Effective Photodynamic Therapy with Ir(III) for Virulent Clinical Isolates of Extended-Spectrum Beta-Lactamase Klebsiella pneumoniae

Background: The extended-spectrum beta-lactamase (ESBL) Klebsiella pneumoniae is one of the leading causes of health-associated infections (HAIs), whose antibiotic treatments have been severely reduced. Moreover, HAI bacteria may harbor pathogenic factors such as siderophores, enzymes, or capsules, which increase the virulence of these strains. Thus, new therapies, such as antimicrobial photodynamic inactivation (aPDI), are needed. Method: A collection of 118 clinical isolates of K. pneumoniae was characterized by susceptibility and virulence through the determination of the minimum inhibitory concentration (MIC) of amikacin (Amk), cefotaxime (Cfx), ceftazidime (Cfz), imipenem (Imp), meropenem (Mer), and piperacillin–tazobactam (Pip–Taz); and, by PCR, the frequency of the virulence genes K2, magA, rmpA, entB, ybtS, and allS. Susceptibility to innate immunity, such as human serum, macrophages, and polymorphonuclear cells, was tested. All the strains were tested for sensitivity to the photosensitizer PSIR-3 (4 µg/mL) in a 17 µW/cm2 for 30 min aPDI. Results: A significantly higher frequency of virulence genes in ESBL than non-ESBL bacteria was observed. The isolates of the genotype K2+, ybtS+, and allS+ display enhanced virulence, since they showed higher resistance to human serum, as well as to phagocytosis. All strains are susceptible to the aPDI with PSIR-3 decreasing viability in 3log10. The combined treatment with Cfx improved the aPDI to 6log10 for the ESBL strains. The combined treatment is synergistic, as it showed a fractional inhibitory concentration (FIC) index value of 0.15. Conclusions: The aPDI effectively inhibits clinical isolates of K. pneumoniae, including the riskier strains of ESBL-producing bacteria and the K2+, ybtS+, and allS+ genotype. The aPDI with PSIR-3 is synergistic with Cfx.

the study protocol and the informed consent form. The clinical isolates of K. pneumoniae were obtained from 122 samples from unrelated patients, received in the bacteriology laboratory of "Hospital el Carmen" (HEC) for a period of six months (2017/2018). The HEC is a complex hospital with 412 beds and serving a population of more than 600,000 inhabitants. All clinical isolates were identified as K. pneumoniae following the protocols of the Institute of Clinical and Laboratory Standards (CLSI) [34]. As controls, the virulent strain (ATCC 43816 KPPR1) of susceptible K. pneumoniae (K2 + , ybtS + , and allS + ) [35] and the MDR KPC + ST258 (KP35) (K2 − , ybtS − and allS − ) [36] strains were also included.

Antimicrobial Susceptibility Testing
The MIC of the antimicrobial agents were determined in 96-well plates by microdilution methodology in cations-adjusted Mueller-Hinton (ca-MHB) broth. Inoculum of 1 × 10 6 colony forming units (CFUs)/mL of each clinical isolate was mixed with decreasing concentrations of each antibiotic and incubated overnight at 37 • C following the CLSI recommendations. The MIC for each antibiotic was determined as the last dilution in which no bacterial growth occurred, and the susceptibility intervals were assigned based on the cutoff points established by the CLSI (2018) for amikacin (Amk), cefotaxime (Cfx), ceftazidime (Cfz), imipenem (Imp), meropenem (Mer), and piperacillin-tazobactam (Pip-Taz). According to the CLSI protocols [34], clinical isolates were strains considered ESBL-producing when resistant to cefotaxime but susceptible to the combination of Cfx/clavulanic acid. Values are presented as the median in mg/L and interquartile range (IQR).

DNA Extraction and PCR Amplification
The total DNA was obtained from stationary bacterial cultures in LB broth using the phenol-chloroform methodology. In brief, pelleted bacteria were suspended in 300 µL of PBS and mixed with 300 µL of phenol: chloroform (25:24 vol:vol); the final mix was stirred well in a vortex and centrifuged at 13,000× g at 4 • C for 15 min. The aqueous phase was mixed 1:1 with chloroform and centrifuged as above. Genomic and plasmidial DNA contained in the aqueous phase was precipitated in 0.6 vol of 2-propanol and sedimented at 13,000× g at 4 • C for 20 min. The nucleic acids were washed twice with 70% ethanol and resuspended in Tris-EDTA buffer (10 mM Tris-HCl, 1 mM disodium EDTA, pH 8.0). PCR reactions were performed, using 0.5 µM of each specific primer pair listed in Table 1, in 1× master mix GoTaq (Promega). The amplification was carried out at a final volume of 20 µL in a Veriti (Applied Biosystem) PCR machine with an initial denaturation step of 10 s at 95 • C, followed by 35 cycles of 10 s at 95 • C, 15 s at 58 • C, and 30 s of extension at 72 • C. A final extension step of 7 min at 72 • C was included, and PCR products were visualized on a 1.7% agarose gel.

K. Pneumoniae Survival to Innate Immunity
To evaluate the survival of K. pneumoniae to the innate immunity, the bactericidal activity of normal human serum (NHS), as well as phagocytosis by human macrophages (MΦ), and polymorphonuclear (PMN) cells was determined. Both serum and leukocytes were obtained from blood samples of healthy voluntary donors who had not taken any antibiotic or anti-inflammatory medication for at least ten days before the day of sampling.

Susceptibility to Normal Human Serum
Serum susceptibility was carried out as before [37]; in brief, 75 µL of pooled NHS were mixed with 25 µL suspension containing 2 × 10 6 CFUs of each isolate in a 96-well plate. As a control, bacterial cultures of each isolate were mixed with PBS. The mixtures were incubated at 37 • C for 3 h, and viable bacteria were enumerated by serial micro-dilution and colony counting on ca-MH agar plates. Serum resistance is expressed as viable bacteria in CFUs/mL compared to untreated isolates controls. The separation of mononuclear and polymorphonuclear cells was performed by centrifugation in a gradient density column of Histopaque (Sigma-Aldrich, St. Louis, MO, USA), following the manufacturer instructions. Monocytes were selected from other mononuclear cells by incubation in RPMI-1640 medium (Sigma-Aldrich) without fetal bovine serum (FBS) and differentiated into macrophages incubating during 7-9 days in RPMI-1640 10% FBS at 37 • C with 5% CO 2 [38]. The macrophage monolayer was infected with 2.5 × 10 7 CFUs with a multiplicity of infection (MOI) of 50:1 of each bacterial isolate and centrifuged at 200× g for 5 min, to synchronize phagocytosis. After 2 h of incubation, cells were washed and incubated for an additional 60 min in RPMI-1640 + 100 µg/mL gentamicin. Macrophages were lysed with 0.1% saponin for 10 min at room temperature, and viable bacteria were enumerated as above. For PMN assays [39], 5 × 10 5 cells were combined with 5 × 10 6 CFUs of each isolate (MOI 10:1) in serum-free RPMI-1640 and synchronize by centrifugation at 524× g for 8 min at 4 • C. After 3 h, PMNs were lysed with 0.1% saponin, and viable bacteria were enumerated. Control groups of non-phagocyted bacteria were included.

Synthesis of the PSIR-3 Compound
The structural and photophysical characterization of the PSIR-3 compound was described previously [40]. The complex synthesized can be described by using the following general formula: [Ir(CˆN) 2 (NˆN)](PF 6 ), where NˆN is the ancillary ligand; and CˆN corresponds to a cyclometalating ligand. In this study PSIR-3 is [Ir(ppy) 2 (ppdh)]PF 6 where ppdh is pteridino(7,6-f)(1,10)phenanthroline-1,13(10H,12H)-dihydroxy and ppy is 2-phenylpyridine [41]. The structure and purities of the compound were confirmed by nuclear magnetic resonance (NMR), Fourier-transform infrared spectroscopy (FTIR), and mass spectroscopy (MALDI-MS) measurements. The absorption spectra were measured in acetonitrile (ACN) solutions using a Shimadzu UV-Vis Spectrophotometer UV-1900. The molar extinction coefficients of the characteristic bands were determined from the absorption spectra. Photoluminescence spectra were taken on an Edinburgh Instrument spectrofluorimeter using ACN solutions of the compounds previously degassed with N 2 for approximately 20 min. The emission quantum yields (Φ em ) were calculated according to the description of the literature [42]. Fluorescence lifetimes were measured by using a timecorrelated single-photon counting (TC-SPC) apparatus (PicoQuant Picoharp 300) equipped with a sub-nanosecond LED source (excitation at 380 nm) powered by a PicoQuant PDL 800-B variable (2.5−40 MHz) pulsed power supply.

Antimicrobial Activity of Photosensitizer Compounds
Stock solutions of 2 g/L of the PSIR-3 compound solubilized in ACN were used to prepare working solutions in distilled water. For the antimicrobial assay, the collection of 118 clinical isolates of K. pneumoniae was used, and the control strains of susceptible K. pneumoniae KPPR1 and the MDR strain ST258 were also included. All bacteria were grown as axenic culture in Luria Bertani broth or agar medium as appropriate. PSIR-3 was mixed in 24-well plates at a final concentration of 4 mg/L for photodynamic experiments, with suspensions of 1 × 10 7 colony forming units (CFUs)/mL of each bacterium, in a final volume of 500 µL of cation-adjusted Mueller-Hinton (ca-MH) broth. Exposure to light was performed for 30 min in a chamber with a white LED lamp at a photon flux of 17 µW/cm 2 . After exposure to light, the CFUs of the viable bacteria were determined by broth-micro dilution and sub-cultured on ca-MH agar plates. Following the recommendations of the Clinical and Laboratory Standards Institute (CLSI 2017) [34], the agar plates were incubated during 16-20 h at 37 • C, in the dark, and colony count was recorded, using a stereoscopic microscope. Control wells with bacteria culture with no photosensitizer or photosensitizer but not exposed to light were also included.

Determination of the Synergy between PSIR-3 and Cfx
The fractional inhibitory concentration index (FIC) value was determined using the following formula [43,44]. MICac is the MIC of a compound A, combined with a compound B, and MICbc is the MIC of the compound B combined with the compound A. The MICa and MICb are the MIC of the A and B compounds alone, respectively. Values in the FIC index ≤0.5 are considered synergistic, and values > 4 are considered antagonistic [44].
To determine the MIC-Cfx combined with each PSs, 1 × 10 7 UFC/mL of ESBLproducing bacteria were aPDI treated for 30 min with 4 mg/L of each PS and mixed with serial dilution (32-0.125 mg/L) of Cfx in ca-MH broth, as above. To determine the MIC-PSs combined with Cfx, 1 × 10 7 UFC/mL of ESBL-producing bacteria was mixed with serial dilution of each PSs (32-0.125 mg/L) and a fixed concentration of 4 mg/L of Cfx, and then it was subjected to aPDI, as above.

Statistical Analyses
Statistical analyses were performed by using the Systat 13.2 software (Systat Software, Inc., San Jose, CA, USA) and GraphPad v6.01 (Prism) software. The X 2 test or Fisher's exact test for categorical variables and the Mann-Whitney U test for continuous non-parametric variables were used. The risk of virulence genes to modify the MIC of the antibiotic was established by determining the odds ratio with a CI: 95%.

Demographic Characterization
This work seeks to demonstrate the capacity of aPDI to inhibiting the growth of clinical isolates of K. pneumoniae, which are diverse in antimicrobial susceptibility, genotype, and virulence. Phenotypic and genotypic characterization was conducted to determine the frequency of genes encoding virulence factors, the MIC values determined for various antibiotics, and susceptibility to innate immune components. From the clinical isolates of K. pneumoniae received in the laboratory, 118 were selected from different unrelated patients. The isolates were recovered mainly from urine samples, 113 (95.8%), and only five from respiratory samples (three endotracheal aspirates (2.5%) and two expectorations (1.7%)). As shown in Table 2, the samples were obtained from 78 (66%) females and 40 (34%) males, from 12 clinical services, with a higher contribution from the emergency room 41 (34.75%), secondly the unit of medicine 24 (20.34%), third ambulatory 16 (13.56%) and Pharmaceutics 2021, 13, 603 6 of 17 in the fourth place urology 14 (11.86%). Of the 118 clinical isolates, 114 came from adults and 4 from pediatric patients. As shown in Figure 1A, the patient's age fluctuated between 7 months and 94 years, with a median (IQR: 25-75%) of 69.5 (54.8-84) years for females, and between 10 months and 92 years, with a median of 68.5 (60.3-80.8) years for males. There are no significant differences in age between genders (p = 0.279 Mann-Whitney U test). The results confirm the infections with MDR and not MDR K. pneumoniae mainly affects the elderly population. The elderly are the most susceptible population to present complications derived from infectious diseases [45].

Antibiotic Susceptibility
In the K. pneumoniae population, the median and IQR (25-75%) of the MIC were determined and expressed in mg/L in a log2 box plot. As shown in Figure 1B, the bacterial

Virulence Gene Frequency
In this work, we select certain virulence genes belonging to families with different mechanisms of action. PCR determined the frequency of carrying genes encoding the virulence factors rmpA, magA, K2, entB, allS, and ybtS for each isolate using specific primers (Table 1). These genes were selected for being representatives of different families of virulence factors. As shown in Figure 1C . The most frequent virulence factor was the entB gene, followed by the ybtS gene. There were no isolates with the magA gene, and only three isolates harbor the rmpA gene. No isolates showed the hypermucoviscosity phenotype (by string test) in agar plates in the population of 118 unrelated isolates.

Correlation of Virulence Factors with Antibiotic Resistance
As shown in Figure 1D, there is a significantly higher frequency of ESBL strains that harbor three or more virulence factors compared to non-ESBL strains (Fisher's; p < 0.026). The non-parametric Mann-Whitney U test (Systat 13 software) was used to determine the association that each virulence gene has on the median MIC value assuming a null hypothesis α = 5%. As shown in Table 3, the rmpA gene was not significantly associated with a modification of the median-MIC of any of the antibiotics analyzed in this study (p > 0.05). On the other hand, the allS gene significantly influenced the median-MIC of Cfx, Cfz, and Pip-Taz (p < 0.05). Similarly, the K2 gene significantly influenced the median-MIC of Cfx and Pip-Taz (p < 0.05). The entB gene significantly affected the median-MIC of Amk and Imp (p < 0.05). The ybtS gene significantly influenced the median-MIC of Cfx (p < 0.006). Finally, the allS gene significantly influenced the median-MIC of Cfx and Cfz antibiotics (p < 0.05). The MIC of the antibiotics more sensitive to the presence of virulence factors were the Cfx (sensitive to K2, ybtS, and allS genes) and Pip-Taz (sensitive to the K2 and allS genes). These data show that, regardless of the patient's conditions, the multi-virulence is effectively an independent risk factor that promotes ESBL-production of clinical populations of K. pneumoniae with a p < 0.026 Fisher's exact test. Values in bold represent a p < 0.05, below the null hypothesis value with α = 5%, which means a significant difference in MIC due to the presence of the virulence factor. Pip-Taz, piperacillin-tazobactam.
As shown in Figure 2, the box plot for each antibiotic stratified by the presence or absence of virulence genes was constructed to verify if the influence is to increase or decrease the median of the MIC. The presence of the entB gene is associated with a decrease in the median-MIC for Amk and Imp. On the other hand, the K2 gene is associated with an increase in the median-MIC for Cfx and Pip-Taz. Similarly, the ybtS gene is associated with an increase in the median-MIC for Cfx. Finally, the allS gene is associated with an increased median-MIC for Cfz and decreased median-MIC for Cfx and Pip-Taz. Some of the virulence factors studied here were associated with the median-MIC modification for several antibiotics. In the K. pneumoniae population tested, the most influencing virulence factors were entB, ybtS, and allS genes. The stratified MICs-box plot made it possible to distinguish how virulence genes modulate antibiotic susceptibility by increasing or decreasing. Remarkably, the sum of the genes of the K2 + , ybtS + , and allS + genes contributes to increasing the median-MICs to values higher than the clinical susceptibility breakpoint established by the CLSI. These increased values occur similarly when efflux pumps are activated in other Gram-negative bacteria [46]. For example, in a population of Pseudomonas aeruginosa, the combined overexpression of the mexA and mexX efflux pump increased the median MIC for ciprofloxacin and cefepime above the cutoff points [46,47].

Association of K2 + , ybtS + , and allS + Virulence Genes to Survive the Innate Immunity
Our results show that the K2 and ybtS virulence genes are risk factors for the production of ESBL and that the allS gene acts as a protective factor. Then, we selected two groups of clinical isolates of K. pneumoniae, which harbor or not these three genes, to assess their susceptibility to innate immunity components. Only three isolates, identified by the numbers 81, 92, and 111, are of the K2 + , ybtS + , and allS + genotype, and unexpectedly all of them are ESBL-producers. From the genotype K2 − , ybtS − and allS − , the isolates 13, 21, and 22 were selected as being non-ESBL. As controls, the susceptible but highly virulent K. pneumoniae KPPR1 (K2 + , ybtS + , and allS + ) and the MDR but less virulent ST258 (K2 − , ybtS − and allS − ) strains were also included.
To determine the bacterial susceptibility to normal human serum (NHS), they were exposed for 3 h, and then the number of viable bacteria was determined by microdilution and plate counting. As shown in Figure 3A, the presence of all three virulence genes, K2 + , ybtS + , and allS + , significantly (** = p < 0.003; two-way ANOVA) increases the survival of ESBL-producing bacteria compared to non-ESBL bacteria lacking all three virulence genes. On average, the serum resistance was improved by four orders of magnitude (4log 10 ). Similar to that observed with serum, ESBL-producing bacteria of the K2 + , ybtS + , and allS + genotype show a significant (**** = p < 0.0001) increase, 4log 10 , in the survival to the MΦ activity, compared to no-ESBL bacteria that lacks these virulence genes ( Figure 3B). Finally, comparable to MΦ, survival to the activity of PMNs of the bacteria of genotype K2 + , ybtS + , and allS + increased significantly (* = p < 0.02), at least one time on average, compared to non-ESBL strains that lack the virulence genes ( Figure 3C). As shown in Figure 3, the virulent control KPPR1 strain was more resistant to serum ( Figure 3D), macrophages ( Figure 3E), and PMN ( Figure 3F) compared to the MDR ST258 K. pneumoniae strain. Consistently the clinical control isolates have shown to be more susceptible to the bactericidal activity of the serum (3D) and the phagocytic activity of MΦ (3E) and PMN (3F). The co-occurrence of harboring multiple genes that encode virulence factors and the ESBL-production leads to enhanced virulence. The ESBL-producing strains of K. pneumoniae of the genotype K2 + , ybtS + , and allS + were more resistant to innate immunity, consistently with studies over MDR populations of K. pneumoniae that increased their 30-day mortality over patients undergoing bloodstream infections [1,48]. Moreover, virulence genes, such as siderophores, which have an essential role in bacterial survival and virulence [49,50], have been previously associated with the MDR-K. pneumoniae [16,51]. In this study, ESBL-producer K. pneumoniae of the K2 + , ybtS + , and allS + genotype shown a survival improvement for killing by PMNs. Previously, the PMN has demonstrated a limited binding and uptake capacity for MDR-K. pneumoniae [39]. The activity of the PMN is the most important cellular component of the innate immune response, essential as the first line of defense against bacterial infections [52].
In this work, we tested a coordination compound characterized by a positive charge in the first coordination sphere ( Figure 4B). The photophysical evaluation of the PSIR-3 performed in acetonitrile solution [40] revealed absorption processes at 375 and 392 nm attributable at the first instance of charge-transfer transitions ( Figure 4C,D) [32]. When the Figure 3. Effect of virulence genes on the susceptibility of ESBL or non-ESBL strains to components of innate immunity. The susceptibility of ESBL-producing K. pneumoniae clinical isolates bearing the virulence genes ybtS + , K2 + , and allS + was determined and compared with non-ESBL producing bacteria that lack the virulence genes. (A,D) Susceptibility to normal human serum (NHS), (B,E) susceptibility to phagocytosis by macrophages (MΦ), and (C,F) susceptibility to phagocytosis by polymorphonuclear cells (PMN). The results of two independent experiments performed in triplicate are shown (n = 6). Viable bacteria were enumerated by colony count on ca-MH agar after serial micro-dilution. The colony forming units (CFUs)/mL values are presented as means +/− SD, on a log 10 scale of treated bacteria (black bars) compared to untreated control bacteria (gray bars). **** = p < 0.0001, ** = p < 0.003 and * = p < 0.02 of two-way ANOVA comparing the proportion of treated/untreated ESBL bacteria with non-ESBL bacteria.

Susceptibility of Clinical Isolates to aPDI with PSIR-3 3.6.1. Photophysical Properties of the PSIR-3 Compound
We have previously shown that Ir(III)-based compounds, such as PSIR-3, have photodynamic antimicrobial activity against imipenem-resistant Klebsiella pneumoniae [32,33]. In this work, we tested a coordination compound characterized by a positive charge in the first coordination sphere ( Figure 4B). The photophysical evaluation of the PSIR-3 performed in acetonitrile solution [40] revealed absorption processes at 375 and 392 nm attributable at the first instance of charge-transfer transitions ( Figure 4C,D) [32]. When the PSIR-3 compound was excited with a wavelength corresponding to the lowest charge-transfer absorption energy, 375 nm, it showed maximum emission at 598 nm ( Figure 4C,D). Figure 4C shows the recorded lifetimes of excited states in 0.32 µs and the calculated quantum yield (Φ em ) in 0.011 [40]. The aPDI activity of the PSIR-3 compound was compared with the positive PS control [Ru(bpy) 3 ](PF 6 ) 2 (bpy = 2,2 -bipyridine) called PS-Ru. According to the literature, the PS-Ru shows a charge-transfer absorption process at 450 nm with maximum emission at 600 nm (excited in 450 nm) in acetonitrile [42], with Φ em of 0.095 [42], and a lifetime registered of its excited state of 0.855 µs ( Figure 4C) [53]. The maximum absorption of PSIR-3 occurs at wavelengths below 400 nm, that although it is more energetic, it penetrates the tissues poorly. Therefore, this compound will activate better if it is directly irradiated, such as in superficial wounds. However, UTI is one of the most common diseases caused by K. pneumoniae, where probes that deliver the light dose within internal surfaces can irradiate the epithelial lining of the bladder [54]. Certain kinds of fiber arrays or inflatable balloons may provide a homogenous light power delivery [55,56]. This catheterization can be applied for inpatients suffering UTI that does not respond to antibiotic treatment.
Pharmaceutics 2021, 13, x FOR PEER REVIEW 12 of 18 PSIR-3 compound was excited with a wavelength corresponding to the lowest chargetransfer absorption energy, 375 nm, it showed maximum emission at 598 nm ( Figure  4C,D). Figure 4C shows the recorded lifetimes of excited states in 0.32 µs and the calculated quantum yield (Φem) in 0.011 [40]. The aPDI activity of the PSIR-3 compound was compared with the positive PS control [Ru(bpy)3](PF6)2 (bpy = 2,2′-bipyridine) called PS-Ru. According to the literature, the PS-Ru shows a charge-transfer absorption process at 450 nm with maximum emission at 600 nm (excited in 450 nm) in acetonitrile [42], with Φem of 0.095 [42], and a lifetime registered of its excited state of 0.855 µs ( Figure 4C) [53]. The maximum absorption of PSIR-3 occurs at wavelengths below 400 nm, that although it is more energetic, it penetrates the tissues poorly. Therefore, this compound will activate better if it is directly irradiated, such as in superficial wounds. However, UTI is one of the most common diseases caused by K. pneumoniae, where probes that deliver the light dose within internal surfaces can irradiate the epithelial lining of the bladder [54]. Certain kinds of fiber arrays or inflatable balloons may provide a homogenous light power delivery [55,56]. This catheterization can be applied for inpatients suffering UTI that does not respond to antibiotic treatment.

Antimicrobial Photodynamic Inhibition of the PSIR-3 over Clinical Isolates
Photodynamic treatment was verified to produce the observed growth inhibition of the 118 clinical isolates of K. pneumoniae compared to the untreated bacteria. The photodynamic activity of the PSIR-3 compound was compared to the activity of the PS-Ru reference compound as a positive control [41,[57][58][59]. As seen in Figure 5, compared to the control of untreated bacteria, photodynamic treatment with 4 µg/mL PSIR-3 inhibits bacterial growth > 3 log 10 (>99.9%) of clinical isolates of K. pneumoniae (**** p < 0.0001; compared to untreated control). The results show that the bactericidal effect produced by PSIR-3 is light-dependent (ns = p > 0.05; compared to the untreated control). These results are comparable with those obtained with the positive control compound PS-Ru, which has shown that bacterial growth inhibition is light-dependent (**** p < 0.0001; compared to the untreated control).
the 118 clinical isolates of K. pneumoniae compared to the untreated bacteria. The photodynamic activity of the PSIR-3 compound was compared to the activity of the PS-Ru reference compound as a positive control [41,[57][58][59]. As seen in Figure 5, compared to the control of untreated bacteria, photodynamic treatment with 4 µg/mL PSIR-3 inhibits bacterial growth > 3 log10 (>99.9%) of clinical isolates of K. pneumoniae (**** p < 0.0001; compared to untreated control). The results show that the bactericidal effect produced by PSIR-3 is light-dependent (ns = p > 0.05; compared to the untreated control). These results are comparable with those obtained with the positive control compound PS-Ru, which has shown that bacterial growth inhibition is light-dependent (**** p < 0.0001; compared to the untreated control).  For the aPDI, the mixture of bacteria with PS was exposed for 1 h at 17 µW/cm 2 of white light. As a control, bacteria combined with the photosensitizers (PSs) not exposed to light (PSIR-3 or PS-Ru) and bacteria not combined with the PSs (control) were included. Colony count enumerated of viable bacteria on ca-MH agar after serial micro-dilution. The CFUs/mL values are presented as means ± SD on a log 10 scale. (B) From the clinical isolates, 66 ESBL-producing bacteria were exposed to aPDI, using PSIR-3 or PS-Ru, and MIC for cefotaxime (Cfx) was performed in triplicate, in ca-MH broth, for 16-20 h. (C) For ESBL-producing bacteria, the MIC for PSIR-3 or PS-Ru, were determined in combination with 4 mg/L of Cfx, performed in triplicate in ca-MH agar for 16-20 h. The MIC values are presented as median ± SD of mg/mL on a log 2 scale. Not significant (ns) p > 0.05 by Student's t-test among treated bacteria compared to control; **** p < 0.0001 by Student's t-test among treated bacteria compared to control.

Synergism between aPDI with PSIR-3 and Cefotaxime
Since PSIR-3 showed synergism combined with imipenem [32], we analyzed whether it shows synergism with cefotaxime in the population of clinical isolates. First, it was determined whether the combined treatment with Cfx increases the inhibition of bacterial growth of aPDI with PSIR-3. The 118 clinical isolates of K. pneumoniae were exposed to the preparation of 4 µg/mL of cefotaxime with 4 µg/mL of PSIR-3 (its MIC). Control bacteria without Cfx and exposure to light were included. As expected, the PSIR-3 compound mixed with cefotaxime significantly (**** p < 0.0001) increased the bactericidal effect on the clinical isolates population from 3 to 6 log 10 reduction ( Figure 5A). As seen before, a significantly increased inhibitory effect was not observed when combining the cefotaxime with the PS-Ru control compound (ns p > 0.05). Secondly, the set of 66 clinical isolates characterized as ESBL-producing K. pneumoniae were treated with PSIR-3 aPDIfor 1 h, and serial dilutions determined the MIC for Cfx in ca-MH broth. As seen in Figure 5B, a significant reduction (**** p < 0.0001) from 8 µg/mL  to 0.17 µg/mL (0.17-0.333) on Cfx-MIC was observed compared to the untreated group. The combined treatment also reduced the PSIR-3-MIC, from 4 to 0.5 mg/L ( Figure 5C). Similar to previously shown with imipenem [32], compound PSIR-3 produced a significant change in Cfx-susceptibility with a fractional inhibitory concentration (FIC) index of 0.15 (Table 4). Figure 5C shows the control compound, PS-Ru, did not significantly change Cfx-susceptibility with an FIC Index 1.58 (Table 4). Because synergy is defined as an FIC index of ≤0.5 [44], the increased inhibitory effect observed when combining PSIR-3 with Cfx is not summative but synergistic. The behavior exhibited by PSIR-3 must be related, as mentioned in previous reports [32,33], to the external substituents bonded to polypyridine ligand structures and affinity to bacterial envelope [41]. This synergism is also comparable to other photosensitizers; for example, rose bengal showed an increase in the susceptibility of Acinetobacter baumannii for a range of antibiotics used along with aPDI [60]. In another example, conventional antibiotics and alternative compounds reported synergism in a murine model for pathogens of the ESKAPE group (Enterococcus faecium, Staphylococcus aureus, Klebsiella pneumoniae, Acinetobacter baumannii, Pseudomonas aeruginosa, and Enterobacter species) [61], using anti-biofilm peptides [62]. For now, it is difficult to accurately calculate the dose of light necessary to activate PSIR-3 effectively; however, we observed in vitro that with low doses of energy (17 µW/cm 2 of white light), it is bactericidal. Compounds with optimum absorbance at higher wavelengths, bordering the 630-750 nm, would improve the exposure of PS to light into the tissues, but being less energetic, the PSs must have a triplet excited state of easy access to promote energy transfer [63]. We, therefore, need to perform a better characterization of its antimicrobial activity when activated with defined wavelengths, before starting in vivo studies, in urinary infection models and to evaluate the need to deliver light through intraurethral catheters [64].

Conclusions
In this work, we saw that multi-drug resistance and virulence are significant factors in clinical isolates of K. pneumoniae. However, the increase in MICs can be neutralized by aPDI, turning resistant strains susceptible. APDI is effective in treating multidrug-resistant bacteria and more virulent strains, as well as strains that combine both characteristics. The aPDI then becomes a great support to antimicrobial therapy in a shortage of new effective antibiotics. The photophysical characterization of PS indicates that its maximum absorption occurs at wavelengths lower than 400 nm, which could constitute a problem for its use in infections of internal organs due to low penetration. In UTI, optical fibers can be used through a catheter to deliver the dose of light [54,55]. Moreover, compounds with optimum absorbance at higher wavelengths, bordering the 630-750 nm, would improve the exposure of PS to light in the tissues.

Data Availability Statement:
The data presented in this study are available on request from the corresponding author. The data are not publicly available because they are confidential data of patients protected by the informed consent protocol.