Identification of Fungi Using Ribosomal Internal Transcribed Space DNA Sequences

I dentification of fungi, especially filamentous fungi, has been a very difficult task. Because of the amount of experience required to accurately identify filamentous fungi to the species level, it has become acceptable to either identify these organisms to the genus level or, in some cases, simply identify them as "molds." Over time, there have been numerous attempts to automate biochemical tests. However, because eukaryotic organisms exhibit far less metabolic diversity than prokaryotic organisms, these systems tended not to have the resolution required to differentiate fungi at the species level.

During the manufacture of sterile and nonsterile pharmaceuticals, the presence of molds in the manufacturing area is often a cause for concern. A well-designed environmental monitoring (EM) program should detect the presence of molds before they can contaminate the product, but on occasion, fungal isolates will be recovered from sterility or media-fill failures. In these cases, it is extremely important to identify the contaminating organism at the species and, possibly, the strain level to track its origin.

A recent review of the US Food and Drug Administration's product recalls illustrates the inability to identify fungal samples at the species or, in many cases, even to the genus level (1). At the conclusion of FDA's investigations, of the 12 fungal isolates responsible for sterile pharmaceutical product recalls, none was identified at the species level, only four were identified at the genus level, and eight were simply designated as mold or yeast. In nonsterile FDA product recalls, 32 products were contaminated with fungi. Of these, three were identified at the species level, none was identified at the genus level, and 29 were identified as mold or yeast.

Disadvantages of current methodology

Although any type of fungal contamination is cause for concern, a species-level identification is needed to provide a definitive root cause as part of an investigation. Fungal identification, especially mold identification, has not been held to the same standards as microbial identification regarding the investigation of contaminated pharmaceutical products. This identification has been done primarily because it is difficult to identify fungi at the species level, not because the identity is not important. The availability of molecular-based identification for fungi—both yeasts and molds—to all pharmaceutical microbiology laboratories has led to the virtual disappearance of these limitations.

For many decades, ribosomal DNA (rDNA) sequences have been used for organismal taxonomic classification. More recently, ribosomal genes have been used to study the phylogenetic relationships of fungi, with some surprising results. The first came in 1993 when Baldauf and Palmer discovered that fungi are more closely related to animals than they are to plants (2), as had been assumed to be the case based on morphology and other phenotypic characteristics. Another realization was that many species of fungi are synonymous with one another, with many names used to describe the same organism. Candida albicans, for example, has 173 synonyms, many of which have been in common usage during different periods of time (3).

Another reason for taxonomic confusion is the fungal life cycle. Many ascomycetes, basidiomycetes, and zygomycetes go through three stages. In each stage, they may have different microscopic and macroscopic characteristics. In the anamorph (or asexual) stage, the cells are haploid, meaning they contain only one copy of each gene. Organisms in this stage can continue to reproduce through mitosis. A special structure called a conidium or sporangiospore is formed that can split from the cell and form a new organism. During the teleomorph (or sexual) stage, the cells are diploid and contain two copies of each gene. The spores created from this type of sexual reproduction are also diploid, having been formed by meiosis from the nuclei of two different organisms. The last stage (dikaryotic) results when sexual reproduction occurs, but the two parent nuclei do not fuse; rather, they coexist within the cell. It is these differences brought on by the particular sexual state that led taxonomists to consider these organisms different species.

Today, through molecular phylogenetics we understand that although an organism may appear and behave differently during different phases of its life cycle, it is still fundamentally the same species—not unlike how a caterpillar eventually turns into a butterfly. The same genetic information reveals itself in very different phenotypic characteristics.

Although fungal taxonomists have used phylogenetic analysis to characterize, classify, and reclassify fungi for many years, this approach has only recently gained popularity for routine identification of fungi in the pharmaceutical manufacturing environment. Phylogenetic analysis has been the preferred approach because, although the technology existed, applications developed that are compliant with current good manufacturing practice (CGMP) and can be validated have only recently been made available through products and contract-service laboratories. Now that the technology is available, the expectation is that it will become more widely adopted as an accepted approach.

A genetic approach

During the past several years, DNA sequencing-based fungal identification systems have developed rapidly. The introduction of MicroSeq (Applied Biosystems, Foster City, CA) as well as the use of DNA sequencing for fungal identification have assured pharmaceutical microbiologists that the technology has the required level of support to provide high quality, accurate, and compliant answers. DNA sequencing provides data that are much more accurate and reproducible than visual characteristics and phenotypic properties. It is independent of growth stage, cultural medium and age of the sample. As a result, companies can get more accurate and reproducible identification results. In fact, FDA recommended the use of genetic methods in its 2004 update to its Guidance for Industry: Sterile Drug Products Produced by Aseptic Processing—Current Good Manufacturing Practice. In this document, FDA states:

Genotypic methods have been shown to be more accurate and precise than traditional biochemical and phenotypic techniques. These methods are especially valuable for investigations into failures (e.g., sterility test; media fill contamination).... Sterility test isolates should be identified to the species level. Microbiological monitoring data should be reviewed to determine if the organism is also found in the laboratory and production environments, personnel, or product bioburden. Advanced identification methods (e.g., nucleic-acid based) are valuable for investigational purposes. When comparing results from environmental monitoring and sterility positives, both identifications should be performed using the same methodology.

Despite overwhelming acceptance and support from the scientific community, pharmaceutical microbiologists, and regulatory agencies, commercially available identification systems can still be improved. The first limitation is the coverage of the database of known fungal species. This coverage should not be confused with the size of the databases; a database filled with hundreds, or even thousands, of species not encountered in the pharmaceutical manufacturing environment adds no value to the identification system. This point has been addressed in recent literature by Rozynek et al., and Hall et al. (4, 5). In both publications, the technology is well accepted, but the coverage of the database reduces the effectiveness of current technology to provide meaningful identifications.

A second limitation of commercially available systems is the choice of gene target with which to build the DNA sequence library. This is a limitation of the application rather than the technology. A gene target is appropriate for phylogenetic analysis when it undergoes enough genetic mutation for there to be observable differences in the DNA sequences of similar but different species. The rate of accumulated nucleotide differences, however, should not be so great that truly related species appear to be more dissimilar than they actually are. This confusion can pose a great challenge in choosing the appropriate target and, in many instances, becomes a case of trial and error. The D2 expansion segment of the Large Subunit (LSU) of the ribosomal gene, as used in the MicroSeq system, typically does a very good job of placing an unknown fungal isolate into the appropriate taxum. However, because of limitations in the resolution of the D2 segment, closely related organisms may have identical, or very similar, DNA sequences.

The D2 expansion segment of the LSU is effective at linking together higher level taxa (e.g., genus, family, order) and can differentiate many species acceptably, but not in all cases. One common example of the inability of D2 to differentiate close, but distinct, species is Komagataella pastoris, and the related K. phaffi and K. pseudopastrois. K. pastoris, formerly known as Pichia pastoris, is an organism frequently used in biological drug production. Because of the amount of genetic manipulation required to introduce genes of interest into this species—as well as all the eventual monitoring involved in scaling up the fermentation from a starter culture—it is extremely important to ensure the same organism is used in each process step. The easiest way to do this is through identification. Unfortunately, the D2 DNA sequence of K. pastoris is very similar to those of K. phaffii and K. pseudopastoris, as only one nucleotide separates the sequences of K. pastoris from the other two species. This is just one of many examples where the D2 DNA sequence was unable to fully identify an isolate to the species level.

Advantages of ITS sequencing

More recently, ribosomal internal transcribed spacer (ITS) regions have been used for fungal systematics and classification. There are two ITS regions in the DNA genes encoding the fungal ribosomal RNA (rRNA) "operon." The first, ITS1, is found between the 18S and the 5.8S rRNA genes. The second, ITS2, is found between the 5.8S and the 28S rRNA genes. The entire rRNA "operon" is transcribed, but, after transcription, the two ITS sequences are excised and, therefore, not used for any functional purpose.

Because ITS sequences are important enough as spacer regions to be maintained by the cell, but not used for any functional purpose, they are allowed to accumulate mutations at a faster rate than the 5.8S, 18S, and 28S rRNA genes. It is this slightly increased rate of accumulated mutations that allows the ITS sequences to provide an improved level of resolution compared with the D2 sequence.

It is generally accepted to sequence the entire stretch of ITS1-5.8S–ITS2 for use in fungal classification. However, for the purposes of routine identification, our laboratory has found that ITS2 alone is usually sufficient for species-level identification.

As an example of the increased resolution of ITS2 sequences compared to D2 sequences, we will revisit the Komagataella problem. When comparing ITS2 DNA sequences of the three Komagataella species, we see many more nucleotide differences. There are 16 differences between K. pastoris and K. phaffii, and 18 differences between K. pastoris and K. pseudopastoris. This increased resolution offers a much more confident identification and allows for small strain-to-strain variability to ensure correct identification.

Phylogenetic analysis has offered fungal taxonomists the ability to reclassify thousands of species to create a more accurate taxonomic system. A huge advantage of DNA sequencing is its ability to classify an isolate based on DNA sequence alone, regardless of the accepted name of the organism or its official name.

Aspergillus brasiliensis is a newly described species that was, in part, created by the transfer of several existing Aspergillus niger strains to A. brasiliensis (6). Most significant in the creation of this new species to pharmaceutical microbiologists was the inclusion of Aspergillus niger ATCC 16404; this organism is cited in several US Pharmacopeia chapters as a quality control (QC) organism, including USP General Chapter <61> Microbial Limits Test—Enumeration and USP General Chapter <71> Sterility Test. Because of the number of pharmaceutical companies performing these tests, it is very important to correctly identify the QC organism. Unfortunately, this is another example where virtually all phenotypic tests, as well as D2 DNA sequencing, are unable to differentiate the two species. Phenotypically, it has always been difficult to differentiate between A. niger strains because of lack of diversity in morphological features, unstable phenotypic characters, and the significant influence of culture conditions on the phenotype (7). Furthermore, D2 sequences add no additional information because the DNA sequences for all observed A. niger and A. brasiliensis strains are identical.

A recent study by Houseknecht et al., looked at the justification for this recent classification (8). The study concluded that although A. niger and A. brasiliensis are very similar, there are ways to differentiate isolates of these species: through evaluation of conidia morphology under high magnification of an electron microscope (the differences are not observable using a light microscope), or through ITS DNA sequencing. The entire ITS1–5.8S–ITS2 DNA sequence shows five differences between the strains of A. niger and A. brasiliensis. When comparing ITS2 sequence alone, there is only one nucleotide difference between the species, but this difference has been shown to be extremely reproducible and is, therefore, considered a diagnostic indicator of the species.

Conclusion

The ability to accurately and reproducibility identify fungi—both yeasts and molds—has been greatly enhanced through comparative DNA sequencing. Fungal taxonomists have used DNA sequences for many years as a basis for reclassification of all fungal taxa and have more recently moved to ITS sequencing as the "gold standard." With the ability to use these powerful technologies as part of a comprehensive EM program, pharmaceutical microbiologists are given one more tool to ensure product safety. As made clear by FDA as part of its Pharmaceutical cGMPs for the 21st Century and process analytical technology initiatives, the agency encourages industry to use new rapid, and accurate technologies.

Michael Waddington is vice-president of business development at Accugenix, 223 Lake Drive, Newark, DE 19702, tel. 302.292.8888, mikewaddington@accugenix.com

References

1. L. Jimenez, "Microbial Diversity in Pharmaceutical Product Recalls and Environments," PDA J. Pharm. Sci. Tech. , 61 (5), 383–399 (2007).

2. S. L. Baldauf and J. D. Palmer, "Animals and Fungi Are Each Other's Closest Relatives: Congruent Evidence From Multiple Proteins," PNAS , 90 (24) 11558–11562 (1993).

3. S.A. Meyer, R.W. Payne, and D. Yarrow, "Candida" in The Yeasts: A Taxonomic Study, C. P. Kurtzman and J. W. Fell, Eds (Elsevier Science B.V., Amsterdam, 1988) pp. 476–477.

4. P. Rozynek et al., "Quality Test of the MicroSeq D2 LSU Fungal Sequencing Kit for the Identification of Fungi," Int. J. Hyg. Environ. Health , 207 (3), 297–299 (2004).

5. L. Hall, S. Wohlfiel, and G. D. Roberts, "Experience with the MicroSeq D2 Large-Subunit Ribosomal DNA Sequencing Kit for Identification of Filamentous Fungi Encountered in the Clinical Laboratory," J. Clin. Microbiol. , 42 (2), 622–626 (2003).

6. J. Varga et al., "Aspergillus brasiliensis sp. Nov., ABiseriate Black Aspergillus Species with World-Wide Distribution," Int. J. Sys. Evol. Micr. , 57 (8) 1925–1932 (2007).

7. E. Rinyu, J. Varga, and L. Ferenczy, "Phenotypic and Genotypic Analysis of Variability in Aspergillus fumigatus," J. Clin. Microbiol., 33 (10), 2567–2575 (1995).

8. J. Houseknecht et al., "Reclassification of ATCC 16404 and ATCC 9642 as Aspergillus brasiliensis ," Pharm. Microbiol. Forum Newsletter,14 (10), 2–8.