Inhibitory effects of calcium or magnesium ions on PDI

,


Introduction
PDI nowadays has a wide range of possible applications. There is plenty of experimental applications in development for example in wastewater treatment [1][2][3][4][5][6][7][8], implementation in antimicrobial coatings [9,10], lowering the microbial load of food and crops [11][12][13][14], decolonization of human skin [15] or in dentistry [16][17][18]. Furthermore, Majiya and colleagues demonstrated sunlight driven water disinfection with a porphyrin immobilized in a chitosan membrane. The researchers successfully reduced the bacterial load by three orders of magnitude and therefore demonstrate a cost-efficient and sustainable method for drinking water disinfection [19].
The efficacy of PDI is frequently studied under laboratory conditions using media like PBS, which are rather uncommon when considering PDI applications under real life conditions. Thus, when comparing PDI efficacies of environmental photodynamic applications with ones from in vitro laboratory studies, it is not surprising that the results of such studies seem to diverge tremendously in some cases.
As mentioned, several fields of application are conceivable for photodynamic inactivation, in which a wide variety of substances, including divalent ions, will inevitably be present. An example of a potential future application outside the laboratory is the antimicrobial treatment of water [19,[22][23][24]. Exemplarily for tap water, water hardness is calculated based on the concentration of calcium carbonate and has the following definition according to the US Geological Service. A concentration of 0 -0.6 mmol l − 1 is considered as soft water, 0.61 -1.2 mmol l − 1 moderately hard water, 1.21 -1.80 mmol l − 1 hard water, and above 1.8 mmol l − 1 very hard water [25]. In Germany, water hardness is divided in three categories termed soft for <1.5 mmol l − 1 , medium from 1.5 to 2.5 mmol l − 1 and hard is > 2.5 mmol l − 1 measured as total CaCO 3 [26]. Worldwide, concentrations of calcium and magnesium ions vary greatly depending on the geological background the water originates from. The concentration of calcium ions in drinking water derived from ground water generally ranges from about 0.025 mmol l − 1 to 2.5 mmol l − 1 with values reported up to nearly 10 mmol l − 1 [27][28][29][30][31][32][33]. Magnesium in drinking water is found all around the world and varies greatly depending on the geographical region. Studies from Sweden found magnesium ion concentrations in drinking water of around 0.065 up to 0.62 mmol l − 1 [34][35][36], reports from Norway mentioned concentrations up to 0.1 mmol l − 1 [37] with a median of around 0.2 mmol l − 1 [38]. Research from England measured values up to 4.56 mmol l − 1 [39] and another study from South-Africa reported on magnesium concentrations up to 2 mmol l − 1 [40].
Another application of PDI is the inactivation of microorganisms in food production and processing [41][42][43]. Approaches of applying PDI towards milk [44] should be taken into focus as divalent ions are inevitably present. The calcium content of milk depends to a certain extent also on the breed of the milked cow [45] or the diet of the cow itself [46]. The various milks commercially available today have quite similar calcium concentrations between 29.5 and 31.56 mmol l − 1 . Yoghurt on the other hand varies in a range of 34.62 to 45.62 mmol l − 1 .
The calcium concentration of raw cheese varies between 98.02 to 299.40 mmol l − 1 [47].
A future promising approach is the treatment of the human skin based on photosensitizer solutions. Although this has been proven to show good initial results, the obtained inactivation values are still lower than when experiments are conducted in controlled liquid environment with H 2 O. For example, within this study, good efficacy of at least 6 orders of magnitude was achieved for SAPYR for 0.72 J cm − 2 and 50 µmol l − 1 . However, on ex vivo skin experiments at least 100 µmol l − 1 were applied in combination with at least 30 J cm − 2 in order to achieve sufficient inactivation [15]. Similar findings were reported by another research group where harsher parameters for efficient inactivation had to be applied in an in vivo model [48]. The differences in the efficacy of these experiments are to a certain extent based on slight experimental differences. However, experiments on skin in general or sweat in particular are by no means similar to pure water. Much more, they resemble complex environments with a variety of substances, even in literature, the found compositions vary greatly [49,50]. Sweat also contains various amounts of calcium and magnesium that inhibit the PDI at least to a certain extent.
Even though the commercial application of PDI in various environments is one of the major aims, it is frequently not sufficiently explored whether or to which extent various ubiquitous substances in these environments may hamper PDI efficacy when using such photosensitizers. Among others, up to date the effects of abundant substances such as calcium or magnesium ions or complex biological molecules remain mostly uninvestigated. Of course, it is known for some photosensitizer that certain chemicals inhibit [21,51] or enhance [52][53][54] the photodynamic process. One of the most prominent molecules in this context is sodium azide acting as a potent physical singlet oxygen quencher [51]. In contrast, there are also studies investigating on effects that promote the photodynamic action in presence of sodium azide [54]. Furthermore, it was recently shown that carbonate and phosphate ions, which are two prominent molecules in most environments, have detrimental effects on the chemical structure of flavin based photosensitizers [21]. Additionally, some research data concerning the photodynamic treatment of milk suggested that calcium and magnesium ions pose some issues in efficacy [44].
Therefore, we hypothesize that ubiquitous bivalent ions might affect the photodynamic process at different stages. In this study, we investigated five different cationic PS with various chemical structures such as a porphyrin, a phenothiazine, two flavins and a phenalenone. The biocidal potential of the different photosensitizers towards several bacteria was evaluated under the influence of various aqueous solutions containing calcium and magnesium in ascending concentrations resembling concentrations found in possible areas of future applications.

Photosensitizers
Methylene blue was purchased from SERVA Electrophoresis GmbH with a minimum dye content of 96%. Methylene blue has a singlet oxygen quantum yield of around 0.50 depending on the applied measurement method [55], providing a mixture of ROS and singlet oxygen that is generated. TMPyP was brought from Sigma-Aldrich with a minimum dye content of 97%. The quantum yield of the porphyrin based photosensitzer is around 0.77 [56], producing chiefly singlet oxygen with minor amounts of other ROS. Besides, an exclusive singlet oxygen producing photosensitizer shortly called SAPYR with a quantum yield of 0.99 [20] was purchased from the TriOptoTec GmbH, the chemical structure of the molecules has been published elsewhere [21]. Additionally, two different flavin based photosensitizers were included with a quantum yield of around 0.75 that was also purchased from Tri-OptoTec GmbH. In general, all light sensitive parts of the procedures were conducted at low light conditions with a maximum radiant flux of 55 µW cm − 2 as described elsewhere [57].

Bacteria
The used bacterial strains were obtained from the German Collection of Microorganisms and cell culture lines DSMZ (Braunschweig, Germany). As a Gram-positive representative Staphylococcus aureus F-182 (DSM 13661) was used. The strain was derived from a clinical isolate from Kansas and exhibits resistance towards methicillin and oxacillin, therefore also considered as MRSA. The Gram-negative organism tested in this study was Pseudomonas aeruginosa Boston 41501 (DSM 1117) initially isolated from a blood culture. As universal culture medium Mueller-Hinton-Bouillon [58] was used on which the bacteria grew over night at 37 • C at 100 rpm.

Ionic solutions
Stock solutions of calcium chloride (CaCl 2 ) and magnesium chloride (MgCl 2 ) were prepared with stock concentrations of 150, 15, 1.5 and 0.15 mmol l − 1 . As a solvent and control served ultra-pure H 2 O with a conductance of 0.056 µS cm − 1 (Milli-Q® Water Treatment System, Merck KGaA, Darmstadt, Germany). The stock solutions were stored in plug-sealed, gas tight glass serum bottles under nitrogen atmosphere in the dark at room temperature. pH was adjusted to 7 using HCl or NaOH. CaCl 2 as well as MgCl 2 were purchased from Carl Roth GmbH + Co. KG (Karlsruhe, Germany) in analytical grade.

Light source
For TMPyP, SAPYR, FLASH-02a and FLASH-06a a blue light source (blue_v, Waldmann GmbH, Villingen-Schwenningen, Germany) was used, while MB was irradiated under a red light source (PDT 1200, Waldmann GmbH, Villingen-Schwenningen, Germany). The applied irradiance for the blue light source was 18 mW cm − 2 and for the red light source 20 mW cm − 2 . The final radiant exposure depended on the time of the application and is represented the product of the applied irradiance in W cm − 2 times the application time in s resulting in J cm − 2 , which are the values given throughout the following.

Spectroscopic analysis
To investigate if aqueous solutions alter the chemical structure of the used photosensitizers, spectroscopic analysis was performed from 300 to 700 nm in a photometer (BMG Labtec, Ortenberg, Germany) with a 96-well microtiter plate (SARSTEDT AG & Co. KG, Nümbrecht, Germany). To rule out light induced reactions, the spectra were recorded before and after illumination with an appropriate light source with defined energy up to 5.4 J cm − 2 . Each reaction was composed out of a total volume of 200 µl with PS concentrations ranging from 0 to 50 µmol l − 1 and ionic solutions in concentrations of up to 75 mmol l − 1 . The obtained transmission was then plotted with OriginLab 2019b (Northampton, USA).

Singlet oxygen production
To evaluate singlet oxygen production in qualitative manners, DPBF (1,3-Diphenylisobenzofurane) assays were carried out. DPBF was purchased from Sigma-Aldrich with a minimum dye content of 97%. DPBF reactions were composed in total as follows: a total volume of 200 µl contained either no PS (internal reference) or 1 to 50 µmol l − 1 PS, 75 mmol l − 1 CaCl 2 or MgCl 2 and 500 µmol l − 1 DPBF which was dissolved in analytic grade ethanol. Assays were conducted as triplicates and measured after a total applied energy of 0, 0.018, 0.036, 0.054, 0.072, 0.09 and 0.18 J cm − 2 with either the blue_v light source or the respective red light source. DPBF fluorescence was then measured utilizing a fluorescence plate reader from BMG Labtech with the excitation wavelength of 411 nm and emission detection at 451 nm. Values obtained for the internal reference (DPBF without PS) were set to 1, relative fluorescence was calculated as rations to the internal reference and the sample (DPBF with PS) and displayed in per cent using OriginLab 2019b software.

Evaluation of the logarithmic bacterial reduction
Bacterial cultures were harvested via centrifugation at 13,000 x g for 7 min. Afterwards, OD 600 was adjusted to 0.6 with a cell density meter (Ultrospec 10, Ammersham Biosciences, Little Chalfont, UK). 1 ml of the cell suspension was transferred to 1.5 ml reaction tubes and centrifuged at 13,000 x g for 7 min. Supernatant was discarded and the remaining pellet was washed in H 2 O three times. After the last washing step, the cells were mixed with 1 ml of either CaCl 2 , MgCl 2 or H 2 O in concentrations of 75 to 0.75 mmol l − 1 . 25 µl of the bacterial cell suspension were mixed with the same volume of PS solutions in ascending concentrations, incubated for 10 min at room temperature under dark conditions with a maximum of 3 µW cm − 2 and afterwards irradiated with a constant energy of 0.72 J cm − 2 .
20 µl of the reaction were transferred into 180 µl Mueller Hinton bouillon after irradiation and cultivated at 37 • C for 48 h. Optical density was measured at 600 nm using a plate reader. The obtained values were then used to calculate bacterial reduction as described elsewhere [59]. The method presented here was initially described as proliferation assay [60] and was adapted in the here presented study for liquid bacterial cultures. Doubling times were calculated for OD 600 at 0.2 and 0.4.

Binding assays
To exclude interactions hindering photosensitizer attachment to bacterial cells, the bacterial cell suspensions were initially adjusted to an optical density of 0.6 at 600 nm. 500 µl thereof were transferred into 1.5 ml reaction tubes, centrifuged, and washed in water as described before. The washed pellet was mixed with 500 µl of the ionic solution and 500 µl of PS in a concentration of 100 µmol l − 1 . The mixture was incubated for 10 min in absolute darkness and centrifuged at 4,500 x g for 10 min. The supernatant was collected and transferred into a cuvette and measured at 444 nm for FLASH-06a, 446 nm for FLASH-02a, 370 nm for SAPYR, 520 nm for TMPyP and 575 nm for MB.

Photostability
Photosensitizers dissolved in H 2 O did not show alterations in the transmission spectra after application of up to 5.4 J cm − 2 radiant exposure, only a marginal loss of concentration of the photosensitizer was observed. (Supplementary Figure 1, Supplementary File 1). The transmission spectra for photosensitizers dissolved in 75 mmol l − 1 CaCl 2 were also not altered after irradiation besides minor concentration losses (Supplementary Figure 2, Supplementary File 2). The concentration decreased in similar amounts as for the water controls. Photosensitizers dissolved in MgCl 2 solutions again showed low photodegradation not exceeding 2 % compared to the non-irradiated controls. Also, the photosensitizers maintained their chemical integrity (Supplementary Figure 3, Supplementary File 3), except for TMPyP as a bathochromic shift was observed. The transmission minimum (Soret band) was shifted to 435 nm and the Q bands were located at 520 to 521 and at 562 to 564 (Fig 1, Supplementary File 3).

Singlet oxygen production
As mentioned before, singlet oxygen production was measured as relative fluorescence of DPBF. The data are additionally given as a table in Supplementary File 4. Lower relative fluorescence hints at more efficient singlet oxygen production while values above 1 are measured, when the photobleaching effect of the reference exceeds the loss of the fluorescence caused by the photosensitizer. Relative fluorescence of DPBF for MB decreased at 10 µmol l − 1 already to values around 0.2 for H 2 O and MgCl 2 solution, while the same relative fluorescence value was achieved for CaCl 2 solution at the highest concentration of PS applied (Fig 2).
DPBF assays of TMPyP showed already drastically lowered relative fluorescence for 1 µmol l − 1 of TMPyP. Application of concentrations as low as 5µmol l − 1 of TMPyP already led to a relative fluorescence of around 0.1, indicating that all DPBF present in the reaction was readily depleted in all cases independent of the used solvents (Supplementary Figure 4B).
DPBF assays for SAPYR (Supplementary Figure 4C), FLASH-02a (Supplementary Figure 4D) and FLASH-06a (Supplementary Figure 4E) showed a similar reduction of the relative fluorescence mostly independent of the used solvents reaching minimal values of 0.1 to 0.2 for 50 µmol l − 1 of applied PS.

Binding assays
MB showed good binding behavior towards S. aureus cells in the presence of H 2 O. However, the binding efficiency decreased with increasing ion concentration (Fig 3A). The measured concentrations for S. aureus are also given as a table in Supplementary File 5. The use of TMPyP showed a comparable but less pronounced effect (Fig 3B). The binding of SAPYR (Fig 3C), FLASH-02a ( Fig 3D) and FLASH-06a (Fig 3E) was almost unaltered in the presence of divalent ions.
The highest amounts of bound PS were measured for FLASH-02a with or without 0.75 mmol l − 1 MgCl 2 showing around 96 µmol l − 1 or 95 µmol l − 1 , respectively. MB bound with 86 µmol l − 1 to S. aureus cells in the case of H 2 O as a maximum value, followed by TMPyP with 81 µmol l − 1 for H 2 O. Most SAPYR was bound for the application of H 2 O with 78 µmol l − 1 and the least amount of PS was found for FLASH-06a with 76 µmol l − 1 for 0.75 mmol l − 1 CaCl 2 not differing significantly from the other measured values for the other experimental conditions.
The measured concentrations for P. aeruginosa are additionally displayed in Supplementary File 6 as a table. Again, MB bound well to P. aeruginosa cells in the presence of H 2 O. As shown for S. aureus, CaCl 2 and MgCl 2 solutions inhibited the binding of MB to the cells drastically (Fig 4A). Descending ionic concentrations led to higher amounts of bound photosensitizer. Furthermore, TMPyP (Fig 4B) showed a similar effect but only in insignificant amounts. As observed for MRSA, the binding of SAPYR (Fig 4C), FLASH-02a ( Fig 4D) and FLASH-06a (Fig 4E) did

PDI of Pseudomonas aeruginosa
The mean logarithmic reduction values for P. aeruginosa resuspended in CaCl 2 are additionally provided as table in Supplementary File 7. PDI at 0.72 J cm − 2 for MB in H 2 O led to bacterial reduction of at least 6 log 10 steps at a PS concentration as low as 10 µmol l − 1 . 0.75 mmol l − 1 CaCl 2 inhibited the PDI of MB and the PDI effect almost disappeared (< 1 log 10    The logarithmic reduction is displayed on the Y-axis while the dark control (DC) and applied PS concentrations are displayed on the X-axis. The different concentrations of the ions are symbolized by various colors indicated in the right corner. Panels A shows results for MB, B for TMPyP, C for SAPYR, D for FLASH-02a, E for FLASH-06a. step) for concentrations of 75 or 7.5 mmol l − 1 CaCl 2 (Fig 5A). In H 2 O 5 µmol l − 1 TMPyP and above led to a bacterial reduction of 6 log 10 steps. CaCl 2 inhibited the photodynamic mechanism for 5 and 10 µmol l − 1 as only a logarithmic reduction around 2 log 10 steps was measured for 7.5 and 0.75 mmol l − 1 CaCl 2 . However, in none of the cases for 75 mmol l − 1 CaCl 2 the efficacy exceeded 1 log 10 step (Fig 5B). The application of SAPYR led to an efficient inactivation at concentrations as low as 10 µmol l − 1 in H 2 O. For 75 mmol l − 1 CaCl 2 (Fig 5C) almost no bacterial reduction was observed. Lower concentrations of CaCl 2 led to an inactivation of 6 log 10 steps for the application of 50 µmol l − 1 SAPYR. 10 µmol l − 1 FLASH-02a and above led to an inactivation of 6 log 10 steps (Fig 5D). Addition of 75 or 7.5 mmol l − 1 CaCl 2 led to no efficient inactivation when FLASH-02a was applied under mentioned conditions, only 0.75 mmol l − 1 showed similar efficacy to the water control (Fig 5D). In H 2 O, concentrations of 10 µmol l − 1 FLASH-06a and above yielded an efficacy of 6 log 10 steps. The application of CaCl 2 did not lead to a reduction that exceeded 2 log 10 steps in any cases (Fig 5E).
A table of the mean logarithmic reduction of P. aeruginosa resuspended MgCl 2 is provided in Supplementary File 8. The application of MgCl 2 had slightly less inhibitory effects on the PDI with MB (Fig 6A) than CaCl 2 . Results obtained for TMPyP with bacteria resuspended in MgCl 2 solutions (Fig 6B) did not differ much from the beforehand presented results for CaCl 2 . The application of 50 µmol l − 1 SAPYR in 75 mmol l − 1 MgCl 2 led to a maximum bacterial reduction of about 1 log 10 step. Lower MgCl 2 concentrations led to a maximum inactivation of around 4 log 10 steps for 7.5 mmol l − 1 MgCl 2 and 6 log 10 steps for 0.75 mmol l − 1 MgCl 2 , respectively ( Fig 6C). P. aeruginosa suspended in 7.5 mmol MgCl 2 solution were inactivated with an efficacy not exceeding 1 log 10 step, 0.75 mmol l − 1 MgCl 2 solution showed a bacterial reduction of around 4 log 10 steps for 50 µmol l − 1 FLASH-02a. 6 log 10 steps were observed for 25 µmol l − 1 FLASH-02a and above in 0.75 mmol l − 1 MgCl 2 (Fig 6D). The experimental outcome of the application of MgCl 2 in combination with FLASH-06a showed a slightly higher inactivation efficacy for 75 mmol l − 1 compared to CaCl 2 . However, for 7.5 mmol l − 1 MgCl 2 the efficacy did not exceed 2 log 10 steps. 0.75 mmol l − 1 MgCl 2 restored an efficacy of 6 log 10 steps for 25 and 50 µmol l − 1 FLASH-06a (Fig 6E).

PDI of Staphylococcus aureus
Additionally, to the mentioned experiments with P. aeruginosa, the same set of conditions were tested for a methicillin resistant S. aureus strain (MRSA). A tabular presentation of the results is provided in Supplementary File 9. The application of MB in H 2 O led to an efficacy of 6 log 10 steps for 25 and 50 µmol l − 1 . 5 and 10 µmol l − 1 led to an efficacy < 3 log 10 steps (Fig 7A). In general, inactivation in the presence of CaCl 2 solution did not show any relevant reduction for 1 µmol l − 1 . The application of 5 µmol l − 1 MB showed a reduction < 2 log 10 steps for 0.75 mmol l − 1 CaCl 2 . For 10 µmol l − 1 MB, the application of 7.5 mmol l − 1 CaCl 2 led to an efficacy of 3.5 log 10 steps. 0.75 mmol l − 1 CaCl 2 showed a reduction for 10 µmol l − 1 MB with 3.1 log 10 steps. 25 µmol l − 1 MB achieved for 7.5 mmol l − 1 CaCl 2 an efficacy of around 3 log 10 steps at most. 50 µmol l − 1 MB did not increase the efficacy for 75 and 7.5 mmol l − 1 CaCl 2 while the application of 0.75 mmol l − 1 CaCl 2 showed an efficacy of 6 log 10 steps (Fig 7 A). The photosensitizer TMPyP showed excellent efficacy in H 2 O for 10 µmol l − 1 and above with an efficacy 6 log 10 steps (Fig 7B). 75 mmol l − 1 CaCl 2 did not lead to efficient inactivation, 7.5 mmol l − 1 CaCl 2 solution showed a reduction of 1.3 log 10 steps for 5 µmol l − 1 and 6 log 10 steps for 10 µmol l − 1 and above. Application of 0.75 mmol l − 1 CaCl 2 showed better efficacy compared to the water control for 1 and 5 µmol l − 1 TMPyP. 25 and 50 µmol l − 1 restored the efficacy of the PDI with 6 log 10 steps. (Fig 7B). SAYPR in H 2 O was capable of an inactivation of 6 log 10 steps for 25 µmol l − 1 and above. However, the application of 75 mmol l − 1 CaCl 2 led to no noteworthy reduction in bacterial viability (Fig 7C). The water control of FLASH-02a showed a reduction of 6 log 10 steps for 25 and 50 µmol l − 1 (Fig 7D), only minor efficacy was achieved for lower concentrations. However, when CaCl 2 solutions were applied, in none of the applied Fig. 6. Diagrams of the calculated logarithmic reduction of Pseudomonas aeruginosa resuspended in MgCl 2 solutions in different concentrations. The logarithmic reduction is displayed on the Y-axis while the dark control (DC) and applied PS concentrations are displayed on the X-axis. The different concentrations of the ions are symbolized by various colors indicated in the right corner. Panels A shows results for MB, B for TMPyP, C for SAPYR, D for FLASH-02a, E for FLASH-06a.
concentrations a noteworthy reduction was achieved (Fig 7D). While the water control for FLASH-06a (Fig 7E) did not differ in significant manners from the data for FLASH-02a, the addition of CaCl 2 aggravates the problems even more, no measurable reduction could be achieved ( Fig  7E).
The obtained results for S. aureus resuspended in MgCl 2 are also  displayed as a table in Supplementary File 10. MB in the presence of MgCl 2 solutions had mediocre inactivation efficacy between 2 and 4 log 10 steps for 5 to 25 µmol l − 1 MB. 6 log 10 steps of bacterial reduction were achieved for 0.75 mmol l − 1 MgCl 2 for the highest applied MB concentration (Fig 8A). S. aureus resuspended in MgCl 2 solutions with TMPyP led to a good overall efficacy as 6 log 10 steps were achieved even for the highest MgCl 2 concentration applied (Fig 8B). For the application of MgCl 2 solution in combination with SAPYR the maximum efficacy obtained was around 2 log 10 steps for the highest SAPYR concentration (Fig 8C). Most dramatically influenced by the application of MgCl 2 were both FLASH-06a and FLASH-02a. For FLASH-02a, the only mentionable observed reduction was that at least for 0.75 mmol l − 1 MgCl 2 at a concentration of 50 µmol l − 1 FLASH-02a a reduction of around 4 log 10 steps (Fig 8D). FLASH-06a in combination with MgCl 2 solutions led to no relevant observable reduction (Fig 8E).

Discussion
The presented results paint quite a clear picture concerning the role of CaCl 2 and MgCl 2 when performing PDI against the bacteria and photosensitizers used. Firstly, the absorption spectra showed that all photosensitizers in pure water were stable upon irradiation with up to 5.4 J/cm 2 . Even the addition of the divalent ions at different concentrations showed no negative effect on photostability of photosensitizers, except for TMPyP in the presence of MgCl 2 (Fig 1). This is not surprising because a porphyrin structure is a pristine chelating agent for divalent ions [61]. The fact that the complexation of bivalent metal ions causes alterations in the spectrum of porphyrins has been described in literature before [62][63][64]. However, such chelating reactions of the porphyrin group are often influenced by specific reaction parameters such as defined pH [65][66][67] or temperatures [68,69]. This might also lead to incomplete complexation reactions, which could also be influenced upon light exposure explaining the different transmission spectra after irradiation. Further possible explanations of this change in absorption behavior in the Q bands might be potential partial cleavage of the methylpyrimidinum groups of TMPyP especially as the side chains of porphyrins seem to rather influence the absorption of the Q bands than of the Soret band [70,71]. The DPBF assays showed an efficient generation of singlet oxygen by all photosensitizers in combination with both divalent ions, except for MB in the presence of CaCl 2 (Fig 2). It seems that singlet oxygen production of TMPyP is even at 1 µmol l − 1 due to its high absorption coefficient so efficient that the relative fluorescence of DPBF dropped by nearly 0.5 for H 2 O or even more for CaCl 2 and MgCl 2 solutions (Supplementary Figure 2B). However, quantitative conclusions concerning the DPBF assays involving TMPyP are limited as both excitation and emission wavelength of DPBF match those of TMPyP to a certain extent. With the TMPyP concentrations applied here it is likely that most photons are absorbed by TMPyP, therefore a concentration dependent comparison of the relative fluorescence should be treated with caution.
These two exceptions might not automatically reduce the efficacy of PDI. TMPyP has a rather high extinction coefficient [72] that even low amounts of functional PS can lead to efficient inactivation which is also reflected by the shown results for the biological inactivation as TMPyP showed the best inactivation efficacy under the given experimental conditions. It is also known that TMPyP with complexed metals is still capable of singlet oxygen production [41]. One of the possible explanations is that the complexation reaction might not be a process that takes place for all TMPyP molecules. Further, a change in pH value in the adjacency of bacterial cells might have stopped or even reverted the complexation.
In most of the PDI applications, the generation of singlet oxygen plays the major role in cell killing [73]. However, the photosensitizers may have the potential to generate not only singlet oxygen, as proven by DPBF assays in the present study. SAPYR shows a singlet oxygen quantum yield with a value of Φ Δ = 0.99 [20], TMPyP Φ Δ = 0.77 [74], the flavins Φ Δ = 0.75 to 0.78 [75], and MB Φ Δ = 0.52 [76]. In particular, the Φ Δ of MB could allow a simultaneous generation of other reactive oxygen species (ROS) that may also yield cell killing. A fact one should keep in mind is especially the potential photodemethylation of diaminomethylgroups as observed for example for photosensitizers like nocathoacin I [77] or methylene blue [78]. For the case of methylene blue, a degradation occurs to azure a or b, leading to reduced singlet oxygen yields and the potential increase of type I reactions [76,79]. However, it seems that calcium and magnesium ions do not favor such demethylation processes in an excessive manner as there are no hints that the spectra of the herein used methylene blue are altered in such ways.
At a glance, the microbial inactivation data, in the absence of the ions, showed an efficient PDI of all tested photosensitizers with a respective concentration of 25 µmol l − 1 yielding a reduction of 6 log 10 steps at low radiant exposure of light (Figs 5-8). Except for TMPyP, the efficacy of all photosensitizers is clearly lower in the presence of elevated concentrations of calcium and magnesium ions (Figs 5-8). A general observable trend was that increased concentrations of CaCl 2 and MgCl 2 led to inhibited inactivation. The effects were most severe for the tested flavins which are also affected by other ions such as carbonate or phosphate [21].
The applied concentrations of CaCl 2 and MgCl 2 resemble the concentrations present in several fields of application.The two lower CaCl 2 concentrations applied within this work, namely 7.5 and 0.75 mmol l − 1 cover the usual calcium concentration in tap water [25][26][27][28][29][30][31][32][33]. Within this mentioned ranges, TMPyP is most efficient against Gram-negatives, followed by SAPYR and FLASH-02a. However, the findings in this study leads to an exclusion of FLASH-06a and MB from its potential use in such water applications. Gram-positives seem to be mostly inhibited by TMPyP again, but now followed by MB. SAPYR and the flavin based PS did not yield sufficient efficacy under the influence of calcium and magnesium ions. However, concerning drinking water applications, Gram-negatives such as Shigella sp., Vibrio sp., Salmonella sp. or Escherichia coli are the more crucial organisms as one of the main causes of contaminated drinking water [80].
The concentrations of magnesium ions usually present in drinking water [34][35][36][37][38][39][40] suggest that the magnesium ions poses less of a problem compared to calcium ion concentrations. Especially Gram-negatives might be readily inactivated in magnesium concentrations below 0.75 mmol l − 1 .
Concerning food applications, potential use in dairy products are the most crucial applications in the light of the herein presented results due to their elevated calcium content [44][45][46][47]. Based on the results of this work, the use of TMPyP might show sufficient success concerning the reduction of the bacterial load while the other PS used here seem to be less promising. However, a publication already observed reduced efficacy of applied PDI and the authors speculated that calcium and magnesium might take a part in the reduced efficacy besides further substances such as proteins or fatty acids [44].
Many researchers have already reported that the outermost layers of bacteria seem to be the target of PDI or at least play a major role in the uptake of the PS. A study by George, Hamblin and Kishen from 2009 revealed for MB that the PS showed lower uptake in the presence of divalent ions [81]. Although this study confirms the findings concerning MB, the other PS show no relevant difference in their uptake or binding behavior. Therefore, the sole differences in uptake and binding behavior of the PS do not explain the drastic differences observed in the microbial efficacy. Concerning cationic photosensitizers it is highly likely that negatively charged LPS molecules in the outer membrane that need calcium and magnesium ions for stability [82] form a positively charged layer surrounding the cell that electrostatically hinder the penetration of the PS up to the outer membrane but bind to outer sugar moieties of the LPS. Especially the fact that ions have a stabilizing effect has been reviewed extensively concerning the use of EDTA [83]. Further, this stability hypothesis is strengthened by a study demonstrating a efficacy-promoting effect of EDTA with zinc phthalocyanine against Gram-negative cells, which are without EDTA not effected by negatively or neutrally charged photosensitizers [84]. Similar might be true for Gram-positive cells as teichoic acids and wall teichoic acids have a certain metal ion binding capacity [85][86][87]. However, these calcium and magnesium ion interactions seem to be not fully understood yet [87].

Conclusion
Although the divalent ions calcium and magnesium have no direct effects on the investigated PS such as chemical degradation their levels for an application in PDI must be kept as low as possible. Therefore, appropriate dilution of the treated liquids or rinsing of surfaces like the human skin with distilled water prior to PDI treatment is highly recommended for future research. Furthermore, several suggestions for the application of photodynamic processes can be given: First, based on several studies that were performed under various conditions, it becomes clear that increased light intensity helps to overcome inhibitory processes, even those of calcium or magnesium ions. Second, higher PS concentrations seem to support the PDI in general. Under these predictions, PDI is an extremely promising antimicrobial treatment for the future, independently on the type of microorganisms or their antibiotic resistances.

Funding
This work was funded by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) -Projektnummer 415812443

Declaration of Competing Interest
The authors declare the following financial interests/personal relationships which may be considered as potential competing interests: Harald Huber reports financial support was provided by German Research Foundation. Wolfgang Baeumler reports financial support was provided by German Research Foundation.