The term “microsporidia” is a nontaxonomic designation commonly used for organisms belonging to the phylum Microspora. This phylum consists of over 100 genera with almost 1,000 species. So far only six genera (Enterocytozoon, Encephalitozoon [including Septata], Pleistophora, Trachipleistophora, Vittaforma, and Nosema) with at least 12 different species belonging to these six genera as well as unclassified microsporidia have been described as pathogens in humans.

For most patients with infectious diseases, microbiological isolation and identification techniques offer the most rapid and specific determination of the etiologic agent (378). This is not a suitable procedure for microsporidia, which are obligate intracellular parasites requiring cell culture systems for growth. Visualization of organisms in cytologic smears, tissue sections, or both is commonly used for diagnosis of infections with microsporidia. However, ultrastructural analysis by transmission electron microscopy is usually necessary for exact species differentiation. This technique may lack sensitivity, and species differentiation can be missed.

During the last 10 years, the detection of infectious disease agents has begun to use nucleic acid-based technologies. Diagnosis of infection caused by parasitic organisms is the last field of clinical microbiology to incorporate these techniques (378). In this paper, we review human microsporidial infections and newly developed molecular techniques for detection, species differentiation, and phylogenetic analysis of microsporidia with special emphasis on species that infect humans.

All microsporidia are obligate intracellular parasites and have no active stages outside their host cells. They are considered to be ancient organisms, evolutionarily placed as an early branch leading from prokaryotes to eukaryotes (64, 65, 285, 362, 363). Microsporidia lack some typical eukaryotic characteristics. The ribosomes (70S), ribosomal subunits (30S and 50S), and rRNAs (16S and 23S) are of prokaryotic size, and the rRNA has no separate 5.8S rRNA (362). Although mitochondria, peroxisomes, and a classical stacked Golgi apparatus are missing, they are true eukaryotes with a nucleus, an intracytoplasmatic membrane system, and chromosome separation by mitotic spindles (362, 363); polyadenylation occurs on mRNA in microsporidia as in every other eukaryotic organism studied to date (381a).

Spore. Microsporidian spores are between 1 and 20 μm long. Species that infect mammals are usually small, with diameters of 1 to 3 μm. The spores have a thick wall, composed of three layers: (i) an electron-dense outer layer called the exospore, which is proteinaceous; (ii) an electron-lucent inner layer called the endospore, which is chitinous; and (iii) a plasma membrane enclosing the cytoplasm, the nucleus (sometimes two nuclei), a posterior vacuole, the polaroplast membranes, and the unique extrusion apparatus. The extrusion apparatus consists of a coiled polar filament and its anchoring disc, which is characteristic of all microsporidia (Fig. 1). The number and arrangement of coils of the polar filament vary among genera and species. Under appropriate conditions inside a suitable host, the polar filament is discharged through the thin anterior end of the spore, thereby penetrating a new host cell and inoculating the infective sporoplasm into the host cell (Fig. 1 and 2) (41, 43, 50, 58).

FIG. 1 Diagram of a microsporidian spore and representative life cycle (merogonic and sporogonic stages vary among different genera).

FIG. 2 Giemsa stain of Nosema algerae spore with an extruded polar tube and with the sporoplasm at the end of the tube. Giemsa stain. Magnification, ×640.

Merogony. In suitable host cells, the sporoplasms that are released from the spores become meronts. Meronts are rounded, irregular, or elongated simple cells with little differentiated cytoplasm, enclosed by a plasma membrane. Meronts may have isolated or diplokaryon nuclei. Inside the host cell, there is a phase of repeated divisions by binary or multiple fissions called merogony. Nuclear division may occur without cell division, resulting in multinucleated plasmodial forms (41, 43, 50, 58).

Sporogony. Meronts develop into sporonts, which are characterized by a dense surface coat. This surface coat later develops into the exospore layer of the spore wall. Sporonts multiply by binary or multiple fission and divide into sporoblasts that will finally develop into mature spores. Sporonts may have isolated or diplokaryon nuclei. Some sporonts divide directly into sporoblasts by binary fission, whereas others become multinucleated plasmodial stages. Sporoblasts are ovoid bodies that will mature to spores by synthesis of spore organelles (Fig. 1) (41, 43, 50, 58).

The term “microsporidia” is a nontaxonomic designation commonly used for organisms belonging to the phylum Microspora, which is contained within the subkingdom Protozoa (50, 337). In 1882 Balbiani classified these parasites as a separate group, “Microsporidies” (12). Before the middle of this century, since knowledge of this group of organisms was fragmentary, classifications of microsporidia were necessarily simple and artificial. Subsequently, the taxonomy of microsporidia has been subjected to several modifications. Major published microsporidian classifications differ considerably in the characteristics used to produce the major divisions within the microsporidia (337). Larsson (214) considered that many features traditionally used for taxonomic systems (for example, the diplokaryon, sporophorous vesicle, meiosis) have evolved independently in several lineages, and therefore seemed not to be useful for phylogenetic analysis. In his classification system, based on differences in ultrastructural morphology, several characters were subdivided into well-defined categories, thereby creating a tree representing the phylogeny of the microsporidia (214). Weiser (376) based his classification only on the nuclear condition of the spores (one nucleus in Pleistophoridida or two nuclei in Nosematidida), whereas Issi (183) used the spore morphology and developmental stages. Until recently, the classification system of Sprague, proposed in 1977 and updated in 1982, was the most widely used (335, 336). In this scheme the microsporidia were divided into two groups, based on the presence or absence of a membrane surrounding the parasites: the Pansporoblastina (membrane present) and the Apansporoblastina (membrane absent). In systems developed during the last decade, it seems that differences in chromosome cycles constitute the most fundamental basis for distinguishing taxa at the highest level (52, 337). Based on this concept, Sprague et al. (337) proposed a comprehensive revision of the classification system in which differences in the nuclear state and their implications for the chromosome cycle were treated as the most fundamental taxonomic characters (Fig. 3). The microsporidia were separated into the Dihaplophasea, which have a diplokaryon in some phase of their life cycle, and the Haplophasea, which have unpaired nuclei in all stages of their life cycle. Phylogenetic trees constructed on the basis of DNA sequence data now show clearly that this and other classification systems do not reflect the true relationships among microsporidia and that the classification of microsporidia should be completely revised.

FIG. 3 Taxonomy of microsporidia infecting humans, using the revised taxonomy of Sprague et al. from 1992 (337), modified in light of the new taxonomic classification of Vittaformae corneae (324), Encephalitozoon intestinalis (10, 166), Trachipleistophora hominis (more ...)

Nucleotide sequence data of the small-subunit (SSU) rRNA of a microsporidian from insects, Vairimorpha necatrix, suggested that microsporidia are very ancient organisms and that the evolutionary developments leading to microsporidia branched very early from those leading to eukaryotes (363). DNA data for protein-encoding genes supported this thesis (35, 64, 65, 186, 187, 387), but phylogenetic trees constructed on the basis of α- and β-tubulin sequences suggested that the microsporidia are close relatives of fungi, which may be evolved degeneratively from higher forms (125, 126, 221). Recently, genes encoding Hsp70 (a heat shock protein or chaperonin) have been identified in the microsporidia Nosema locustae, V. necatrix, Encephalitozoon hellem, and Encephalitozoon cuniculi, and phylogenetic analyses have shown unequivocally that these genes are most closely related to those encoding Hsp70 proteins from the mitochondria of other eukaryotes, suggesting that microsporidia may be evolved degeneratively from higher forms (159, 173, 193, 251a, 275a). This possible degenerative evolution is discussed in more detail below.

SSU and large-subunit (LSU) rRNA and protein-encoding DNA sequences are now available for several microsporidian species, including six species infecting humans. The taxonomy of microsporidia will be amended significantly in the near future when these and newly generated nucleotide sequence data are considered for future classification systems. For example, molecular analyses have led to the reclassification of Septata intestinalis into Encephalitozoon intestinalis (10, 166). However, this reclassification is still controversial on the basis of ultrastructural data and rules of taxonomy (49). Several phylogenetic trees based on DNA sequence data have been suggested recently (9–11, 125, 233, 280, 364, 365, 381, 396) and are discussed below.

FIG. 4 Transmission electron micrograph of duodenal epithelium from an HIV-infected patient heavily parasitized with Enterocytozoon bieneusi. Several cells are infected with merogonic and sporogonic stages, and one cell is infected with darkly staining spores. (more ...)

E. bieneusi was first detected by Modigliani et al. (245) and described in detail by Desportes et al. (99) in 1985 following examination of a 29-year-old Haitian AIDS patient with chronic diarrhea who lived in France. A similar case was described in the United States in the same year (119). Since then, the number of reported cases has steadily increased in Europe, North and South America, Africa, and Australia (7, 36, 53, 55, 56, 60, 72, 83, 87, 96, 103, 124, 127, 130, 131, 136, 142, 244, 246, 263, 325, 389). The parasite usually infects intestinal enterocytes of HIV-infected patients but has been also detected in lamina propria cells of small-bowel biopsy specimens, biliary tree, gallbladder, liver cells, pancreatic duct, and tracheal, bronchial, and nasal epithelia (93, 129, 165, 250, 278, 279, 310, 367, 368).

The second and last member of the family Enterocytozooidae is Nucleospora salmonis. This microsporidium was originally described by Hedrick et al. in 1991 (169), but shortly thereafter Chilmonczyk et al. (67) described it as Enterocytozoon salmonis. Ultrastructural examinations showed morphological similarities between E. bieneusi and N. salmonis, with Nucleospora exhibiting most of the distinguishing morphological characteristics of the family Enterocytozoonidae. However, in contrast to E. bieneusi, N. salmonis grows in the nucleus rather than in the cytoplasm of cells and parasitizes fish rather than humans (120, 197, 386). Desportes-Livage et al. (101) further described several ultrastructural differences in the development of these two genera. Based on rRNA sequence data first generated by Barlough et al. (13), rules of taxonomy, and the morphology and intranuclear location of the organism, it has been suggested that in the absence of significant reasons for the suppression of the generic name Nucleospora, the original name N. salmonis rather than E. salmonis is valid (120).

FIG. 5 Transmission electron micrograph of duodenal epithelium from an HIV-infected patient infected with Encephalitozoon cuniculi. Two meronts with single nuclei, four sporonts with a thickened plasma membrane and highly developed endoplasmatic reticulum, and (more ...)

FIG. 6 Transmission electron micrograph of an Encephalitozoon cuniculi spore in nasal discharge from a patient with AIDS and chronic rhinosinusitis. The spore contains a polar tubule with six coils lying in a single row and is coated with an electron-dense exospore. (more ...)

E. cuniculi was the first microsporidium to be recognized as a parasite of mammals. First found in rabbits in 1922 (388), this microsporidium was named by Levaditi et al. in 1923 (220). Subsequently, it has been detected in many mammalian hosts, including humans (50, 51). To date, E. cuniculi is the best studied of the microsporidian species, and much of what is known about the pathogenesis of microsporidial disease has been derived from studies of this organism.

Two pathogenic species of Encephalitozoon which infect humans, E. cuniculi and E. hellem, are morphologically similar by light and electron microscopy and can be distinguished only by antigenic, biochemical, or nucleic acid analysis (104, 112). Several cases of Encephalitozoon infection were reported to occur in patients with and without AIDS prior to 1991. Light and/or electron microscopic analysis indicated that these infections appeared to be due to E. cuniculi. However, in 1991 Didier et al. (104) used biochemical and antigenic methods to describe a new species of Encephalitozoon, E. hellem, which had been found in three patients with AIDS. Since all subsequently published cases of Encephalitozoon infections in humans appeared to be caused by E. hellem (113, 162, 178, 211, 304–306, 369), there was some doubt whether E. cuniculi did in fact cause human infection (293). However, in 1995 De Groote et al. (95) and Franzen et al. (144) described two homosexual men with AIDS and disseminated E. cuniculi infection; identification was confirmed by an immunofluorescence assay and by DNA identification. Recently, E. cuniculi has been detected in several HIV-infected patients (97, 177, 179, 242, 271, 374).

A third Encephalitozoon species, E. intestinalis, infecting HIV-infected patients, was first described in 1992 by Orenstein et al. (266, 268) as a microsporidium with ultrastructural similarities with the genus Encephalitozoon. It was later classified as a new genus and species, Septata intestinalis by Cali et al. (47) on the basis of ultrastructural differences. Based on rRNA sequence data, it has been suggested that this organism be placed in the genus Encephalitozoon and renamed Encephalitozoon intestinalis (10, 166). This reclassification is still controversial (49), as discussed below. E. intestinalis shows a unique parasite-secreted fibrillar network surrounding the developing parasites, so that the parasitophorous vacuole appears septate (Fig. 7) (46, 47, 59, 266).

FIG. 7 Transmission electron micrograph of duodenal epithelium of an HIV-infected patient infected with Encephalitozoon intestinalis. One meront with two nuclei, four sporonts with thickened plasma membrane, and four spores are separated by amorphous material (more ...)

Although microsporidia of the genus Nosema are widespread parasites, only a few human infections with Nosema spp. have been reported. A case of systemic infection occurred in a 4-month-old thymus-deficient infant (235). At autopsy, numerous mature and immature microsporidian spores measuring 4.0 to 4.5 by 2.0 to 2.5 μm with nuclei in diplokaryon arrangement and 10 to 12 coils of the polar tubule were found. No other developmental stages were documented, but the features of the spores supported its assignment to the genus Nosema as a new species, Nosema connori (235, 334). A microsporidium species infecting the corneal stroma of a 39-year old man from Ohio was named Nosema ocularum (36, 39, 44). Spores were lying freely in direct contact with the host cell cytoplasm and measured 3.0 by 5.0 μm with 9 to 12 coils of the polar tubule (36, 44, 45).

Another microsporidium infecting muscle cells of a 31-year-old HIV-infected patient was described by Cali et al. (48). Development took place in direct contact with the muscle cell cytoplasm, and the organisms contained one or two diplocaryotic pairs of nuclei. The spores measured about 2.5 to 2.9 by 1.9 to 2.0 μm, with 7 to 10 turns of the polar tubule. These features are most closely aligned with the genus Nosema, and this organism is currently named “Nosema-like microsporidian” (48).

Another Nosema-like microsporidium was identified in fecal material of a patient with AIDS (239). Because all the parasites were located in partially digested striated muscle cells, it was concluded that this did not represent a true infection (239).

The genus Trachipleistophora was established for a microsporidium responsible for a fourth case of myositis, this time in a patient with AIDS; organisms were found in corneal scrapings, skeletal muscle, and nasal discharge (138). These parasites were cultivated in vitro and in athymic mice (180). Meronts had two to four nuclei and divided by binary fission. In sporogony, the surface coat became separated from the plasma membrane and formed a sporophorous vesicle. The parasite differed from the genus Pleistophora, because no multinucleate sporogonial plasmodium was formed at any stage. Thus, this organism was placed in a new genus and named Trachipleistophora hominis (180).

Recently, two cases of infection with a Pleistophora-like microsporidian, which also seems to be a species of Trachipleistophora, have been reported (271, 390). Sporogony distinguishes this parasite from T. hominis since two different types of sporophorous vesicles and spores are formed (390), and the parasite has recently been classified as a new species T. antropophtera (356a).

One of the Pleistophora-like microsporidia involving skeletal muscles (69), which was described before T. hominis was described as a new species, resembles T. hominis, whereas other Pleistophora-like microsporidia (216) may be different (180, 181).

Microsporidium ceylonensis was identified in a corneal ulcer of an 11-year-old Tamil boy from Sri Lanka. The spores measured 1.5 by 3.5 μm, and no meronts or sporonts were seen (6, 50). Microsporidium africanum was detected in corneal stroma of a 26-year-old woman from Botswana suffering from a perforated corneal ulcer (50, 277). Spores with 15 to 16 turns of the polar tubule measured 4.5 by 1.5 μm, and no developmental stages of the parasite were seen.

Many other genera in several invertebrate phyla and in all five classes of vertebrates have been described (39, 50, 58). The number of named and unnamed species, now approaching 1,000 and belonging to nearly 100 genera, certainly represents only a small fraction of the total diversity. Examination of new hosts will continue to increase the number of microsporidian genera and species (58).

TABLE 1 Case reports of microsporidiosis in patients not infected with HIV with normal or unknown immunestatus

TABLE 2 Case reports of microsporidiosis in patients not infected with HIV but withimmunodeficiency

More than 1,000 cases of microsporidiosis have been documented, the majority with E. bieneusi, in HIV-infected patients (14, 30, 53, 72, 87, 99, 103, 109, 110, 118, 136, 142, 207, 244, 246, 263, 278, 279, 326, 344). Between 2 and 50% of HIV-infected patients with severe immunodeficiency and CD4 cell counts below 100/μl and otherwise unexplained diarrhea are infected, depending on the study group and method of diagnosis (72, 130, 131, 136, 142, 207, 212, 246, 263). When patients do not suffer from diarrhea, E. bieneusi is only rarely reported (127, 282, 283). Rabeneck et al. (282, 283) observed no significant difference in the occurrence of microsporidiosis in patients with (18 of 55 [33%]) and without (13 of 51 [25%]) chronic diarrhea. However, these findings were not duplicated by other investigators, and support for a pathogenic role for microsporidia is based on its identification, often as the sole pathogen, in several hundred patients worldwide. It seems likely that, as with other parasites, a relationship exists between the intensity of infection and clinical illness. Because intestinal microsporidiosis may be a common infection in humans that can exist latently (148, 350, 353), microsporidia are most likely to cause disease if the immune status of a host is such that the infection cannot be controlled. However, quantitation of E. bieneusi spores in stool specimens is not correlated with intensity of diarrhea (71).

Infections with other microsporidian species have been reported less frequently, but more than 100 cases of human infections with Encephalitozoon spp. have been documented (68, 95, 102, 113, 121, 122, 136, 143, 144, 195, 215, 238, 243, 247, 266, 291, 297–299, 304–306, 328, 369, 373, 374). Most of these cases were due to E. intestinalis or E. hellem (68, 113, 121, 122, 136, 143, 195, 238, 243, 247, 266, 291, 297, 298, 304–306, 328, 369, 373), but recently E. cuniculi was detected in several HIV-infected patients as well (1, 95, 144, 177, 179, 242, 374).

Human infections with other species (N. connori, N. ocularum, V. corneae, Pleistophora spp., T. hominis, T. antropophtera, M. ceylonensis, and M. africanum) have occurred only in a few patients so far (6, 44, 48, 50, 69, 92, 138, 161, 216, 235, 277, 375) and these infections may represent only random opportunistic events.

Microsporidia are released into the environment via stool, urine, and respiratory secretions. Persons or animals infected with microsporidia are possible sources of infection. Experimental Encephalitozoon infections of several animals by the oral, tracheal, and rectal routes have been reported (80, 153, 209, 313). Person-to-person transmission of microsporidia may be significant. In a case-control study, intestinal microsporidiosis was associated with male homosexuality, thereby suggesting sexual routes of transmission (181a). Person-to-person transmission was suspected in an HIV-seronegative partner of an HIV-infected man with intestinal microsporidiosis due to E. intestinalis (139). Another male patient with microsporidial urethritis had a sexual partner with diarrhea due to intestinal microsporidiosis (27).

Whether microsporidiosis in humans is a zoonosis is unknown, and no direct proof of transmission from animals to humans has been documented, with the exception of one case where a 10-year-old girl seroconverted after close contact with a dog infected with E. cuniculi (239a). Animal reservoirs of microsporidia infecting humans have been confirmed recently (Table 3). E. cuniculi is commonly found in several mammals (50, 51), and Encephalitozoon spp. have occasionally been found in lovebirds (50, 196). E. hellem was recently detected in birds (parrots) (28), E. intestinalis has been found in different mammalian animals (donkey, dog, pig, cow, and goat) in Mexico (32a), and Enterocytozoon bieneusi has been found in stool samples of pigs and dogs in Switzerland (98) and in simian immunodeficiency virus-infected macaques (234). E. hellem infection of birds, E. bieneusi infections of pigs, dogs, and monkeys, and E. intestinalis infection of different mammals were confirmed by molecular techniques with rRNA data (32a, 98, 234). Molecular analysis of different human, rabbit, dog, mouse, and blue fox E. cuniculi isolates showed that all isolates from humans were of the same subtype as isolates from dogs and rabbits (97, 98, 108, 111, 181, 236). This fact supports the hypothesis that human infections with E. cuniculi may be a zoonosis (97, 98, 111, 181, 236).

TABLE 3 Animal hosts of humanmicrosporidia

Arthropods are the most common hosts of microsporidia, and experimental infections of mice by a mosquito microsporidium (Nosema algerae) have been accomplished (342, 345). Whether insect microsporidia might infect humans is unknown.

Several microsporidia have been found in surface water samples (8), but whether human microsporidiosis is a waterborne disease is unknown. Results from recent studies involving molecular techniques seemed to indicate the presence of E. intestinalis, E. bieneusi and V. corneae in raw sewage, tertiary effluents, surface water, and groundwater in France and the United States (123a, 332a); there is one report of a presumably waterborne outbreak in Lyon (France) during summer 1995 (78a). Risk factors for intestinal microsporidiosis also suggest water as the source of infection (133, 181a). In a case-control study, the only two factors associated with intestinal microsporidiosis were swimming in a pool and male homosexuality, both suggesting that the mode of transmission is fecal-oral (181a). However, no seasonal variation in the prevalence of microsporidiosis in HIV-infected patients was seen over 4 years in southern California, suggesting a constant presence of microsporidia in the environment rather than a seasonal association with recreational water use or seasonal contamination of the water supply (75a).

Microsporidiosis is truly an emerging infectious disease with a rapidly broadening clinical spectrum of diseases. The spectrum of diseases includes gastrointestinal, pulmonary, nasal, ocular, muscular, cerebral, and systemic infections. Microsporidiosis should be considered in the differential diagnosis of HIV-related symptomatic disease of virtually all organ systems (Table 4) (271).

TABLE 4 Clinical manifestations of human microsporidialinfections

Enterocytozoon bieneusi. Intestinal infections with microsporidia have been found mainly in HIV-infected patients, and most infections have been due to E. bieneusi. It is most common in patients with severe immunodeficiency and a CD4 cell count below 100/μl (60, 127, 130, 131, 142, 246, 332). The parasites cause a severe, nonbloody, nonmucoid diarrhea with up to 10 or even more bowel movements per day, slowly progressive weight loss, and malabsorption of fat, d-xylose, and vitamin B₁₂ (60, 78, 127, 130, 131, 204, 205, 212). Intestinal infection is associated with lactase deficiency and a reduced activity of alkaline phosphatase and α-glucosidase at the basal part of the vilus and with reduced villus height and a vilus surface reduction (301). Diarrhea appears gradually and may continue for months. Patients are often reluctant to eat and may complain of nausea (60, 78, 204, 205 212). Some patients have intermittent diarrhea, but only a few excrete microsporidial spores without having diarrhea (282, 283, 326, 330). In groups of patients with chronic diarrhea who were negative for other enteric pathogens, the prevalence of E. bieneusi was between 7 and 50% (127, 130, 131, 207, 246, 263). Some patients have coinfections with other pathogens (30, 171, 370); cryptosporidia are the most common pathogens coinfecting patients with intestinal microsporidiosis (157, 171, 370).

E. bieneusi infection of the biliary tract, with or without cholecystitis, is responsible for some of AIDS-related cholangiopathies which are not explained by Cryptosporidium spp. or cytomegalovirus infection (33, 201, 240, 278, 279).

Dissemination of E. bieneusi is very uncommon, but it has been detected in duodenal lamina propria cells (310), in bronchoalveolar lavage fluid and transbronchial biopsy specimens (93, 367), and in nasal sinuses of HIV-infected patients (129, 165, 184).

Intestinal infections with E. bieneusi in non-HIV-infected patients have been reported in only 15 patients in Germany, Switzerland, France, Spain, Zambia, and the United States (96, 154a, 167, 249, 284, 295, 296, 329): one patient had Crohn’s disease (96), an African woman had a cardiac-valve graft (96), one patient had a congenital disorder of the lymphatic system (154a), another had a low CD4 cell count of unknown origin (365a), and two patients were immunosuppressed due to treatment associated with liver and heart-lung transplantation (284, 296). The other nine patients were otherwise healthy (Tables 1 and 2). Although patients usually presented with self-limiting diarrhea, some patients were treated with albendazole or metronidazole (284, 296).

Encephalitozoon spp. Similar to Enterocytozoon bieneusi, E. intestinalis causes an enteritis with diarrhea, weight loss, and malabsorption (68, 70, 78, 121, 136, 143, 207, 212, 247, 266). Besides intestinal infections, these parasites may infect the biliary tract and gallbladder, resulting in cholangitis and cholecystitis (385). Disseminated infections occur regularly and involve heavy infections of the urinary tract including the kidneys (68, 121, 143, 150, 247, 268). Left untreated, small bowel infection with E. intestinalis can lead to perforation and peritonitis (331). E. cuniculi only occasionally infects the gastrointestinal tract, and its pathogenicity in humans is unknown. Franzen et al. (144) described an AIDS patient with a widely disseminated E. cuniculi infection including the gastrointestinal tract but with no accompanying gastrointestinal symptoms. Weber et al. (374) described a second patient with disseminated infection due to E. cuniculi who had no gastrointestinal symptoms but who had microsporidian spores in the stool samples.

Among persons not infected with HIV, only three cases of intestinal infection with an Encephalitozoon spp. have been reported (139). A 36-year-old HIV-seronegative homosexual man was asked to provide stool for examination after E. intestinalis was demonstrated in stool samples of his HIV-infected partner. E. intestinalis was detected in two of seven stool samples from the non-HIV-infected man and again 4 months later, together with Isospora belli, when he became mildly symptomatic after a trip to Brazil (139). Two other patients were travelers presenting with chronic diarrhea, and microsporidian spores were detected in their stools (286). Molecular identification of microsporidian species as E. intestinalis was based on PCR amplification of an SSU rRNA sequence. Albendazole treatment led to the elimination of spores in the stool, but the clinical signs persisted.

Other species. A Nosema-like microsporidium has been identified in fecal material of a patient with AIDS (239). The parasites were located in partially digested striated muscle cells, suggesting that infected animal musculature had been ingested. It was concluded that this represents an incidental finding rather than a true infection (239).

Peritonitis due to E. cuniculi was described in a 45-year-old HIV-infected man with a CD4 cell count of 57/μl (392). The patient presented with a 13-kg weight loss over the course of a year and was treated with trimethoprim-sulfamethoxazole because of Pneumocystis carinii pneumonia. After the end of therapy, he developed renal failure and a tumorlike mass was recognized in the abdomen. He died, and at limited autopsy microsporidia consistent in ultrastructure with E. cuniculi were discovered within areas of mixed nongranulomatous inflammation in sections of the omentum magnum (392). The reports of these two cases were published before E. hellem was described as a new species. In both instances, diagnosis was made only on an ultrastructural basis, so that the exact species identification is uncertain.

A second case of fulminant hepatic failure caused by microsporidial infection with an Encephalitozoon sp. was reported in a 43-year-old homosexual man with AIDS (322). He suffered from microsporidial diarrhea 2 months prior to development of fulminant hepatitis. The patient died before albendazole became available. The autopsy revealed disseminated microsporidial infection involving the liver, gallbladder wall, and a mediastinal lymph node.

Both E. bieneusi and E. intestinalis have been detected in nonparenchymal liver cells of several HIV-infected patients, but the patients did not show any signs of hepatitis (14, 268, 278, 279).

Disseminated Trachipleistophora antropophtera infection involving several organ systems including the liver and the pancreas was reported in an 8-year-old HIV-infected girl with seizures and cerebral lesions. This patient died after empirical antitoxoplasma therapy (271, 390).

Encephalitozoon spp. In HIV-infected patients, keratoconjunctivitis may be caused by all three Encephalitozoon spp. (E. hellem, E. cuniculi, and E. intestinalis) (44, 45, 102, 104, 105, 113, 144, 152, 211, 224–226, 238, 243, 291, 306, 319, 369). Most patients present with bilateral conjunctival inflammation and also exhibit bilateral punctate epithelial keratopathy, leading to decreased visual acuity. The keratoconjunctivitis is often asymptomatic or moderate but can be severe; it rarely leads to corneal ulcers (225).

Other species. Keratitis with corneal stroma infection was described in an otherwise healthy 45-year-old man from South Carolina who developed decreased vision in his left eye during an 18-month history of unilateral progressive central keratitis (92). There was no history of prior trauma. Corneal biopsy revealed microsporidia invading deep into the corneal stroma. This organism was successfully propagated in vitro and was named Nosema corneum (317). On the basis of ultrastructural data, it is now in a new genus and has been renamed Vittaforma corneae (324).

A microsporidium was found to be responsible for the symptoms in a 39-year old man from Ohio who developed blurred vision and irritation in his left eye. His visual symptoms persisted despite the discovery and surgical removal of a foreign body (36, 39). A subsequent biopsy of the persistent corneal ulcer revealed organisms with typical microsporidian ultrastructure; the species was named Nosema ocularum (44).

Trachipleistophora hominis was found in the corneal scrapings of an HIV-infected patient with disseminated infection who suffered from myositis and keratoconjunctivitis (138).

In 1973 and 1981, two cases with corneal involvement were documented in a 11-year-old Tamil boy from Sri Lanka with a corneal ulcer and a 26-year-old woman from Botswana suffering from a perforated corneal ulcer (6, 277). Both otherwise healthy patients did not have HIV infection. The genera could not be determined, and the organisms were named Microsporidium ceylonensis and Microsporidium africanum, respectively (50).

Pulmonary microsporidial infection was also found in a 27-year-old woman from India with chronic myeloid leukemia undergoing allogenic bone marrow transplantation (194). The patient died of a fungal infection, and the diagnosis of pulmonary microsporidiosis was reached only postmortem. Ultrastructural examinations confirmed the organism to be a microsporidium, but taxonomic classification could not be done because the organism could not be identified as any of the known pathogenic species of microsporidia (194).

Chupp et al. (69) reported a 33-year-old Haitian man with AIDS who was admitted to the hospital with fever, cough, and diffuse myalgias and weakness (69, 281). A Pleistophora-like microsporidium was detected in muscle cells in a biopsy specimen from the right quadriceps. A similar case was reported by Grau et al. (161) in a 35-year old HIV-infected Spanish man who originated from The Gambia. The patient suffered from myositis with fever, myalgia, and progressive weakness. Microsporidian spores were detected in a muscle biopsy specimen.

In an Australian patient who presented with a severe, progressive myositis associated with fever and weight loss, Pleistophora-like microsporidia were demonstrated in corneal scrapings, skeletal muscle, and nasal discharge (138). The organisms were cultivated in vitro as well as in athymic mice. Since these parasites differed from Pleistophora, the new genus and species Trachipleistophora hominis was established (180).

A Nosema-like microsporidium was detected by Cali et al. in a biopsy specimen from the left quadriceps of a 31-year-old patient with AIDS and myositis (48).

Encephalitozoon spp. Two cases of disseminated Encephalitozoon infection with cerebral involvement were reported in a 9-year-old Japanese boy and in a 2-year-old Columbian boy. Both patients suffered from cerebral symptoms such as headache, vomiting, spastic convulsions, and convulsive seizures. Encephalitozoon-like organisms were found in urine from both patients and in cerebrospinal fluid from one patient. The exact species differentiation of these two parasites is uncertain (21, 237) (see “Systemic infections” below).

Cerebral microsporidiosis due to E. cuniculi was recently described by Weber et al. (374) in a 29-year-old HIV-infected man with a CD4 cell count of 0 cells/μl. The patient was hospitalized because of headache, visual and cognitive impairment, nausea, and vomiting. Magnetic resonance imaging scans showed right maxillary sinusitis and multiple small, contrast-enhanced lesions in the hippocampal, mesencephalitic, and intracortical regions. Examination of cerebrospinal fluid showed microsporidial spores, which were also detected in sputum, urine, and stool specimens. The microsporidium was cultivated in vitro and was classified as E. cuniculi by Western blot analysis, ribotyping, and sequencing of the rRNA intergenic spacer region (374). A similar case was reported by Mertens et al. (242) in a 25-year-old HIV-infected woman. Microsporidian spores, classified as E. cuniculi by immunohistochemistry and PCR, were detected in the brain, heart, kidneys, urinary bladder, spleen, lymph nodes, adrenals, and trachea at autopsy.

Other species. Cerebral involvement with Trachipleistophora antropophtera was reported in two AIDS patients, a 33-year-old man and an 8-year-old girl, with seizures and cerebral lesions, who died after empirical anti-toxoplasma therapy (20, 271, 390). At autopsy, a pansporoblastic microsporidium was seen in several organ systems including the brain (271, 390).

Urethritis. Two cases of urethritis associated with microsporidia were found in patients with AIDS who suffered from urethritis, sinusitis, and diarrhea (27, 77). Encephalitozoon-like spores were detected in a smear of expressed urethral pus as well as in stool samples, nasal discharge, sputum, and urine of one patient (77) and in stool samples of the second patient (27). Both patients were treated with albendazole, and the symptoms disappeared.

Prostatic abscess. A prostatic abscess due to E. hellem was found in an AIDS patient with disseminated E. hellem infection (308). The prostate was of normal size with a 1.5- by 1.8-cm central periurethral abscess containing necrotic prostatic tissue. Tissue Gram stain revealed gram-positive microsporidian spores, which were identified as E. hellem by an indirect fluorescence assay.

Tongue ulcer. A shallow 1-cm ulceration on the dorsum of the tongue was observed in an HIV-infected patient with severe immunodeficiency (15 CD4 cells/μl) and disseminated infection due to E. cuniculi (95). Spores were identified in several samples and in soft tissue beneath the tongue ulcer. The microsporidian was identified as E. cuniculi by immunofluorescent staining, in vitro cultivation, and molecular analysis of the SSU rRNA gene by PCR. The patient was treated with albendazole, and the symptoms resolved within 2 weeks (95).

Skeletal involvement. Only two cases of skeletal involvement with microsporidia have been found in patients with AIDS (19). Both patients suffered from disseminated microsporidial infections. In one patient the mandible and associated soft tissues were involved. Species identification was not done.

Cutaneous microsporidiosis. One case of nodular cutaneous microsporidiosis that resolved with oral clindamycin therapy was found in an HIV-infected patient (200a). Underlying osteomyelitis that also resolved after therapy was not proven to be caused by the microsporidia. Species differentiation by PCR techniques was not successful.

Encephalitozoon spp. The first case of documented human microsporidial infection was a case of disseminated Encephalitozoon infection in a 9-year-old Japanese boy who suffered from recurrent fever, headache, vomiting, and spastic convulsions reported in 1959. Encephalitozoon-like organisms were found in cerebrospinal fluid and urine. He was treated with sulfisoxazole and penicillin and recovered (237).

A similar case occurred in 1984 in a 2-year-old Columbian boy who lived in Sweden. He had convulsive seizures, and gram-positive organisms consistent with an Encephalitozoon sp. were found in urine. Anti-E. cuniculi antibodies (immunoglobulin G [IgG] and IgM) were detected in serum samples (21).

Disseminated infections with all three Encephalitozoon spp. are now increasingly recognized in severely immunosuppressed HIV-infected patients, usually in those with CD4 cell counts below 100/μl (95, 121, 144, 150, 162, 215, 247, 268, 297, 304, 305, 328, 369). The spectrum of disease has expanded to include keratoconjunctivitis, bronchiolitis and pneumonia, sinusitis, nephritis, urethritis, cystitis, prostatitis, hepatitis, peritonitis, gastroenteritis, and cholangitis, but there are clear differences in the typical distribution pattern for each microsporidian species: E. hellem parasitizes mainly the keratoconjunctiva, urinary tract, nasal sinuses, and bronchial system; on the other hand, E. intestinalis appears to be confined mainly to the gastrointestinal and biliary tract with dissemination to the kidneys, eyes, nasal sinuses, and sometimes the respiratory tract; finally, E. cuniculi causes widely disseminated infections involving nearly all organ systems, but the clinical manifestations vary substantially, ranging from no symptoms to severe disease (144, 150, 162, 215, 247, 268, 304, 305, 328, 369).

Other species. In 1973, Nosema infection and Pneumocystis carinii pneumonia were diagnosed at autopsy in a 4-month-old athymic male infant (235). Shortly after birth, the child developed diarrhea, vomiting, fever, dyspnea, weight loss, and mechanical ileus. Laparatomy and several antibiotics failed to alter the clinical course, and at autopsy sporoblasts with mature and immature spores of a Nosema sp. were seen in almost all tissues examined except the spleen (235). The parasite was named Nosema connori (334).

T. hominis was reported as the cause of myositis in a 34-year-old HIV-infected man; parasites were recognized in corneal scrapings, skeletal muscle, and nasal discharge (138). The newly recognized pansporoblastic microsporidium, T. antropophtera caused disseminated infection involving the brain, heart, kidneys, pancreas, thyroid, parathyroid, liver, bone marrow, lymph nodes, and spleen in an HIV-infected 8-year-old child (271, 390). T. antropophtera infection of the brain was also seen in a 33-year-old HIV-infected male in whom autopsy was limited to the brain (20, 271, 390).

Successful treatment of microsporidiosis in immunodeficient patients is limited. Several in vitro culture systems and animal models have been used to identify potential antimicrobial agents for treatment of microsporidiosis. Different drugs control the levels of microsporidial infection in invertebrate hosts; these include fumagillin, an antibiotic produced by Aspergillus fumigatus, and itraconazole for control of Nosema apis in honey bees and other microsporidia in weevils (54). However, in vitro investigations with Nosema bombycis showed no effect of itraconazole and metronidazole on the number of cells infected or on the spore harvest (54). On the other hand, albendazole had marked effects on these parameters, and several ultrastructural changes in the parasites were noted (54, 163).

Other in vitro models used to evaluate drug efficacy included E. cuniculi, E. hellem, and E. intestinalis (16, 117, 141, 168, 218, 219, 380). These studies showed that albendazole, fumagillin, 5-fluorouracil, sparfloxacin, oxibendazole, and propamidine isethionate inhibited E. cuniculi growth in cell cultures. Chloroquine, pefloxacin, azithromycin, rifabutin, and thiabendazole were partially effective at high concentrations. Arprinocid, metronidazole, minocycline, doxycycline, itraconazole, and difluoromethylornithine were not evaluable, since the concentrations that inhibited microsporidia were also toxic for the cells in the cell culture. Pyrimethamine, piritrexim, sulfonamides, paronomycin, roxithromycin, atovaquone, flucytosine, toltrazuril, ronidazole, and ganciclovir were ineffective (16). Spore germination of E. hellem and E. intestinalis was inhibited by nifedipine, metronidazole, and nitric oxide donors (168), and E. hellem spore germination was also inhibited by cytochalasin D, demecolcine, and itraconazole (218). TNP-470, a semisynthetic analogue of fumagillin, was highly effective against all three Encephalitozoon spp. and V. corneae in cell cultures (82, 111a). A fluorescent probe, designated calcein, and confocal microscopy have been used to demonstrate drug-induced effects in Encephalitozoon-infected green monkey kidney cells, and both albendazole and fumagillin caused different types of parasite changes (219). In vivo efficacy of albendazole, fumagillin, and TNP-470 against E. cuniculi has been demonstrated in experimentally infected SCID mice, athymic mice, and rabbits (82, 210, 314). Unfortunately, long-term in vitro cultivation of Enterocytozoon bieneusi has not been feasible so far; therefore, a direct assay of the effects of agents on this parasite is not yet practicable.

Based on these in vitro studies, several drugs have been used to treat microsporidial infections in humans. Until recently, blinded, placebo-controlled comparative trials were lacking. Therefore, most clinical experience in the therapy of human microsporidiosis consists of only anecdotal observations. Several case reports and small case series have shown that albendazole was highly effective for treatment of Encephalitozoon infection in HIV-infected patients and led to impressive clinical improvement and eradication of the parasites (1, 77, 95, 109, 121, 143, 144, 150, 162, 185, 211, 215, 247, 248a, 269, 328, 373). However, since some patients relapsed after therapy, maintenance therapy may be necessary for these patients (248a, 373). Symptomatic improvement with reduction of clinical findings was also achieved with topical fumagillin (113, 158, 314) in several HIV-infected patients with microsporidial keratoconjunctivitis due to Encephalitozoon species. Resolution of E. hellem infection of the corneal epithelium of an AIDS patient with itraconazole was also reported (391), but this finding could not be duplicated (113). Itraconazole seems to be ineffective in preventing Enterocytozoon bieneusi infection in HIV-infected patients (2). Keratoconjunctivitis due to an Encephalitozoon species in another AIDS patient responded to topical dibromopropamidine isethionate (238).

Infections due to E. bieneusi are much more difficult to treat, and currently there is no acceptable treatment. Pneumocystis carinii prophylaxis with co-trimoxazole seems to have no influence on the prevalence of intestinal microsporidiosis in HIV-infected patients, suggesting that this drug may be ineffective (3). Improvement or disappearance of diarrhea caused by E. bieneusi has been reported after treatment with metronidazole, but repeated biopsies showed that microsporidia persisted (127, 128), and other investigators did not observe any response after treatment with metronidazole (114, 136). The efficacy of albendazole for E. bieneusi infections is controversial (257). Diarrhea may improve in 50 to 60% of patients (29, 103, 114), but persistence of organisms was seen in posttreatment biopsy specimens despite several ultrastructural changes in the parasite (29, 31, 114). Other trials with albendazole showed much lower response rates (142, 172, 246). Double-blind placebo-controlled trials with albendazole are in progress. Despite a remarkable clinical response with atovaquone in symptomatic AIDS patients with intestinal E. bieneusi infection, there was no apparent decrease in the parasite burden in either stools or biopsy specimens (5). A double-blind placebo-controlled study with atovaquone is also under way. Similarly, azithromycin treatment showed only partial effect on diarrhea in E. bieneusi-infected patients, and microsporidial infection persisted on repeat biopsy and stool examinations (172). Improvement of diarrhea with clearance of microsporidian shedding in stool was observed in three E. bieneusi-infected patients treated with furazolidone (116). Purified fumagillin was also able to clear E. bieneusi infection from stool as well as intestinal biopsy specimens in three patients, but the drug is toxic and causes thrombocytopenia in nearly all patients (248). These observations must be confirmed by treatment of more patients with these two drugs. As mentioned above, TNP-470, a synthetic analogue of fumagillin that is less toxic, is as effective as fumagillin against several microsporidian species in vitro and in athymic mice and holds promise as a new antimicrosporidial compound (111a).

Of note, elevated tumor necrosis factor alpha levels in stool samples have been demonstrated in patients with microsporidial diarrhea. Therefore, the anti-tumor necrosis factor alpha agent thalidomide has been used to treat diarrhea due to E. bieneusi. Some patients responded to this therapy, but again this observation must be confirmed in controlled trials (320, 321). Symptomatic improvement has been achieved with octreotide, but this drug has no effect on the parasites (325).

Combination antiretroviral therapy that includes a protease inhibitor can restore immunity to microsporidia. The use of potent antiretroviral therapy in patients with advanced HIV infection can improve symptoms due to microsporidiosis and in some cases leads to disappearance of the parasites (61, 75b, 140, 160a). However, the rapid time to release in patients with declining CD4 lymphocyte counts suggest that the microsporidial infections are not eradicated.

Since no effective therapy is available for E. bieneusi infection whereas infections with Encephalitozoon spp. respond very well to albendazole therapy, exact species differentiation of microsporidia infecting humans is absolutely necessary.

Diagnosis of human microsporidiosis is dependent on the identification of spores in clinical samples, e.g., stool specimens, duodenal or bile juice, urine, conjunctival smears, bronchoalveolar lavage fluid, sputum, nasal discharge, or biopsy tissues. The detection of spores in clinical samples, however, is a laborious, challenging, and time-consuming task because the tiny organisms can easily be missed. Originally, definitive diagnosis of microsporidiosis required transmission electron microscopy, but during the last few years new staining methods, suitable for light microscopy, have been developed. Microsporidia have now been found in virtually every tissue and body fluid in humans.

Although the diagnosis and identification of microsporidia by light microscopy have greatly improved during the last few years, species differentiation is usually impossible by these techniques. Immunofluorescent staining techniques have been developed for species differentiation of microsporidia, but antibodies used in these procedures are available only at research laboratories so far. Similarly, cell culture systems can be used for in vitro cultivation of microsporidia, but this is not a suitable technique for routine use because it is laborious and time-consuming.

A variety of serological tests has been developed to detect antibodies to microsporidia, but the sensitivity and specificity of these tests are unknown. Also, these tests are not suitable methods to diagnose infections in immunosuppressed persons.

Recent success in nucleotide sequencing of various microsporidia has now led to the application of new molecular techniques for the diagnosis of human microsporidiosis.

The ultrastructural characteristics of microsporidian species found in humans are summarized in Table 5.

TABLE 5 Morphological characteristics of microsporidia infectinghumans

Detection of microsporidia by transmission electron microscopy is highly specific, but the technique may lack sensitivity, especially when performed on body fluids and stool samples. However, large studies to evaluate the sensitivity of transmission electron microscopy are lacking (62, 73). Likewise, sample preparation and examination are laborious and time-consuming (62, 73).

Cytologic diagnosis and stool examination. Microsporidian spores have been detected in several body fluids and stool samples (264, 347). The outer layer of the spore wall (exospore) is proteinaceous, and the inner layer (endospore) is chitinous (26, 50) so that Gram, Giemsa, and trichrome stains, as well as fluorescent dyes, have been advocated for staining microsporidian spores (74). Gram and Giemsa stains are not suitable for cytologic diagnosis because they do not differentiate between microsporidia and other elements present in body fluids or stool specimens that can be confused with microsporidian spores (75). Microscopic examination of body fluids to diagnose microsporidia did not become routine until special chromotrope-based and fluorescent stains were developed. Nowadays these chromotrope- and/or fluorochrome-based stains are used in several modifications.

The chromotrope-based stain developed by Weber et al. (366) has markedly improved spore detection in stool samples and body fluids. In this technique, involving a modification of the trichrome stain with a concentration of chromotrope 2R that is 10 times higher than that used in the trichrome stain, microsporidian spores stain characteristic pinkish red. Usually the spores have a characteristic appearance when examined under high-power magnification (×1,000). The spore wall stains intense red, and some spores show a distinct beltlike stripe that grids the spores diagonally or equatorially (Fig. 8) (366). This staining technique is lengthy, and spores are difficult to detect if only a few are present in the sample (94). Several modifications (changes in temperature and staining time [203], decrease in the phosphotungstic acid level, and substitution of a color-fast counterstain [294]) of the chromotrope-based stain have been suggested for speeding the process and for better contrast between the spores and the background. The improved hot Gram-chromotrope technique provides some real advantages in staining microsporidia for light microscopy. In this procedure, samples are stained in solutions of crystal violet and iodine used in Gram’s stain and then in a modified chromotrope solution heated to 50 to 55°C. With this stain, microsporidian spores are stained dark violet against a pale green background and the total staining time is shortened to 5 min (251).

FIG. 8 Encephalitozoon cuniculi spores in conjunctival swab from an HIV-infected patient with disseminated infection. Modified chromotrope-based stain. Magnification, ×870.

The epifluorescence of microsporidian spores stained with optical brighteners was a second major breakthrough for detecting spores in stool samples and body fluids (74, 75, 228, 348, 349). Of these fluorescent stains, Uvitex 2B, Calcofluor white, and Fungifluor (a formulation of Calcofluor white in 10% KOH) are the stains of first choice, although Calcofluor white produces somewhat greater background staining (74). The staining procedure is easy and quick, but examination requires a fluorescent microscope with a 350- to 380-nm excitation filter and a high-magnification objective lens (magnification, ×1,000). Fluorochromes bind to the chitin of the endospore, and when excited under UV light, the bound dye fluoresces brightly in the visible spectrum (74). Spores are identified by their size, shape, and fluorescent staining properties (Fig. 9) (74, 348). Several modifications of the fluorescent stain with Uvitex 2B, originally described by van Gool et al. (348), were introduced to use on touch preparations (74, 356), brush cytologic specimens (270), and smears of several body fluids (74, 107) and for staining of paraffin-embedded material (74, 146). Uvitex 2B was used a number of years ago for detecting nonhuman microsporidia in tissue sections (355).

FIG. 9 Encephalitozoon intestinalis spores derived from in vitro culture. Fluorescence microscopy after Uvitex 2B stain. Magnification, ×1,000.

Although fluorescent stains seem to be more sensitive than chromotrope-based trichrome stains, they may lead to some false-positive results due to the similarity in staining of small yeast cells (74, 75, 94, 228). In addition, other studies have not shown superior sensitivity of the fluorescent stains over the chromotrope-based stains (107, 182). Most authors conclude that the two techniques should be used simultaneously to enhance performance and to provide greater accuracy, especially for patients with light infections (94, 107, 182).

Because the number of microsporidia in clinical samples can be very small, centrifugation of body fluids may be necessary. Whether concentration techniques are useful for detection of microsporidian spores in stool specimens remains controversial. Some authors reported that concentration techniques such as the formalin-ethyl acetate concentration procedure or different flotation methods led to a substantial loss of microsporidial spores and thus to false-negative results (62, 366). In contrast, others found that concentration techniques such as the water-ether sedimentation or centrifugation of KOH-treated stool samples increased the yield of microsporidian spores (62, 349). Therefore, one of these concentration techniques should be used for stool samples.

Histologic diagnosis. Because of the small size of the spores, reliable visualization of microsporidia by light microscopy depends on a distinct contrast between spores and other cellular contents. Routine hematoxylin-and-eosin stained parasites in tissue are easily overlooked even by experienced pathologists (241). Tissue Gram stains such as Brown-Brenn or Brown-Hopps seem to be very useful for reliable identification of microsporidia in paraffin-embedded tissue sections (206, 267). Microsporidian spores are Gram variable, but with these stains they stain dark blue or red against a faint brown-yellow background (206). Fluorescent staining techniques with optical brighteners are easy and quick to perform on tissues, and the sensitivity of these stains seems to be very high (Fig. 10) (74, 146). A major advantage of these stains is that they can be combined with other staining techniques (74). Silver stains such as the Warthin-Starry stain (136, 137) or a modified chromotrope-based trichrome stain (160) are preferred by some, but both techniques are time-consuming and interpretation of sections may be difficult.

FIG. 10 Paraffin-embedded duodenal biopsy specimen from a patient with AIDS with intestinal Enterocytozoon bieneusi infection. The microsporidial spores are easily visualized within the enterocytes. Fluorescence microscopy after Uvitex 2B stain. Magnification, (more ...)

Giemsa, chromotrope 2R, or fluorochrome staining of touch preparations of intestinal tissue and of endoscopic brush cytologic specimens is useful in the diagnosis of intestinal microsporidiosis, but these techniques require fresh material (15, 74, 270, 288, 309).

Semithin sections of resin-embedded biopsy material, stained with a variety of different stains (hematoxylin and eosin, Giemsa, toluidine blue, methylene blue-azure, and basic fuchsin), are useful methods for visualization of spores and tissue stages of microsporidia. However, resin embedding of biopsy specimens is not routinely used in most laboratories (Fig. 11) (263, 267, 273, 309).

FIG. 11 Resin-embedded semithin (1-μm) section of duodenal mucosa from a patient with AIDS and intestinal Enterocytozoon bieneusi infection. Epithelial cells contain spores of Enterocytozoon bieneusi. Toluidine blue stain. Magnification, ×800. (more ...)

Microsporidia have been successfully cultivated in a number of mammalian cell lines including monkey and rabbit kidney cells (Vero and RK13), human fetal lung fibroblasts (MRC-5), MDCK cells, and several other cell lines (50, 102, 177–179, 275, 350, 354). Species that have been cultivated in vitro from a variety of human specimens include E. hellem (102, 178, 359), E. cuniculi (95, 177, 179, 312), E. intestinalis (122, 151, 350, 360), V. corneae (317), and T. hominis (180). Attempts to culture Enterocytozoon bieneusi from small intestinal biopsy specimens laden with merogonic stages and spores by using a range of cell lines and pretreatments have had limited success (54); to date, E. bieneusi has been propagated only in short term cultures (6 months) (361). In the culture systems (human lung fibroblasts and Vero monkey kidney cells) used, E. bieneusi seems to exert a greater cytotoxic effect than has been observed with cell cultures of Encephalitozoon spp. The inability to grow E. bieneusi in a continuous-culture system may reflect a need of this organism for some specific nutritional requirements that are not provided by the cell cultures used so far (361).

Animal models for E. bieneusi infection are difficult to establish. Attempts to infect immunosuppressed gnotobiotic piglets, gamma interferon knockout mice, and SCID mice treated with anti-gamma interferon monoclonal antibodies with E. bieneusi have been unsuccessful. Recently, experimental oral transmission of E. bieneusi to simian immunodeficiency virus-infected rhesus monkeys has been reported (343).

Immunofluorescent antibody tests involving polyclonal antisera against E. hellem, E. cuniculi, and E. intestinalis produced in mice or rabbits showed that several species of microsporidia demonstrated immunological cross-reactivity (4, 259, 359, 360, 397). This cross-reactivity of the polyclonal antisera against Encephalitozoon spp. was used for the detection of several microsporidian species including E. bieneusi in various clinical samples by using different immunofluorescence tests (4, 259, 359, 360, 397). However, the cross-reactivity of the antisera limit their use as diagnostic tools because species differentiation is not possible with these reagents.

Several monoclonal antibodies which recognized E. hellem (4, 359), E. cuniculi, or E. intestinalis (17) were developed. Most of these monoclonal antibodies are specific to spore antigens (4, 227a, 359), whereas other researchers used polar tube protein-reactive monoclonal antibodies in combination with monoclonal antibodies that recognize the surfaces of spores (17). Some of these monoclonal antibodies are species specific, whereas others react against spore walls or the polar tubes of several microsporidian species (227a).

Mono- and polyclonal antibodies are useful tools for species differentiation of microsporidia in different clinical samples (307), but antibody-staining techniques may be less sensitive than other techniques. Didier et al. (107) compared a chromotrope-based stain, a fluorescent stain containing Calcofluor white, and a fluorescent polyclonal antibody stain. The fluorescent polyclonal antibody stain was the least sensitive method for detecting microsporidial spores in stool samples, urine, and duodenal fluid (107). Therefore, antibodies should be used for species differentiation in samples only when the initial diagnosis of microsporidiosis by using fluorescent stains with optical brighteners and/or chromotrope-based stains has been made (Fig. 12) (107). E. bieneusi-specific antibodies have not been developed so far.

FIG. 12 Indirect immunofluorescent staining of Encephalitozoon cuniculi spores in nasal discharge of an HIV-infected patient with disseminated infection, using polyclonal anti-Encephalitozoon cuniculi antiserum. Magnification, ×400.

Serologic surveys for antibodies to E. cuniculi have suggested a possible link between exposure to a tropical environment and infection with microsporidia. In patients with malaria and schistosomiasis, the microsporidial seropositivity rate was 4.7 and 9.1%, respectively (176). A study of homosexual men in Sweden reported that 10 of 30 persons (33%) were seropositive for antibodies to E. cuniculi; all the seropositive patients had at some time visited a tropical area (22, 176). Explanations for this apparent relationship remain speculative, and clinicopathologic correlations have not been reported for any of these serologic surveys. Of interest, a high seroprevalence of antibodies against E. intestinalis were observed among Dutch blood donors (24 of 300 [8%]), pregnant French women (13 of 276 [5%]), and patients with various infectious and noninfectious diseases (6 of 150 [4%]) (353).

For E. bieneusi, which has not been propagated in long-term cell cultures so far, specific serologic assays remain unavailable. Whether antigens of Encephalitozoon spp. cross-react with those of E. bieneusi has not been determined exactly so far (259, 302, 379, 397).

However, serologic methods are not useful as diagnostic tools for microsporidiosis because at least half of the serum samples from persons without a history of microsporidial infection may have positive titers and immunosuppressed persons may have a poor response to antigen challenge.

Molecular studies of microsporidia are in their infancy. A limited number of genes have been located, and only a few DNA sequences have been reported in a few microsporidian species so far.

Compared with other eukaryotes, microsporidia have extremely small genomes, often in the size range of bacterial genomes. Pulsed-field gel electrophoresis studies of the karyotypes of several microsporidian species showed that the haploid genome usually ranges from only 5.3 to 19.5 Mb, which represents the largest microsporidian genome measured to date in Glugea atherinae (23, 24, 252). However, the haploid genome of E. cuniculi was estimated to be only 2.9 Mb, which is smaller than all the other microsporidian genomes. This size is about half that of the previously smallest known microsporidian genome, 5.3 Mb in Nosema locustae (23). Eleven chromosomal DNA bands obtained by pulsed-field gel electrophoresis with DNA isolated from E. cuniculi spores ranged only from 217 to 315 kb. However, the chromosome number of E. cuniculi was larger than that of a Vairimorpha sp. and Nosema costelytrae, both of which have only eight chromosomes (231). The very small genome in E. cuniculi may be related to the early divergence of microsporidia, but it may be also a derived characteristic, perhaps related to the highly adapted parasitic lifestyle of this organism.

Chromosomal localization of genes in microsporidia were done for only a few genes in a limited number of species. Eight genes in E. cuniculi have been localized so far. Each of the 11 chromosomes of E. cuniculi contains rDNA, which strongly contrasts with the location of rDNA within a single but rather large chromosome (760 kb) in Nosema bombycis, an insect microsporidium with a genome of 15.3 Mb (25). Further analysis of eluted DNA from each E. cuniculi chromosome by restriction enzyme digestion and rDNA hybridization showed identical patterns, suggesting the presence of a single rDNA unit copy per chromosome (25). However, most probes were assigned to single chromosomes, indicating a prevalence of low-copy-number nucleotide sequences within the very small genome of E. cuniculi. Both β-tubulin and aminopeptidase genes were located on two different chromosomes (the β-tubulin gene on chromosomes II and III, and the aminopeptidase gene on chromosomes I and VIII), whereas thymidylate synthase, dihydrofolate reductase, and serine hydroxymethyl transferase genes were located only on chromosome I, cdc2-like and ERCC6-like genes were located only on chromosome VIII (25), and an Hsp70 gene was located only on chromosome XI (275a).

The first sequence data of SSU-rRNA of a microsporidium, V. necatrix, were reported by Vossbrinck et al. in 1987 (363) (accession no. M24612). Subsequently, PCR amplification with primers complementary to conserved sequences of the V. necatrix SSU rRNA gene has been used to amplify and sequence DNA fragments of the SSU rRNA gene from several microsporidia that infect humans (164–166, 359, 364, 365, 393–396). Several nucleotide sequences of SSU and LSU rRNA genes including the intergenic spacer regions of different microsporidian species, including six species infecting humans (E. bieneusi, E. cuniculi, E. hellem, E. intestinalis, T. hominis, and V. corneae), have been published and are accessible via the GenBank and EMBL databases (October 1998) (Table 6).

TABLE 6 Human microsporidian genes inGenBanka

The rRNA sequences of E. bieneusi were first reported by Zhu et al. and deposited in the GenBank database under accession no. L07123 and L20290 (395). Gene amplification was performed with DNA extracted from human intestinal biopsy specimens from HIV-infected patients with electron microscopy-confirmed E. bieneusi infection. One primer pair was in a conserved region of the SSU rRNA (V1 and 1492), and a second was in a conserved region of the SSU rRNA (530f) and LSU rRNA (580r). Primer V1 is based on the sequence of the SSU rRNA gene of V. necatrix at the 5′ end, and primer 1492 is based on rRNA sequences highly conserved across many organisms at the 3′ end. An amplification product of about 1,300 bp was obtained, cloned, and sequenced (395). Resequencing of the SSU and LSU rRNA regions of different E. bieneusi isolates by other researchers (89, 165, 289, 361) (accession no. L16868, and AF024657) showed a 2.3 to 2.7% dissimilarity even in conserved regions with respect to the sequences reported by Zhu et al. Some of these differences may be caused by polymorphisms, but many are probably due to unresolved sequencing artifacts in the sequence reported by Zhu et al. (395).

E. intestinalis SSU rRNA sequences were reported by several researchers (10, 166, 360, 394, 396). Zhu et al. used primer pair V1 and 1492 to amplify a 1,300-bp DNA fragment from intestinal biopsy specimens with electron microscopy-confirmed E. intestinalis infection (394, 396). The DNA fragment was cloned and sequenced, and the obtained sequence was deposited in the GenBank database under accession no. L19567 (394, 396). Resequencing of the SSU rRNA region of different E. intestinalis isolates by other investigators (10, 166, 360) (accession no. U09929, L39113, L16866, and L16867) showed several dissimilarities to the sequence reported by Zhu et al. Although some may be caused by polymorphisms, dissimilarities are also reported in conserved regions of the SSU-rRNA, thus indicating that the E. intestinalis sequence of Zhu et al. may contain several sequencing errors.

The E. cuniculi SSU rRNA sequence was also first reported by Zhu et al. and was deposited in the GenBank database under accession no. Z19563 (393). They amplified the SSU-rRNA gene with primer pair V1 and 1492 from rabbit kidney cells (RK-13) infected with E. cuniculi in vitro. A 1,200-bp DNA fragment was amplified, cloned, and sequenced (393). Several further SSU and LSU rRNA sequences of different E. cuniculi strains have been deposited in the GenBank database, and there are significant differences between the SSU rRNA sequence determined by Zhu et al. and that obtained by others (10, 164, 359, 364) (accession no. L07255, L17072, L39107, X98469, X98470, and X98467) which includes also bases in conserved regions of the rRNA gene. Again, some dissimilarities may be caused by polymorphisms, but dissimilarities in conserved regions indicate that the sequence reported by Zhu et al. contained several sequencing errors.

The complete E. hellem SSU rRNA sequence data were first described by Visvesvara et al. (359). They used a primer pair based on nucleotides 1 to 21 (Micro-F [the nucleotide sequence of this primer is identical to that of primer V1]) and 1244 to 1218 (Micro-R) of the V. necatrix SSU rRNA gene to amplify, clone, and sequence the SSU rRNA gene of E. hellem, E. cuniculi, and E. intestinalis (359, 360). The obtained sequences are deposited in the GenBank database under accession no. L19070, L17072, and U09929.

SSU rRNA sequences of V. corneae (accession no. L39112 and U11046) and T. hominis (accession no. AJ002605) are available via the GenBank database as well.

Vossbrinck et al. (364) used highly conserved primers 530f in the SSU rRNA and 580r and 228r in the LSU rRNA to amplify the 3′ end of the SSU rRNA gene, the intergenic spacer, and the 5′ end of the LSU-rRNA gene of E. hellem (accession no. L13331) and E. cuniculi (accession no. L13332). The same primers were used by Zhu et al. (396) to amplify the same region of SSU and LSU rRNA genes including the intergenic spacer of E. bieneusi (accession no. L20290) and E. intestinalis (accession no. L20292) (396). The 3′ end of the SSU-rRNA gene, the intergenic spacer, and the 5′ end of the LSU-rRNA gene of several microsporidian species found in humans have now been sequenced by many researchers and are available via the GenBank or EMBL database (Table 6).

To date, complete LSU rRNA sequences have been determined only for E. cuniculi (accession no. AJ005581) and N. apis (accession no. U97150). The sizes were estimated to be about 2,487 and 2,480 bp, respectively, which is in the size range of prokaryotic 23S rRNA; this shortening is essentially due to truncation of divergent domains, with most of them being removed (158a, 275b). Most variable stems of the conserved core are also deleted, reducing the LSU rRNA to only the structural features preserved in all living cells (275b).

β-Tubulin sequences of E. cuniculi (accession no. L31807) and E. intestinalis (accession no. L47274) and α- and β-tubulin sequences of E. hellem (accession no. U66908, L47271, and L31808) were amplified with primers corresponding to conserved sequences (125, 126, 188, 189, 192, 221). Southern blots indicate that E. cuniculi, E. hellem, and E. intestinalis possess a single β-tubulin gene copy and that the three β-tubulin sequences are very similar to each other (125, 188, 189).

α-Tubulin sequences of Nosema locustae (accession no. U66907) and Spraguea lophii (U66906) have been determined and are available via the GenBank or EMBL databases (192).

DNA isolation from microsporidian spores is more difficult, and harsh conditions must be used to destroy the spore wall. Mechanical disruption of the spores with glass beads in combination with proteinase K digestion is commonly used (109, 110, 134). Some authors recommend additional incubation with enzymes that resolve chitin (lyticase or chitinase) (134) or boiling the samples to release DNA from spores (260, 262).

If stool samples are used, treatment of stool with 0.5% sodium hypochlorite, 10% formalin, or 1 M KOH is recommended to remove PCR inhibitors (134, 190, 222), whereas others have reported that inhibition can be nullified by simple dilution of the samples (262).

TABLE 7 Primer pairs and hybridization probes used for diagnosis and species differentiation of humanmicrosporidia

TABLE 8 Primer pairs used for phylogenetic analysis of humanmicrosporidia

Primer pairs and hybridization probes for E. bieneusi. PCR diagnosis of E. bieneusi was first reported by Zhu et al. (395). Primer pair V1 and EB450 amplified cloned E. bieneusi SSU rRNA sequences and DNA from E. bieneusi-infected tissues but also amplified E. hellem DNA from cell cultures (81, 145, 149, 381, 395). Specificity was tested with DNA prepared from Toxoplasma gondii, Trypanosoma cruzi, Escherichia coli, Saccharomyces cerevisiae, E. intestinalis, V. necatrix, E. cuniculi, Glugea stephani, N. locustae, N. bombycis, Pleistophora, and E. hellem and with gastrointestinal biopsy specimens infected with E. cuniculi, E. intestinalis, Cryptosporidium spp., Giardia lamblia, Mycobacterium avium, and Candida albicans (81, 145, 381, 395). Only weak amplification of DNA prepared from E. hellem was seen (395). In other studies, these primers did not amplify DNA from E. hellem (81). Based on the SSU rRNA sequence of E. hellem, amplification should not occur. It is likely that the amplification observed by Zhu et al. (395) was due either to contamination of the original E. hellem culture with spores of E. bieneusi (which were being cultivated in the same incubator) or to contamination of the culture with a plasmid containing the cloned E. bieneusi SSU rRNA gene (81). The V1-EB450 primer pair reliably amplifies E. bieneusi DNA from gastrointestinal biopsy specimens from patients with light and electron microscopy-confirmed E. bieneusi infection (81, 145, 149, 381, 395) and from stool specimens from patients with light microscopy-confirmed intestinal microsporidiosis (262, 339a). Further evaluation of this primer pair gave negative results with one bile sample that was confirmed to contain E. bieneusi by electron microscopy and with E. bieneusi derived from a short-term culture (89). Resequencing of the SSU rRNA region of different E. bieneusi isolates by other researchers showed a 2.3 to 2.7% dissimilarity to the sequence reported by Zhu et al. (395), resulting in a base mismatch of primer EB450 (position 9 of the EB450 primer has G in place of C) (89). The upstream primer V1 is not specific for E. bieneusi but is directed against a conserved sequence for many microsporidian species (381). The base mismatch of primer EB450 combined with this lack of specificity may lead to negative results with some samples. An internal 30-mer oligonucleotide, EB150, complementary to a region of the amplicon produced by V1 and EB450 has been used as a probe for Southern blot hybridization analysis of the PCR products (81, 145, 149, 381, 395) by isotopic and nonisotopic procedures (145, 149, 395). Probe EB150 hybridized with E. hellem DNA when amplified by V1 and EB450 (149, 395); again, it is likely that this was due either to contamination of the original E. hellem culture with spores of E. bieneusi or to contamination of the culture with a plasmid containing the cloned E. bieneusi SSU rRNA gene (81).

The primer pair V1-EB450 and a second pair, V1-Mic3 (Mic3 is located at positions 445 to 427 of the SSU rRNA gene of E. bieneusi), have been used to confirm E. bieneusi infection in simian immunodeficiency virus-infected rhesus monkeys. The specificity of the PCR products was confirmed by DNA sequencing and restriction fragment length polymorphism analysis with MnlI and DdeI restriction endonucleases (343). The primer pair V1-Mic3 has also been used to amplify a 446-bp fragment from the E. bieneusi SSU rRNA gene of HIV-infected patients with intestinal microsporidiosis due to E. bieneusi (63).

The primer pair EBIEF1 and EBIER1 described by Da Silva et al. (89) amplified cloned E. bieneusi SSU rRNA sequences and DNA from bile fluid, a duodenal aspirate from an AIDS patient with electron microscopy-confirmed E. bieneusi infection, and a short-term culture of an E. bieneusi isolate. Specificity was tested with human DNA as well as with cloned microsporidian SSU rRNA coding regions of 13 species of microsporidia, and positive amplification was shown only with the E. bieneusi cloned template. PCR products generated with this primer pair were analyzed only by ethidium bromide-stained gel analysis, and the identity of the PCR products was not confirmed by Southern blot hybridization, DNA sequencing, or restriction enzyme digestion (89). Estimating the size of the PCR-generated fragment by gel electrophoresis alone is usually not sufficient to confirm the specificity of a PCR. Every primer that is nonspecifically elongated at its 3′ end because of unwanted annealing to partially complementary sequences not only is lost for specific priming but also enhances further nonspecific amplifications. This may lead to false-positive results, which should be ruled out by the confirmation techniques mentioned above.

An E. bieneusi-specific primer pair and hybridization probe were designed by Schuitema et al. (303) and further evaluated by David et al. (91) and Liguory et al. (222). The primer pair and probe were used for species differentiation for samples that were found to be PCR positive by using another primer pair, which amplifies different microsporidian species (see below) (91), and for diagnosis of intestinal infection with E. bieneusi by using DNA from stool samples (222). Examination of 31 stool specimens from 26 patients with intestinal E. bieneusi infection, 6 stool specimens from 3 patients with E. intestinalis infection, and 61 stool specimens from 45 patients without intestinal microsporidiosis showed a sensitivity and specificity of 100% (222). When this primer pair was used with target DNA from 25 gastrointestinal biopsy specimens from patients with confirmed E. bieneusi infection, 23 samples were found to be positive; no amplification was seen with DNA from three biopsy specimens from patients with confirmed E. intestinalis infection (91).

The unique rRNA intergenic spacer sequence of E. bieneusi was used to design a primer pair, Eb.gc and Eb.gt, which amplified a 210-bp DNA fragment (357). This primer pair was used to amplify DNA from gastrointestinal biopsy specimens, stool samples, and duodenal aspirates of patients with E. bieneusi infection but not from material infected with E. intestinalis, Isospora belli, Cryptosporidium spp., or G. lamblia. The identity of the amplified PCR products was not confirmed by Southern blot hybridization, DNA sequencing, or restriction enzyme digestion. Again, false-positive results which may occur during unwanted annealing of primers to partially complementary sequences should be ruled out by the confirmation techniques mentioned above.

To date, in situ hybridization has only occasionally been used to diagnose infections with microsporidia. A 446-bp PCR-product, amplified with the E. bieneusi-specific primer pair V1-Mic3, was labeled with digoxigenin–11-dUTP by a random-priming reaction and used for in situ hybridization of paraffin-embedded jejunal biopsy specimens from patients with intestinal E. bieneusi infection (63). The same technique was used with necropsy-derived tissue samples from simian immunodeficiency virus-infected monkeys after experimental E. bieneusi infection (343). Histologic examination of tissue sections failed to identify microsporidial infection, but in situ hybridization with the E. bieneusi probe revealed characteristic supranuclear staining of scattered villous-tip epithelial cells within the jejunal mucosa, which was also confirmed by electron microscopy (343).

Primer pairs and hybridization probes for E. intestinalis. A primer pair, V1 and SI500, described by Weiss et al. (381) amplified cloned E. intestinalis SSU rRNA sequences as well as DNA from E. intestinalis-infected tissues, body fluids, stool samples, and cell cultures (81, 148–150, 260, 262, 381). These primers did not amplify cloned E. bieneusi SSU rRNA sequences or DNA from E. bieneusi- or E. cuniculi-infected tissue samples (81, 148). The specificity of this primer pair was further tested with DNA prepared from a variety of microsporidian species, S. cerevisiae, and E. coli, and no amplification was observed (81). To confirm the identity of the PCR products, an internal 18-mer oligonucleotide (SI60), that was used by Weiss et al. as upstream primer in combination with the reverse primer SI500 (381), was used for isotopic and nonisotopic Southern blot hybridization by others (81, 148–150). Another internal 30-mer oligonucleotide was used as a hybridization probe by others (262). Although the primer pair V1-SI500 reliably amplified DNA products of the correct size from tissues from patients with electron microscopy-confirmed E. intestinalis infection, Franzen et al. (148) detected a PCR product in biopsy specimens from patients with E. bieneusi infection and in a biopsy specimen from a patient with E. cuniculi infection. The PCR products hybridized with the E. intestinalis-specific probe (SI60), and partial sequencing of the amplified DNA fragments showed high homology (96 to 100%) to published E. intestinalis sequences. This confirmed that the amplified PCR products were derived from the SSU rRNA gene of E. intestinalis and provided genetic evidence for latent E. intestinalis infection in these patients (148). These results were reinforced in another study which used primer pair V1-SI500 in combination with the hybridization probe SI60 (149). A total of 46 HIV-infected patients with diarrhea were studied, and PCR gave positive results for E. intestinalis in 10 patients and indicated five cases of double infection with E. intestinalis and E. bieneusi. Histologic examination by light microscopy showed microsporidian spores in all cases, but light microscopy was unable to distinguish between species in almost all cases; unfortunately, electron microscopy results were available only for 3 of the 10 patients (149). Primer pair V1-SI500 in combination with the hybridization probe SI60 was also used to detect E. intestinalis DNA in blood from an HIV-infected patient with disseminated infection (150).

Primer pair SINTF1 and SINTR, described by Da Silva et al. (90), amplified cloned E. intestinalis SSU rRNA sequences and DNA from a duodenal-jejunal segment from a patient with AIDS and suspected intestinal microsporidiosis and from eight different cultured samples of E. intestinalis. Specificity was tested with cloned microsporidian SSU rRNA coding regions of E. cuniculi, E. hellem, E. intestinalis, E. bieneusi, and V. corneae; positive amplification was shown only with the E. intestinalis SSU rRNA as the template. This primer pair was also used for species differentiation of microsporidia in stool samples from several animals (32a). PCR products generated by this primer pair were analyzed only by ethidium bromide-stained gel analysis, and the identity of the PCR products was not confirmed by Southern blot hybridization, DNA sequencing, or restriction enzyme digestion. Again, false-positive results cannot be ruled out without confirmation as mentioned above.

An E. intestinalis-specific primer pair and hybridization probe were designed by Schuitema et al. (303) and were further evaluated by David et al. (91) and Liguory et al. (222). The primer pair and probe were used for species differentiation of samples found to be PCR positive with a primer pair that amplified different microsporidian species in a previous PCR (see below) (91) and for diagnosis of intestinal infection with E. intestinalis by using DNA from stool samples (222). Examination of 6 stool specimens from 3 patients with E. intestinalis infection, 31 stool specimens from 26 patients with intestinal E. bieneusi infection, and 61 stool specimens from 45 patients without intestinal microsporidiosis showed a sensitivity and specificity of 100%, but the number of stool samples from patients with E. intestinalis infection was very small in this study (222). When this primer pair was used with target DNA from three gastrointestinal biopsy specimens from patients with confirmed E. intestinalis infection, all samples were found to be PCR-positive and no amplification was seen with DNA from 25 biopsy specimens from patients with confirmed E. bieneusi infection. Again, the number of biopsy samples from patients with E. intestinalis infection was very small in this study (91).

Another E. intestinalis-specific primer pair, V1 and Sep1, has been used to exclude E. intestinalis infection in patients with E. bieneusi-infected stool samples (63) and in simian immunodeficiency virus-infected rhesus monkeys with experimental E. bieneusi infection (343).

Primer pairs for E. hellem and E. cuniculi. Two primer pairs specific for either E. cuniculi (positions 344 to 364 and 872 to 892 of the E. cuniculi SSU rRNA sequence [ECUN-F and ECUN-R]) or E. hellem (positions 358 to 378 and 884 to 904 of the E. hellem SSU rRNA sequence [EHEL-F and EHEL-R]) were designed for species-specific amplification of DNA fragments by Visvesvara et al. (359, 360). These two primer pairs were used for species differentiation of cultured organisms and organisms present in various clinical samples (95, 297, 298, 359, 360).

Two primer pairs, originally designed for genus- and species-specific detection of Echinococcus multilocularis, surprisingly also amplified DNA of the SSU rRNA gene of E. cuniculi (154, 255). The PCR products amplified with these two primer pairs were sequenced, and it was shown that the DNA of E. cuniculi carried segments that are characteristic for the genus Echinococcus and the species Echinococcus multilocularis. Perhaps E. cuniculi infected host animals of E. multilocularis or E. multilocularis larvae, so that the DNA of E. cuniculi may unknowingly have been cloned and sequenced in a former analysis, or some genome sequences may be shared between E. cuniculi and E. multilocularis (154, 255).

General primer pairs and hybridization probes for several microsporidian species. Primer pair PMP1 (the nucleotide sequence of this primer is identical to that of primer V1) and PMP2 amplified four microsporidian pathogens that infect humans (E. bieneusi, E. hellem, E. cuniculi, and E. intestinalis) (134). This primer pair was useful for the detection of microsporidian DNA from cultured organisms and stool samples as well as from several other clinical samples including gastrointestinal biopsy specimens (123, 134). These primers also amplified V. corneae from culture (134). To confirm the identity of PCR amplicons, restriction enzyme digestion of amplified PCR products with PstI and HaeIII was used. Since E. bieneusi does not have a PstI restriction site in the amplified region but PstI cuts the Encephalitozoon amplicons into two fragments, E. bieneusi could be easily differentiated from Encephalitozoon spp. However, E. intestinalis could not be differentiated from E. cuniculi (134), thus limiting the use of this primer pair for species differentiation. The same primer pair, PMP1-PMP2, was used by Dowd et al. (123), and the detection sensitivity was two spores in vitro culture-derived spores. Restriction fragment length polymorphism analysis after digestion of the PCR product with PstI was used to differentiate E. bieneusi from Encephalitozoon spp., whereas species differentiation of Encephalitozoon spp. was done by DNA sequencing (123). The same authors used this primer pair for the detection of E. intestinalis, E. bieneusi, and V. corneae in environmental water samples as well (123a).

A similar approach was chosen by Raynaud et al. (286), who used the primer pair C1 (partial sequence of primer V1) and C2 to amplify a conserved region of the SSU rRNA gene of four microsporidia found in humans (E. bieneusi, E. hellem, E. cuniculi, and E. intestinalis). To confirm the identity of the amplified 1,200-bp PCR products, restriction enzyme digestion with HindIII and HinfI was used. The amplified PCR products displays one HindIII restriction site in E. bieneusi and none in any Encephalitozoon spp. On the other hand, Encephalitozoon spp. could be differentiated on the basis of the number of HinfI restriction sites: one for E. cuniculi, two for E. hellem, and three for E. intestinalis. The technique was used to confirm E. intestinalis infection diagnosed by light microscopy in two non-HIV-infected travelers (286).

Another primer pair that amplified DNA from four known human microsporidian species was designed by Schuitema et al. (303) and further evaluated by David et al. (91). For species differentiation, digoxigenin-labeled E. bieneusi- and E. intestinalis-specific probes were used in a nonisotopic hybridization assay. Two species-specific primer sets for E. bieneusi and E. intestinalis were used in a second PCR (91, 222, 303). About 10 copies of recombinant plasmids carrying amplified SSU rRNA genes from E. bieneusi and E. intestinalis were detected by using the primer pairs and the hybridization assay. Of 28 intestinal biopsy specimens from patients with electron microscopy-confirmed infection due to E. bieneusi or E. intestinalis, 26 were considered PCR positive, and species differentiation was correctly done in all cases. No false-positive results were seen in 23 intestinal biopsy specimens from patients without intestinal microsporidiosis (91).

A set of pan-Encephalitozoon primers, int530f and int580r, selected from sequences conserved between E. cuniculi and E. hellem, was first described by Schuitema et al. (303) and was further used by Didier et al. (108–111). This primer pair amplified a product of approximately 1,000 bp that includes a large portion of the SSU rRNA gene, the intergenic spacer, and a small portion of the LSU rRNA gene of E. hellem, E. cuniculi, and E. intestinalis, but did not amplify DNA from culture-derived V. corneae or intestinal biopsy-derived E. bieneusi as the target (108). Species differentiation was done by Southern blot hybridization with a nonradioactive, chemiluminescent 3′-oligolabeling and detection system involving species-specific hybridization probes specific for E. cuniculi, E. hellem, and E. intestinalis (109). The specificity of the three hybridization probes was determined by using culture-derived E. cuniculi, E. hellem, E. intestinalis, and V. corneae isolates (109), but the sensitivity was not determined, and the clinical utility is limited because the primer pair did not amplify E. bieneusi DNA.

A nested PCR assay which detects four microsporidian species that infect humans (E. bieneusi, E. hellem, E. cuniculi, and E. intestinalis) and an additional four species not known to be pathogenic to humans (V. necatrix, V. lymantriae, Ameson [Nosema] michaelis, and Ichthyosporidium giganteum) was described by Katzwinkel-Wladarsch et al. (190, 191). Stool samples spiked with E. cuniculi spores and stool samples, nasal discharge, urine, sputum, and cerebrospinal fluid from patients with light microscopy-confirmed E. bieneusi, E. hellem, E. cuniculi, or E. intestinalis infection were assayed. Two upstream primers, MSP-1 and MSP-3, complementary to the 3′ region of the SSU rRNA gene of the microsporidian species mentioned above and four downstream primers, MSP-2B and MSP-4B for E. bieneusi and MSP-2A and MSP-4A for the other mentioned microsporidian species, complementary to the 5′ region of the LSU rRNA gene were used (190, 191). E. bieneusi was distinguished from Encephalitozoon spp. because the PCR products differed in size, and E. cuniculi could be distinguished from E. intestinalis by restriction endonuclease digestion with MnlI, which resulted in distinct restriction fragment patterns. The results were confirmed by DNA sequence analysis, and the generated sequences were homologous to previously published DNA sequences. The limit of detection varied between 3 and 100 spores per 0.1 g of stool (190). Comparison of this PCR assay with light microscopy on 34 specimens from 31 HIV-infected patients produced identical results in 82% of specimens (28 of 34) (191). Of the discrepant results, four samples were microscopy negative and PCR positive and two were microscopy positive and PCR negative. Species differentiation was 100% accurate. These primer pairs were also used for amplification of E. bieneusi DNA sequences from fecal samples of pigs with suspected E. bieneusi infection confirmed by light microscopy (191). The intergenic spacer region of the amplified PCR products was sequenced and compared with published sequences of E. bieneusi, and the DNA sequence of the amplified intergenic spacer region (accession no. U61180) showed 97% identity to the corresponding sequence of E. bieneusi (191).

Another nested PCR assay which detects four microsporidian species that infect humans (E. bieneusi, E. hellem, E. cuniculi, and E. intestinalis) was developed by Kock et al. (202). A PCR with upstream primer Mic3U and downstream primer Mic421U was used to amplify 410- to 433-bp DNA fragments of the SSU rRNA gene of all the above mentioned microsporidia. A nested PCR with upstream primer Mic266 and downstream primers Eb379, Ec378, Eh410, and Ei395 was used to amplify species-specific, 113- to 134-bp DNA fragments. The assay was tested with culture-derived spores of E. cuniculi and E. hellem and with stool samples spiked with culture-derived spores. Stool samples from HIV-infected patients with electron microscopy-confirmed E. bieneusi (n = 8), Encephalitozoon spp. (n = 2), and E. intestinalis (n = 1) infection and intestinal biopsy specimens from three patients with electron microscopy-confirmed E. bieneusi (n = 2) and E. intestinalis (n = 1) infection were also examined (202). Without exception, the PCR assay verified electron microscopy-confirmed infections in stool samples and biopsy specimens of all patients as well as in spiked stool samples and tissue cultures (202).

Strain differentiation of Encephalitozoon spp. and E. bieneusi. Molecular techniques have confirmed the existence of different strains of E. cuniculi. Didier et al. (108) first demonstrated differences in the intergenic spacer region of eight E. cuniculi isolates from mice, rabbits, and dogs. Two primer pairs, int530f-int580r and 530f-580r, were used to amplify the 3′ end of the SSU rRNA gene, the intergenic spacer, and the 5′ end of the LSU rRNA gene of E. cuniculi (108). Using a double-stranded DNA heteroduplex mobility assay and restriction fragment length polymorphism after digestion with the restriction endonuclease FokI, they found differences in the isolates from mice, rabbits, and dogs. After DNA sequencing of the intergenic spacer region, the isolates were shown to differ by a small repetitive sequence of 5′-GTTT-3′. This sequence was repeated twice in two isolates from mice, three times in three isolates from rabbits and in one isolate from a mouse, and four times in the two isolates from domestic dogs. These differences were used to characterize a rabbit strain (strain I), a mouse strain (strain II), and a dog strain (strain III). The three strains also differed antigenically by sodium dodecyl sulfate-polyacrylamide gel electrophoresis and Western blotting.

Hollister et al. (181) also studied the intergenic spacer region of three isolates of E. cuniculi: a putative mouse isolate (strain D. Owen [accession no. X98467 and X98468]), a dog isolate (strain Stewart [accession no. X98469]), and a human isolate (strain Donovan [accession no. X98466 and X98470]) established in vitro from the urine of a patient with AIDS (1, 177, 179). In the intergenic spacer region, four tetranucleotide repeats 5′-GTTT-3′ were found for the dog and human isolates and three were found for the mouse isolate (181). Another human E. cuniculi isolate from a patient with AIDS (95) also corresponded to dog strain III, with four tetranucleotide repeats (111, 189).

Deplazes et al. (97) used primers that corresponded to positions 1 to 19 (which are the first 19 bases of primer V1) and 1277 to 1299 of the SSU rRNA gene sequence of E. cuniculi to amplify SSU rRNA gene fragments from organisms cultured from six HIV-infected patients infected with E. cuniculi, E. hellem, and E. intestinalis and from nine rabbits infected with E. cuniculi. PCR products were cleaved with restriction enzymes (MboI and HpaII) and analyzed on ethidium bromide-stained gels. Restriction fragment length polymorphisms of all E. cuniculi isolates from humans did not differ from the rabbit isolates (97). Further analysis of the intergenic spacer region of these six E. cuniculi isolates from humans and of different E. cuniculi isolates of animal origin was done by Mathis et al. (236). Five E. cuniculi isolates from Swiss patients were found to be homologous to isolates from rabbits, whereas the sixth isolate, from a patient from Mexico, corresponded to dog strain III (236).

In the light of these data (summarized in Table 9), dogs or rabbits might be the source of human infection. However, it is difficult to understand how host-related differences can be maintained in microsporidial isolates when strains can be transmitted among different types of hosts, at least experimentally. Too few isolates of E. cuniculi have been obtained and analyzed so far to assess host specificity or to determine if humans are at risk of infection by exposure to dogs or rabbits. More data on strains from many different hosts including humans are needed, before the epidemiology of E. cuniculi is clarified.

TABLE 9 Number of 5′-GTTT-3′ repeats in the intergenic spacer region of different E. cuniculi isolates from different hosts and geographicalareas

E. hellem has been found once in a bird, and E. intestinalis has recently been detected in several mammals. It remains unclear if these animals are relevant reservoirs of the parasites. Double-stranded DNA heteroduplex mobility assays, restriction fragment length polymorphisms after digestion with the restriction endonucleases, and DNA sequencing of the intergenic spacer between the SSU rRNA and the LSU rRNA of different E. hellem and E. intestinalis isolates from different geographical regions (Europe and America) showed no differences among these isolates (109, 110, 364). However, Schnittger et al. recently reported genotypic differences of the SSU and LSU rRNA coding regions in E. intestinalis (GenBank accession no. Y11611). Again, too few isolates of E. hellem and E. intestinalis have been analyzed so far to determine if there are different strains of these species. Additional isolates must be analyzed to help define the epidemiology of these microsporidian species.

Three distinct genotypes of E. bieneusi have been identified by Rinder et al. (289) by amplification of a 484-bp DNA fragment between primers MSP-3 and MSP-4B from 12 stool samples of eight HIV-infected patients (289). Nine polymorphic sites were found in the 243-bp intergenic spacer region of the rRNA gene. The genotypes were stable in samples taken during 11 weeks of infection. The intergenic spacer rRNA E. bieneusi sequence reported by Zhu et al. (accession no. L20290) (396) was not considered to be a fourth genotype by the authors. This sequence is only 86.8% identical to those of the other three genotypes; this may be caused not only by polymorphisms but also by several sequencing errors. Liguory et al. (222a) identified four genetically unrelated lineages among 78 different isolates of E. bieneusi from 65 patients with intestinal microsporidiosis. Strain differentiation was done by restriction fragment length polymorphism with NlaIII and Fnu4HI after amplification of the intergenic spacer region with the primer pair Eb.gc and Eb.gt. Type I strains were found in the majority of samples (78%), whereas type II, III, and IV strains were found only in 12, 5, and 5% of the samples, respectively. The same strains were found in sequential stool specimens from the same patients. DNA sequences were determined only for strain I, which showed 100% homology to the E. bieneusi sequence of Zhu et al. (accession no. L20290) (396), and for strain II, which showed seven mismatches with this sequence (97% identity) (222a). rRNA sequences of E. bieneusi isolates from humans, a pig (accession no. U61180), and rhesus macaques (accession no. AF023245) exhibit additional variability over the intergenic spacer region and the 5′ end of the LSU rRNA gene (62a). The rRNA sequence of the pig belongs to the type IV lineage, which has been found in only 5% of humans (222a).

When rRNA genes are used to discriminate among strains, several problems must be considered. The presence of different genotypes for individual isolates does not automatically warrant their consideration as different strains, and different copies of a gene within one genome may not be identical (289). It is not clear whether different prevalences of strains in humans reflect different sources of strains that lead to human infection or are related to the pathogenicity of strains in particular types of hosts. However, comparisons of different strains from humans, animals, and environmental sources will provide a better understanding of the epidemiology of microsporidial infection.

Comparison of molecular techniques with light microscopy. In-house validations of PCR protocols for the detection of microsporidia, especially by those who developed them, are generally satisfactory to excellent. However, until recently none of these techniques had been validated in a blinded, externally controlled fashion. In a multicenter evaluation of the detection of microsporidia in stool specimens by light microscopy and PCR, quality parameters were determined with identical sets of 50 stool samples for six laboratories using light microscopy and six laboratories using their own PCR protocol (290). A total of 14 clinical samples contained E. bieneusi spores, and 14 samples were spiked with spores of E. hellem, E. cuniculi, and E. intestinalis at different concentrations ranging from 10² to 10⁶ spores/g; 22 samples were negative. The average overall sensitivity was 67% (36 to 96%) for laboratories using PCR and 54% (25 to 71%) for laboratories using light microscopy, and the specifities were 98 and 95%, respectively. A uniform detection limit of between 10⁴ and 10⁶ spores/g of stool was apparent for light microscopy, while PCR detected concentrations as low as 10² spores/g. However, the greatest differences were found between individual laboratories rather than between the two techniques used (290). These apparent differences in quality between laboratories suggest that individual laboratories first should improve their established technique as much as possible and not give up one method in favor of the other. Nevertheless, it will certainly be helpful to have a second technique available when confirmation of the first is desired. Since the threshold of detection for light microscopy seems to be between 10⁴ and 10⁶ spores/g of stool, epidemiological studies involving only light microscopy may give prevalence data which did not reflect the true prevalence of microsporidia. Application of PCR to stool samples offers an approach that can be adapted for processing large numbers of specimens to define the true prevalence of intestinal microsporidiosis in human populations and to determine whether microsporidia are present in the general population (272).

Because all the participating laboratories in this trial remained anonymous, no recommendations about DNA isolation techniques, primer pairs, or PCR methods can be given. Nevertheless, testing all clinical samples with separate species-specific primer pairs is a time-consuming task. An efficient approach for PCR detection of microsporidia in clinical samples would involve using general microsporidian primer pairs that amplify several species. The species of microsporidia detected could then be determined by PCR with species-specific primer pairs. Primers V1 and PMP2 amplify microsporidian DNA of different microsporidian species. Primer V1 is used by several researchers, and primer PMP2 has also been evaluated by different groups (123, 134, 149). Other published general microsporidian primer pairs (190, 202, 286, 303) have not been evaluated by others so far. The primer pairs V1-EB450 and V1-SI500, although constructed by using SSU rRNA sequences which appear to contain several sequencing errors, are best evaluated for diagnosis of microsporidiosis due to E. bieneusi and E. intestinalis. Several researchers have published studies with these primer pairs, usually with excellent results (81, 145, 148–150, 260, 262, 339a, 381, 395).

16S rRNA genes are found in all prokaryotic organisms and accumulate mutations at a low, constant rate over time. Hence, they may be used as “molecular clocks” (258, 387). Highly variable portions of the SSU-rRNA genes provide unique signatures for any organism and useful information about possible relationships (387). Several groups have been using molecular approaches to develop taxonomic systems for microsporidia based on rRNA genes and also on protein-coding DNA sequences. This analysis has yielded insights into new relationships among microsporidia, which are being continuously integrated into a revised taxonomy. For example, it has been shown that Nosema trichoplusiae is a synonym for Nosema bombycis based on the sequence of the SSU rRNA coding region (276). Therefore, new molecular techniques have led not only to the discovery of new organisms but also to the elimination of an old organism.

The first microsporidian DNA sequence data were reported by Vossbrinck et al. (363). The SSU rRNA gene of the insect microsporidian V. necatrix was sequenced, and a phylogenetic tree was then constructed. The root of the eukaryotic tree lay between the V. necatrix rRNA sequence and those of all other eukaryotic lineages for which sequences were available, showing that the V. necatrix sequence is consistent with a very early branch in the eukaryotic line of descent (363).

rRNA sequences for phylogenetic construction of microsporidia pathogenic to humans were also first used by Vossbrinck et al. (364). They used a highly conserved primer pair, 530f in the SSU rRNA and 580r in the LSU rRNA, to amplify a DNA fragment of about 1,350 bp that contains the 3′ end of the SSU rRNA, the intergenic spacer, and the 5′ end of the LSU rRNA of E. hellem, E. cuniculi, and V. corneae obtained from tissue culture (364). Restriction digests of the amplified DNA fragments with SauIIIA, EcoRI, DraI, and HinfI showed clear differences among the three species. Three regions of the amplified DNA fragment were sequenced with primer 228r (in the LSU rRNA) in addition to primers 530f and 580r. The sequence difference between E. hellem and E. cuniculi was 23.1%, while the sequence difference between the two Encephalitozoon spp. and other studied families (Vairimorpha and Ichtyosporidium) ranged between 54.2 and 56.2% (364). These data indicate that E. hellem and E. cuniculi are different organisms which are closely related.

The same primers (530f, 580r, and 228r) were used by Zhu et al. (396) for phylogenetic analysis of E. bieneusi, E. intestinalis, and A. michaelis. E. bieneusi and E. intestinalis DNA was extracted from intestinal biopsy specimens from patients with electron microscopy-confirmed microsporidial infection; spores of A. michaelis were obtained from the muscle of blue crabs (Callinectus sapidus). Following amplification and sequencing of PCR product, sequences were aligned with those of several other microsporidian species including E. hellem and E. cuniculi. By using three sequenced regions, Zhu et al. calculated that there was a 16.7% sequence difference between E. hellem and E. cuniculi and that there were 21.9 and 26.4% sequence differences between E. intestinalis and E. cuniculi and between E. intestinalis and E. hellem, respectively (396). These differences calculated by Zhu et al. (396) differed from those reported by Vossbrinck et al. (364), perhaps due to differences in alignment or to the version of PAUP (phylogenetic analysis using parsimony) used. Based on their data, Zhu et al. (396) stated that E. cuniculi, E. hellem, and E. intestinalis are closely related, most likely at the family level, with E. cuniculi and E. hellem belonging to the same genus and E. intestinalis belonging to the Encephalitozoonidae family but as a separate genus. Much larger differences were found between the other microsporidia. Since E. bieneusi differed from E. intestinalis by 45.7%, from E. cuniculi by 43.2%, and from E. hellem by 42.2%, it is clear that E. bieneusi is placed in a separate family (396).

Baker et al. presented a phylogeny of various species of microsporidia based on sequence data of the whole SSU rRNA gene (10). DNA sequences either were obtained by PCR with primer pair 18f (partial sequence of primer V1) and 1537r, complementary to conserved regions of the SSU rRNA gene at the 3′ and 5′ ends, followed by sequencing with several internal primers or were obtained from the GenBank database. Results of parsimony analysis with PAUP suggested that the 16 studied microsporidian species could be divided into four distinct groups referred to as the Ichtyosporidium group, the Encephalitozoon group, the Endoreticulatus group, and the Vairimorpha/Nosema group (Fig. 13). Delineation of Vairimorpha/Nosema as a group was additionally supported by LSU rRNA data obtained from several other species of Nosema and Vairimorpha (9). From the branching pattern in Fig. 13 and the sequence difference between V. corneae (formerly N. corneum) and N. bombycis (27.9%) it seems clear, that V. corneae is unrelated to the other Nosema spp. Baker et al. (9, 10) placed V. corneae in the Endoreticularis group because the parasite is most closely related to Endoreticulatus schubergi (7.3% distance). This supports the reclassification of N. corneum to a new genus as V. corneae which was done based on the ultrastructure of developmental stages in the livers of experimentally infected atymic mice (the diplokaryotic arrangement of the nuclei was the only character that conformed to the description of the genus Nosema) (324). Baker et al. (10) placed in the Encephalitozoon group the genera which Cali et al. placed in the family Encephalitozoonidae (47). By using parsimony (branch-and-bound option of PAUP), E. intestinalis was most closely related to E. cuniculi (Fig. 13), and by using distance methods (unweighted pair group method with arithmetic mean [UPGMA]), E. hellem was most closely related to E. intestinalis (mean distance values, E. intestinalis/E. hellem = 0.043, E. intestinalis/E. cuniculi = 0.059, E. hellem/E. cuniculi = 0.066), but in no case was E. hellem most closely related to E. cuniculi (10). Because of the close relationship between the three Encephalitozoon spp., Baker et al. proposed that E. intestinalis, originally described as Septata intestinalis by Cali et al. (47), be designated Encephalitozoon intestinalis and that the morphological definition of the genus be changed to reflect this designation (10).

FIG. 13 Tree representing the phylogenetic relationship of several microsporidian species as determined by SSU rRNA sequence analysis. This was the most pasimonious tree found by using the branch-and-bound option of PAUP. Reprinted from reference 101 with permission (more ...)

Similar conclusions were drawn by Hartskeerl et al. on the basis of SSU rRNA sequence data (166). The SSU rRNA gene of E. intestinalis, E. cuniculi, E. hellem, and E. bieneusi was amplified with primers complementary to conserved regions at the 3′ and 5′ end of the SSU rRNA gene (164, 166). Restriction enzyme digestion of the amplified DNA with HphI and BsiHKAI produced different restriction fragment length polymorphism patterns for all the species examined. The degree of sequence homology was established by a pairwise comparison with the Bestfit program. SSU rRNA gene sequences of E. cuniculi and E. hellem showed a lower degree of homology (89.7%) than did those of E. cuniculi and E. intestinalis (90.7%) or E. hellem and E. intestinalis (91.4%). The E. intestinalis sequence was only approximately 70% identical to those of microsporidia of other genera such as E. bieneusi or V. necatrix. Although sequence dissimilarity values are not suitable methods for phylogenetic analysis, the authors concluded that E. cuniculi, E. hellem, and E. intestinalis belong to the same genus and proposed the reclassification of Septata intestinalis to Encephalitozoon intestinalis (166).

The data reported by Hartskeerl et al. (166) and Baker et al. (10) differed significantly from those of Zhu et al. (396). Hartskeerl et al. and Baker et al. placed E. intestinalis within the genus Encephalitozoon, while in the phylogeny of Zhu et al., E. intestinalis was placed in a genus of its own (as originally proposed by Cali et al. [47]). There are several possible explanations for this. First, Zhu et al. used sequence data of the 3′ and 5′ ends of the SSU and LSU rRNA genes and the intergenic spacer for their phylogenetic analysis, whereas Hartskeerl et al. and Baker et al. used the sequence of the whole SSU rRNA gene. However, Zhu et al. stated that their results were unchanged when the whole SSU rRNA gene sequence was used. Second, the alignment methods differed. Phylogenetic analysis depends on comparison of homologous characters, and only regions in which exact alignment is possible should be used. Third, and this may be the most important point, the E. cuniculi, E. hellem, and E. intestinalis sequences obtained by Zhu et al. (393–396) differed significantly even in conserved regions of the rDNA from those reported by Baker et al. (10) and by Hartskeerl et al. (164, 166).

Cali et al. (49) did not agree with the reclassification of Septata intestinalis into Encephalitozoon intestinalis. The sporogony of this species is tetrasporous (47). On the other hand, there are no supporting published data for tetrasporous development of E. cuniculi, the type organism from which the genus was established. Until E. cuniculi is demonstrated to be tetrasporous, the genus Encephalitozoon cannot be modified to include this feature, thus excluding Encephalitozoon (Septata) intestinalis. They also noted that the morphological presence of secretions from E. intestinalis and the lack of such secretions in E. cuniculi was not addressed in the reclassification of E. intestinalis by Hartskeerl et al. (166) and Baker et al. (10). In other microsporidia, this feature has been given taxonomic value. For example, some genera in the family Pleistophoridae were created on the basis of variations in secretion (49). Based on these data, Cali et al. excluded E. intestinalis from the genus Encephalitozoon until E. cuniculi is shown to be tetrasporous (49).

Whereas molecular phylogenies suggest that the family Encephalitozoonidae should contain one genus, morphologic differences between E. hellem, E. cuniculi, and E. intestinalis suggest that three genera should be used. In cell culture, E. hellem displays variable development (49, 102); some parasites occur within a parasitophorous vacuole, while others develop in direct contact with the host cell cytoplasm. This behavior has never been observed for E. cuniculi. Such differences have been used historically for genus-level classification in microsporidia. Consequently, Cali et al. (49) proposed the removal of E. hellem from the genus Encephalitozoon and its reclassification in a genus of its own, creating three genera in the family Encephalitozoonidae. This would satisfy both morphological and molecular criteria (49).

Docker et al. (120) compared sequences of the SSU rRNA gene, the intergenic spacer region, and the LSU rRNA gene of N. salmonis with that of E. bieneusi amplified with primer pair 530f and 580r; the two sequences differed in composition by 19.8%. This genetic divergence between the two species was sufficiently large to place the two species into two different genera in the family Enterocytozoonidae (120, 197), as originally proposed by Hedrick et al. (169). This placement is supported by morphological data. Although N. salmonis exhibits most of the distinguishing morphological characteristics of the family Enterocytozoonidae, the distinctively different hosts and sites of development support the placement of Nucleospora and Enterocytozoon into separate genera within the family Enterocytozoonidae (101, 120, 197).

Another SSU rRNA-based phylogeny of microsporidia has been published by Malone and McIvor (233). Although they included several other microsporidian species in their analysis than were analyzed by Baker et al. (9, 10) the calculated phylogenetic tree, using the neighbor-joining method, showed a branching pattern comparable to the phylogeny of Baker et al. (9, 10) for the species used in both analysis.

The phylogenetic relationships of microsporidia were also analysed by riboprinting (280). The SSU rRNA and the intergenic spacer region of different microsporidia from fish, insects, and shrimp were amplified and digested with restriction enzymes. The generated riboprints were analyzed, and phylogenetic trees were constructed by using the MIX and DOLLOP programs of the PHYLIP package, in which each fragment is considered a different character. While sequencing gives more detailed information on the phylogenetic relationship of different species and genera, riboprinting is a faster and cheaper way of determining the identity and phylogeny of microsporidia. This approach will work at the genus level, but DNA sequencing should be used to define the relationships of different species.

α- and β-tubulin sequences obtained by amplification with primer pairs directed against conserved sequences from different microsporidian species (E. hellem, E. intestinalis, E. cuniculi, N. locustae, and Spraguea lophii) have been used by several authors to construct phylogenetic trees (125, 126, 188, 221). The β-tubulin sequences from E. hellem, E. intestinalis, and E. cuniculi were closely related, but that of E. intestinalis was more closely related to that of E. cuniculi than was that of E. hellem (221), reinforcing the reclassification of E. intestinalis. The generated α- and β-tubulin trees were nearly identical in topology, but in several important aspects these trees are inconsistent with phylogenetic trees constructed on the basis of SSU rRNA sequence data. The few SSU rRNA-based molecular trees for which microsporidian data are available suggest that microsporidia are one of the most ancient eucaryotic lineages (10, 11, 155, 217, 363), but the α- and β-tubulin trees suggest that the microsporidia are close relatives of fungi, which invites speculation that microsporidia evolved degeneratively from higher forms (126, 192, 221).

Phylogenetic analysis of the complete nucleotide sequence of the genes putatively encoding translation elongation factors 1α and 2 from the fish microsporidium Glugea plecoglossi suggests that G. plecoglossi represents the earliest branch of the eukaryotes among the analyzed eukaryotic species (186, 187). This reinforces the hypothesis that microsporidia are extremely ancient eukaryotes that diverged before the occurrence of mitochondrial symbiosis rather than evolving by degeneration from higher forms.

One of the original reasons behind the proposal that microsporidia are ancient eukaryotes is their lack of mitochondria (64). The presence of mitochondrion-derived genes reveals that the ancestor of a lineage contained a mitochondrion, even though an organelle has not been identified in the lineage (193). Recently, genes encoding a chaperone protein, Hsp70 (70-kDa heat shock protein), have been identified in the microsporidia N. locustae, V. necatrix, E. hellem, and E. cuniculi (159, 173, 275a). Phylogenetic analyses have shown close relationships to Hsp70 proteins from mitochondria of other eukaryotes. In addition to the detection of genes encoding Hsp70 proteins in microsporidia, tRNA synthetase genes consistent with the presence of mitochondria have been found in microsporidia (381a). The simplest interpretation of these data is that microsporidia have lost mitochondria while retaining genetic evidence of their past presence. Interestingly, these microsporidian Hsp70 sequences appear as a sister group of fungi in Hsp70 phylogenies, supporting the list of reasons to believe that microsporidia are closely related to fungi, as a highly specialized group that adapted extensively to an exclusively intracellular parasitic mode of life by degenerative evolution (193, 251a).

The concept that microsporidia are ancient eukaryotes was also supported by their apparent lack of spliceosomal introns. The fact that the microsporidia N. locustae and V. necatrix possess core spliceosomal components (U2 and U6 small nuclear RNAs) (115, 133a) is consistent with the recent divergence of the microsporidia. It has been suggested that microsporidia lost most of their spliceosomal introns during the radical genome size reduction accompanying the adaptation to an intracellular parasitic lifestyle (133a).

Phylogenetic analysis of 19 eukaryotic and archeal LSU rRNA sequences, including the LSU rRNA sequence of E. cuniculi, produced different trees depending on the maximum-likelihood method used. FastDNAml produced a tree in which three fast-evolving lineages including microsporidia are predicted to have diverged early from other eukaryotic lineages, whereas the BASEML method placed the Encephalitozoon sequence in a relatively late-emerging position with a very high long terminal branch, indicating a high rate of evolution (275b).

Why do different microsporidian genes suggest contradictory phylogenies? The fact that a phylogeny can be obtained for a group of organisms does not guarantee that it really reflects the evolutionary history (381a). Much of our evolutionary thinking is currently based on the phylogeny of SSU-rRNA, which may be biased by the long-branch attraction artefact. Like any other single-gene phylogeny, it may be misleading in some aspects but correct in others. The phylogenies presented above clearly point to the dangers of placing too much confidence in the phylogny of a single gene, and more DNA sequence data of protein-encoding sequences from microsporidia will be needed to answer this question.

Several drugs, notably colchicine, vinca alkaloides, and benzimidazoles, are effectors of microtubule polymerization. Among these agents, the benzimidazoles are unique in being selectively toxic to certain lower eukaryotes. The β-tubulin subunit has been identified as the primary target of benzimidazole. Benzimidazoles such as albendazole act by disrupting microtubules through β-tubulin binding, and six different amino acid residues have been implicated in benzimidazole susceptibility by mutational analysis of fungi (125, 188). Mutations in His-6, Phe-167, Glu-198, Phe-200, and Arg-241 confer resistance to the derivate benomyl, while a mutation in Ala-165 confers thiabendazole resistance (221). Both E. hellem and E. cuniculi β-tubulins include His-6, Phe-167, Glu-198, Phe-200, and Arg-241, predicting that these microsporidia would be susceptible to benzimidazoles (125, 188). In addition, E. hellem and E. intestinalis β-tubulins include Val-268, and mutational and comparative sequence analyses with the yeast S. cervisiae suggest that Val-268 also plays a role in benzimidazole susceptibility (126).

The genomes of microsporidia, especially E. cuniculi, are very small. A genomic library of the smallest chromosome (chromosome I) of E. cuniculi is already available, and systematic sequencing of this chromosome will be completed in the near future (25, 124a). Complete mapping and gene sequencing of the whole genome of E. cuniculi are in progress (124a), and this will be useful in advancing our knowledge of physiologically important genes and control regions of coding and noncoding sequences in microsporidia.

rRNA data have suggested that microsporidia are one of the most ancient eukaryotic lineages. On the other hand, the relatedness of microsporidia to fungi and animals, as suggested by the α- and β-tubulin and mitochondrion-type Hsp70 sequences, invites speculation that microsporidia evolved degeneratively from higher forms. Examination of protein-coding sequences from microsporidia will be of interest to answer this question and is under way.

The taxonomy of microsporidia has undergone several changes during the last few years and will continue to change significantly in the near future when new DNA-based data are incorporated into new classification systems. As a first step, molecular data suggest that the family Encephalitozoonidae should contain only one instead of two genera, but morphological data and rules of taxonomy demand three genera. However, there is no longer justification for two genera within the family Encephalitozoonidae.

Different strains of E. cuniculi and E. bieneusi have already been identified, and more isolates from different hosts and environmental samples are currently being examined to substantiate the belief that human microsporidiosis is a zoonotic or waterborne disease. With respect to other species, no different strains have been identified so far, but examination of different isolates from different geographic areas and hosts by molecular techniques is under way.

Animal hosts have been identified for E. cuniculi and, most recently, for E. hellem, E. intestinalis, and E. bieneusi. Further examination of different hosts and environmental samples by the highly sensitive molecular techniques is under way. These studies may lead to the identification of further animal reservoirs for microsporidia which infect humans. The highly sensitive molecular techniques may be also helpful for identifying asymptomatic carriers of microsporidia.

At present only limited data have been published comparing molecular techniques for diagnosis of microsporidia with traditional methods of determining the sensitivity and specificity of these new techniques. Nevertheless, these studies indicate that the sensitivity and specificity of molecular techniques can be very high. One international, blinded multicenter study comparing traditional light-microscopic techniques with PCR has been published recently (290), but further studies are needed.

Microsporidia are an important cause of disease in HIV-infected patients and are now increasingly also recognized as pathogens in non-HIV-infected patients with or without immunosuppression. Therefore, methods for the diagnosis and species differentiation of microsporidia should be available in every laboratory performing parasitological examination of clinical specimens.

Although molecular techniques, especially PCR, seem to be very sensitive and specific and are useful for the diagnosis and species differentiation of microsporidia, large comparative blinded studies determining sensitivity and specificity are lacking. Therefore, screening of clinical samples for microsporidia by these techniques is not recommended at the moment. For the detection of microsporidia in clinical samples, chemofluorescent stains or chromotrope-based stains should be used as the methods of first choice. Nevertheless, accurate taxonomic classification of microsporidia is essential because the different genera and species exhibit different biological and epidemiological characteristics, which influence approaches to prevention and treatment. Application of PCR methods offers an approach that can be adapted for processing large numbers of specimens to define the epidemiological extent of microsporidian infection in human populations (272). Therefore, identification of microsporidia by light microscopy should be followed by confirmation and specifies differentiation with molecular techniques.

As is true for many other new and emerging pathogens, we have just scratched the surface of a complex and evolving relation between the phylum Microspora and humans (382). Application of molecular techniques to diagnosis, species differentiation, and phylogentic analysis of microsporidia will lead to an enormously increased knowledge of these organisms in the near future.

We acknowledge and appreciate the assistance of Britta Franzen during preparation of the manuscript.

Portions of this work were supported by the Köln Fortune program/Faculty of Medicine, University of Cologne, Cologne, Germany.