1 Introduction

The quantity and diversity of viruses in mammalian and avian reservoirs cause zoonosis to be common and unpredictable [1, 2]. The first modern examples are the AIDS pandemic originated by HIV-1 (lentivirus) transfer from chimpanzees to humans in west central Africa one century ago [3], and the 1918 influenza pandemic, known as the Spanish flu, associated with a virus having gene sequences closely related to avian H1N1 viruses [4]. More recently, in 2002–2003, the Severe Acute Respiratory Syndrome Coronavirus (SARS-CoV) epidemic started in China from a coronavirus of bat origin [5], and in 2012 bats were again suspected of transferring, through dromedary camels, the Middle East Respiratory Syndrome Coronavirus (MERS-CoV) and trigger another epidemic [5]. The SARS-CoV-2, first reported in 2019 in China, has conspicuous similarities with bat coronavirus [6,7,8] and was the cause of COVID-19 pandemic. Infection is promoted by tight binding between SARS-CoV-2 transmembrane spike (S) glycoprotein [9] and its receptor, the human angiotensin-converting enzyme 2 (hACE2) [10], which is highly expressed in nasal epithelial cells [11]. Changes in genetic codes of viruses currently infecting humans and the emergence of other pandemic viruses is likely, if not inevitable.

Extraordinary drug repurposing efforts were made to identify candidate therapeutic drugs for COVID-19 in the chemical space of drugs previously approved for other therapeutic indications [12,13,14]. Several antivirals exhibited substantial activity in vitro, and remdesivir was the first to be approved by the FDA for COVID-19 patients [15]. Early 2022, molnupiravir joined the very short list of FDA approval for COVID-19 [16]. Drug repurposing is challenging because drug candidates are selected by their specificity for one target and compounds exhibiting target promiscuity are deprioritized for drug development. The specificity of existing drugs explains the meager therapeutic options for pre- and post-exposure prophylaxis options for COVID-19 emerging from drug repurposing. Although the design of new molecules for specific targets in SARS-CoV-2 should yield better therapeutic options, the classical discovery and development process of new antivirals usually takes much longer than the duration of recent pandemics [17]. A most remarkable exception was the development of nirmatrelvir [18], which went from the therapeutic concept to FDA approval for COVID-19 in less than two years [19]. The specificity of antivirals comes with a high risk of resistance [20]. Broad-spectrum antivirals with low toxicity would change the management of COVID-19 and the approach to future pandemics.

Lentivirus (HIV-1), influenza (H1N1) and coronavirus (MERS-CoV, SARS-CoV-2) are enveloped particles that bud from the plasma membrane of eukaryotic cells. Targeting the viral envelope, and in particular its lipids, would take advantage of the difference between static viral envelopes and biogenic cellular membranes with reparative capacity [21], and would provide a strategy to minimize the selection of resistant strains [22]. Antimicrobial photodynamic therapy (aPDT) is more effective inactivating enveloped viruses than other viruses [23], which suggests that it can target the viral envelope. aPDT combines a photosensitizer molecule with low toxicity in the dark, light of a wavelength absorbed by the photosensitizer and molecular oxygen [24, 25]. When the photosensitizer is electronically excited, it can transfer its energy (or an electron) to molecular oxygen generating singlet oxygen (or superoxide ion, which subsequently may generate other reactive oxygen species, ROS) [26]. The oxidative stress produced by singlet oxygen is experienced within 200 nm of the excited photosensitizer because singlet oxygen lifetime is only 3 µs in aqueous media [27]. In fact, the volume explored by each singlet oxygen molecule is twice that of a SARS-CoV-2 virus [28]. Hence, the mechanism of aPDT does not require tight binding between the photosensitizer (drug) and its target. This is a major advantage because the photoinactivation of microorganisms is of broader scope than antiviral or antibiotic medicines, but comes with the challenge of being mostly a local treatment. Consequently, aPDT has been more successful addressing localized infections.

Small-sized meso-imidazolyl porphyrins with positive charges surrounding the macrocycle proved to be very potent aPDT photosensitizers using light at 415 nm [29, 30]. Following this lead, we searched for related red-light absorbing macrocycles, because longer wavelengths offer deeper tissue penetration. Herein we report the computational design and synthesis under multigram-scale of cationic imidazolyl chlorin derivatives for aPDT of SARS-CoV-2. We show that dicationic 5,15-bis(1,3-dimethylimidazol-2-yl)chlorin strongly absorbs light at 650 nm and selectively photo-inactivates virions at sub-micromolar concentration. We obtained PCR amplification inhibitions > 99.999% of the viral titers of clinical samples under conditions that ensure ~ 100% viability of human epidermal keratinocytes, and show that this can be achieved in a few seconds. This selectivity is assigned to the proper choice of photosensitizers size and charge, which favors their association with virions, and to a photodynamic effect that damages critical components for infection. The high potency of this dicationic meso-imidazolyl chlorin photosensitizer and its selectivity towards enveloped viruses, combined in a broad-spectrum antiviral approach that avoids the development of drug resistance, offer a valuable tool for the management of pandemics originated by influenza, HIV, cytomegaloviruses, coronaviruses and other enveloped viruses.

2 Results and discussion

2.1 Quantum chemical calculations

We reported that size and positive charge distribution in imidazolyl porphyrins impacts their efficacy of photodynamic inactivation of bacteria [29]. Although small-size cationic imidazolyl porphyrins proved to be very good aPDT photosensitizers in vitro, they lack the strong electronic absorption in the red/near-infrared that is desirable for clinical applications. In order to overcome this limitation, this work focused on the properties of chlorins, which are known to absorb one order of magnitude more light in the 630–660 nm range than the analogous porphyrins [31]. Given the interest in cationic imidazolyl chlorins, we selected 5,15-bis(1,3-dimethylimidazol-2-yl)chlorin (IC-H-Me2+), 5,15-bis(1-methyl-3-ethylimidazol-2-yl)chlorin (IC-H-Et2+) and 5,15-bis(1-methyl-3-(1,1,1-trifluoroethyl)imidazol-2-yl)chlorin (IC-H-FEt2+) for ab initio studies aiming at the evaluation of the effect of peripheral substituents on the development of charge asymmetries. The electron-withdrawing fluorinated substituents and the electron-donating ethyl groups are expected to impact differently on local charge densities. A substantial increase in charge asymmetry could favor interactions with phospholipids of the virus envelope.

Asymmetrically substituted chlorins have atropisomers and the two atropisomers of IC-H-Et2+ and of IC-H-FEt2+ have nearly identical energies. Fig 1 compares calculated positive charge densities using a color code where blue is a higher positive charge density and red is the lower positive charge. When atropisomers exist, the atropisomer with the larger substituents on the same side of the macrocycle was selected for the representation. The other atropisomers are shown in Fig. S1 and lowest energy conformers are presented in Fig. S2. The trifluoroethyl substituent does not lead to a discernable increase of positive charge density in the macrocycle. Moreover, the two atropisomers of the trifluoroethyl imidazolyl chlorin may have different photosensitizing abilities, as shown for redaporfin [32]. Based on positive charge densities, there is no compelling evidence to prefer chlorins larger than IC-H-Me2+ for aPDT of viruses, but we nevertheless prepared the precursor porphyrins of the chlorins in Fig. 1 to determine if any of them had advantageous photochemical properties for photoinactivation of viruses.

Fig. 1
figure 1

Electronic density maps of di-cationic imidazolyl chlorins, without counterions, illustrating atropisomers with larger substituents on the same side of the macrocycle, from total SCF density mapped with ESP, isovalue = 0.0004 at B3LYP/6-31G(d,p) level, in atomic units (e/a03). IC-H-Me2+: 5,15-bis(1,3-dimethylimidazol-2-yl)chlorin; IC-H-Et2+: 5,15-bis(1-methyl-3-ethylimidazol-2-yl)chlorin; IC-H-FEt2+: 5,15-bis(-1methyl-3-(1,1,1-trifluoroethyl)imidazol-2-yl)chlorin. Blue represents higher positive charge density and red represents lower positive charge density

Full size image

2.2 Preparation of photosensitizers

We developed a multi-gram synthesis of 5,15-bis(1-methylimidazol-2-yl)porphyrin (IP-H) based on the two-step Lindsey method [33, 34], where trifluoroacetic acid was used as catalyst to promote the condensation of dipyrromethane with 1-methylimidazol-2-carboxyaldehyde, followed by quinone oxidation [29]. This afforded 1.6 g (8% yield) of isolated IP-H.

The dicationic porphyrin derivative 5,15-bis(1,3-dimethylimidazol-2-yl)porphyrin IP-H-Me2+ was then prepared by reaction with iodomethane, in dimethyl formamide (DMF), at 40 ºC, under microwave irradiation (Pmax = 125 W) for 20 min. After precipitation with diethyl ether the solid was washed again with diethyl ether, affording IP-H-Me2+ in 82% yield (Scheme 1). The synthesis of 5,15-bis(1-methyl-3-ethylimidazol-2-yl)porphyrin (IP-H-Et2+) was performed under similar microwave conditions but using iodoethane, at 70 ºC for 30 min, yielding 79%. We found that cationization of IP-H with 1,1,1-trifluoro-2-iodoethane did not give any IP-H-FEt2+, which may be attributed to the high transition state energy of SN2 reactions involving β-fluoroethyl iodides [35]. Instead, using 1,1,1-trifluoro-3-iodopropane at 150 ºC and 30 min in similar microwave conditions, we obtained 5,15-bis(1-methyl-3-(1,1,1-trifluoropropyl)imidazol-2-yl)porphyrin (IP-H-FPr2+), albeit with moderate yield (59%).

Scheme 1
scheme 1

Route to di-cationic 5,15-bis(1,3-dimethylimidazol-2-yl)porphyrin and chlorin derivatives

Full size image

Below we show that the photochemical properties of the smaller dicationic methyl imidazolyl porphyrin are more advantageous for aPDT. We pursued exclusively with the synthesis of the cationic 5,15-bis(1,3-dimethylimidazol-2-yl)chlorin IC-H-Me2+ because: (i) ab initio calculations did not reveal substantial differences in positive charge density between the three chlorins, (ii) asymmetric cationization leads to atropisomers of similar stabilities, (iii) cationization with ethyl or trifluoropropyl derivatives requires higher temperatures and affords lower yields, (iv) the molar absorption coefficient and singlet oxygen quantum yield of IP-H-Me2+ are higher than for the other dicationic imidazolyl porphyrins.

The classic Whitlock method [36] and the more sustainable solvent-free methodology [37], both using activated p-toluenesulfonyl hydrazide, have been successful in the synthesis of a variety of chlorins [38].

However, their use in the reduction of imidazolyl porphyrins, becomes unpractical due to cumbersome purifications. Therefore, we replaced p-toluenesulfonyl hydrazide as the imide source by NH2NH2.H2O. The effects of aprotic solvent polarity and of excess hydrazine were studied, without catalyst, by the sequential addition of 10 + 10 equivalents of NH2NH2.H2O to DMF or to o-xylene solutions. Qualitative assessment based on UV–Vis indicated that moderate amounts of chlorin were formed (Table S1, entry 1 and 3). Additionally, we performed these reactions using a large excess of hydrazine (40 equivalents), under inert atmosphere, to avoid chlorin re-oxidation. Although more chlorin was formed, a larger contamination with bacteriochlorin was observed (Table S1, entries 2 and 4).

With the view of improving chlorin selectivity, we evaluated the effect of the solvent (DMF or o-xylene) in the synthesis of IC-H (Scheme 1), using Pd/C as catalyst of hydrated hydrazine activation [39] [40, 41]. Qualitative evaluation by UV–Vis showed that DMF under aerobic conditions afforded the most promising results (Table S1, entries 5–6). Thus, the desired chlorin was prepared with the optimized method (DMF, Pd/C, 40 equivalents of NH2NH2.H2O) and after 2 h it was precipitated from the reaction mixture with the addition of diethyl ether. After filtration, IC-H was purified by flash chromatography to afford 41% isolated yield of porphyrin:chlorin (25:75), mixture as determined by 1H NMR (Fig. S11). The desired cationic 5,15-bis(1,3-dimethylimidazol-2-yl)chlorin was prepared by reaction of IC-H dissolved in DMF with iodomethane, under inert atmosphere, at 30 ºC, for 4 h. IC-H cationization reaction was not performed under microwave irradiation in order to avoid chlorin oxidation. IC-H was precipitated from the reaction mixture by adding a solution of CH2Cl2/pentane (1:1). The precipitated was filtered-off, washed with CH2Cl2/pentane and dried, yielding the desired 5,15-bis(1,3-dimethylimidazol-2-yl)chlorin IC-H-Me2+ with 82% isolated yield and no appreciable contamination with porphyrin as demonstrated by 1H NMR and mass spectrometry (Figs S13 and S14). Attempts to prepare IC-H-Me2+ directly by reduction of IP-H-Me2+ were unsuccessful.

2.3 Photosensitizer properties

Table 1 presents the major photophysical properties of the photosensitizers synthesized in this work. The UV–Vis spectra of the dicationic porphyrins in DMSO:water (2:30, v:v) are typical of porphyrin derivatives (Fig. S16) and follow Beer-Lambert law in the 1.8–20.0 µM concentration range, suggesting negligible aggregation. The fluorescence excitation spectrum of IC-H-Me2+ is also consistent with the absence of aggregation (Fig. 2A). Photodecomposition quantum yields (\(\Phi\)pd) were determined exposing the porphyrins to LED light (5 mW at 630 nm) for sufficient time to deliver up to 80 J while measuring their absorption spectra (Fig. S16). The dicationic species did not exhibit measurable photodecomposition, \(\Phi\)pd < 10–7. Steady-state singlet oxygen phosphorescence was obtained exciting porphyrins or chlorins in ethanol and singlet oxygen quantum yields (\(\Phi\)∆) were obtained using methylene blue as reference (Fig. S17). Table 1 shows that IC-H-Me2+ has a high \(\Phi\)∆ and a high molar absorption coefficient (ε) in the phototherapeutic window. The related porphyrins with ethyl (IP-H-Et2+) or trifluoropropyl (IP-H-FPr2+) substituents show a tendency to increase the fluorescence quantum yields and decrease the singlet oxygen quantum yields. This is not desirable for aPDT and was mentioned above as one of the reasons why we did not pursue the synthesis of the analogous chlorins. Fig. S16 shows that its UV–Vis spectrum of IC-H-Me2+ in ethanol does not change with exposure to 80 J at 660 nm (13 mW LED, FWHM = 19 nm). This water-soluble and remarkably photostable dicationic imidazolyl chlorin has a high \(\Phi\)∆ and intense absorption in the phototherapeutic window (Fig. 2A), which are excellent properties for aPDT.

Table 1 Photosensitizers and their photophysical and photochemical properties in ethanol
Full size table
Fig. 2
figure 2

a UV–Vis absorption (red, normalized for absorption at 651 nm) fluorescence excitation (dotted line) and fluorescence (blue, normalized for 658 nm) spectra of IC-H-Me2+ in ethanol. b Cytotoxicity in the dark against HaCaT after 24 h of incubation. c Phototoxicity for a light dose of 5 J cm–2 against HaCaT cells after 1 h of incubation. d Phototoxicity for a light dose 5 J cm–2 against HaCaT cells after 24 h of incubation

Full size image

2.4 Cytotoxicity and phototoxicity against human cells

Cytotoxicity and phototoxicity of IC-H-Me2+ were evaluated in human epidermal keratinocytes (HaCaT) and in human embryonic kidney (HEK-293 T) cells, as well as in HEK cells overexpressing the hACE2 gene (HEK-293 T-hACE2), after 1 h, 12 h or 24 h of incubation, to evaluate possible side effects of aPDT of virus (Figs. 2b-d, 3 and S18). The onset of cytotoxicity in the dark after 24 h of incubation is > 100 µM. The onset of phototoxicity is ~ 2 µM for HaCaT cell after 1 h of incubation and ~ 0.4 µM for HEK-293 T after 12 h of incubation. The low toxicity of IC-H-Me2+ at short incubation times is readily understood considering that at early incubation times ROS are generated outside the cell or in the cell membrane, and damage is essentially restricted to the cell membrane. However, eukaryotic cells are quite resilient to membrane damage because they evolved to include membrane repair mechanisms that restore membrane integrity after injury [42]. The onsets of cytotoxicity and phototoxicity provides the limits to perform selective aPDT of virus.

Fig. 3
figure 3

Infection of HEK-293 T-hACE2 cells with lentiviral pseudovirus vectors (LVPs) after 72 h of incubation. a) Cell viability. b) Bioluminescence of the luciferase assay relative to cell viability. HEK = cells only, all the other cases also have cells but they are in combination with other variables; ICa = [IC-H-Me2+] = 0.4 µM; ICb = [IC-H-Me2+] = 0.8 µM; L = 4 J cm–2 at 650 nm. * p < 0.05

Full size image

2.5 aPDT of lentivirus

Lentiviral pseudovirus vectors (LVPs) are safe, versatile and very useful to investigate the impact of therapies on the ability of viruses to infect cells. Our LVPs are derived from HIV and used four major plasmids, two of them encoding packaging genes, one the envelope including S protein, and the other encoding the reporter gene Luciferase. LVPs were produced in HEK293T cells [43], using polyethylenimine. In this study, 8.5 × 103 HEK-293 T-hACE2 cells were seeded in 96 multi-well plate with DMEM high glucose. After 24 h, cells were exposed to 60 ng/p24 LVPs (positive control), to LVPs with IC-H-Me2+, to LVPs with IC-H-Me2+ and light (4 J cm–2), or just to PBS (negative control) with DMEM high glucose, without Phenol Red. These LVPs show a special ability to infect HEK-293 T-hACE2 cells since they overexpress the hACE receptor of the pseudovirus S protein. After 24 h of cell exposure and infection with LVPs, the cell medium was replaced. Forty-eight hours later, the luciferase assay gives a luminescent intensity proportional to the degree of infection. Figure 3a shows that LVPs, LVPs with light, and LVPs with IC-H-Me2+do not affect cell viability. There is a small decrease in cell viability when the highest concentration of IC-H-Me2+ is combined with light and virus. This decrease in cell viability is consistent with the data in Fig. 2d.

Figure 3b shows luciferase luminescence normalized by cell viability, to correct for the trivial absence of luminescence from dead cells. As expected, controls without LVP do not show any luminescence because the lentivirus must be present and infect the cells to transfect the luciferase reporter gene. The average luminescence from the cells infected with LVP was taken as reference and set to the maximum luminescence intensity (100%). Cells infected with LVP and exposed to light (LVP + L) or incubated with different concentrations of IC-H-Me2+ (LVP + ICa, LVP + ICb) showed average luminescence statistically indistinguishable from that of cells infect with LVP alone. Remarkably, we observed a dramatic reduction of luminescence when the virus is treated with IC-H-Me2+ and light prior to incubation with HEK-293 T-hACE2 cells. The infectivity is reduced to 50% of LVPs alone for [IC-H-Me2+] = 0.4 µM and 4 J cm–2, and all detectable infection was eliminated doubling the chlorin concentration (0.8 µM). The light dose used here can be delivered in 20 s with a clinical laser of 200 mW cm–2, without thermal effects. This offers opportunities for fast photodisinfection without toxicity to human cells.

It is tempting to assign the susceptibility of these viruses to aPDT to changes in the viral envelope possibly related to singlet oxygen oxidation of C = C double bonds present in their unsaturated phospholipids, which increase the rigidity of the envelope [44]. The fusion between viral and cell membranes is an obligatory step in cell infection by enveloped viruses and requires appropriate membrane fluidity [45]. The oxidation of unsaturated phospholipids and concomitant decrease in fluidity may result in inhibition of membrane fusion and infection.

2.6 aPDT of SARS-CoV-2

The clinical relevance of aPDT with IC-H-Me2+ was investigated using clinical samples of SARS-CoV-2 collected from the nasopharynx of COVID-19 patients admitted in the University of Coimbra Hospital Center (CHUC), after their informed consent, according to a protocol approved by the CHUC ethical committee [46]. SARS-CoV-2 samples transported in viral transport medium to a biosafety-II lab were split in control and aPDT experiments. aPDT was completed less than 15 min after the addition of IC-H-Me2+ to SARS-CoV-2 in 24 multi-well plates. The amount of SARS-CoV-2 in the samples was evaluated by RT-qPCR using a calibration curve to quantify the reduction of the viral titer with the treatment. Figure 4 shows that the viral titer drops below the sensitivity of our RT-qPCR equipment (amplification inhibition > 99.999%) at 0.4 µM with 5 J cm–2. Within seconds of illumination, aPDT with IC-H-Me2+ can damage SARS-CoV-2 RNA to the point that it is no longer amplified in PCR. This photodisinfection mechanism complements viral envelope targeting mentioned above. Although the processes are entirely different, it is interesting to compare IC-H-Me2+ with remdesivir. This antiviral has a half-maximum effective concentration (EC50) of 0.12 µM and eliminates all detectable SARS-CoV-2 infection at 2.4 µM [13]. A more direct comparison with the photodisinfection efficacy of IC-H-Me2+ can be made with a commercial methylene blue formulation, which requires concentrations higher than 15 µM and ~ 100 J to achieve the same level of photodisinfection [46]. IC-H-Me2+ offers the same level of photodisinfection as methylene blue with a concentration two orders of magnitude lower and a light dose one order of magnitude lower (considering the areas of the wells, the total light dose per well is 9.5 J for a 5 J cm–2 radiant exposure).

Fig. 4
figure 4

aPDT of clinical samples with different light doses. a 5 J cm–2. b 1.4 J cm–2. V = virus only; V + L = virus and light; V + C = virus and IC-H-Me2+; V + L + ICa,b = virus, light and IC-H-Me2+ where ICa [IC-H-Me2+] = 0.4 µM or ICb = [IC-H-Me2+] = 0.8 µM. Control gathers the various positive controls

Full size image

3 Conclusions

Using rational design, this study identified a new class dicationic imidazolyl chlorin photosensitizers for aPDT of viruses. A synthetic route was developed for these imidazolyl chlorins using hydazine.H2O catalyzed by C/Pd as reducing agent, followed by methylation of imidazolyl group. This allowed for the synthesis of IC-H-Me2+ in high yields. We showed that IC-H-Me2+ has excellent properties for aPDT of microorganisms, namely small size, intense absorption in the red, photostability and high yield of singlet oxygen generation. IC-H-Me2+ does not exhibit cytotoxicity in the dark at least up to 100 µM. aPDT of lentivirus and SARS-CoV-2 with IC-H-Me2+ shows that it allows for a broad-spectrum approach to virus inactivation, possibly targeting viral membrane components, the S protein and viral RNA.

aPDT is still underexploited in the clinical management of viral infections, although its use to treat bacterial infections is very well established [47] and the clinical evidence endorsing its use in periodontal disease [48] and in infected diabetic foot ulcers [25] is impressive. The most successful examples of antiviral PDT applications remain the photoinactivation of HIV-1 in blood products using methylene blue [49] and the treatment of herpes and warts [50,51,52,53]. Very recently a clinical trial on SARS-CoV-2 with methylene blue showed that intranasal photodisinfection reduced viral infectivity in clinical nares swabs of 90% of the subjects testing positive to SARS-CoV-2 [54] and another clinical trial with the same product showed that the probability of becoming PCR negative 7 days after treatment was increased with PDT [55]. The upper oral-nasopharyngeal area is the portal of entry of SARS-CoV-2 and this virus is found primarily in the upper respiratory track in the first 5 days after infection. Timely local photodisinfection of the upper oral-nasopharyngeal area may reduce, or suppress, the spread of the disease.

Most viruses interact with cells of the host mucosal epithelium in the respiratory track, the gastrointestinal track or the genital track, which they use as portals of entry in the human body. If they infect and replicate only within cells at the site of infection, such as the human papillomavirus, they cause localized infections that can be treated with aPDT at any time using minimally invasive devices. If they spread to other sites within the body, such as SARS-CoV-2, they will cause systemic infections but these can still be managed with aPDT during the incubation and prodromal periods, and possibly also for a few additional days.

The virus photoinactivation efficacy of IC-H-Me2+ is unprecedented for a red-light absorbing photosensitizer. Sub-micromolar concentrations of IC-H-Me2+ together with red light doses that can be delivered through optical fibers in a few seconds, reduce very high SARS-CoV-2 loads below the detection limit of RT-qPCR. aPDT with IC-H-Me2+ may be performed by trained personnel in one minute to reduce viral load, curb transmission, delay dissemination, reduce symptoms and give more time to the host immune system to mount an effective response against the infection. aPDT with IC-H-Me2+ is unlikely to be susceptible to mutations in the virus because it does not depend on the affinity to a specific target. It can be expected to have broad applications in the inactivation of known enveloped viruses and of viruses that may cross species barriers in the future. aPDT with IC-H-Me2+ expands the medical armamentarium in the prophylaxis and treatment of viral infections.

4 Materials and methods

4.1 Electronic structure calculations

Molecular structures of di-cationic imidazolyl chlorins were optimized at the DFT level of theory, using the B3LYP hybrid functional and the standard 6-31G(d,p) basis set, as discussed elsewhere [29]. Gamess [56] was used for geometry optimization while electron density maps were calculated with Gaussian09 [57]. The results of the calculations are presented in terms of electronic density maps to illustrate positive charge distributions (see Fig. S1). When atropisomers were found, the energies of the lower energy conformers were calculated (see Fig. S2).

4.2 Synthetic methods

All reagents and solvents were purchased from Sigma-Aldrich, Fluorochem and Acros Organics. All reagents were used as supplied, without further purification. Air sensitive reactions were handled under nitrogen or argon atmosphere, in a vacuum system, using Schlenk techniques. The solvents were distilled and degassed prior to use, according to standard procedures. 1H NMR spectra were recorded on a Bruker Avance 400 spectrometer at 400 MHz. 1H NMR spectra were acquired in CD3OD, DMSO-d6, acetone- d6 and the signals were referenced using the solvent peak as internal standard. High-resolution mass spectrometry (HRMS) analyses were conducted on a Bruker Microtof and a Bruker QqTOF Impact II.

Multi-gram synthesis of 5,15-bis(1-methylimidazol-2-yl)porphyrin (IP-H). Dipyrromethane (13.27 g; 92 mmol) was dissolved in 17.2 L of dichloromethane and introduced onto a 20 L reactor. Next, 1-methylimidazol-2-carboxyaldehyde (10.13 g; 92 mmol) was added. The solution was bubbled with nitrogen along 15 min. After this period, trifluoroacetic acid (TFA) (4.20 mL; 56 mmol) was added. The solution was left with stirring at room temperature (25 ºC), for 24 h, protected from light. The porphyrinogen oxidation was performed with tetrachloro-p-benzoquinone (28.56 g; 116 mmol). The reaction mixture was left at room temperature, for 2 h under stirring, protected from light. The temperature was then risen to 60 ºC and left for 1 h. Finally, triethylamine (31.96 mL; 229 mmol) was added and the reaction was maintained under stirring, for another 30 min. The solvent was evaporated to dryness. The crude was dissolved in the minimum amount of dichoromethane (100 mL) and a silica gel plug was performed, using initially dichloromethane as eluent and finally a mixture of dichlomethane/methanol (methanol up to 10%). The product was recrystallized from a mixture of dichloromethane:heptane. 5,15-bis(1-methylimidazol-2-yl) porphyrin IP-H was obtained in 8% yield (1.6 g). 1H NMR (400 MHz, ((CD3)2CO), δ, ppm: atropisomer mixture 10.67–10.63 (m, 2H), 9.69–9.67 (m, 4H), 9.11–9.10 (m, 4H), 7.83–7.81 (m, 2H), 7.67–7.65 (m, 2H), 3.67 (s, 3H), 3.61 (s, 3H), -3.30 to -3.16 (m, 2H). See Fig. S3. ESI–MS m/z: obt. 471.2041 [M + H]+; calc. for [C28H23N8]+ 471.2040. The syntheses of IP-H in the mg scale was previously described [29].

General procedure for cationization. 5,15-bis(1-Methylimidazol-2-yl)porphyrin (50 mg, 0.11 mmol) in 0.2 mL of DMF was placed in a 5 mL microwave vial and 10 equivalents of the corresponding alkyl iodide was added. The mixture underwent microwave irradiation with Pmax = 125 W using different time intervals and temperatures depending on the alkyl iodide used. The evolution of the reaction was monitored by TLC. The crude obtained was precipitated using diethyl ether and the solid was filtered. Then, it was re-dissolved in dichloromethane and a second precipitation was performed using diethyl ether. The resulting solid was filtered, washed with diethyl ether and dried.

5,15-bis(1,3-dimethylimidazol-2-yl)porphyrin diiodide (IP-H-Me2+). 5,15-bis(1-Methylimidazol-2-yl)porphyrin was dissolved in DMF and iodomethane (0.068 mL, 1.1 mmol) was added. The reaction was left under microwave irradiation at 40 ºC, for 20 min. After precipitation, 5,15-bis(1,3-dimethylimidazol-2-yl)porphyrin diiodide, IP-H-Me2+, was obtained in 82% yield (45 mg). 1H NMR (400 MHz, DMSO-d6) δ, ppm: 10.98 (s, 2H), 9.96 (d, J = 4.7 Hz, 4H), 9.26 (d, J = 4.7 Hz, 4H), 8.52 (s, 4H), 3.75 (s, 12H), -3.55 (s, 2H). ESI–MS m/z: 250.1218 [M-2I]2+; calculated for [C30H28N8]2+/2 250.1213. See Figs. S4 and S5.

5,15-bis(1-methyl-3-ethylimidazol-2-yl)porphyrin diiodide (IP-H-Et2+). 5,15-bis(1-Methylimidazol-2-yl)porphyrin was dissolved in DMF and iodoethane (0.088 mL, 1.1 mmol) was added. The reaction was left under microwave irradiation at 70 ºC, for 30 min. After precipitation, 5,15-bis(1-methyl-3-ethylimidazol-2-yl)porphyrin diiodide IP-H-Et2+ was obtained in 79% yield (46 mg). 1H NMR (400 MHz, DMSO-d6)δ, ppm: mixture of atropisomers 10.98 (s, 2H), 9.96 (d, J = 4.8 Hz, 4H), 9.26 (d, J = 4.8 Hz, 4H), 8.65–8.62 (m, 2H), 8.58–8.56 (m, 2H), 4.06–3.96 (m, 4H), 3.78–3.72 (m, 6H), 1.18–1.09 (m, 6H), -3.55 (s, 2H). ESI–MS m/z: 264.1451 [M-2I]2+/2; calculated for [C32H32N8]2+/2 264.1370. See Figs. S6 and S7.

5,15-bis(1-methyl-3-(1,1,1-trifluoropropyl)imidazol-2-yl)porphyrin diiodide (IP-H-FPr2+). 5,15-bis(1-Methylimidazol-2-yl)porphyrin was dissolved in DMF and 1,1,1-trifluoro-3-iodopropane (0.129 mL, 1.1 mmol) was added. The reaction was left under microwave irradiation at 150 ºC, for 30 min. After precipitation, 5,15-bis(1-methyl-3-(1,1,1-trifluoropropyl)imidazol-2-yl)porphyrin diiodide, IP-H-FPr2+, was obtained in 59% yield (43 mg). 1H NMR (400 MHz, CD3OD) δ, ppm: mixture of atropisomers 10.92 (s, 2H), 9.87 (d, J = 4.7 Hz, 4H), 9.16 (br s, 4H), 8.61–8.55 (m, 2H), 8.50–8.45 (m, 2H), 4.48–4.40 (m, 4H), 3.92 – 3.86 (m, 6H), 2.76–2.59 (m, 4H); 19F NMR (377 MHz, CD3OD) δ, ppm: -66.72 to -66.75 (m, 6F). ESI–MS m/z: obtained 664.2467 [M-2I]+; calculated for [C34H30F6N8]+ 664.2492. See Figs. S8, S9 and S10.

5,15-bis(1-methylimidazol-2-yl)chlorin (IC-H). 5,15-bis(1-Methylimidazol-2-yl) porphyrin (100 mg; 0.2 mmol) was dissolved in DMF (2.5 mL). Then, Pd/C (5%) (10 mg) and hydrazine (NH2NH2.H2O) (0.67 mL; 8 mmol) were added. The reaction was left at 70 ºC for 2 h with stirring and was monitored by UV–Vis spectroscopy. Once the reaction was completed, the reaction mixture was filtered to retain the solid Pd/C and diethyl ether was added to induce precipitation of the crude chlorin. The solid was filtered under vacuum and purified by silica gel flash chromatography using a gradient of solvents; starting from dichloromethane, to remove small impurities, then a solvent mixture (dichloromethane/methanol/triethylamine; 100:4:0.1) was used to remove unreacted porphyrin. The last fraction eluted with a solvent mixture of dichloromethane/methanol/triethylamine (100:5:0.5) was collected. After evaporation of the solvent and drying, 41 mg of a solid characterized by 1H NMR as a mixture of IC-H (75%) and IP-H (25%) was obtained (41% yield). 1H NMR (400 MHz, ((CD3)2CO), δ, ppm: (mixture of atropisomers; 75% chlorin content) 10.12–10.11 (m, 1H), 9.42–9.41 (m, 1H), 9.31–9.29 (m, 1H), 9.13–9.11 (m, 2H), 8.88–8.87 (m, 1H), 8.52–8.51 (m, 1H), 8.37–8.36 (m, 1H), 7.74–7.51 (m, 4H), 4.84–4.79 (m, 2H), 4.67–4.61 (m, 1H), 4.27–4.21 (m, 1H), 3.69–3.57 (m, 6H), -1.49 (s, 1H), -1.80 (s, 1H). ESI–MS m/z: obt. 473.2195 [M + H]+; calc. for [C28H25N8]+ 473.2202. See Figs. S11 and S12.

5,15-bis(1,3-dimethylimidazol-2-yl)chlorin diiodide (IC-H-Me2+). 5,15-bis(1-Methylimidazol-2-yl) chlorin IC-H (71 mg; 0.15 mmol) was dissolved in DMF (1 mL) and an excess of iodomethane (0.46 mL; 7.39 mmol) was added. The reaction was kept at 30 ºC for 4 h, under argon atmosphere. The reaction was followed by TLC and UV–Vis spectroscopy. The product was precipitated with CH2Cl2 and pentane and the solid was filtrated and dried under vacuum. The 5,15-bis(1,3-dimethylimidazol-2-yl)chlorin diiodide IC-H-Me2+ was obtained in 82% yield (93 mg). 1H NMR (400 MHz, CD3OD), δ, ppm: 10.31–10.23 (m, 1H), 9.55–9.48 (m, 2H), 9.32–9.22 (m, 2H), 8.95–8.89 (m, 1H), 8.47–8.39 (m, 2H), 8.29 (s, 1H), 8.25–8.24 (m, 1H), 8.20–8.13 (m, 2H), 4.97–4.95 (m, 2H), 4.55–4.38 (m, 2H), 3.86–3.83 (m, 12H). ESI–MS m/z: obt. 251.1293 [M-2I]2+; calc. for [C30H30N8]2+/2 251.1291. HPLC purity is > 95%, as the average of the areas of the chromatograms with detection at 385 and 405 nm. See Figs. S13, S14 and S15.

4.3 Photophysical measurements

The absorption coefficients were determined according to Beer-Lambert’s law. For each compound, a minimum of 6 solutions were prepared in concentrations ranging from 10–7 to 10–6 M corresponding to absorbance values between 0.1 and 1.0. Fluorescence experiments were performed in a spectrometer PerkinElmer LS 45 and Horiba Fluorog. Porphyrin samples were excited at 420 nm, using tetraphenylporphyrin (TPP) as a reference in toluene, \(\phi\)F = 0.11 [58], with optical absorption ca. 0.01 at the Soret band. Fluorescences of samples were determined in ethanol, applying a correction factor for the refractive index in the determination of fluorescence quantum yield. Chlorin samples were excited either at 652 nm or 635 nm, using oxazine 170 as reference in ethanol, \(\phi\)F = 0.579 [59]. The absorptions of samples and references were below 0.01 at the excitation wavelengths. Fluorescence excitation of IC-H-Me2+ in ethanol was collected with detection of emission at 700 nm.

Fluorescence quantum yields were determined by integrating the emission spectra, in the wavenumber scale, following the equation:

$${\phi }_{F}^{s}=\frac{{n}_{s}^{2}}{{n}_{r}^{2}}\frac{(1-{10}^{-{A}^{r}})}{(1-{10}^{-{A}^{s}})}\frac{\int s}{\int r}{\phi }_{F}^{r}$$

where, s and r mean sample and reference.

In order to determine the photodecomposition quantum yield, 3 mL of solutions either in ethanol, 2% DMSO in water, or cell culture medium were irradiated by LED light sources (Fig. S16). A LED emitting at 630 nm and total output power of 5 mW was employed for porphyrins. A LED emitting at 660 nm and total output power of 13 mW was employed for chlorins. The solutions were kept stirring during the experiment and optical absorption spectra recorded at various time intervals. The quantum yield was calculated applying the equation [26]

$${\phi }_{PD}=\frac{hc}{\lambda P(1-{10}^{-A})}\frac{V{N}_{A}}{\varepsilon l}\frac{\Delta A}{\Delta t}$$

The photodecomposition quantum yield was determined by measuring the difference between initial and final optical absorption ΔA and irradiation time Δt. In the equation h is Planck’s constant, c the speed of light, NA Avogadro’s number, λ the light source wavelength, P the total power, V the volume of solution in the cuvette, A the absorption of the compound in the wavelength of light source and ε and l the molar absorption coefficient and cuvette optical path, respectively.

Singlet oxygen quantum yields (\(\Phi_{\Delta }\)) were determined by collecting the steady-state phosphorescence spectrum of oxygen in the spectrofluorometer Horiba Fluorolog (Fig. S17). The phosphorescence was collected using a photomultiplier Hamamatsu R5509-42, with voltage set to 1750 V, cooled to 193 K in a liquid nitrogen chamber. The cutoff filter Newport 10LWF-1000-B was used to avoid fluorescence light. Methylene blue was employed as a reference, \(\Phi_{\Delta }\) = 0.50 [60]. Excitation was either at 625 nm for porphyrins or 650 nm for chlorins. Optical absorption in both wavelengths was ca. 0.3. All samples were prepared in ethanol. Singlet oxygen quantum yields were calculated using the same procedure and equation as for fluorescence.

The light doses used to induce photodegradation were corrected considering the overlap between the emission spectra of the light sources and the absorption spectra of the compounds [61].

4.4 Toxicity in vitro

20.000 HaCaT (human epidermal keratinocytes) cells and 8.5 × 103 HEK-293 T (human embryonic kidney) cells were seeded in 96 multi-well plate in RPMI or DMEM medium, respectively, without phenol red. After 24 h, cells were incubated with several concentrations of IC-H-Me2+ for the desired incubation times and illumination was performed with a 660 nm LED (13 mW). Cell viability was assessed 24 h later using the Alamar Blue assay. See Fig. S18 for HEK-293 T cells.

4.5 HEK-293 T-hACE2 cells

Plasmids pRSV–Ver, pCMV- r8,92, and pRP[Exp]-CMV-human beta globin intron > {S (2020, deltaC19)-3xFLA (Vector Builder, Neu-Isenburg, Germany) were mixed with the Luciferase reporter transgene. HEK-293 T cells were then transfected with a previously defined proportion of each plasmid and PEI at 1 mg/mL. After pseudoviral production and purification, quantification was performed using HIV-1 P24 Antigen ELISA kit from ZeptoMetrix. Pseudoviral validation was efficiently performed in previous studies (data not shown) by infecting HEK-293 T permanently expressing human angiotensin I-converting enzyme 2 (hACE2) (CoronaAssay-293 T cells (hACE2)—#: CACL-0012, Vector Builder)—(HEK-293 T-hACE2) and HEK-293 T cells with the produced LVs at different dilutions for 72 h.

4.6 SARS-CoV-2 clinical samples and RT-qPCR

The origin of the SARS-CoV-2 clinical samples is the same as that reported in earlier work, and the RT-qPCR methods employed in this work were also identical to those reported earlier [46]. The concentration range used in the calibration curve was 2 × 102 – 3 × 106 copies/ml, which corresponded to a range of 30 to 16 cycles. A calibration curve was made for each one of the analyses of the samples with freshly prepared standards. Fig. S19 presents a typical calibration curve. The lowest number of cycles measured for any sample was Ct = 22.84 (3.84 × 104 copies/ml). After treatment with IC-H-Me2+ and 5 J cm–2, the RNA of SARS-CoV-2 became undetectable. In our system, the highest number of cycles measured for any sample was Ct = 43.89 (2.33 × 10–2 copies/ml, by extrapolation from the calibration curve) [46]. Hence, in this case we set the final virus titer as 2 × 10–2 copies/ml. The virus titer is different from one patient to another. The percentage inhibition of amplification was calculated as (1–Qt/Qctr) × 100%, where Qt and Qctr correspond to virus titer in treated and control samples.

4.7 Statistical analysis

In vitro results are shown as Mean and Standard Error of the Mean. Statistical differences between populations were assessed with Student’s t-test for unpaired data with unpaired variance: *, p ≤ 0.05; **, p ≤ 0.01; ***, p ≤ 0.001.