Historically, morphologic methods have played an important role in detecting new microorganisms, and they are still crucial for diagnosing infections caused by agents not routinely cultured, such as Mycobacterium leprae (40). Because microscopic examination of stained smears from biologic fluids or tissue imprints is usually rapid and easy, it has often been performed in patients who have an unexplained disease, although its interpretation is subjective and its sensitivity and specificity are generally low. The first evidence for the responsibility of Ehrlichia species in humans with an acute febrile illness was provided by examining blood smears stained with a Romanowsky stain, in which these as-yet-uncultivated organisms could be observed forming intracytoplasmic morulae within leukocytes (6,7). Borrelia burgdorferi were first observed in Giemsa-stained smears from midgut diverticula of ticks (19). Examination of smears can also be helpful when multiple organisms are cultured from a nonsterile site, as microbial culture alone, as well as molecular detection, cannot distinguish between colonization or asymptomatic shedding and tissue invasion: in such a situation, the morphology of the predominant organism visualized in the tissue sections can suggest the true causative agent (40).

Although individual bacteria generally are not detected in hematoxylin and eosin (H&E)-stained tissue sections, exceptions do exist. Clumps of finely particulate basophil material were seen in H&E-stained sections of bacillary angiomatosis and subsequently identified as Bartonella (41). In H&E-stained sections of gastric biopsy specimens that show acute gastritis, curved bacteria consistent with Helicobacter pylori may be seen in the layer of mucus on the crypt epithelium (25). Moreover, as histopathologic damage and causal microorganisms usually have a long-established association, microscopic examination of H&E-stained tissue sections during the course of an unexplained disease may lead to hypotheses about the nature of the etiologic agent (40).

Gram stain has also proven useful to routinely diagnose H. pylori and H. heilmanii in the gastric mucosa of patients with gastritis, as well as that of B. henselae in cardiac valves (10,24,25). Silver impregnation is among the most useful methods for detecting bacteria, especially for that stained weakly with a tissue Gram stain. Thus, bacillary angiomatosis lesions were found to contain clusters of bacilli on Warthin-Starry staining 2 years before the etiologic role of B. henselae was elucidated. With the same stain, this bacterium was also detected in cardiac valves of patients with endocarditis (Figure 2) (41). The first observation of Whipple agent was reported in 1907 by George Whipple in silver-stained sections of a lymph node, although the author did not link this observation with the cause of the disease (2).

Special stains have also played a role in establishing the etiologic role of new bacteria. Gimenez’ and Pinkerton’s stains allowed the detection of rickettsial organisms in tissue sections from patients with acute febrile disease (40).New mycobacteria were initially detected by using Ziehl-Nielsen, Kinyoun, or auramine O stains. For example, in an HIV-infected boy, examination of a retroperitoneal lymph node showed granuloma with large numbers of intracellular acid-fast bacilli that were later characterized as a new Mycobacterium species, M. genavense (34). Morphologic techniques, indeed, do not allow specific identification of the detected organisms. Despite this limitation, the approach consisting of detecting infectious lesions and agents by using cytologic and histologic examination appeared to be sometimes more valuable than the cultural or molecular techniques (40).

Broad-spectrum media allowed several previously unrecognized gram-positive bacteria, such as novel corynebacteria or Staphylococcus species, as well as novel beta-Proteobacteria, to be isolated, mainly from blood or pus of patients (18,37,38). The first isolation of B. elizabethae, B. quintana, and B. henselae was also achieved on blood agar (15). Use of Campylobacter-selective medium allowed novel Campylobacter and Helicobacter species to be grown from stools and rectal swabs, respectively (23), and provided further evidence for the association between C. jejuni infection and Guillain-Barré syndrome (44). For Campylobacter spp., selective, antibiotic-containing media could be satisfactorily replaced by nonselective blood agar, provided stool specimens had been filtered with a cellulose acetate membrane (23). Newly recognized serotypes of enterohemorrhagic Escherichia coli were isolated on MacConkey-sorbitol agar from stools or urine of patients with hemolytic-uremic syndrome (26). For Vibrio cholerae O:139 and most novel Vibrio species, the most convenient, highly selective medium was thiosulfate-citrate-bile salts sucrose agar (33).

The usefulness of broad-spectrum media should not obscure the fact that some emerging bacteria would not have been isolated without specific media. Buffered charcoal-yeast extract (BCYE) agar facilitated the recovery of most novel Legionella species, as well as Afipia broomeae and A. clevelandensis, from human respiratory sources (12,29) The first cultivation of Borrelia burgdorferi was achieved in 1981 in a modified Kelly medium (19). In 1994, 20 years after the first attempts, the Kelly growth medium itself allowed first cultivation of B. recurrentis from the blood of an Ethiopian patient with louse-borne relapsing fever, and B. duttonii, agent of East African tick-borne relapsing fever, was isolated for the first time in 1999 in BSK II medium (21). Generally, combining different types of medium, using both solid and liquid media, increases the effectiveness of culture, perhaps because of a preference of the bacterium for one type of medium over another or simply from the increased sensitivity obtained by culturing a large volume of specimen. For example, B. elizabethae and B. henselae were detected in BACTEC blood culture medium before inoculation in blood agar (13,15). Isolation of most novel Mycobacterium species required both solid- and liquid-specific media (34).

While more expensive and less easy to use than artificial media, animal models can provide certain advantages not available with artificial media. For example, until recently, inoculation to mice was the only means available to propagate B. duttonii (21). Today, animals are still necessary for isolating organisms such as Treponema pallidum or Mycobacterium leprae. Animal inoculation can help to reduce the contaminant flora. Thus, a combination of passage in guinea pigs and subsequent transfer into embryonated eggs was the key for isolating L. pneumophila from lung autopsy specimens (31). Embryonated eggs themselves have been recognized as a standard for rickettsial isolation, allowing, for example, the first isolation of Astrakhan fever rickettsia (45).

Cell culture is easy to use and may be very sensitive. Isolation of T. whipplei was obtained from valve and duodenal biopsy specimens by using human embryonic lung fibroblasts (HEL) (2,36). Ehrlichia chaffeensis and R. japonica were grown from blood samples of patients on canine macrophage cells (Figure 4) and African green monkey cells, respectively (9,46). Cultivation of facultative intracellular bacteria also was facilitated by cell culture. L. pneumophila has been isolated by using HEL cells while inoculated BCYE and agar plates remained sterile (47). With a bovine endothelial cell line, B. quintana was isolated for the first time from cutaneous biopsy material of a bacillary angiomatosis patient (48). Such enhanced sensitivity is a major advantage for an infection with low levels of bacteremia or when limited biopsy material is available (49). Indirectly, HEL cells also provided the first evidence for the role of a toxic factor in pseudomembranous colitis, which could be neutralized by clostridial antiserum. This observation led to the discovery of Clostridium difficile as the responsible agent (50).

The search for appropriate media that could allow the growth of still uncultivatable or unrecognized bacteria has led us to try coculture with nonmammalian cells. Cell lines from toads (XTC-2) have been used in our laboratory to grow Rickettsia felis, a flea-associated Rickettsia pathogenic for humans (8). Coculture with arthropod cells will probably enhance our ability to detect intracellular, arthropod-transmitted bacteria. For example, tick cells (IDE8) have been used to grow the agent of human granulocytic ehrlichiosis (7). Cocultivation of samples with free-living amebae has allowed recovery of otherwise uncultivatable microorganisms from patients and the environment. This technique provided evidence for the role of several Legionella or Legionella-like species and of Parachlamydia acanthamoeba as etiologic agents of community-acquired pneumonia (51).

For noncultured organisms, molecular techniques have been proposed to identify isolated bacterial genes directly from clinical specimens. These techniques, however, are quite difficult to use and can identify only a few, short genetic fragments (3). On the other hand, by providing pure microbial cell mass, culture enables genes to be identified in high numbers through recombinant chromosomal libraries built from the extracted DNA. Genes identified in this fashion can then be utilized as more refined diagnostic tools. For example, Rickettsia mongolotimonae and R. slovaca were associated with human disease on the basis of amplification of a species-specific rOmpA gene fragment from skin biopsy specimens (11,12). DNA probes developed after isolation of Chlamydia pneumoniae enabled this organism to be detected by in situ hybridization in coronary atherosclerotic plaques (62). Further, molecular subtyping of cultured strains has offered new perspectives for epidemiologic studies. Thus, comparison of nucleotide sequences of 16S rDNA, OspA, and Fla genes for different strains of B. burgdorferi provided phylogenetic data that consistently supported the division of B. burgdorferi sensu lato into three geographically distinct genotypes, which were subsequently shown to have different pathogenic potentials (63). Correlation between genotypes and biologic characters is a key to understanding the pathophysiology of bacterial diseases.

Because of the importance of organisms such as H. pylori, M. genitalium, and C. pneumoniae as emerging human pathogens and the value of complete genome sequence information for drug discovery and vaccine development, the complete nucleotide sequences of these three organisms has been determined by the whole-genome random sequencing method as described initially for Haemophilus influenzae. Sequence analyses allowed identification of several predicted coding regions that included genes required for DNA replication, transcription and translation, DNA repair, cellular transport, and energy metabolism (64). With the availability of complete genome sequences, further assessment of microbial genetic diversity is possible; based on the large number of sequence-related genes encoding outer membrane proteins, H. pylori was predicted to use recombination as a mechanism for antigenic variation and adaptative evolution (65). As the genome sequences of new bacterial species or strains are determined, comparative genomics will be an increasingly useful method to provide insights into physiologic differences among microorganisms (64).