Introduction

Aspergillus fumigatus is a ubiquitous saprophytic fungus associated with a variety of diseases including allergic manifestations and chronic infections in the immunocompetent population and acute invasive localized or disseminated aspergillosis in the immunocompromised host. The estimated burden of disease is 3–500,000 for the acute invasive infections, 3 million with chronic pulmonary aspergillosis and 4 million with allergic bronchopulmonary aspergillosis [1]. Azole antifungal drugs are the cornerstone in the antifungal treatment of aspergillosis due to the clinical superiority of voriconazole for invasive infections and the fact that this group is the only oral option for patients with allergic or chronic forms of aspergillosis treated outside the hospital [25]. Itraconazole is often the primary choice for the allergic and chronic aspergillosis, voriconazole is the first line agent for invasive infection and posaconazole is licensed for prophylaxis and salvage treatment in the immunocompromised host. Second line options are amphotericin B formulations and echinocandins (Fig. 1) [4, 5].

Fig. 1
figure 1

Drug targets of agricultural fungicides (left) and antifungal agents used in human medicine for the treatment of aspergillosis (right)

Azole resistance in Aspergillus spp. may be intrinsic or acquired. Intrinsic resistance is characteristic for some of the sibling species of the A. fumigatus that are not easily identified in the routine laboratory (e.g. A. lentulus and A. udagawae) and acquired resistance is seen in A. fumigatus isolates at rates most commonly not exceeding 5 % but with significant variation (zero to 61 %) depending on geographical location, case mix and method for resistance detection [6, 7••, 8].

Early and appropriate antifungal treatment is associated with lower failure rates in patients with acute invasive aspergillosis and not surprisingly azole resistance has been associated with a poorer outcome [2, 3, 9, 10, 11•]. Often the diagnosis of azole resistant aspergillosis is difficult or delayed. This is in part due to cultures having low diagnostic sensitivity and hence, in many cases no fungal isolate is available for susceptibility testing. However, even if the culture is positive, susceptibility testing is unfortunately not routinely performed at many centres despite recommendations to do so and despite the fact that azole breakpoints have recently been established [6, 12•, 13, 14••]. Therefore, understanding of the clinical relevance of azole resistance and when it should be suspected and tested for is of utmost importance and the fundamental basis for improving management of this disease. In this review we attempt to address azole resistance in Aspergillus with particular focus on the link between the azole use in agriculture and the risk for acquiring azole resistant Aspergillus disease in humans.

Azole Resistance in Aspergillus

Azoles inhibit the ergosterol biosynthetic pathway by binding to its target enzyme lanosterol 14-α demethylase encoded by the CYP51A gene. This enzyme belongs to the cytochrome P450 family and is required for converting lanosterol to ergosterol, an essential component of the fungal cell membrane (Fig. 1). This results in the accumulation of 14-α methyl sterols and impaired cell membrane integrity [15, 16]. Multiple mechanisms of acquired azole resistance in A. fumigatus have been suggested and include: 1) target gene mutations, 2) target gene up-regulation, 3) up-regulation of efflux pumps, 4) reduced membrane permeability and 5) other mechanisms (Table 1).

Table 1 A summary of currently described azole resistance mechanisms in clinical isolates of A. fumigatus

Azole Resistance in Aspergillus Isolates from Azole Exposed Patients Only

A number of cyp51A mutations have been detected in isolates with wild type azole susceptibility phenotype, whereas others are associated with mono- or multi-azole resistance (“hot spot mutations” [13]; Tables 1 and 2)). cyp51A knock-out mutant data, heterologous transformation analysis, molecular dynamic simulations and studies on site-directed mutagenesis, protein folding and homology modelling have assisted in establishing a role for the gene in azole resistance [17•, 18, 19, 20•, 21•], and identifying, confirming and predicting mutations conferring (cross-)resistance [20•, 21•, 22, 23]. Mutational cyp51A hot spot codons such as G54, G138, M220, Y431, and G448 are considered definitely involved in azole resistance and have been reported frequently from multiple centres (Table 2).

Table 2 cyp51A alterations found in A. fumigatus with and without azole resistance or reduced susceptibility

None of these mutations, however, are consistently present in clinically resistant isolates, and azole resistant isolates do not always exhibit cyp51A mutations [20•, 24, 25, 26•]. Relatively consistent azole susceptibility profiles have been described for isolates with hot spot mutations (Table 2). Most mutations confer itraconazole-resistance (Table 2), whereas pan-azole resistance typically has been reported in isolates with G138C or M220K alterations (Table 2). Acquired multi- or pan-azole resistance, where hot spot mutations have emerged during azole therapy over a period of less than a year, has been described [11•, 2730].

Azole-Resistant A. Fumigatus Found in Azole Naïve Patients

Azole resistant A. fumigatus has been found in azole naïve patients and has been shown to harbour one of three different resistance mechanisms: 1) a 34 bp tandem repeat (TR34) in the CYP51A promoter region coupled with a L98H substitution in the CYP51A gene, 2) a 46 bp tandem repeat (TR46) coupled with Y121F/T289A substitutions in the CYP51A gene, or 3) a G432S substitution in the CYP51A gene. Of particular interest is the TR34/L98H genotype, which was initially detected at a Dutch centre in 12/13 itraconazole-resistant patient isolates [31••]; this mutation was subsequently found in the vast majority of itraconazole-resistant isolates from other Dutch centres [20•, 32] and associated with an up-regulation of the target enzyme level (mediated by the tandem repeat) and by decreased affinity for azoles (mediated by the substitution) in combination leading to the pan azole resistant phenotype [21•]. Using site-directed mutagenesis, Snelders et al. [21•] showed that the multi-azole resistant phenotype could not be induced exclusively by introducing either the TR34 or L98H mutation, indicating that the multi-azole resistant phenotype associated with L98H is dependent on the TR34 in the promoter region.

Link to the Agricultural Use of Azoles

Dominance of a single resistance mechanism as the one observed in the Netherlands is difficult to explain by resistance development in individual azole-treated patients, since a wide array of different resistance mechanisms is normally found in patients with resistance after long term treatment [11•, 26•, 33••]. Apart from being associated with multi-azole resistance in clinical isolates from azole naïve as well as exposed patients [31••, 33••, 34•, 35•], TR34/L98H has also been found in azole resistant, environmental isolates in Denmark and The Netherlands [32, 36•]. This raised the hypothesis that azole resistant patient isolates may not only result from longstanding azole therapy resulting in selection for resistant mutants, but be acquired directly from the environment [23]. Three additional observations support this notion. First, A. fumigatus isolates of identical microsatellite short tandem repeat (STR) genotype with and without azole resistance have been found in individual patients suggesting selection in vivo. But notably, although patients harbouring a TR34/L98H as well as a susceptible isolate have been described, these isolates have never had the same STR genotype, suggesting “double infection” with unrelated A. fumigatus isolates and not in vivo selection of resistance [33••]. Second, five specific azole fungicides (propiconazole, bromuconazole, epoxiconazole, difenoconazole and tebuconazole) show a molecular structure very similar to the medical triazoles, adopt similar poses while docking the target enzyme, have activity against wild type A. fumigatus but not against azole-resistant TR34/L98H-positive isolates and have all been introduced in The Netherlands between 1990 and 1996 directly preceding the isolation of the first TR34/L98H in 1998 [23, 37••]. And third, tebuconazole has been shown to be able to induce tandem repeats in the promoter region of CYP51A under laboratory conditions [37••].

In order to fully understand the nature and development of antifungal resistance, it is necessary to bear the following factors in mind [38]: Antifungal drug resistance (ADR) is not transmitted from person to person, 2) ADR is not conveyed by lateral gene transfer (as seen among bacteria via plasmids), 3) ADR has developed rapidly over the past few years, 4) ADR development under azole pressure (e.g. during therapy) presumably occurs during asexual reproduction, which is much more likely to occur in patients with chronic aspergillosis and aspergillomas than in patients with acute invasive aspergillosis, and 5), insertion of a tandem repeat acting as a transcriptional enhancer under azole pressure might be introduced more often during sexual reproduction requiring the presence of two opposite mating types which occurs mainly in the environment in A. fumigatus (teleomorph: Neosartorya fumigate) [39]. These factors have major implications for the interpretation on reports on cyp51A mutations, and may also to some extent predict the pattern of mutations seen in particular cohorts of patients in particular geographic areas. Hence, the finding of TR34/L98H mutations in azole resistant clinical isolates in patients with invasive aspergillosis in a country where this combination of mutations occurs in the environment may not be surprising. On the other hand, azole resistant isolates from patients with chronic aspergillosis and azole therapy may harbour either different, sporadic cyp51A mutations or the TR34/L98H genotype, depending on exposure and relative fitness of competing mutants. Conspicuously, TR34/L98H was recently found in 55.1 % of culture-negative, PCR positive sputum samples from patients with allergic bronchopulmonary/chronic pulmonary aspergillosis [7••]. If TR34/L98H mutants are less fit than wild type isolates and non-TR34/L98H mutants and thus more difficult to culture, this could explain why they were detectable only by PCR [7••]. This raises concern, since susceptibility testing is dependent on cultured isolates. On the other hand, the TR34/L98H genotype appears to be at least as fit in the environment as suggested by its clonal expansion across the Netherlands [40] and virulence studies in the animal model have failed to detect loss of virulence [41].

Recent reports of additional cyp51A mutants, namely TR46/Y121F/T289A, found in both clinical and environmental isolates in the Netherlands [42••], and a G432S mutant in an azole-naïve patient in France [43••] add support to the hypothesis that azole resistance acquired in the environment is often but not exclusively associated with upregulation of CYP51A induced by the presence of a tandem repeat in the promoter region. As described below this is also the case in fungal plant pathogens. Interestingly, both the Y121 and the G432 codons in these two resistance genotypes are also recognised as hot spot codons involved in azole fungicide resistance in the fungal plant pathogen Mycospherella graminicola (corresponding to the G460 and Y137 codons in this organism) [44].

Azole-Resistance in Other Aspergillus Spp.

All Aspergillus spp. are intrinsically resistant to fluconazole [45]. Whereas wild-type A. fumigatus sensu stricto is susceptible to other azoles, intrinsic multi/pan-azole resistance may operate in some morphologically similar species in the section Fumigati [46]. Thus, azole resistance has been reported in A. lentulus, which is characterised by primary CYP51A dependent resistance, and in A. fumigatiaffinis, Neosartorya pseudofischeri, and A. viridinutans [36•, 4648]. However, recognition of these cryptic species happened only recently, and their respective roles in clinical aspergillosis and potential contribution to resistance problems remain to be further clarified.

Acquired resistance in other Aspergillus spp. has only been sporadically investigated and reported. However, since some of these species (e.g. A. terreus, A. flavus and A. nidulans) exhibit reduced susceptibility to amphotericin B [4956], the importance of azole susceptibility surveillance of such species should not be underestimated. We recently demonstrated elevated azole MICs for two A. terreus isolates [57, 58•], one of which had a cyp51A M217I mutation (equivalent to M220I in A. fumigatus) [58•]. An S240A alteration in CYP51C was recently associated with clinical voriconazole-resistance in A. flavus [59•]. Itraconazole resistance in species belonging to the A. niger complex is not unusual [60, 61] but has not so far been linked to any particular cyp51A gene mutation [61]. The A. ustus complex includes A. calidoustus, which has been detected in transplant patients and appears to be intrinsically pan-azole resistant [49]. Resistance profiles of other Aspergillus species rarely reported as causes of clinical aspergillosis were recently reviewed by van der Linden et al. [62]. Noticeably, resistance conferred by cyp51A mutations coupled to tandem repeats or in azole naïve patients have so far only been demonstrated for A. fumigatus isolates, indicating that this type of resistance acquired in the environment may not yet be a significant issue except in A. fumigatus.

Control Practice in Agriculture

Fungal plant pathogens cause disease in many agricultural and horticultural crops compromising yield and quality [63]. Yield losses in the range of 10 % to 30 % are not uncommon. The approach for infection control varies significantly between countries and over the seasons as many plant pathogens are crop and climate associated, often with the most severe attacks in wet seasons. Effective fungicides have been available for more than 30 years and fungicides are today commonly used for many crops. Depending on the crop and local risks for attack the number of treatments may vary between 5 and 15 times/year in orchard crops and potatoes to 0–4 in cereal crops (Fig. 2). Fungicide use in European cereal crops and in wheat in particular is the largest market for fungicides worldwide [64] (Figs. 2 and 3).

Fig. 2
figure 2

Global market-shares of the various agricultural fungicides in 2010 (a), and porportional use of individual crops (b). Compounds and classes are indicated as relative proportions of the total market value (11,475 mill $) (personal communication Phillips McDougall, 2010)

Fig. 3
figure 3

Countries where A. fumigatus with the TR34/L98H have been reported (dots) and percentage of agricultural fungicide use by continent (3 % used outside the regions shown) [7••, 8, 11•, 22, 32, 33••, 34•, 35•, 64, 97, 98, 99•]

Several classes of fungicides are available for plant protection including triazoles, strobilurins, morpholines, SDHIs and chloronitriles (Figs. 1 and 2). Fungicide resistance has been reported for the majority of fungicides although less commonly for the multisite inhibitors. Concerns related to human health specifically for the azoles have included the risk of endocrine side effects following exposure of farmers and green house workers from preparing spray mixtures or handling azole treated plants. Recently documented selection for azole resistance in human pathogenic fungi adds to this concern.

Triazoles in Agriculture and for Material Preservation

Azole fungicides constitute the most widely used class of antifungal agents for the control of fungal plant diseases (Fig. 2) [64] (personal communication Phillips McDougall, 2010). In agriculture, the first azoles (triadimefon and imazalil) were introduced in 1973 [65], and triazoles have been widely used since the beginning of 1980. In comparison with several other groups of fungicides the field performances of azoles have been relatively stable, suggesting that emergence of acquired resistance in fungal plant pathogens has been limited (www.FRAC.info; [66]). Triazoles are also commonly used for material preservation, but no official statistical information is available to verify to what extent. Examples are tebuconazole and propiconazole which are both used to protect the surface of materials or objects such as paints, plastics, sealants, wall adhesives, binders, papers, art works, wood, and for the preservation of fibrous or polymerised materials, such as leather, rubber, paper and textile products. Additionally tebuconazole is used for preservation and remedial treatment of masonry or other construction materials (EU regulation).

A total of 25 different triazoles/imidazoles have been developed for agricultural crops [65]. The products are applied either as a seed treatment or as foliar applications (sprayed on growing plants), the latter potentially applying a greater selection pressure on other fungal pathogens. In addition to protecting against fungal plant pathogens, triazole compounds may offer plant growth-regulating properties [67] and the ability to protect plants against various environmental stresses [68, 69]. Each new triazole often offered some new advantages in basic activity, spectrum, persistency or mobility in the crop. Initially triadimenol followed by propiconazole and prochloraz were most commonly used, whereas today these have largely been replaced by more potent triazoles including tebuconazole, metconazole, epoxiconazole and prothioconazole. Three of these are among the five azole fungicides which have been associated with a high potential for selecting the TR34/L98H A. fumigatus genotype as described above [37••].

Even though other groups of chemicals e.g. strobilurins and SDHI fungicides (Fig. 1) [70], have been made available, problems related to resistance have been so significant that they are no longer appropriate for control of major diseases in many crops [7173]. Hence, azoles alone or combinations of several agents (typically including at least one azole) are used in order to limit further selection of resistance [71, 74, 75].

Azole Resistance Mechanism in Fungal Plant Pathogens

Triazole resistance has over the years appeared in several plant pathogenic fungi. Field resistance was first reported for the cucumber pathogen Sphaerotheca fuliginea [76], and subsequently in several other pathogens like Penicillium digitatum [77], Blumeria graminis f.sp. hordei [78], Venturia inaequalis [79], Rhynchosporium secalis [80], and Mycosphaerella graminicola [71].

In several European countries a 10–100 times loss of susceptibility in vitro of M. graminicola populations has been reported over the last 20 years [71, 73, 81, 82]. Four azole resistance mechanisms have been found, most of which are identical to those described for A. fumigatus above: 1) point mutations in CYP51, 2) upregulation of target gene production, 3) efflux pumps, and 4) altered sterol biosynthesis; the latter has been found only in laboratory selected mutants and thus will not be dealt with any further here.

Point Mutations in Target Gene CYP51

A variety of different point mutations has been found in the CYP51 gene in plant pathogens but at a relatively low prevalence. Initially, Blumeria graminis f.sp. hordei and Uncinula necator with an Y136F alteration were associated with triazole resistance [83]. Since then, a total of 22 different alterations have been verified in M. graminicola [75]. Amino acid sequence alignment of CYP51 from M. graminicola and Candida albicans showed eight identical alterations in azole resistant isolates of the two organisms, while other eight unique alterations were found specifically for M. graminicola [75]. Moreover, alterations often accumulate in a single isolate leading to a stepwise shift in resistance which is also typical for C. albicans but apparently not for A. fumigatus [71, 84]. For example the alterations Y459D/C, G460D, and Y461H have each been linked to low level resistance, whereas I381V in combination with one of these resulted in a significantly higher level of resistance to all azoles [85]. Sequence alignment for M. graminicola and A. fumigatus identifies three codons found in azole resistant isolates of both species, namely Y137, Y459 and G460 in M. graminicola and Y121, Y431 and G432 in A. fumigatus, respectively [42••, 43••, 44]. Notably, two of these have been involved in azole resistance in azole naïve patients as described above [42••, 43••], whereas the third (Y431) was found in a patient with chronic aspergillosis and bilateral aspergilloma, and was shown to have been selected for in vivo [11•] (Fig. 4). The latter observation suggests that some alterations can be induced by fungicide as well as human azole use.

Fig. 4
figure 4

Alignment of amino acid (AA) sequences of Cyp51 from A. fumigatus (Genbank accession no. AAF32372) and M. graminicola (Genbank accession no. ACI29117) where alterations have been associated with azole resistance. References: Leroux and Walker, Pest Management Sci (2011) and as in Table 2. = Codons associated with azole resistance in azole exposed patients. = Codons associated with azole resistance in azole naïve patients or in M. graminicola; * associated with a tandem repeat in the promoter region of A. fumigatus

The European population of M. graminicola is currently dominated by two molecular types, one being tebuconazole susceptible (V136A) and another being tebuconazole resistant (A379G, I381V and ΔY459/G460) [71, 8587]. This has had major impact on the field performances of tebuconazole but less impact on other azoles such as epoxiconazole and prothioconazole [75].

Over Expression of the CYP51 Gene

Over expression linked to insertions or duplications in the promoter region of CYP51 resulting in elevated intracellular levels of the target enzyme has been detected in different plant pathogens with reduced azole susceptibility [88, 89]. Specifically for M. graminicola an insertion in the promoter of CYP51 was found coupled with a cyp51 alteration I381V in several isolates, but the increase in cyp51 expression remains to be demonstrated experimentally [44].

Role of Efflux Pumps

The simultaneous resistance to a variety of structurally unrelated toxic compounds is most commonly caused by upregulation of efflux pumps and was initially described in plant pathogens by De Ward et al. [90]. This has been studied for azoles in Botrytis cinerea [91], Pyrenophora tritici repentis [92] and M. graminicola [84]. Pump inhibitors like e.g. promazine slightly increase the susceptibility to azoles [87] and the biological potential for efflux resistance exists in the population of M. graminicola, but the genes identified have not so far been identified in field isolates [75].

In France, isolates of M. graminicola cross resistant to azoles, thiolcarbamates and SDHI’s have been reported, suggesting a combination of mutations in CYP51 and overexpression of drug efflux transporters to be involved (Figs. 1 and 2) [44]. Azole resistant A. fumigatus without CYP51A mutations have been found in up to 40 % of the resistant isolates in the clinical setting [26•]. To what extent efflux pumps may operate in these isolates and if so to what extent such may be induced by human azole medication or environmental use remains to be understood.

Future Prospects for the Control of Fungal Plant Pathogens

Although the number of fungicide classes from the perspectives of human medicine appears impressive, many are not true options for use as single agents either due to resistance having already emerged or because the risk when used as single agent is too high. Hence, azoles will remain the most commonly used class in cereal crops for the foreseeable future in agriculture. Various initiatives have been undertaken by European authorities as well as by the industrial community (Fungicide Resistance Action Committee, FRAC) to promote practises that reduce risk of selection of resistance although consensus as to how has not been established [82, 9395]. How this will proceed and to which extent it may influence selection of resistance in human pathogens is yet to be seen.

Conclusions

Substantial data today support that azole resistance in A. fumigatus has been induced during long term azole treatment in individual patients but also occurs in naïve patients due to the selection for resistant mutants in the environment. The first azole resistant environmental strain (the TR34/L98H genotype) has spread throughout The Netherlands since 1998 and now accounts for between 6 % and 12.8 % of clinical A. fumigatus in this country illustrating the fitness and competitiveness of this genotype [96]. Subsequently, TR34/L98H has also been detected in many other West-European countries including clinical isolates from Denmark [33••], Norway [32], the UK [7••, 11•], Belgium [97], France [97, 98, 99•], and Spain [22], and in the Asia-Pacific region including India [35•] and China [34•]. Noticeably, West-Europe and Asia-Pacific represent the regions with the highest and second-highest fungicide use in a global perspective, respectively (Fig. 3) [8, 33••, 35•, 36•]. Additionally, two more resistant genotypes have recently been reported in azole naïve patients in France and The Netherlands, and thus again in West-Europe accounting for 37 % of the global fungicide market [42••, 43••]. Importantly, susceptibility testing is not routinely performed in many centres and the resistance rates reported may therefore very well represent the tip of the iceberg. We have yet to see if resistance emerges in S-America where the fungicide use has increased over the recent years and to which extent the resistance rates increase further in the parts of the world where it is already present. Important players are the fitness of the resistant phenotypes and the pattern of azole fungicide use where not only amounts but also choice of individual compounds plays an important role.

Obviously, these changes have clinical implications. Whereas acquired resistance in clinical practise may be expected after long term treatment, it is important to realise that azole susceptibility is not obligate in the azole naïve patients with aspergillosis. This suggests susceptibility testing should be performed in all patients with Aspergillus infection requiring antifungal therapy and highlights the need for better diagnostics improving the culture positivity rate and establishing alternative options for culture negative cases like direct detection of prevalent environmental mutants by PCR [100••]. Moreover, initial combination therapy may be considered in areas with higher prevalence of environmental azole resistant isolates for patients with severe infection. And finally, surveillance studies in both the clinical setting and the environment should be conducted in order to provide updated local data on susceptibility rates.