Photodynamic eradication of intratumoral microbiota with bacteria‐targeted micelles overcomes gemcitabine resistance of pancreatic cancer

Increasing evidence suggests that intratumoral microbiota plays a pivotal role in tumor progression, immunosurveillance, metastasis, and chemosensitivity. Particularly, in pancreatic ductal adenocarcinoma, tumor‐resident Gammaproteobacteria could transform the chemotherapeutic drug gemcitabine (Gem) into its inactive form, thus rendering chemotherapy ineffective. Herein, a strategy for selectively eradicating intratumoral bacteria was described for overcoming Gem resistance in a pancreatic cancer animal model. An antimicrobial peptide was linked with photosensitizer through a poly (ethylene glycol) chain, which can self‐assemble into micelles with a diameter of ∼20 nm. The micelles could efficiently kill bacteria under light irradiation by inducing membrane depolarization, thereby inhibiting Gem metabolism. In a bacteria‐resident pancreatic cancer animal model, the selective photodynamic eradication of intratumoral bacteria was demonstrated to efficiently reverse Gem resistance. This research highlights antibacterial photodynamic therapy as a promising adjuvant strategy for cancer therapy by modulating intratumoral microbiota.


INTRODUCTION
Despite substantial advances in cancer treatment, pancreatic ductal adenocarcinoma (PDAC) remains a disease of near uniform mortality with a 5-year overall survival of 9%. [1]he location of the pancreas is unique, and close to the digestive tract, which leads to a poor prognosis for surgery.
In addition, PDAC is less sensitive to radiotherapy. [2]Therefore, chemotherapy is one of the most common treatment modalities for PDAC.Gemcitabine (Gem) has been accepted as the standard first-line therapy for patients with advanced PDAC.Since it was proven to be superior to 5-fluorouracil for the treatment of PDAC in 1997, no single agent has been able to surpass Gem in terms of therapeutic efficacy, and even the combination chemotherapy is mostly based on Gem. [3,4]However, the clinical benefit response rate of PDAC patients to Gem was only 23.8% and Gem resistance develops within a few weeks of chemotherapy initiation. [5,6]herefore, developing effective strategies to overcome Gem resistance is highly desired.
Increasing evidence suggests that tumor-associated microbiota is an intrinsic component of the tumor microenvironment across at least 33 major cancer types. [7]0] In particular, Geller et al. found that bacteria colonized in PDAC, mainly Gammaproteobacteria, could produce a long isoform of bacterial enzyme cytidine deaminase (CDD L ) to metabolize Gem into its inactive form, 2′,2′difluorodeoxyuridine(dFdU). [11]This implies the presence of Gammaproteobacteria in PDAC may contribute to the Gem resistance.Although systemic administration of antibiotics could potentially reverse Gem resistance by eradicating intratumoral Gammaproteobacteria, this treatment may disturb the balance of gut microbiota, which has been shown to modulate the anticancer drug efficacy. [12,13]In addition, the growing problem of antibiotic resistance has made conventional antibiotic therapy less efficient.Therefore, there is a critical unmet need for a novel antibacterial technique that can precisely modulate intratumoral microbiota while not affecting gut microbiota. [14][22][23][24][25] However, the application of these photosensitizers in PDT has been limited by their poor water solubility and lack of selectivity to distinguish bacteria and mammalian cells.
A widely accepted strategy for bacteria-targeted therapy or imaging has been to label antimicrobial peptides, which were originally found as essential components of the innate immune systems to combat microbial threat. [26]One typical example is ubiquicidin (UBI), an antimicrobial protein that is conserved in eukaryotic cells from yeast to humans.UBI(29-41) with a sequence of TGRAKRRMQYNRR is a fragment of UBI protein found to possess potent antimicrobial activity against antibiotic-resistant bacteria.The UBI(29-41) peptide can preferentially bind to bacteria but not to mammalian cells.[29] Therefore, the conjugation of UBI(29-41) peptide with photosensitizers was expected to achieve selective photodynamic eradication of intratumoral bacteria.
Herein, a chlorophyll-derived photosensitizer, pyropheophorbide-a (Ppa), was conjugated with UBI(29-41) peptide through a monodisperse poly(ethylene glycol) (PEG) chain (Scheme 1).The obtained amphiphilic conjugates could self-assemble into micelles (UPPM) in aqueous solution.The UBI(29-41) peptide endows UPPM with excellent bacteria-targeting capability.Under light irradiation, UPPM could inhibit bacteria growth by inducing bacterial membrane damage, thereby inhibiting the Gem metabolism.In the bacteria-resident subcutaneous PDAC animal model, the bacteria-targeted photodynamic therapy with UPPM could eradicate intratumoral bacteria while not affecting gut microbiota.The effect of chemotherapy with Gem after photodynamic eradication of intratumoral bacteria was significantly enhanced compared with the control groups.

RESULTS AND DISCUSSION
Ppa was selected for photodynamic bacteria eradication due to its high singlet oxygen quantum yield and molar extinction coefficient in the near-infrared region. [30,31]To improve the water solubility of Ppa, a monodisperse heterofunctional PEG chain was utilized to conjugate Ppa with UBI(29-41) peptide to obtain Ppa-PEG-UBI polymer (Figures S1 and  S2).For comparison, Ppa-PEG without UBI(29-41) peptide was also synthesized.Due to the excellent water solubility of the PEG chain, Ppa-PEG-UBI, and Ppa-PEG could self-assemble into micelles in aqueous solution, denoted as UPPM and PPM, respectively.Under transmission electron microscopy (TEM), UPPM and PPM showed uniform spherical morphology with an average size of 19.75 ± 3.18 and 19.97 ± 2.35 nm, respectively (Figure 1A).UV-vis absorption spectra revealed that both UPPM and PPM in phosphate-buffered solution (PBS) exhibited blueshifted Soret (∼387 nm) and redshifted Q bands (∼681 nm) compared to Ppa, Ppa-PEG, and Ppa-PEG-UBI monomer (Figure 1B), which suggested the generation of both Haggregates and J-aggregates. [32]The fluorescence spectra revealed that there was much less fluorescence quenching for UPPM than PPM in PBS (Figure 1C).The Zeta potential of UPPM and PPM was determined to be 20.7 ± 3.0 and −1.1 ± 1.9 mV, respectively, which further confirmed the successful modification of UBI(29-41) peptide containing many positively charged amine acid residues in UPPM.The ROSgenerating capability of UPPM was further evaluated using the commercially available probe Singlet Oxygen Sensor Green.As shown in Figure 1E, under laser irradiation, UPPM exhibited a highly improved production of singlet oxygen compared with free Ppa and PPM in PBS.
To investigate the potential of UPPM for photodynamic eradication of microbiota resident in the PDAC, Escherichia coli Nissle 1917 (EcN) was selected as the representative of intratumoral bacteria.EcN belongs to Gammaproteobacteria, which is the main bacterial phylum colonized in PDAC and accounts for Gem resistance of PDAC by producing CDD L . [11,33]EcN has been widely utilized as a delivery vehicle for cancer therapy or diagnosis, due to its tumortargeting capability and ease of genetic manipulation. [34,35]oreover, EcN has been approved as a probiotic drug in many countries for the treatment of gastrointestinal diseases.To facilitate bacterial tracking, EcN was electrotransformed with two plasmids expressing superfolder green fluorescence protein (sfGFP) and luxCDABE cassette, respectively (Figure S3). [36]The tagged EcN can be localized with fluorescence imaging in vitro or bioluminescence imaging in vivo (Figure S4).
Due to the UBI(29-41) peptide modification, UPPM was expected to specifically bind with negatively charged bacterial cell membranes.As shown in Figure 2A, the red fluorescence of Ppa in UPPM exhibited good colocalization with the green fluorescence of EcN, whereas almost no PPM was observed to colocalize with EcN.Subsequently, flow cytometry was performed to quantify the selectively binding efficiency of UPPM with EcN (Figure 2B,C).The percentage of Ppa-positive EcN in the UPPM group was about 15-fold higher than that of PPM.Taken together, these results suggested the excellent bacteria-targeting capability of UPPM, making it potential for further therapeutic applications.To quantitatively evaluate the photodynamic antibacterial activity of UPPM, the bacteria viability was determined by the plate count method.EcN were incubated with UPPM and PPM at various concentrations from 0 to 12 μM for 1.5 h at 30 • C, respectively, followed by light irradiation.Subsequently, the corresponding bacteria were homogeneously spread across the surface of antibiotic-selective lysogeny broth agar plates and a standard plate counting assay was conducted to evaluate the bactericidal ability of UPPM and PPM.As shown in Figure 2D,E, the light irradiation alone did not reveal a significant impact on bacterial viability.Under light irradiation, UPPM at a concentration as low as 9 μM could nearly kill bacteria completely.On the contrary, no significant bacteria-killing was observed with PPM at a concentration as high as 12 μM.Moreover, without light irradiation, both UPPM and PPM did not induce any change in bacterial viability at a concentration as high as 100 μM (Figure S5).These results highlighted the potential of UPPM for antibacterial therapy triggered by a UBI(29-41) peptide and light irradiation dual-targeted mechanism.
The positively charged UBI(29-41) peptide was known to selectively bind with anionic bacterial cell membrane. [26]herefore, the bacteria-targeted photodynamic therapy with UBI(29-41) peptide was expected to result in ROS-induced damage to the bacteria cell membrane. [37]To determine the change of bacterial membrane after the treatment with UPPM, the membrane potential was monitored by flow cytometry with a fluorescent membrane-potential indicator DiOC 2 (3).DiOC 2 (3) exhibits green fluorescence in all bacteria, but it becomes more concentrated in the cytosol of healthy bacteria maintaining a high membrane potential, leading to its self-associate and the red-shifted fluorescence emission.The ratio of red and green fluorescence was used to illustrate EcN membrane potential.EcN cells were treated with UPPM and PPM under dark or light irradiation, followed by DiOC 2 (3) staining and flow cytometry analysis.A proton ionophore (CCCP), which could induce thorough membrane depolarization and a low ratio of red to green, was set as a positive control.Compared with PPM under light irradiation and UPPM under dark, there were higher counts of bacteria in the Q3 region for bacteria treated with UPPM under light irradiation (Figure 2F).In addition, UPPM under light irradiation exhibited the lowest red/green ratio compared to other groups (Figure 2G), confirming that bacteria-targeted UPPM induced bacteria membrane depolarization and significant damage.
As a Gammaproteobacteria, EcN expresses CDD L mediating the deamination of Gem to produce dFdU.With high-performance liquid chromatography, the conversion of Gem to dFdU can be quantitatively monitored (Figure 3A and Figure S6).After incubating 5 × 10 6 /mL EcN with 50 μM Gem, the concentration of Gem decreased with time going on while increasing dFdU were formed (Figure 3B).At 14 h, nearly all the Gem were converted into dFdU.Due to the potent photo-bactericidal capability of UPPM, the Gem metabolism could be inhibited with the increase of UPPM concentration, while Gem metabolism was not significantly affected with PPM (Figure 3C and Figure S7).We further evaluated the anti-cancer effect of metabolized Gem by EcN under different treatments in murine pancreatic cancer Panc02 cell line.As expected, without UPPM treatment, Gem-containing media that was incubated with EcN for 14 h lost the Panc02-killing capability (Figure 3D).With the increasing UPPM concentration, the Panc02-killing capability of Gem was reserved.On the contrary, the viability of Panc02 cells was not significantly changed with PPM.Collectively, these results demonstrated that the photobactericidal capability of UPPM could efficiently inhibit Gem metabolism and reserve the anti-cancer capability of Gem.
To evaluate the therapeutic efficacy of UPPM in vivo, a murine EcN-bearing tumor model was established by simultaneously injecting EcN bacteria and Panc02 cells subcutaneously in the left limb of the Balb/c nude mice.With the luxCDABE cassette transformed in EcN, the EcN bacteria in mice can be monitored with bioluminescence imaging.The bacteria were stably colonized in the tumor as evidenced by the in vivo bioluminescence imaging and ex vivo fluorescence imaging of tumor sections (Figure S8).In addition, EcN also translocated in the gut of the mice (Figure S8A).The tumortargeting ability of UPPM was evaluated with fluorescence imaging using the fluorescence emission of Ppa.As shown in Figure 4, the accumulation of UPPM in tumors was significantly higher than that of PPM, due to the bacteria-targeting capability of UBI(29-41) peptide.The fluorescence of UPPM reached a maximum at the 5-h post-injection and this time point was utilized for further therapeutic applications.
We further evaluated the efficacy of photodynamic bacteria eradication in overcoming Gem resistance using the EcN-bearing tumor model.The tumor-bearing mice were randomly divided into five groups: (a) UPPM+L+Gem; (b) PPM+L+Gem; (c) UPPM+L; (d) Gem; (e) Saline.For the UPPM+L+Gem group, UPPM at a dose of 5 mg/kg Ppa was injected on Day 0. Five hours post UPPM injection, the tumor area was irradiated with a 671 nm laser at the power of 100 mW/cm 2 for 5 min.After laser irradiation, Gem was intraperitoneally injected at a dose of 50 mg/kg, followed by two additional Gem injections on days 4 and 9 (Figure 5A).As expected, the bacteria in the tumor can be eradicated completely with the UPPM-mediated photodynamic therapy in the UPPM+L+Gem and UPPM+L groups (Figure 5B,C).On the contrary, intratumoral bacteria in the PPM+L+Gem group only decreased slightly compared with the Gem and Saline group.Due to the excellent performance of photodynamic antibacterial eradication with UPPM, the tumor volume decreased significantly with the following chemotherapy using Gem in the UPPM+L+Gem group (Figure 5D-F).However, in the Gem and PPM+L+Gem group, the tumor grew rapidly owing to the Gem metabolism by EcN.These results further confirmed that bacteria-targeted photodynamic eradication of intratumoral bacteria could inhibit Gem metabolism and enhance the chemotherapeutic efficacy.
During the treatments, the body weight of the mice was not changed significantly (Figure S9).At the end of the treatments, the major organs were collected for histological examination, and no noticeable abnormalities including inflammation, cell necrosis, and apoptosis were observed (Figure S10).The renal function and liver function of the mice after UPPM treatment were evaluated with blood chemistry analysis (Figure S11).No significant changes were found compared with the mice treated with saline alone.We further evaluated the impact of photodynamic bacteria eradication on the gut microbiota using the bioluminescence imaging of EcN in the intestine of mice (Figure S12).Compared with the treatment with antibiotics, the gut bacteria were not affected after the photodynamic eradication of intratumoral bacteria, while the antibiotics killed all the bacteria in the gut and tumor.Considering the importance of gut microbiota, the selective photodynamic eradication of intratumoral bacteria with UPPM should be safer (Scheme 1).

CONCLUSION
In conclusion, we demonstrated photodynamic antimicrobial therapy as an efficient way of reversing Gem resistance of pancreatic cancer, which is a common issue in the therapy of pancreatic cancer.application of photodynamic antimicrobial therapy in large and deep tumors, laser light needs to be delivered with optical fibers that can be inserted into tumors under imaging guidance. [38]

EXPERIMENTAL SECTION
Detailed experimental materials and methods can be found in the Supporting Information.

S
C H E M E 1 (A) Schematic illustration of the preparation process of pyropheophorbide-a (Ppa)-PEG-UBI micelles; (B) Schematic illustrating the mechanism of bacteria-targeted photodynamic antimicrobial therapy in overcoming gemcitabine (Gem) resistance.

F
I G U R E 1 Characterization of UPPM and PPM.(A) TEM images of UPPM and PPM, respectively.Scale bar = 100 nm.(B) Absorption spectra and (C) fluorescence emission spectra of UPPM, PPM, and pyropheophorbide-a (Ppa) in different solutions (Ppa, 10 μM).(D) Surface zeta potential of UPPM and PPM.(E) The fluorescence intensity of SOG changes as a function of laser irradiation time in the presence of PBS, free Ppa, PPM, or UPPM.F I G U R E 2 Photodynamic antibacterial activity mediated by UPPM.(A) Fluorescence imaging for the specific binding of superfolder green fluorescence protein (sfGFP)-labeled Escherichia coli Nissle 1917 (EcN) with UPPM and corresponding Pearson's correlation coefficient.Scale bar = 10 μm.****p < 0.0001.(B,C) Flow cytometry analysis of the specific binding of sfGFP-labeled EcN with UPPM.n = 3, ****p < 0.0001.(D) Lysogeny broth (LB) agar plates of sfGFP-labeled EcN incubated with UPPM or PPM under different concentrations with laser irradiation.Scale bar = 20 mm.(E) Quantitative analysis of the bacteria viability in (D).n = 4, ****p < 0.0001.(F) Membrane potential measurements for EcN upon treatment with PPM and UPPM with/without light irradiation, via detecting the fluorescence of DiOC 2 (3) by flow cytometry.(G) Quantitative analysis of the red/green fluorescence ratio in (F).Bacterial membrane potential was indicated by the ratio of cells that emit red fluorescence to those with green fluorescence, n = 3, **p < 0.01.

F I G U R E 3
Photodynamic eradication of bacteria inhibited gemcitabine (Gem) metabolization and enhanced the cytotoxicity of Gem.(A) Highperformance liquid chromatography (HPLC) chromatogram of Gem and dFdU.(B) Quantification of the remaining Gem and dFdU production after 50 μM Gem was incubated with 5 × 10 6 /mL Escherichia coli Nissle 1917 (EcN) for the indicated time.(C) Quantitative analysis of remaining Gem after incubation for 14 hours with bacteria which were treated by UPPM+L, PPM+L.L: Laser irradiation.(D) The viability of Panc02 cells incubated with 5 μM Gem, or metabolized Gem by bacteria which were treated by UPPM+L, PPM+L.Data are shown as mean ± s.e.m; *p < 0.05; ***p < 0.001; ****p < 0.0001 as determined by Student's t-test.

F
I G U R E 4 (A) Time-lapse fluorescence imaging of UPPM and PPM in subcutaneous panc02-bearing mice and (B) quantification of mean fluorescence intensity of tumor.(n = 5, *p < 0.05).The circled area indicated the position of the tumor.(C) Ex vivo fluorescence imaging of main organs after 24 h and (D) quantification of the fluorescence signals of isolated organs.(n = 5, *p < 0.05).

F I G U R E 5
Improving the anticancer activity of gemcitabine (Gem) through photodynamic bacteria eradication mediated by UPPM.(A) Balb/C nude mice were subcutaneously injected with Pan02 and bacteria.After 10 days (indicated as day 0), photodynamic bacteria inhibition was performed and Gem was i.p. injected on days 0, 4, and 9. (B) In vivo bioluminescence images of subcutaneous bacteria-colonized pancreatic tumor-bearing mice before and after various treatments.The red dashed circles indicated the overlay of tumor and bacteria.(C) The average radiance of the bioluminescence of bacteria colonized in tumor in (B).(n = 3, *p < 0.05) (D) Images of bacteria-colonized pancreatic tumor-bearing mice at timed intervals.(E) Tumor growth curves of various treatments.(n = 3) (F) The tumor volumes of various treatments on day 18.Data shown as mean ± s.e.m; **p < 0.01; ***p < 0.001 as determined by Student's t-test.
more applications for cancer therapy owing to its high efficiency and spatiotemporal precision in modulating intratumoral microbiota.Admittedly, this research has some limitations.Due to the diversity of intratumoral microbiota and heterogeneity of pancreatic cancer, the efficiency of photodynamic antimicrobial therapy should be further evaluated in more clinically relevant animal models with more diverse intratumoral bacteria and different cell lines.In addition, for the potential clinical