Abstract
The possibility that patients diagnosed with Alzheimer’s disease (AD) have disseminated fungal infection has been recently advanced by the demonstration of fungal proteins and DNA in nervous tissue from AD patients. In the present study, next-generation sequencing (NGS) was used to identify fungal species present in the central nervous system (CNS) of AD patients. Initially, DNA was extracted from frozen tissue from four different CNS regions of one AD patient and the fungi in each region were identified by NGS. Notably, whereas a great variety of species were identified using the Illumina platform, Botrytis cinerea and Cryptococcus curvatus were common to all four CNS regions analyzed. Further analysis of entorhinal/cortex hippocampus samples from an additional eight AD patients revealed a variety of fungal species, although some were more prominent than others. Five genera were common to all nine patients: Alternaria, Botrytis, Candida, Cladosporium, and Malassezia. These observations could be used to guide targeted antifungal therapy for AD patients. Moreover, the differences found between the fungal species in each patient may constitute a basis to understand the evolution and severity of clinical symptoms in AD.
Keywords
INTRODUCTION
Disseminated mycoses are implicated as causative agents or as risk factors for Alzheimer’s disease (AD) [1, 2]. Overwhelming evidence has demonstrated that fungal components can be found in postmortem brain tissue from patients diagnosed with AD, but not from control patients. For example, proteomic analyses revealed several peptides that unequivocally correspond to fungal proteins [1]. Moreover, fungal DNA could be detected by PCR and DNA sequencing, establishing the existence of a variety of fungal species in a single AD patient [1, 2]. Additionally, fungal macromolecules such as polyglucans, proteins, and DNA could also be found in peripheral blood and cerebrospinal fluid (CSF) from AD patients [3, 4]. Strikingly, direct visualization of fungal structures in brain sections from AD patients was possible using specific rabbit polyclonal antibodies raised against a variety of fungi [2, 5]. These fungal structures correspond to yeast-like cells and hyphae and were found in neural cells both intra- and extracellularly. All these observations are consistent with the hypothesis that disseminated mycoses exist in AD patients. Whether these mycoses represent a risk factor or are the etiological agents of AD remains unknown. Nevertheless, further support to the concept that mycoses are present, even before the appearance of the disease, comes from the findings of enhanced chitinase detected in blood and CSF samples from AD patients [6–9]. In addition, the finding that amyloid-β (Aβ) peptide possesses potent antifungal activity and is involved in the innate immune response against microbial infections [10, 11] supports the notion that amyloid plaques represent an immune response to putative infections in AD patients.
The human microbiome consists of a variety of microbiota, including a great diversity of bacteria and fungi [12]. It has been estimated, however, that only 0.1–1.0% of the microbiome correspond to fungi [13, 14]. The human fungal microbiome, also known as the mycobiome, includes all fungal species detected in humans, of which over 390 species have been identified in different anatomical locations including skin and mucoses, the respiratory tract, the oral cavity, and the digestive tract [15, 16]. Clearly, different parts of the body have different mycobiomes, which might reflect the chemical environment or differences in pH tolerance. Importantly, variation in the composition of the mycobiome occurs between individuals. Moreover, the mycobiome can change throughout the lifetime of an individual, is strongly affected by diet, and can be sculpted by disease. Identification of the commensal fungal communities in the different anatomical locations is important to understand the cross-talk between these species and the immune system [17, 18]. The mycobiome of the gastrointestinal tract has been analyzed in detail [16]. In total, 335 species belonging to 158 genera have been found in the human digestive tract and the oral cavity; of these, 221 species are exclusive to the intestine, 88 are found only in the oral cavity, and 26 species are common to both. Notably, most of these species cannot be cultured and must be identified by molecular techniques. Thus, from the 247 species found in the digestive tract, only 59 could be grown in culture and the remaining species were identified by molecular analysis [16]. Studies of the oral mycobiome in healthy subjects identified fifteen fungal genera, with Candida and Cladosporium being the most common [19]. However, Malassezia and Epicoccum were identified as the most abundant genera of the oral mycobiome in another study [20]. Malassezia is also the most abundant genus in human skin, followed by Penicillium and Aspergillus [21]. The varied mycobiota detected in human skin and mucoses do not infect internal tissues, blood, or CSF in healthy individuals because of the action of the immune system [17, 18]. An important concept is that the human mycobiome shapes the immune system to control and prevent potential fungal infections. A large variety of fungi have been isolated from human internal tissues under pathological conditions [22]. The aim of the present study was to identify the different fungi that exist in brain tissue from AD patients and to evaluate whether some of these species are common to all patients.
MATERIALS AND METHODS
Description of AD patients
Samples were evaluated from patients diagnosed with AD. The age and gender of the subjects are listed in Supplementary Table 1. All samples were supplied by a brain bank (Banco de Tejidos CIEN, Madrid) and were analyzed anonymously. The transfer of samples was carried out according to national regulations concerning research on human biological samples. The Ethics Committee of the Universidad Autónoma de Madrid approved the study. In all cases, written informed consent was obtained.
All brain samples were provided by the same laboratory. All of them were processed according to a common postmortem protocol followed by Banco de Tejidos CIEN. Briefly, rapid neuropathological autopsy was performed upon call by the donor’s proxies (mean postmortem interval, 4.5 h). Immediately after extraction, the right half of the brain was sliced and frozen at –80°C. The left half was fixed by immersion in phosphate-buffered 4% formaldehyde for at least 3 weeks. A full neuropathological study was performed in the left half brain after fixation. Neuropathological diagnosis and staging of all disease entities was performed according to consensus criteria. Various neuropathological variables related to AD, vascular, Lewy and TDP (TAR DNA-binding protein) pathologies in addition to the presence of hippocampal sclerosis were recorded for full classification of cases. Samples from the frozen tissue were obtained with sterile instruments taking all measures to avoid contamination in a laminar flow hood.
DNA extraction from frozen CNS tissue
DNA was extracted from frozen samples of different CNS regions using the QIAmp Genomic DNA Isolation Kit (Qiagen) as follows: 20 μl proteinase K (>600 mAU/ml) and 180 μl of buffer ATL were added to 25 mg of brain tissue, followed by pulse-vortexing for 15 s. Digestion was carried out at 56°C for 1–3 h with agitation. Subsequently, 200 μl of buffer AL was added to each sample followed by vortexing for 15 s and incubation at 70°C for 10 min. A 200-μl volume of ethanol was then added to each sample followed by vortexing for 15 s. The mixture was applied to the QIAamp Mini spin column and centrifuged at 8,000 rpm for 1 min. Then, 500 μl of buffer AW was applied to the column followed by centrifugation at 8,000 rpm for 1 min. After a final wash step with 500 μl of buffer AW2 (14,000 rpm for 3 min), samples were collected in 40 μl distilled water and DNA was quantified in a NanoDrop® ND-1000 UV-Vis spectrophotometer. Negative controls included three samples of tri-distilled filtered water.
Nested PCR
A number of measures were used to avoid PCR contamination including the use of separate rooms and glassware supplies for PCR set-up and products, aliquoted reagents, positive-displacement pipettes, aerosol-resistant tips, and multiple negative controls. DNA samples obtained from frozen CNS tissue were analyzed by nested PCR using several primer pairs. Primer design for amplication of the internal transcribed spacer (ITS) regions of ribosomal DNA has been described in detail [5]. The first PCR was carried out with 4 μl of DNA incubated at 95°C for 10 min followed by 30 cycles of 45 s at 94°C, 1 min at 57.3°C, and 45 s at 72°C. Oligonucleotides used in the first PCR were forward ITS1 (external 1 5′): 14485′GTTCTGGGCCGCACGGG 3′1465 and reverse ITS1 (external 1 3′): 106R5′GGCAAAGATTCGATGATT3′88R. The second PCR was performed using 0.5 μl of the product obtained in the first PCR and ITS1 (internal 1) primers for 30 cycles of 45 s at 94°C, 1 min at 55°C, and 45 s at 72°C. The oligonucleotides used were forward ITS1 (internal 1 5): 17715′TCCGTAGGTGAACCTGCGG3′1790 and reverse ITS1 (internal 1 3′): 50R5′GCTGCGTTCTTCATCGATGC3′30R. A separate PCR assay was designed to amplify the ITS-2 region. The first PCR assay was carried out with 4 μl of DNA incubated at 95°C for 10 min followed by 20 cycles of 45 s at 94°C, 1 min at 52°C, and 45 s at 72°C. Oligonucleotides used in the first PCR were forward ITS-2 (external 5′): 1525′TTTCAACAACGGATCTC3′169 and reverse ITS-2 (external 3′): 8585′AGTACGGGATTCTCACCCTC3′838. The second PCR was carried out with 0.5 μl of the product obtained in the first PCR and ITS-2 (internal) primers for 25 cycles of 45 s at 94°C, 1 min at 55°C, and 45 s at 72°C. Oligonucleotides used were forward ITS-2 (internal –5′): 2745′GCATCGATGAAGAACGCAGC3′295 and reverse ITS-2 (internal –3′): 572R5′TCCTCCGCTTATTGATATGC3′552R. Amplified DNA products were analyzed by agarose gel electrophoresis and stained with ethidium bromide. PCR products were sequenced by Macrogen.
Next-generation sequencing
The yeast ITS1 region is highly variable both in length and in nucleotide sequence, and for this reason it is used in metagenomic next-generation sequencing (NGS) studies. In this present study, the region between the internal 1 primers was amplified with specific primers joined to linker sequences in a first round of PCR (specific product of ∼300 nt). A second PCR was performed on this product using fusion primers containing Illumina and linker sequences. These PCR products were sequenced on a MiSeq sequencing platform (Illumina). PCR and sequencing were performed by the Genomics Unit at the Scientific Park of Madrid. Between 677,391 and 989,072 paired end reads were obtained with length range 35–301 bases with a mean of 145–203 bases (see Tables 1 and 2). Quality analyses were performed over reads using FastQC software (http://www.bioinformatics.babraham.ac.uk/projects/fastqc/). All sequences have been submitted to European Genome-phenome Archive with the following access number EGAS00001002228.
Representation of species over 1% from different brain regions from one AD patient
Representation of species over 1% from different brain regions from one AD patient
LFC, external frontal cortex; CEC, cerebellar cortex; ERH, entorhinal cortex/hippocampus; CP, choroid plexus. (B) species obtained by blast analysis. (Q) species obtained by qiime analysis.
Representation of species over 1% from eight AD patients
(B) Species obtained by blast analysis. (Q) Species obtained by qiime analysis.
Computational analysis
As a first step, paired reads were joined into one single sequence using a Qiime script (join_paired_ends.py), which discards those pairs that do not align between them or that have low quality. Next, the adapters from the sequences were deleted using Cutadapt and all sequences with a length shorter than 35 bp were discarded. On average, 68.34% of joined reads were recovered in every sample. Once sequence set-up was ready, we performed a metagenomic-type analysis that consisted of several steps (http://nbviewer.ipython.org/github/biocore/qiime/blob/1.9.1/examples/ipynb/Fungal-ITS-analysis.ipynb). As a reference, we used the most recent version of the Qiime Fungal ITS data-base (ftp://ftp.microbio.me/qiime/tutorial_files/its_12_11_otus.tgz).
Sequences clustering
The sequences of all samples were grouped to define the Operational Taxonomic Units (OTUs) using the pick_open_reference_otus.py workflow (http://qiime.org/) with a percentage identity of 97%. We obtained a total number of 1.968 OTUs of the total analysis of the 10 samples.
Principal component analysis
Bray-Curtis distance matrix and the weight score of each principal component were calculated using the QIIME script core_diversity_analyses.py. The 3D plot model of the principal component analysis was done with the package scatterplot3d in R.
Identification of no blast hits OTUs
According to the taxonomical classification, we found that on average 39% of the matches corresponded to “No Blast” hit. For this reason, an additional standard Blast search analysis was performed.
RESULTS
Identification of fungal species in different CNS regions of one AD patient by NGS
We have recently investigated the range of fungal species that can be detected in the CNS of one AD patient using a classical PCR/sequencing approach [2]. The results obtained revealed that several fungi could be identified in the same patient and could vary depending on the CNS region examined. Our goal was to analyze the fungal DNA sequences in these different regions of the CNS using NGS. Compared with classical PCR and sequencing, NGS represents a more powerful technique, which additionally provides a semi-quantitative analysis of the different species of fungi identified. As a first approach, four CNS regions of a single patient were examined: lateral frontal cortex (LFC), cerebellar cortex (CEC), entorhinal cortex/hippocampus (ERH), and choroidal plexus (CP). These regions were selected because the evolution of the neuropathology of AD indicates that it starts by the ERH and is then spread to other areas of the CNS. DNA was extracted from frozen brain tissue and PCR of the ITS1 regions was accomplished using universal primers as described in Materials and Methods. The variability of the yeast ITS1 region makes it suitable for metagenomic NGS studies. The amplicons obtained were sequenced using the Illumina platform. Computational analyses of the sequences using Qiime and Blast classified the data of the four CNS samples into phylum, class, order, family, and genus (Fig. 1A, B). The fungal species identified by Qiime and Blast representing >1% of the total sequences are shown in Table 1. Interestingly, many species were identified using NGS and the percentage of the amplicons corresponding to these species is also shown in Table 1. Several species were common to the four CNS regions analyzed, such as Cryptococcus curvatus (ranging from 8.6% to 55.2% depending on the region) and Botrytis cinerea (ranging from 1% to 3.2%). In addition, several species could be detected in three regions, including Alternaria alternata (1.1% to 2%) and Davidiella tassiana (3.4% to 29.4%). The total number of species that could be detected in a single CNS region was remarkable. Additionally, a large number of species could be detected that represented between 0.1% –1% of the total species in all regions (Supplementary Table 2). The variety of fungi found in this one patient and in the other patients tested (see below) argues against possible contamination in the experimental workflow.
Fungal distribution in different brain regions from AD patient AD1. Computational analyses of the sequences obtained on the Illumina platform using Qiime analysis classified the data into phylum, class, order, family and genus. The uncultured fungus species obtained from Qiime were analyzed by Blast. Panel A shows the results obtained by Qiime analysis. Panel B shows the results obtained by Blast of uncultured fungus. (A) Ascomycota; (B) Basidiomycota.
Search for fungal DNA in control samples
Recently, we reported the presence of fungal enolase, tubulin, and chitin in AD brains and also in control subjects, albeit to a lower extent [23]. Moreover, fungal antigens are also detected in Corpora amylacea from AD patients and to a lower extent from control samples [24]. These findings suggested a lower burden of fungal infection in control subjects. This possibility was not totally unexpected since bacterial DNA has also been found in brain tissue from control subjects [25]. Although fungal DNA is not usually detected by classical PCR in control brain tissue, it was of interest to analyze fungal DNA using NGS, and much more powerful technique. To this end, DNA was extracted from LFC and ERH samples from one control. NGS was accomplished as indicated above. Interestingly, a great variety of fungal species were also apparent in these two samples, as listed in Supplementary Table 3. The phylum, class, order, family, and genus of these species are shown in Supplementary Figure 1. Notably, some differences in the species that appear in the AD patient and in the control subject are apparent. However, additional studies should be required to identify the mycobiome that can be found in individuals that did not suffered a neurodegenerative disease. The NGS technique does not allow comparison of the burden of the fungal infection in AD patients and in the control samples. This comparison is achieved much better by immunohistochemistry, as we have already reported [23, 24]. These studies indicated that the level of fungal antigens were certainly lower in control subjects as compared to AD patients. Nevertheless, it seems possible that both bacterial and fungal DNA can also be detected in brain tissue from controls using this powerful technique. Most probably, the progression of the disease may be related to increased infection along the time.
Amplification of fungal DNA by nested PCR from different AD patients
Our next goal was to analyze the mycobiome present in brain tissue from several AD patients. To this end, DNA was extracted from frozen tissue corresponding to the ERH of eight AD patients (AD2–AD9), as listed in Supplementary Table 1. Initially, we performed nested PCR directed to amplify ITS1 and ITS2 regions, as previously described [1, 3]. We were able to identify different amplified products, which were determined to correspond to different fungal species by Sanger sequencing (Fig. 2A, B). Notably, species of the genera Cladosporium, Cryptococcus, Malassezia, and Penicillium were identified using this method. However, the majority of these species were detected in only one patient, again arguing against the possibility that they represent a contamination common to all samples. Indeed, as shown in Fig. 2, the controls for nested PCR and DNA extraction were negative and no products were amplified. This indicates that no contamination of the PCR assay or the DNA extraction method occurred.
PCR analysis of ERH brain region from eight AD patients. PCR was carried out as described. A) schematic representation of fungal rRNA genes (18S, 5.8S, and 28S rRNA) and the ITS1 and ITS2 sequences. Location of the primers employed for the different nested PCRs: external 1 and external ITS2 employed in the first PCR; internal 1, internal ITS2 employed in the different second PCR. B) PCR analysis of ERH regions using primers internal 1 to amplify ITS1. C) PCR analysis of ERH regions using primers internal ITS2. The species identified by Macrogen are listed below. Control PCR, PCR without DNA; CE, control of DNA extraction without DNA.
Fungal species found in brain tissue from different AD patients revealed by NGS
To gain further insight on the variety of fungi that can be identified in different AD patients, NGS was performed using DNA extracted from frozen tissue of patients AD2–AD9. Identified fungal DNA sequences that represented >1% of the total are listed in Table 2. Notably, while the majority of these species differed between patients, some were common to all eight patient samples tested, as illustrated for Botrytis cinerea (1.9% to 3.5%), uncultured Candida (6.9% to 10.2%), and uncultured Malassezia (1.1% to 2.1 %). In addition, a large number of species were detected representing 0.1% to 1% of the total in all patients (Supplementary Table 4). The fungal species representing >0.1% of total were classified into phylum, class, order, family, and genus, as depicted in Fig. 3A and B. The genera common to all eight patients were Alternaria, Botrytis, Candida, Cladosporium, Fusarium, Malassezia, Penicillium, and Sporidiobolus.
Fungal distribution in ERH brain samples from eight AD patients. Computational analyses of the sequences obtained on the Illumina platform using Qiime analysis classified the data into phylum, class, order, family and genus. The uncultured fungus species obtained from Qiime were analyzed by Blast. Panel A shows the results obtained by Qiime analysis. Panel B shows the results obtained by Blast of uncultured fungus. (A) Ascomycota; (B) Basidiomycota; (C) Chytridiomycota.
By including the ERH region of patient AD1 with the other eight AD patients (AD2–AD9), the genera common to all were Alternaria, Botrytis, Candida, Cladosporium, and Malassezia (Fig. 4). Interestingly, these genera form part of the human mycobiome. By considering the percentage of each genus found in the different samples the most prominent ones were found to be Cryptococcus (11.6%) and Candida (8.6%).
Percentage of the most representative genera in ERH samples of AD1 and ERH regions from AD1 and eight AD patients (AD2–AD9). The average percentage of genera common to all samples considering the percentage above 0.1% in the different patients of this study.
Finally, comparison of the data obtained from AD and control samples were carried out by principal component analysis. Figure 5A shows this analysis of the four samples of AD1 (LFC, CEC, ERH, and CP) versus the two control samples (LFC and ERH). The principal components of LFC, ERH, and CP from AD1 appeared clustered, while the CEC was apart, as occurred with the two control samples. Most notably, this analysis applied to the ERH samples of the eight AD patients and the control provided evidence that a relationship existed between the AD samples, whereas the ERH control appeared far apart (Fig. 5B).
Principal component analysis of AD and control samples. 3D principal component analysis scatter plot of AD patients and control subjects. A) Analysis for the four AD1 samples collected from different brain regions including LFC, CEC, ERH, and CP, and one control including LFC and ERH regions. B) Analysis for the AD2-AD9 samples isolated from the ERH area and control ERH.
DISCUSSION
The main concept arising from the present study is the diversity of the mycobiome found in brain samples from AD patients. Previous studies have shown that the gastrointestinal mycobiome of individuals is diverse and can evolve to respond to changing environmental factors such as age, diet, and lifestyle, among others [13, 16]. The fungal diversity we found by comparing the brain mycobiome of different AD patients may be important to understand the pathology of the disease. It is well known that both the evolution of AD and the severity of clinical symptoms strongly vary between patients. If AD is provoked by fungal infection, then it is quite possible that neurological symptoms and the pathology of the disease might depend on the mycobiome present in the brain, and variation in the clinical symptoms and the evolution of the disease in individual patients can be a result of specific fungi infecting the CNS. The finding that two brain samples from one control subject also contain fungal DNA suggests that the level of infection is also very important in the disease. It has been observed that brain tissue from control subjects also contains bacterial DNA [25]. Therefore, it is possible that fungal infection increases in some individuals due to a depletion in the immune system or to other susceptibility traits.
The existence of mixed fungal infections in all AD patients analyzed here is also worthy of mention. The possibility that one initial infection at the portal of entry of one fungus expedites further colonization by other fungi should be considered. Indeed, some fungal species can evade the immune response and facilitate the entry of additional fungi [26–28]. For instance, fungal polysaccharides have immunomodulatory properties and can strongly influence the local immune response, whereas these macromolecules can be beneficial for fungal colonization [29–31].
In light of the above, the determination of the portal of entry of the mixed fungal infection would be an important next step to mechanistically understand the colonization process. Clearly, the best candidate would be the olfactory nerve, since it is well established that AD neuropathology begins in the entorhinal cortex and slowly disseminates to other areas of the CNS [32]. If this is indeed the case, the mycobiome of the nasal cavity might presumably constitute the infectious reservoir that reaches the CNS through the terminal of the olfactory nerve. An alternative source of CNS infection could be the mycobiota of the oral cavity, with fungi perhaps passing into the bloodstream through small wounds in the gums. Yet another possibility is that fungi pass through wounds in the skin or traverse the intestinal mucosae. Closer examination of the fungal species found in the CNS from AD patients and their comparison with the human mycobiome in different body regions may shed some light on the portal of entry of this infection. In this regard, Candida, Cladosporium, and Cryptococcus are well known human pathogens able to infect the CNS [33–37]. Malassezia is a prominent commensal of the oral cavity [20] and is also found in the skin, and can be pathogenic for humans [15, 38]. Botrytis cinerea is a plant pathogen which can infect a variety of fluids and vegetables [39], and in this manner could easily reach the saliva and the human digestive tract [15]. Furthermore, species of the genus Alternaria are also plant pathogens that produce mycotoxins, such as alternariol [40]. Alternaria can be found in the nasal and oral cavities as well as in skin, vagina, and conjunctiva [15]. To fully understand the precise neuropathology of mixed fungal infections, it is also important to consider that a number of species, mainly belonging to the genera Aspergillus, Penicillium, Fusarium, and Alternaria, can produce a variety of mycotoxins.
Our findings may have a bearing on the use of antifungal therapy. The arsenal of clinically active antifungal compounds that reach the CNS is limited [41]. While the triazole fluconazole has little toxicity and is well distributed throughout the CNS [42, 43], it has a narrow spectrum of action and resistance can arise when it is chronically administered. The best current choice is represented by the echinocandins, for example caspofungin, in combination with voriconazole [44–47], and successful treatment of CNS fungal infections has been achieved using these compounds [48]. Therefore, if AD patients are infected with a variety of fungal species, the best way to eradicate them should be combined therapy that after several months could be changed to alternate therapy. This would serve to reduce the likelihood of resistance and the spectrum of action will be widened. The implementation of antifungal therapy to AD patients will serve to answer the most important question: is disseminated fungal infection the cause of AD, a risk factor or just a mere consequence of neurodegeneration?
Footnotes
ACKNOWLEDGMENTS
The financial support of Pharma Mar, S.A. is acknowledged. We also acknowledge an institutional grant to Centro de Biología Molecular “Severo Ochoa” from the Fundación Ramón Areces. We thank the brain bank (Banco de Tejidos CIEN, Madrid) for providing the frozen brain samples. We are especially indebted to the Genomics and Next-Generation Sequencing service of the Centro de Biología Molecular Severo Ochoa, and in particular to the Computational Analysis Team.
