities and 10 EI pool isolates from pool water samples.
To increase comparability to other studies, the widely used reference strains PAO1 and PA14 were included.
Our methods are described in detail in Kaiser et al. (2017) [10].
To determine bio lm formation, test isolates adjusted to the exponential growth phase and a McFarland standard of 0.25 were grown for 18 h in M63 minimal medium in 96-well microtiter plates. Bio lm formation was determined using the optical density of crystal violet bound in the bio lm matrix, by staining the bacterial bio lm with crystal violet and eluting in ethanol. A clinical isolate of the XDR group was used as a positive control and reference isolate. This was done for the bio lm-index spontaneous without the addition of antibiotics and for the bio lm-indices MPMX with meropenem (MPM) at concentrations of 1 mg/L, 4 mg/L and 125 mg/L. Furthermore, a stress response for each MPM concentration was generated, which was calculated as log10 (bio lm-index spontaneous /bio lmindex MPMX ).
The in vitro tness was determined in a competitive growth assay with a reference isolate. Stationaryphase test and competitor isolates, adjusted to a McFarland index of 0.5, were incubated together for 24 hours at 37°C with constant shaking. The ratio of the two isolates was determined by comparing colony growth on a blood agar plate and an agar plate containing imipenem with the formulae log10(cfu test isolate /cfu competitor ). The reference strain PA14 was chosen for the XDR group, and an XDR isolate with balanced in vitro tness to PA14 was chosen for the other groups. A high initial bacterial concentration was used to emphasize assertiveness.
Serum susceptibility was determined using a killing assay. A suspension with the test isolate, adjusted to a statuary phase and CFU count of 2 X 10^3 cells/ml, and a 10% normal human serum was incubated for 30 min at 37°C, and the CFU reduction was determined and compared to a control sample with 10% heat inactivated normal human serum (30 min; 56°C). CFU counts were detected using the pour plate method on TSA after incubation for 48 h at 37°C. PMN susceptibility was determined using a killing assay under the same conditions, including serum susceptibility with inactivated human serum and a polymorphonuclear neutrophil leukocytes (PMN) concentration of 10,000/ml and compared to the control without PMN using the poor plate method.
PMNs were puri ed from fresh human whole blood taken from healthy volunteers and separated using Polymorphprep (Axis-Shield, Oslo, Norway).

Results
Phenotypical properties observed are shown in Table 2. Spontaneous bio lm formation was signi cantly higher in VET isolates, such as XDR and EI isolates, than in human non-XDR isolates. Within the EI isolates, we observed an increased spontaneous bio lm formation for EI pool isolates compared to EI hospitals and EI community isolates.
The stress response showed both isolates with induced and inhibited bio lm formation within each group at all MPM concentrations. Although generally less pronounced in the XDR isolates, inhibition predominated in all groups and MPM concentrations, with except for the cfXDR isolates at the subinhibitory level of 1 mg/L and the endemic strain ES-2 only at the highest used concentration of 125 mg/L.
A signi cantly higher in vitro tness was observed in XDR isolates than in all other groups (varying pvalues) except for the cfXDR isolates, which had the second lowest in vitro tness. The VET isolates showed higher tness than both EI and non-XDR isolates, with a signi cant difference in EI isolates (p<0.001).
VET isolates, such as XDR isolates, showed signi cantly higher serum susceptibilities than EI (p<0.0001) and non-XDR isolates (p<0.01). The cfXDR and XDR blood isolates had particularly high serum susceptibilities, although for these groups, increased exposure to the immune system must be assumed.
In terms of PMN susceptibility, the non-XDR group showed signi cantly lower values (varying p-values) compared to the other groups.
The genetically determined virulence factors are shown in Table 3. Note that cif, toxA and exoT were present in almost all isolates, and lasA and exoY were present in most of the isolates. The exoS and exoU genes were always exclusive.

Discussion
With regard to some properties showing values at the margins of the distribution, our VET isolates differed distinctly from our EI and non-XDR isolates, which presumably re ecting a generally less adapted and specialized phenotype of Pa. Therefore, a higher degree of phenotypical adaption to certain reservoirs can be assumed.
Increased spontaneous bio lm formation does not seem to be an uncommon adaptation to harsh environmental habitats, such as those of EI pool isolates or isolates adapted to habitats within mammalian hosts, such as our cfXDR isolates or possibly our VET isolates. A connection between adaptation and elevated spontaneous bio lm formation has been shown for Pa [15,16] and for other species, such as Staphylococcus aureus [17,18] and Acinetobacter baumannii [19]. In addition, bio lm formation has been shown to be related with both factors associated with persistence and virulence, like higher resistance to disinfection [20,21], an essential role in pathogenesis in various infection [22], an association with persistent infections, especially in cystic brosis patients [23] and increased mortality in hospitalized patients with bloodstream infections [24].
Because of their different antimicrobial resistance levels, comparing the stress responses of XDR isolates to other isolates is a challenge; however, the wide range of phenotypes within these groups is striking.
The diverse manifestations and the characteristic stress response of ES-2 indicate that inducible bio lm formation is presumably much more important in vivo than is represented in the literature and such bio lm formation may demonstrate even wider variations when other stimulants are considered [10,[25][26][27][28][29].
In particular, the capacity to build a high quantitative amount of bio lm is one of the outstanding properties of the two endemic strains and the cfXDR isolates studied herein, which underscore the connection between bio lm formation and persistence. Interestingly, these groups seem to be adapted to different environments, such as lung tissue for the cfXDR group and probably the hospital setting for endemic strains. Groups showing high spontaneous bio lm formation may be related via the pathogenesis of infections with these isolates. Isolates showing high spontaneous bio lm formation in vitro may behave in a similar way in vivo and build bio lms more readily than isolates that show lower in vitro levels of spontaneous bio lm formation.
There may also be a connection between spontaneous bio lm formation and high serum susceptibility in XDR and VET isolates. This is in sharp contrast to what is observed for EI and non-XDR isolates. In clinical isolates with high serum susceptibility -most obviously XDR blood isolates, one of the most susceptible groups to normal human serum and likely the group with the highest exposure -there must be a mechanism to compensate for this disadvantage, namely high bio lm formation. Furthermore, bio lm formation has been connected to reduced motility [15,30], so a connection with pathogenesis is possible. For these isolates, the route of infection is likely the spread of established bio lm communities or early bio lm manifestation by recently transferred Pa, rather than active colonisation.
Although we observed higher in vitro tness values for VET isolates compared to EI and non-XDR isolates, the high value obtained for XDR isolates was remarkable. Previous studies have indicated a negative association between inserted antibiotic resistance and in vitro tness [31][32][33]. Nevertheless, compensation mechanisms have been demonstrated [32,34]. Interestingly, in our study, those groups that seem to be phenotypically adapted at higher levels also show higher in vitro tness values. An important exception are the cfXDR isolates, possibly because this group went through the adaptation process in a habitat with lower microbiological competitive pressure, given that signi cant geno-and phenotypical alterations have been shown in CF-Isolates [35,36]. The intermediate in vitro tness observed in VET isolates may be an early form of this adaption. An important methodical aspect may be that these tests were carried out with a high initial bacterial count, such that the high in vitro tness observed may be the consequence of assertiveness rather than a high division rate.
The differences in the frequencies of nan1, which is associated with respiratory manifestation and adhesion to certain cells [9,37], and lasB, which is associated with cell invasion [38], may also indicate an alteration of the path of infection in XDR isolates. The similarity between the XDR isolates and the water groups, which both had higher exoU proportions, is interesting. It appears that exoU frequency increases in living environments that are near humans.
The XDR group demonstrated characteristics that are particularly associated with persistence and, as a result, an alteration in virulence properties. In contrast to the non-XDR group, there appears to be a shift in the path of infection in the XDR group towards the passive spread of bio lms. Interestingly, the XDR respiratory isolates, which have particular clinical importance, did not show properties as distinct as those of most of the other XDR subgroups, which may re ect a pathogenesis speci c to respiratory infections. This explanation may even apply to XDR wound isolates, which resemble most of the non-XDR group in their properties.
Within EI, we observed an elevation of spontaneous bio lm formation as well as exoU frequency in EI pool isolates compared to EI hospitals and EI community isolates. These properties seem to align with our observations for XDR and VET isolates, the closer environments they are obtained from is associated with people. Given the living environment of Pa from which we obtained our XDR isolates, adaptation to a human-associated environment seems more likely than direct adaptation to human tissue. One major exception is the cfXDR group, which shows unique properties and is presumably phenotypically adapted directly to human respiratory tissue [35,39,40].
In this study, veterinary isolates resembled XDR isolates and particularly XDR blood isolates with regard to many properties, particularly the combination of elevated bio lm formation, high serum susceptibility and genetically determined virulence factors. This may be the result of advanced adaptation in both groups, most likely to an environment close to the mammalian host in regard to the pathogenesis of infection or to the mammalian host itself. Whether this special phenotype is a consequence or a prerequisite for colonisation cannot be answered without further studies. However, it is disturbing that the veterinary isolate resembles the presumably most virulent XDR group obtained from clinical material. Notwithstanding the reasons why veterinary isolates show this pattern of features, ampli cation of potentially more virulent Pa strains, potentially leading to elevated zoonotic risks in immunocompromised and cystic brosis patients, must be considered.

Declarations
Ethics approval and consent to participate: All used isolates were routinely collected in the microbiology laboratory of the Heidelberg University Hospital or the Institute for Hygiene and Infectious Diseases of Animals of the University of Giessen and stored at -70°C. The current study thus is descriptive of those isolates. Data collected from patients was anonymized and restricted to possible clinical symptoms of infection. Ethical approval and informed consent statements were therefore not required.

Consent for publication: Not Applicable
Availability of data and material: All data and materials are fully available without restriction. All relevant data are within the manuscript and its Supporting Information les.

Competing interests:
The authors have no relevant nancial or non-nancial interests to disclose.

Funding:
The authors received no nancial support for the research, authorship, and/or publication of this article.