Our study demonstrated that polymyxin antibiotics have, aside from their antibactericidal activity, an antifungal activity, especially against multidrug-resistant Candida, Rhodotorula and Cryptococcus yeast isolates, but also against resistant filamentous fungi, such as Scedosporium, Rhizopus and Lichtheimia species. The MICs obtained against Candida spp. ranged between 64 and 128 μg/ml for colistin and mainly between 16 and 64 μg/ml for PMB. The latter is in concordance with PMB MICs already reported by Zeidler et al and Zhai et al studies [7, 24] confirming the validity of our results and the reproducibility of the used technique. Although we obtained a MIC of PMB equal to 16 μg/ml against C. neoformans, a lower MIC (MIC = 8 μg/ml) against this species has been previously described . This could possibly be explained by the resistance phenotype of the strain tested in our study. Moreover, colistin has not been shown, in the early 1970s, to be an effective molecule against three strains belonging to the genus Rhodotorula . However, to the best of our knowledge, neither colistin MICs against C. neoformans nor PMB MICs against R. mucilaginosa have been reported elsewhere.
On the other hand, the filamentous fungi isolates tested in this study constitute the most common emerging cause of human mould infections with an increase being reported from various geographical sites , including particularly Aspergillus and Fusarium spp. Here, colistin and PMB MICs ranging from 16 to 64 μg/ml have been observed against the F. oxysporum and F. Solani strains which are resistant to almost all azole antifungals and amphotericin B. The same range of MICs has been reported in previous study where 12 Fusarium spp. were tested against PMB but not against colistin . The absence of colistin and PMB efficacy against several Aspergillus spp. has been reported in various studies with MICs > 256 μg/ml [24, 25] which are similar to our results. Despite that Aspergillus spp. are remaining the first cause of mould infections, mucormycosis is increasingly reported in immune-compromised patients and is associated with an elevated rate of mortality (40–70%) even under an appropriate therapy . Among mucoralean pathogens, Rhizopus is the main frequently identified genus in human infections. In our study, colistin and PMB MICs against R. oryzae are one to two folds higher than those reported in earlier studies [15, 24]. Indeed, in Ben-Ami et al study, the colistin MICs against the fourteen clinical Rhizopus spp. tested were variables and ranged between 16 and 32 μg/ml, but MICs of antifungal agents were not mentioned . So the discordance of MIC results between the Ben-Ami et al study and our work could be explained by the eventual high resistance level of our strain to azole agents which can be due to the over expression of efflux-pumps and/or other mechanisms . It is important to mention that PMB MIC against R. oryzae was equal to 64 μg/ml in this study compared to 32 μg/ml obtained by Zhai et al study .
Finally, among the pertinent emerging fungal pathogens shown by several studies, Scedosporium and Lomentospora spp. are often notified . They can induce a broad range of diseases; from colonisation in cystic fibrosis patients (for Scedosporium spp.) to disseminated severe infections in immuno-compromised hosts (for Lomentospora prolificans). Although, the colistin MIC against S. apiospermum strain tested here is within the colistin MICs range previously described by Schemuth et al, the colistin MIC obtained for L. prolificans was higher than that described in this previous study (32 μg/ml versus 12 μg/ml) . Nevertheless, it is worthy to note that MICs90 were used by Schemuth et al  whereas MICs100 were used in our study. Finally, to the best of our knowledge, colistin and PMB activities against Lichtheimia corymbifera have not been previously reported.
In human studies, a single dose of 75 to 150 mg of colistin produced bioactive serum colistin concentrations ranging from 6 to 18 μg/ml; higher serum colistin concentrations (13 to 32 μg/ml) were measured during the prolonged therapy of patients with cystic fibrosis . Therefore, the obtained MICs of colistin and PMB are difficult to be achieved with IV administration, mainly due to their renal and neurological toxicities and the risk of frequent selection of bacterial resistant strains.
However, the efficacy of polymyxin molecules on a large number of MDR fungi can be considered advantageous to treat bacterial and fungal co-infections that occur frequently in immunocompromised patients  and cystic fibrosis (CF) patients. Chronic bacterial and fungal colonization of the respiratory tract secretions is the main cause of morbidity and mortality in CF patients. Therefore, it would be helpful to use a treatment that is active on both bacteria and fungi in this context. It is worthy to note that, in clinical practice, colistin is administered by inhalation in CF patients as prophylaxis and also as a treatment against Pseudomonas aeruginosa infection . In addition, aerosolised colistin treatment, is used in ventilator-associated pneumonia (VAP) cases caused by MDR bacteria in intensive care unit setting . Interestingly, in a recent in vivo study, Landersdorfer et al  observed high epithelial lining fluid and low plasma colistin concentrations following the administration of only a pulmonary dose through jet nebulization, confirming a benefit of the local administration of colistin in comparison to its IV treatment . Moreover, a prospective study conducted on 18 patients with chronic lung disease showed that nebulized colistin is effective and improves the quality of life, without presenting side effects and without selecting colistin-resistant isolates in treated patients . So high-dose nebulized colistin could be proposed against pulmonary life-threatening MDR fungi, without increasing colistin plasma concentration, and thus avoiding colistin’s toxicity.
Similar to CF cases, the use of polymyxin antibiotics can improve the poor prognosis of fungal keratitis, due to the emergence of MDR fungal pathogens, particularly Fusarium spp. , and to the limited ocular penetration of antifungals . Notably, PMB can be formulated for ophthalmic use , which is described as a highly effective drug on bacterial corneal ulcerations . Moreover, the use of such antimicrobial agent constitutes a potential alternative treatment that may improve the outcome in some critical infections caused by MDR fungi, such as the recent MDR Fusarium keratitis-case report in a 46-year-old man who was still declining even the maximal therapeutic support and therapeutic keratoplasty .
Several approaches could be used to overcome the development of antifungal resistance in the treatment of fungal diseases. Aside from the discovery of new effective agents, one realistic alternative option would be to enhance the activity of existing agents. Combination therapies exploit the chances for better efficacy, decreased toxicity and reduced development of drug resistance . A previous study demonstrated an in vitro synergy between colistin and echinocandins in several pathogenic yeasts, namely C. albicans, C. glabrata, C. tropicalis, C. parapsilosis and C. krusei, as well as in fluconazole-resistant C. albicans strains .
To the best of our knowledge, no previous studies have tested the activity of colistin in combination with other antifungal agents against fungi of the genera Rhodotorula and Lichtheimia.
A high decrease of colistin’s MICs was observed when it was combined with azoles (with fluconazole against R. mucilaginosa and with itraconazole against either C. albicans or L. corymbifera, Table 2 and Fig. 3). It is well known that the main mechanism of action of azoles is the inhibition of enzymes that transform lanosterol into ergosterol, a major lipid of the fungal membrane. This inhibition alters both the permeability and fluidity of fungal membrane . On the other hand, polymyxins are well known for weakening the outer membrane in Gram-negative bacteria and the disruption of its permeability leading to a leakage of intracellular components . Therefore, and as supported by PI staining results (Fig. 2), it is likely that antifungal azoles ease the polymyxins’ action and add a potential damage to the fungal membrane which results in a synergistic potency of the combined drugs. Moreover, colistin MIC values significantly decreased from 128 μg/ml to 1 μg/ml and from 32 μg/ml to 0.5 μg/ml against C. albicans and L. corymbifera respectively when it was associated with amphotericin B. The association of the fungal membrane permeabilization induced by amphotericin B via ion channel formation  with the probable membrane damage occurred by colistin could explain the decrease of MICs and the synergistic effect between colistin and amphotericin B.
Thus, despite the elevated MICs of colistin found in our work against multidrug-resistant yeast and moulds, the use of colistin, in combination with other antifungal agents, remains an excellent way to avoid the development of fungal resistance and to decrease the antifungal effective concentration usually used in clinical settings [16, 22].
Colistin is one of many AMPs already used in clinical settings . So, in addition to the colistin-antifungal combination evaluated in this study, other AMPs could further be tested to potentiate the antifungal activity of existing antifungal compounds. For example, Wakabayashi et al, previously described the synergistic effect of lactoferin, a human antimicrobial peptide, with clotrimazole against C. albicans . Moreover, lactoferin induced an important decrease of all azoles’ MICs tested against azole-resistant Candida spp. . Consequently, natural or synthetic AMPs, have been identified as an original therapeutic alternative that could be investigated by medical researchers and pharmaceutical companies. Using the same approach which was used herein, another AMP, less toxic than polymyxins such as bacitracin or gramicidin analogues, could be tested as monotherapy or in association with antifungals against MDR fungi.