Cynthia K. Warner

Human Granulocytic Ehrlichiosis in Massachusetts

Two new tick-borne zoonoses have recently emerged as threats to public health in North America. Both are caused by infection with Ehrlichia species, obligate intracellular bacteria that localize within the phagosomes of leukocytes. The spectrum of illness varies from mild to severe; about one third of patients require hospitalization. Persons exposed to ticks may present with a spotless spotted fever or a disease similar to Lyme disease that does not feature erythema and is accompanied by fever, chills, severe headache, myalgia, malaise, and nausea. A maculopapular or petechial rash, however, may be present [1, 2]. Laboratory findings include thrombocytopenia, leukopenia, and abnormal hepatic function test results. Treatment with tetracyclines induces defervescence within 48 hours. Ehrlichia first attracted attention in 1935 as pathogens of dogs [3] and shortly thereafter as potential zoonotic agents. An infection similar to mononucleosis was reported in western Japan [4], and the ehrlichial cause of Sennetsu fever was subsequently confirmed by subinoculating mice with patient blood and observing the characteristic inclusions (morulae) within mononuclear cells. The index case of human ehrlichiosis in North America was discovered in a tick-exposed Arkansas resident in 1986 [5]. Morulae were seen within the patients monocytes, and his serum reacted against antigens of E. canis, the agent of tropical canine pancytopenia. In 1990, a new species, closely related to E. canis, was cultivated from a febrile Arkansas patient and designated E. chaffeensis [6, 7]. About 400 illnesses caused by this agent have since been confirmed in 30 states. Nine of these cases were fatal. Since 1990, morulae have been discovered in the granulocytes of 12 febrile patients from northern Minnesota and Wisconsin [8]. The serum of these patients did not react with E. chaffeensis antigen. In addition, an analysis of sequences obtained from a polymerase chain reaction (PCR) assay using universal eubacterial 16S ribosomal DNA (rDNA) primers indicated that the pathogen was closely related to, if not conspecific with, E. phagocytophila and E. equi [9], agents of ovine tick-borne fever and equine ehrlichiosis, respectively. Two of the 12 reported cases were fatal. This infection has been designated human granulocytic ehrlichiosis because morulae are restricted to polymorphonuclear leukocytes. The public health importance, especially the geographic distribution, of human granulocytic ehrlichiosis remains undescribed because of the novelty of the disease and because confirmatory diagnostic tests are not widely available. We describe the definitive identification of this ehrlichia organism from a patient in New England, a discovery that suggests that human granulocytic ehrlichiosis may infect residents of certain sites outside the upper midwestern United States. Case Report A 68-year-old woman presented to the Nantucket Cottage Hospital in Massachusetts on 1 November 1994 with a 5-day history of fever, confusion, headache, and fatigue. She had been in her usual state of good general health and had no significant medical history other than a previous tick bite in June 1994. She took no medications routinely and had no allergies. The patient had removed an unidentified tick from her left flank approximately 5 days before symptoms developed. The tick had not been saved. She had noted no rashes or other signs or symptoms after the tick bite. At presentation, the patient appeared flushed and was somewhat confused as to the time and place, but was otherwise alert and neurologically intact. Her oral temperature was 40 C, her pulse rate was 88 beats/min, and her blood pressure was 110/60 mm Hg. Physical examination showed no evidence of rash or cellulitis at the site of the tick bite and no other remarkable findings. No organomegaly was noted. An automated complete blood count showed an abnormal leukocyte count (3200 109/L) and platelet count (40 000 109/L). A manual differential count showed 32% segmented neutrophils, 27% bands, 34% lymphocytes, 6% monocytes, and 1% metamyelocytes. The hematocrit was 0.38. A chemical analysis of the blood showed low potassium and calcium levels and slightly elevated urea nitrogen levels (22 mg/dL). Liver enzyme levels were elevated: alanine aminotransferase level, 0.07 kat/L; aspartate aminotransferase level, 0.12 kat/L; and lactate dehydrogenase level, 7.80 kat/L. Infection with Babesia microti was suspected, and a Giemsa-stained thin peripheral blood smear was carefully scrutinized. No intraerythrocytic organisms were noted, but an inclusion was noted in a band neutrophil. The patient was suspected of having Lyme disease or ehrlichiosis, and doxycycline (100 mg orally twice a day for 10 days) was started pending confirmatory analysis at the Harvard School of Public Health. The patient defervesced within 24 hours and had an uneventful recovery. An anticoagulated venous blood sample (ethylenediaminetetracetate, 10 mM) obtained at presentation and kept refrigerated for 2 days was examined for evidence of ehrlichiae. Buffy-coat smears stained with the Wolbach Giemsa variant for rickettsiae [10] showed rare inclusions (12 on 5 slides) that were morphologically consistent with ehrlichial morulae within bands and metamyelocytes (Figure 1). The patients serum contained borderline IgM (titer, 1:64) against E. chaffeensis antigen by the indirect fluorescent antibody test [6] but did not contain IgG. No reactivity was noted to antigens of Babesia microti (indirect fluorescent antibody test [11]) or Borrelia burgdorferi (immunoblot [12]). Figure 1. Morula (arrow) within a metamyelocyte from the patients blood (buffy-coat smear, Wolbach Giemsa stain; original magnification, 1250). In a room in which intact Ehrlichia species or their DNA had never been manipulated, an aliquot of the anticoagulated blood was pipetted into a sterile cryotube using a barrier-filtered micropipettor. This sample was sent to the Centers for Disease Control and Prevention for analysis by PCR and automated cycle sequencing. By using a modification of the previously described nested PCR procedure with ehrlichia-specific primers [13] for the 16S rRNA gene, an 1158-base pair amplification product was obtained and sequenced. The sequence of this amplification product (GenBank accession number U23038) did not differ from the sequence of the 16S rRNA gene amplified from the persons in Minnesota and Wisconsin who had diagnoses of human granulocytic ehrlichiosis [9]. Convalescent serum samples obtained on 17 November 1994 and 10 January 1995 continued to show borderline IgM reactivity to E. chaffeensis antigen. Specific IgG to E. chaffeensis was first detected in the 10 January sample (titer, 1:128). Immunoglobulin G reactivity to antigen of E. equi, which has been used as a surrogate for the agent of human granulocytic ehrlichiosis, was not seen in the serum sample obtained at the original presentation (1 November). However, IgG reactivity was present at a titer of 1:64 in the 17 November sample and at a titer of 1:256 in the 10 January sample. No reactivity to antigens of Rickettsia rickettsii, R. typhi, or Coxiella burnetii was seen in the acute and convalescent serum samples (indirect fluorescent antibody test). Seroconversion for Lyme disease or babesiosis was not observed. Discussion An elderly tick-exposed patient from Nantucket, Massachusetts, presenting with a febrile illness and nonspecific symptoms, was initially suspected of having Lyme disease or babesiosis. Laboratory test results indicated both thrombocytopenia and leukopenia, which we use as markers for B. microti infection (unpublished data), and a Giemsa-stained thin blood smear was carefully scrutinized. A suspicious inclusion detected within a granulocyte led to a careful search for ehrlichiae, and classic morulae were detected by examining buffy-coat smears. Ehrlichial DNA was amplified and sequenced, and the identity of the agent was confirmed by comparing this sequence with those amplified from patients from the upper midwestern United States who were accessioned in GenBank. Our report describes the first definitive identification of this pathogen in a New England resident. A case of human ehrlichiosis was reported from nearby Cape Cod in 1992 [14]; in retrospect, this case may be considered to have been caused by the agent of human granulocytic ehrlichiosis [15]. Serum from that patient was reported to strongly react (IgG titer >1:512) to antigens of E. chaffeensis. In contrast, serum samples from the original Minnesota and Wisconsin cases of human granulocytic ehrlichiosis have no reported reactivity with E. chaffeensis [8]. However, convalescent serum samples from several cases of human granulocytic ehrlichiosis that have recently been diagnosed and confirmed by molecular analysis in Minnesota apparently react with antigens of E. chaffeensis (Persing DH, Mayo Foundation. Personal communication). The degree of cross-reactivity between E. chaffeensis and the agent of human granulocytic ehrlichiosis may eventually be described as more cases of human granulocytic ehrlichiosis are analyzed. However, the presence of granulocytic morulae and the fact that the agent has the same molecular identity as the pathogen identified in the upper midwestern United States establish the diagnosis of human granulocytic ehrlichiosis in our patient. Whether the agent constitutes an antigenic variant remains to be determined. The agent of human granulocytic ehrlichiosis may frequently infect residents of sites where deer ticks are abundant but may be diagnosed as a rashless Lyme disease. Human granulocytic ehrlichiosis may be successfully treated with tetracyclines, and a response to treatment might affirm a presumptive diagnosis of Lyme disease. In 1991, we found that 11% of serum samples obtained from residents of a well-studied Lyme-disease focus in Cape Cod reacted with antigen of E. chaffeensis [16], and

The first isolation, in vitro propagation and genetic characterization of Ehrlichia canis in Israel

Ehrlichia canis, the etiologic agent of canine ehrlichiosis, was isolated in Israel from a naturally infected dog with acute signs of the disease. The organism designated E. canis 611, was passaged experimentally to a beagle, from which it was propagated in primary canine monocytes. The organism was then grown in vitro in a continuous canine cell line, DH82. Nine beagles subsequently injected with whole E. canis-infected blood all developed typical symptoms of ehrlichiosis. An indirect immunofluorescence antibody test to E. canis was developed and compared with a commercial kit, revealing a good correlation between the two assays. Transmission electron microscopy of DH82 cells infected with the Israeli strain of E. canis (611), revealed organisms similar to those described in the literature: two different forms of morulae appeared, one tightly, the other loosely, packed. The 16S rRNA gene sequence obtained from the Israeli Ehrlichia isolate was compared with other isolates, E. canis Oklahoma and E. canis Florida. The Israeli strain 16S rRNA had three nucleotide differences from the Oklahoma isolate, and four nucleotide differences from the Florida isolate, in addition to one nucleotide gap in each. The Israeli isolate was found to be 0.54% different from the Oklahoma strain, and 0.61% different from the Florida strain. There are the same magnitudes of differences displayed by the other most closely related group in the phylogenetic tree, namely Ehrlichia equi, Ehrlichia phagocytophilia and the human granulocytic ehrlichia.

Development and use of specific polymerase reaction for the detection of an organism resembling Ehrlichia sp. in white-tailed deer.

The role of white-tailed deer (Odocoileus virginianus) in the epidemiology of Ehrlichia chaffeensis and the agent of human granulocytic ehrlichiosis (HGE) is not fully understood, and diagnostic procedures may be complicated by the recent detection of 16S rDNA sequence from an Ehrlichia sp.-like organism in wild deer. A specific forward primer (DGA) and an Ehrlichia spp. reverse primer (GA1UR) were constructed to amplify this new, distinct Ehrlichia sp.-like 16S rDNA. The DGA primer, a forward primer specific for E. chaffeensis (DCH), and a forward primer specific for the E. phagocytophila genogroup (GE9f) were each used with GA1UR in nested polymerase chain reactions to amplify 16S rDNA sequences from control samples containing the deer Ehrlichia sp.-like organism, E. chaffeensis, or the HGE agent. Primer pairs DGA/GA1UR and DCH/GA1UR specifically amplified 16S rDNA sequences from the corresponding target organism, whereas GE9f/GA1UR amplified 16S rDNA sequence from both the HGE agent and the deer Ehrlichia sp.-like organism. With a nested PCR using DGA/GA1UR and DCH/GA1UR on DNA extracted from white blood cells from 62 deer from 10 populations in four U.S. states, we observed a high prevalence (65%) of 16S rDNA sequences of the deer Ehrlichia sp.-like organism, and a low prevalence (5%) of the E. chaffeensis sequence. In this field survey, E. chaffeensis-reactive antibodies detected by indirect fluorescence assays were associated (P < 0.001) with PCR evidence of the deer Ehrlichia sp.-like organism, but not E. chaffeensis. Infestations of Amblyomma americanum also were associated (P < 0.001) with PCR evidence of the deer Ehrlichia sp.-like organism. The potential for serologic cross-reactions and non-specific PCR products arising from the deer Ehrlichia sp.-like organism should be considered when evaluating the role of deer and their ticks in the epidemiology of ehrlichial pathogens of humans.

Procedures for reproducible detection of rabies virus antigen mRNA and genome in situ in formalin-fixed tissues

Procedures allowing the reproducible in situ detection of rabies virus antigen and RNAs (both genome and message) in formalin-fixed tissue are described. These procedures can be used on sequential tissue sections and thereby permit comparison of results from tests detecting both antigen and RNA in the same tissue. This antigen-detecting procedure has also been used to identify both the phylogenetically distant rabies viruses from silver-haired bat and vampire bat and the rabies-related viruses Mokola, Duvenhage, and Lagos bat. One of the critical steps in these procedures is the digestion (and the resulting exposure of the target molecules) with proteinase K. These methods may be useful for the identification of other viruses of public health importance. Because in many situations only formalin-fixed tissue is available for postmortem diagnosis, the technical ability to identify a virus antigen and nucleic acid in such tissues greatly extends potential diagnostic capabilities.

A comparative study of the fluorescent antibody test for rabies diagnosis in fresh and formalin-fixed brain tissue specimens

Many diagnostic methods have been used to detect rabies virus antigen. The preferred method for routine diagnosis of rabies in fresh or frozen brain tissues is the fluorescent antibody test (FAT). In this study, the FAT was used to evaluate the rabies status of fresh/frozen brain specimens from more than 800 rabies-suspected cases, in more than 14 different species of animals. A comparable brain specimen from each case was fixed in 10% buffered formalin and examined by the FAT. The evaluation of rabies status between fresh and formalin-fixed tissues was in agreement in more than 99.8% of the cases. When fresh tissue is not available for testing, these results validate the use of this procedure for routine diagnosis of rabies in formalin-fixed brain tissues.

Use of anti-glycoprotein monoclonal antibodies to characterize rabies virus in formalin-fixed tissues.

Seventy anti-rabies virus monoclonal antibodies (Mabs) were tested for reactivity with rabies and rabies-related viruses in formalin-fixed (FF) tissues. Forty-three of the Mabs were directed against the glycoprotein and 27 were directed against the nucleocapsid as determined by enzyme immunoassays and neutralization tests. Twenty of the anti-glycoprotein Mabs and one of the anti-nucleocapsid Mabs reacted with the rabies challenge virus strain (CVS) in FF tissue. These 21 Mabs were screened against other lyssaviruses in FF tissues: five rabies virus strains (coyote, skunk, raccoon, red bat, and silver-haired bat), and four rabies-related viruses (Australian bat lyssavirus, Duvenhage virus, Lagos bat virus, and Mokola virus). One of the anti-glycoprotein Mabs was reactive with all the virus strains screened. Another of the anti-glycoprotein Mabs reacted with all of the rabies virus strains tested, but not with any of the rabies-related virus strains tested. The remaining Mabs had reactivity patterns that could be useful for characterizing lyssaviruses in FF tissues.