University of Nebraska - Lincoln DigitalCommons@University of Nebraska - Lincoln

U.S. Department of Veterans Affairs Staff Publications U.S. Department of Veterans Affairs

1994

Haemophilus parainfluenzae : Application of a Molecular Approach for Identification of athogenicP Bacterial Species

Kamal A. Hamed Stanford University School of Medicine

Philip R. Dormitzer Stanford University School of Medicine

Catherine K. Su Stanford University School of Medicine

David A. Relman Stanford University School of Medicine, [email protected]

Follow this and additional works at: https://digitalcommons.unl.edu/veterans

Hamed, Kamal A.; Dormitzer, Philip R.; Su, Catherine K.; and Relman, David A., " parainfluenzae Endocarditis: Application of a Molecular Approach for Identification of athogenicP Bacterial Species" (1994). U.S. Department of Veterans Affairs Staff Publications. 2. https://digitalcommons.unl.edu/veterans/2

This Article is brought to you for free and open access by the U.S. Department of Veterans Affairs at DigitalCommons@University of Nebraska - Lincoln. It has been accepted for inclusion in U.S. Department of Veterans Affairs Staff Publications by an authorized administrator of DigitalCommons@University of Nebraska - Lincoln. 677

Haemophilusparainfluenzae Endocarditis: Application of a MolecularApproach for Identificationof PathogenicBacterial Species

Kamal A. Hamed, Philip R. Dormitzer, Catherine K. Su, From the Division of Infectious Diseases and GeographicMedicine, and David A. Relman Department of Medicine, Stanford University School of Medicine, Stanford;and the Departmentof VeteransAffairs Medical Center, Palo Alto, California

Haemophilus parainfluenzae is both a human oropharyngeal commensal bacterium and a cause of serious invasive disease. The fastidious growth characteristics of this organism and the poor specificity of traditional methods for species identification are likely to have led to inaccura- cies in the diagnosis of infections caused by H. parainfluenzae and related organisms. We report a case of H. parainfluenzae endocarditis in which confusion related to microbial identification was resolved by the analysis of 16S ribosomal RNA sequences. Rapid identification was facilitated by amplification of 16S ribosomal DNA directly from cultured cells with use of the polymerase chain reaction and by direct DNA sequence determination of the amplified product. This proce- dure is potentially useful for the identification of fastidious bacterial pathogens by reference laboratories.

Haemophilusparainfluenzae is a common inhabitant of lation or identification,especially of Haemophilusparaphro- the upperrespiratory tract as well as an occasionalpathogen. philus and H. parainfluenzae.These organismsremain diffi- In one study, it was isolated from the throats of all of 10 cult to cultivatein the laboratoryand usually requireat least healthy childrenand all of 10 healthy adultsand constituted 1 week for identificationto the species level. Theiridentifica- 74% of cultivated pharyngealHaemophilus organisms [1]. tion has been based on metabolic or biochemical features Among Haemophilusspecies, this organism is the second that may be variableor nonspecific [5]. most frequentlyisolated pathogen after Haemophilus influen- Species and, in some instances, strains can be reliably zae. Reported disease syndromescaused by H. parainfluen- identifiedby the analysis of certain specific gene sequences zae include endocarditis,meningitis, pharyngitis, epiglottitis, -in particular,those of the ribosomalRNA (rRNA) genes. , dental , , brain abscess, This type of approachhas been appliedclinically to the iden- , empyema, septicemia, septic arthritis, osteo- tificationof"unculturable" microbial pathogens and fastidi- myelitis, soft-tissueabscess, ,hepatic abscess, and ous microorganismssuch as the mycobacteria[6-9]. In this urinarytract infection [2, 3]. paper,we use a case of endocarditisto illustratehow a 16S Endocarditiscaused by Haemophilusspecies accounts for rRNA-based approachcan be used to expeditiouslyresolve more than half of all cases caused by a group of fastidious confusion regardingthe identificationof H. parainfluenzae. coccobacillaryorganisms known as the HACEKgroup. The other in this are organisms group actinomyce- Materials and Methods temcomitans, hominis, , and Kingellakingae. HACEK organisms, in turn,cause more Determinationof 16S rRNAsequence. A partialsequence than half of all cases of gram-negativebacterial endocarditis of the 16S rRNA gene of the bacterialisolate-referred to as [4]. It is likely that many cases of endocarditiscaused by UNK-from the case describedbelow was obtained as fol- these organismshave been classified as culture-negativeor lows. The bacterialcell mass fromseveral minute colonies of attributedto the wrong species because of difficultiesin iso- a blood-bottlesubculture was digestedovernight at 55?C in 10 mM Tris-HCl (pH 7.5), 1 mM EDTA, 1%Laureth-12 (Mazer Chemicals, Gurnee, IL), and 0.2 mg of proteinase K/mL (BoehringerMannheim, Indianapolis),as previously Received 5 November 1993; revised 18 April 1994. described After inactivationof the K at 94?C Financial support:Lucille P. MarkeyCharitable Trust (to D.A.R., a Lu- [6]. proteinase cille P. MarkeyBiomedical Scholar); Stanford PMGM Director'sResearch for 10 minutes, DNA from 1-/gLand 10-/1laliquots of the Fund, sponsoredby SmithKlineBeecham (to D.A.R.); and StanfordDepart- lysatewas amplifiedby the polymerasechain reaction(PCR) ment of Microbiologyand ImmunologyNational ResearchService Award with use of the bacterial 16S rRNA T32-AI07328 (to P.R.D.). broad-range primers Reprintsor correspondence:Dr. David A. Relman, Palo Alto Veterans 8FPLand 806R [10, 1 1]. The PCR productswere separated AffairsMedical Center, 154T, 3801 MirandaAvenue, Palo Alto, California by agarosegel electrophoresisand purifiedwith the Magic 94304. PCR Preps System (Promega Corporation,Madison, WI). ClinicalInfectious Diseases 1994;19:677-83 The DNA were with ? 1994 by The Universityof Chicago.All rightsreserved. purified fragments directly sequenced 1058-4838/94/1904-0006$02.00 the Taq polymerase-basedfmol sequencing system (Pro- 678 Hamedet al. CID 1994;19(October) mega). The sequencingprimers 50F (5'AACACATGCAAG- time yielded CSF containing6 red blood cells/mm3,9 leuko- TCGAACG3')and 512R (5'CGAATTCCCGCGGCTGCT- cytes/mm3(100% monocytes), 27 mg of protein/dL,and 58 GGCACGGA3') were based on conserved bacterial 16S mg of glucose/dL; the blood glucose level was 85 mg/dL. rRNA sequences [11, 12]. The sequencing reactions were Gram stainingof the CSF revealedno organisms. chased with terminal deoxynucleotidyltransferase(Gibco On admission the patient was lethargic and somewhat BRL, Gaithersburg,MD), and 35S-labeledsequencing prod- confused;her temperaturewas 39?C, her blood pressurewas ucts were analyzed by separationon buffer gradient poly- 100/68 mm Hg, and her pulse was 108/min, with orthostatic acrylamide-ureagels and autoradiography[13]. changes. Her fundi were unremarkableand her neck was Multiple-sequencealignment was performedwith Pileup, supple. Examinationof her mouth revealedgood dental hy- a programin the GeneticsComputer Group Sequence Analy- giene with multiple eruptingwisdom teeth. A grade 1/6 sys- sis SoftwarePackage, version 7.1.1 [14]. Pileup uses a simpli- tolic murmurwas heard by one examiner. The lungs were fication of the progressivealignment method to createalign- clear to auscultationand percussion. Hepatosplenomegaly ments and a dendrogram[15]. The sequences of 16S rRNA was detected. The patient was oriented to time, place, and genes from membersof the family ,including person, and she had no focal neurologicalfindings. The rest a representativeof each subclusterdefined by Dewhirstet al. of the resultsof her physicalexamination were normal. [16], and from Escherichiacoli were obtainedfrom GenBank Initiallaboratory findings included a peripheral-bloodleu- (Bethesda,MD) and used as a data set for the alignment.The kocyte count of 17,000/mm3,with 65%polymorphonuclear species of origin for these sequences(as listed by Dewhirstet cells and 22%band forms;a hematocritof 32.0%;a platelet al. [16]), the straindesignations, and the GenBankaccession count of 84,000/mm3; and the following serum levels: so- numbers are as follows: A. actinomycetemcomitans,ATCC dium, 135 mmol/L; potassium, 2.7 mmol/L; chloride, 95 29522, M75036; H. aphrophilus,ATCC 33389T, M75041; mEq/L;bicarbonate, 30 mEq/L;albumin, 2.7 g/dL; and lac- H. influenzae,ATCC 33391, M35019 and M59433; Actino- tate dehydrogenase,952 IU/L. Serum levels of creatinine, bacillus seminis, ATCC 15768, M75047; Pasteurellaspe- alkaline phosphatase,aspartate aminotransferase, and ala- cies, CCUG 18782, M75055; Pasteurellamultocida, NCTC nine aminotransferasewere within normallimits. Urinalysis 10322, M35018 and M59769; Haemophilus parasuis, revealedmicroscopic hematuria. A chest roentgenogramwas NCTC 4557, M75065; Actinobacillussuis, ATCC 15557, normal. Blood was drawn for cultures,and therapywas be- M75071; Actinobacillusureae (listed by GenBankas Pasteur- gun with parenteralbroad-spectrum antibiotics. ella ureae), Henrikson 3520/59T, M75075; Actinobacillus On the second day of hospitalization,the patient devel- species (listed by GenBank as Actinobacilluscapsulatus), oped photophobia,increased lethargy, and disorientationto CCUG 19799, M75067; Actinobacilluspleuropneumoniae, time. She also had meningismusbut exhibitedno other neu- ATCC 27088T, M75074; Haemophilusparaphrohaemolyti- rological abnormalities.A second lumbar puncture docu- cus, NCTC 10670T,M75076; Haemophilusducreyi, ATCC mented a total CSF leukocyte count of 9/mm3, with 61% 33922, M75079; Pasteurella haemolytica, NCTC 9380T, polymorphonuclearcells, 14%lymphocytes, and 25%mono- M75080; H. parainfluenzae,ATCC 33392T, M75081; H. cytes. No red blood cells were detected in CSF; the protein parainfluenzae(listed by GenBank as H. paraphrophilus), level was 24 mg/dL and the glucose level was 50 mg/dL. The ATCC 29242, M75082; Haemophilusspecies, strain 202, glucose concentrationin blood was 136 mg/dL. The initial M75077; Pasteurellapneumotropica, NCTC 8141T, M75083; diagnosiswas meningoencephalitisof probableviral origin, and E. coli, K-12/MG1655, JO1695. Basepositions were iden- and intravenousacyclovir was added to the regimen. How- tifiedby the numberingestablished for E. coli 16S rRNA [17]. ever, magnetic resonance imaging of the head revealed hy- In orderto ensure unambiguousalignment, bases 94-489 of perintenseareas that did not display enhancementwith ga- each sequencewere used for analysis. dolinium in the left basal ganglia and the left middle The entire procedure-from cell digestion to sequence cerebellarpeduncle on a T2-weightedimage. Transesopha- analysis-consumed -48 hours; however, this procedure geal echocardiographyperformed on the same day showed can be significantlyshortened, as will be describedbelow. mitralvalve prolapse,a large vegetationon the anteriormi- tral leaflet, and possible posteriormitral annular abscess. By this time the systolic murmurheard on admission had in- Case Report creasedto grade 3/6, with radiationto the axilla. A 24-year-oldwoman with nontropicalsprue was admit- On the fifth day of hospitalization,one of the cultures of ted to StanfordUniversity Medical Center with a 10-day his- blood drawn at admission became positive for small gram- tory of nausea, emesis, and headacheand a 4-day historyof negativerods with positivecatalase but negativeoxidase reac- fever.She had no historyof abdominalpain, diarrhea,photo- tions. These organismswere preliminarilyidentified as A. phobia, neck stiffness, or head trauma.The day before ad- (Haemophilus) actinomycetemcomitans.At this time two mission the patient had undergonea neurologicalexamina- transient Janeway lesions were described on the palmar tion that yielded normal results. A lumbarpuncture at that aspect of the fingers.The patient continued to be lethargic CID 1994; 19 (October) H. parainfluenzae Endocarditis 679

Anaerobic/F). After 24 hours, pinpoint colonies were ob- served on chocolate agarplates only; gram stainingrevealed thin, delicate gram-negativerods. After 48 hours, these colo- nies became gray; in addition, fine pinpoint colonies were observed on Brucellaagar plates. An based on commercial tetramethylphenylenediaminereagent (BBL, Cockeysville,MD) was negative. The star-likecolonial mor- phology and puff-like growth on the side of a brain-heart infusion broth tube that are sometimes seen with A. actino- mycetemcomitanswere not observed.Repeat oxidase testing of subculturedcolonies gave a positive result. The organism requiredV Factor but not X Factor and was catalase-posi- tive. It was subsequently identified as H. parainfluenzae biotype II, both by the RapID NH System(Innovative Diag- nostic Systems, Atlanta) and by biochemical testing recom- Figure.Mitral-valve 1. leaflet vegetations from a woman sube- mended by the Centers for Disease Control and Prevention Figure 1. Mitral-valveleaflet vegetations from a woman subse- (CDC). (However, the isolate did not ferment glucose and quently shown to have Haemophilusparainfluenzae endocarditis. sucrose,probably because of poor growth.)The organismdid Smallgram-negative bacilli were foundwithin these vegetations not produce f3-lactamase.The MICs for the strain were as of Dr. Donald of Stan- (courtesy Regula,Department Pathology, follows: , 0.25 Mug/mL;, <0.015 gtg/ fordUniversity School of Medicine). mL; and chloramphenicol, <0.5 Mg/mL. This isolate was studied in further detail, as described under Re- sults. and to have high-gradefever, but she had neither clinical evidence of heartfailure nor congestive electrocardiographic Results evidence of conduction defects. On the sixth day of hospital- ization, there was an abruptonset of intense left-upper-qua- 16S rRNA sequencedetermination. The bacterial blood drant abdominalpain. A computerizedtomogram of the ab- isolate (UNK) culturedfrom the case just describedwas ini- domen with contrast showed multiple low-density regions tially identifiedby the clinical microbiologylaboratory as an throughoutthe spleen-the largestmeasuring 4 cm X 5 cm oxidase-negativestrain ofA. actinomycetemcomitansand was -that were consistent with infarctions. The scan also re- subsequentlyreclassified as an oxidase-positivestrain of H. vealed hepatosplenomegaly.The following day the patient parainfluenzae.To resolve this discrepancy, we sought to underwentmitral valve replacementwith a bileafletmechan- identify the isolate by sequence analysis. A sequence com- ical prostheticvalve. Loosely attached vegetations 2-7 mm prising411 base pairsof the bacterial16S rRNA gene (E. coli in diameterwere found on both leaflets of the native valve 16S rRNA positions 79-489) was determined unambigu- (figure 1). Microscopicexamination of the valvularmaterial ously for UNK (figure2). This sequence was depositedin the revealed numeroussmall gram-negativebacilli. Genome Sequence DataBase(Los Alamos National Labora- The patient'spostoperative course was generallyunevent- tory, Los Alamos, NM) under accession number L26363. ful, with resolutionof fever on the second day. No CSF cul- 16SrRNA gene sequencesfrom positions 94-489 were avail- turesyielded growth.The patientreceived intravenous ampi- able for all members of a sequence data set drawn from E. cillin (12 g daily) for 3 weeks;because of the developmentof coli and from representativemembers of the Pasteurellaceae an ampicillin-associatedrash, this drugwas subsequentlyre- (listed in Materialsand Methods)and could be aligned unam- placed with intravenousceftriaxone (2 g daily), which was biguously with the UNK sequence. In a phylogenetic tree administeredfor an additional 3 weeks. In addition, intrave- constructedwith this alignment, the UNK sequence and the nous (2 mg/kg daily) was given for a total of 4 only two available H. parainfluenzae16S rRNA sequences weeks. The patient was doing well 4 months after the initial formed a coherent, monophyletic lineage that was distinct episode. from all other taxon lineages (data not shown). The UNK An identical organismwas eventuallyisolated from four of sequence was 99.2% identical to the sequence from one eight paired sets of blood cultures. Growth was detected in strain of H. parainfluenzae(GenBank accession number, the aerobic bottle of each set 48-96 hours after inoculation M75082) and was 98.7%identical to the sequence from the as well as in one anaerobicbottle that was subculturedafter 7 second strain(M75081). The two publishedH. parainfluen- days. A Bactec 9240 system (Becton-Dickinson, Sparks, zae sequences were 99.5%identical to each other in this re- MD) was used with enriched soybean-casein digest broth gion. The 16S rRNA gene sequence from H. aphrophilus bottles (BACTEC PLUS Aerobic/F and BACTECStandard (M75041) was the next most closely related to the UNK 680 Hamedet al. CID 1994;19 (October)

80 90 100 110 120 ECO ...... C...... T C . . . AAC A ...... t C . . . C ...... N HAP A G T a . . . T A C T . A . . C ...... N ...... N HPI UNK GAAGCTTGCTTCTTTGATGACGAGTGGCGGACGGGTGAGTAATGCTTGGGA

130 140 150 160 170 ECO . A . . G C . . G ...... T ...... G ...... A AAC ...... HAP .. N.g ...... G...... C ...... HPI . . N . N ...... UNK ATCTAGCTTATGGAGGGGGATAACTACGGGAAACTGTAGCTAATACCGCG

180 190 200 210 220 2. of 16S ECO .. ACG ... CA... CC.... A. G ...... T . ... Figure Sequences AAC . . GG..... C...... C...... NT ..... G ... GA... rRNA genes from UNK (the bac- HAP ...... terialblood isolate froma woman HPI . . T G ...... UNK TAGAATCGGGAGATGAAAGTGTGGGACCTTCGGGCCACATGCCATAGGAT subsequentlyshown to have Hae- mophilus parainfluenzae endo- carditis), , and 230 240 250 260 270 ECO .T GA ...... C.. G ..... C ...... T ..G... selected membersof the Pasteur- AAC . . A ...... T ...... G . . . G C T...... ellaceae. Positions 79-489 are HAP . . A ...... G . ... T. . . CT ...... HPI.....N. shown, numbered according to UNK GAGCCCAAGTGGGATTAGGTAGTTGGTGAGGTAAAGGCTCACCAAGCCGA the systemfor E. coli.The Interna- tional Union of Biochemistrynu- cleotide code is with 280 290 300 310 320 employed, ECO . . . . . C ...... A ...... lower-caseletters denoting ambig- AAC ...... G ...... C . uous nucleotides in se- HAP ...... C ...... published HP I ...... N ...... N quences, Ns denoting undeter- UNK CGATCTCTAGCTGGTCTGAGAGGATGACCAGCCACACTGGGACTGAGACA mined bases, and dots denoting identity with the UNK sequence. 330 340 350 360 370 ECO denotes E. coli (GenBank ECO ... T ...... C. accessionnumber, J01695); AAC, MC N. HAP ...... N N N ...... Actinobacillusactinomycetencomi- HPI ... . N. .. . . N...... N ...... tans (M75036); HAP, Haenophi- UNK CGGCCCAGACTCCTACGGGAGGCAGCAGTGGGGAATATTGCGCAAT GGG lus aphrophilus(M75041); and HPI, H. parainfluenzae(M7508 1). 380 390 400 410 420 The UNK partial 16S rRNA se- ECO ...... T...... T ...... has been in the AAC quence deposited HAP Genome Sequence DataBase(ac- HPI cession number, UNK GCAACCCTGACGCAGCCATGCCGCbTGAATGAAGAAGGCCTTCGGGTTGT L26363).

430 440 450 460 470 ECO ...... A. CG. G...... G. G. A. AG...... CCT.T.C.C.. AAC ...... TT ...... TTgG. GT ...... GC A. . CCAA. . HAP ...GA ...... TTG. . GTG ...... GCGCAACAA.. HPI ...... N ...... G . . A . G . . . . UNK AAAGTTCTTTCGGTAGCGAGGAAGGCATTTAGTTTAATAAACTGGATGAT

480 ECO ...... C. AAC ...... A HAP ...... T HPI UNK TGACGTTAAC

sequence, with 92.9%identity. The sequence fromA. actino- 16S rRNA secondarystructures [ 18]. The lattersubstitutions mycetemicomitans(M75036) was 89.9% identical to the preserve canonical and noncanonical base pairing within -- UNK sequence. these structures(position 473, A-U G-U; position 475, Three nucleotide differencesbetween the UNK sequence G-U > A-U). Selected comparisonswith use of more com- and either or both of the H. parainfluenzaesequences oc- plete 16S rRNA gene sequences (1,483 and 1,479 bases in curred at non-base-pairedpositions (numbered 182, 183, length) from the H. parainfluenzaestrains indicate that the and 469) of 16S rRNA. The two other nucleotidedifferences regionfrom position 94 to position 489 shows more variabil- between the UNK sequence and both H. parainfluenzaese- ity than does the more complete sequence (data not shown). quences occurredat base-pairedpositions within established Since 16S rRNA sequences from differentspecies generally CID 1994; 19 (October) H. parainfluenzae Endocarditis 681

Table 1. Characteristics differentiating HACEK organisms.

Factor required for Expressionor biochemicalactivity of indicatedenzyme growth Productionof acid from Indole Lysine Nitrate Organism X V production Urease Oxidase Catalase decarboxylase reduction Glucose Lactose Sucrose

Haemophilus parainfluenzae - + V V + V V + + + Haemophilus paraphrophilus - + - - + - - + + + +* - - - V - - + + + Actinobacillus actinomnycetemcomitans - - - - V + - + + - - - + - + - - - + + Eikenella corrodens - - - - + - + + kingae - - - - -+ -+

NOTE. Key: + = more than 90%of strainspositive; - = more than 90%of strainsnegative; V = 10%-90%of strainspositive. * Requirementfor Factor X is reportedat initial isolation but is lost on subculture.

differby > 1.5%[19], our similarityscores as well as our den- factor requirements,enzymatic activities, and fermentation drogramare consistent with the placement of UNK within patterns(table 1). These tests often require 10-14 days and the species boundariesfor H. parainfluenzae. may give misleading or variable results, depending on the degree of bacteriallaboratory passage (e.g., the dependence of H. on FactorX the of the Discussion aphrophilus [22]), growthstage organism (a point particularlyrelevant to microorganisms Russell and Fildes [20] firstreported human disease due to that are difficultto cultivate in the laboratory),the manner H. parainfluenzaein 1928, some 6 yearsafter the description in which the test is performed(e.g., the oxidase assay in the of the species by Rivers[21 ]. H. parainfluenzaeis a fastidious case we have reportedherein), and naturallyoccurring dif- organism whose isolation often requiresa long incubation ferencesamong strains.Several commercial kits are available period, specialized media, or terminal subculture.Colonies for relatively rapid identificationof Neisseriaspecies, Hae- of H. parainfluenzaemay grow to 3 mm in diameterafter mophilusspecies, and Moraxellacatarrhalis (e.g., the HNID, incubation at 35?C-37?C for 24 hours and may appearei- API NH, and RapID NH systems) [23, 24]; however, addi- ther flat, grayish,translucent, and smooth or roughand wrin- tional tests are sometimesneeded for the identificationof H. kled. Gram staining may reveal pleomorphicgram-negative parainfluenzae,other Actinobacillusspecies have not been bacilli. In the identificationof H. parainfluenzae,a require- adequatelyevaluated, and misidentificationsdo occur. ment for Factor V should be establishedby the documenta- We reviewed the medical literature from 1966 to July tion of satellitegrowth around nicotinamide adenine dinucle- 1993 and found 37 reportsdescribing 73 cases of H. parain- otide-containing disks on nutritionally deficient medium, fluenzae endocarditis[25-30]. (A complete list of references and the lack of a requirementfor Factor X should be estab- is availablefrom the authorsupon request.)The mean dura- lished by the porphyrintest, with b-aminolevulinicacid as tion of incubationof primaryblood culturesuntil the detec- substrate.Many clinical laboratoriesreport all strains that tion of microbiologicalgrowth (reported in 54 cases) was 6.6 requireonly Factor V as H. parainfluenzae.Thus, it is possi- days, with a rangeof 1 day to 21 days. The intervalfrom the ble that the true prevalenceof infection with this organism detection of microbiologicalgrowth in liquid media to the has never been establishedand that many cases reportedas identificationof H. parainfluenzaewas not stated in most due to H. parainfluenzaemay in fact be caused by other spe- instancesbut was as long as 12 days in one case report[31]. cies that requireonly Factor V, such as H. paraphrophilus. In 48 of the 73 cases, the method used for identificationto Confirmationof species identity is sought only in certain the species level was given. In 11 cases, speciesidentification circumstancesby most laboratoriesand is based on growth- was describedonly in terms of requirementsfor Factors X 682 Hamedet al. CID 1994;19 (October) and V; the results of hemolysis assessment were also in- ships among this group of organisms(data not shown). On cluded in two of these cases. Both growth-factorrequire- the other hand, this region would not be adequate for the ments and biochemicalactivities (as outlined in table 1) were phylogenetic analysis of all clinical bacterial isolates. The reportedin 37 cases. In 10 of these cases, the identity of H. UNK sequence is more closely related to the two available parainfluenzaewas confirmedby the CDC. 16S rRNA sequencesfor H. parainfluenzaethan to any other The identificationof fastidiousmicrobial pathogens in gen- sequences;it differsfrom each of the formertwo sequencesat eral is a problem of potentially wide magnitude. Even if a fewer than 1.5%of positions. A phylogenetic tree places fastidious microorganismcan be reliably propagatedin the these three strains of H. parainfluenzaein a monophyletic laboratory, microbial phenotypic characteristicsare often groupthat is distinctfrom other species.Thus, our data dem- variableand thereforeunreliable for speciesidentification; in onstrateconvincingly that the clinical isolate UNK belongs addition, their determinationmay require lengthy periods. within the species H. parainfluenzae.The minor sequence The identificationof H. parainfluenzaeby traditionalmeth- heterogeneitythat we observed may indicate strain-related ods illustratesthese points. Many of the phenotypiccharac- differences[9] or differencesamong the multiple copies of teristicsused to distinguishamong Haenophilus species and the rRNA operon in some organisms[35]. other HACEKorganisms can varyamong strainsof the same Although our procedure-from cell digestionto sequence purportedspecies (table 1), in part because assay methods analysis-consumed -48 hours, it could easily have been are poorlystandardized. In addition,these phenotypesproba- shortened. Specifically,the time requiredfor cell digestion bly do not provide a means for unified and unambiguous could have been reduced to 1 hour, and one of the auto- classificationof membersof the family Pasteurellaceae[ 16]. mated DNA-sequencingsystems, which permit much more Small-subunit(16S) rRNA sequenceanalysis may provea rapidcollection of data [36, 37], could have been used. With rapidand specificapproach to the identificationof fastidious these modifications,the total time requiredfor species iden- ~ microorganisms.These sequences can be amplifieddirectly tificationmight be reducedto 12 hours. from infected host tissue-with consequent elimination of The clinical impactof this information,while limited,may the need for laboratorycultivation-and from minute quan- include more precisepredictions of the courseof diseaseand tities of culture material [6, 10]. Small-subunitrRNA se- the outcome of therapy. For example, given the preceding quences currentlyform the basis for a revised approachto discussion,the literatureconcerning H. parainfluenzaeendo- microbial taxonomy [11] and have been used to establish carditismust be viewed with some caution, since some iso- phylogenetic relationshipsamong many ,including lates classifiedwithin this species may in actualityhave been membersof the genus Mycobacterium[9, 32, 33], members membersof other species or genera (e.g., Actinobacillus).In of the familyPasteurellaceae [ 16, 34], and severalpreviously addition, a delay in species identification may permit the uncharacterizedhuman pathogens [6-8]. Analyses of the growth of relatively large valvular vegetations associated Pasteurellaceaehave demonstrated four major clusters of with the HACEK organisms,which-given the particularly strains,three of which correspondapproximately to the gen- friablenature of these vegetations-may lead to a high fre- era Haemophilus,Actinobacillus, and Pasteurella.However, quency of embolization in cases of HACEK endocarditis. some members of each genus appear to have been errone- Direct amplification of microbial small-subunitrRNA se- ously classifiedby traditionalmethods. For example, A. ac- quences from blood may furtherreduce such delays in diag- tinomnycetemcomitansshould probably be reclassifiedas a nosis. Althoughthe technicalexpertise required for this kind species of Haemophilus[16]. Conversely, members of the of approachmay not be availablein most clinical laborato- species H. parainfluenzae,which seems to includegenetically ries, the use of automatedDNA-sequencing procedures and heterogeneousstrains, are more closely related to Actinoba- simplifiedprotocols may allow regionalreference centers to cillusthan to Haemophilus.Some strainsof H. parainfluenzae offersmall-subunit rRNA-based identification of diverse,fas- may have been originally misidentified-e.g., as H. par- tidious microbialisolates in the near future. aphrophilus-and vice versa. More recent 16S rRNA-based analysis suggests that H. parainfluenzaemay consist of a monophyletic cluster distinct from other known related or- Acknowledgments ganisms [34]. The authorsthank TerryL. Lewis for valuableadvice and 16S rRNA from a culti- We analyzed a partial sequence expertiseand HarryB. Greenbergand SuzanneM. Matsuifor vated bacterialpathogen purportedto be H. parainfluenzae technicalsupport. in orderto clarifyambiguities about its identity and to dem- onstrate the potential usefulness of this approach for the identification of fastidious microorganismsto the species References level. The of the 16S rRNA that we is region gene analyzed 1. KuklinskaD, Kilian M. Relative proportionsof Haemophilusspecies in one of the most variablefor membersof the Pasteurellaceae the throat of healthy children and adults. Eur J Clin Microbiol and is sufficientto reproducethe basic phylogeneticrelation- 1984;3:249-52. CID 1994; 19 (October) H. parainfluenzae Endocarditis 683

2. Hand WL. Haemophilusspecies. In: MandellGL, Douglas RG Jr, Ben- 21. Rivers TM. Influenza-likebacilli; growth of influenza-likebacilli on nett JE, eds. Principlesand practice of infectious diseases. 3rd ed. media containingonly an auto-clave-labilesubstance as an accessory New York:Churchill Livingstone, 1990:1729-33. food factor. Bull Johns Hopkins Hosp 1922;33:429-31. 3. BlackCT, KupferschmidJP, West KW, GrosfeldJL. Haemophiluspar- 22. Kilian M. A taxonomic study of the genus Haemophilus,with the pro- ainfluenzaeinfections in children, with the reportof a unique case. posal of a new species. J Gen Microbiol 1976;93:9-62. Rev Infect Dis 1988;10:342-6. 23. Janda WM, BradnaJJ, Ruther P. IdentificationofNeisseria spp., Hae- 4. Geraci JE, Wilson WR. Endocarditisdue to gram-negativebacteria. mophilusspp., and other fastidiousgram-negative bacteria with the Report of 56 cases. Mayo Clin Proc 1982;57:145-8. MicroScanHaemophilus-Neisseria identification panel. J Clin Micro- 5. Kilian M. Haemophilus.In: Balows A, HauslerWJ Jr, HerrmannKL, biol 1989;27:869-73. IsenbergHD, Shadomy HJ, eds. Manual of clinical microbiology. 24. BarbeG, BabolatM, BoeufgrasJM, Monget D, FreneyJ. Evaluationof 5th ed. Washington, DC: American Society for Microbiology, API NH, a new 2-hour system for identificationof Neisseriaand 1991:463-70. Haelophilus species and Moraxellacatarrhalis in a routine clinical 6. Relman DA, Loutit JS, Schmidt TM, Falkow S, Tompkins LS. The laboratory.J Clin Microbiol 1994;32:187-9. agent of bacillaryangiomatosis-an approachto the identificationof 25. Chunn CJ, Jones SR, McCutchanJA, Young EJ, Gilbert DN. Haeemo- N unculturedpathogens. Engl J Med 1990;323:1573-80. philus parainfluenzaeinfective endocarditis. Medicine (Baltimore) 7. BottgerEC, Teske A, KirschnerP, et al. Disseminated"Mycobacterium 1977;56:99-113. infection in with AIDS. Lancet genavense" patients 1992;340:76-80. 26. GeraciJE, WilkowskeCJ, Wilson WR, WashingtonJA II. Haemophi- 8. Relman DA. The identificationof unculturedmicrobial J pathogens. lus endocarditis. Report of 14 patients. Mayo Clin Proc Infect Dis 1993;168:1-8. 1977;52:209-15. 9. R, Wilson KH. Molecular of the Frothingham phylogeny Mycobacte- 27. Lynn DJ, Kane JG, ParkerRH. Haeimophilusparainfluenzae and in- avium demonstrates divisions.J riumi complex clinically meaningful fluenzae endocarditis:review of forty cases. Medicine (Baltimore) Infect Dis 1994;169:305-12. 1977;56:115-28. 10. Relman DA. Universalbacterial 16S rDNA and amplification sequenc- 28. Ellner JJ, Rosenthal MS, LernerPI, McHenryMC. Infectiveendocar- In: DH, Smith TF, Tenover FC, White TJ, eds. ing. Persing Diagnos- ditiscaused by slow-growing,fastidious, gram-negative bacteria. Med- tic molecular and microbiology:principles applications. Washing- icine (Baltimore) 1979;58:145-58. ton, DC: American for 1993:489-95. Society Microbiology, 29. Jemsek JG, GreenbergSB, Gentry LO, Welton DE, Mattox KL. Hae- 11. Woese CR. Bacterialevolution. MicrobiolRev 1987;51:221-71. mophilusparainfluenzae endocarditis. Two cases and review of the 12. Lane DJ, Pace B, Olsen GJ, Stahl DA, ML, Pace NR. Sogin Rapid literaturein the past decade. Am J Med 1979;66:51-7. determinationof 16S ribosomal RNA sequences for phylogenetic 30. RaucherB, Dobkin J, Mandel L, EdbergS, Levi M, Miller M. Occult analyses. Proc Natl Acad Sci USA 1985;82:6955-9. polymicrobialendocarditis with Haemophilusparainfluenzae in intra- 13. SambrookJ, Fritsch EF, Maniatis T. Molecularcloning: a laboratory venous drugabusers. Am J Med 1989;86:169-72. manual.2nd ed. Cold SpringHarbor, NY: Cold SpringHarbor Labo- 31. Gordon AM, Love WC. Endocarditisdue to Haemophilusparainfluen- ratoryPress, 1989. zae. J Med Microbiol 1970;3:550-4. 14. Devereux J, HaeberliP, SmithiesO. A comprehensiveset of sequence 32. Stahl DA, Urbance JW. The division between fast- and analysisprograms for the VAX. Nucleic Acids Res 1984;12:387-95. slow-growing species correspondsto natural relationshipsamong the mycobac- 15. Feng DF, Doolittle RF. Progressivesequence alignmentas a prerequi- teria. J Bacteriol1990;172:116-24. site to correctphylogenetic trees. J Mol Evol 1987;25:351-60. 33. PitulleC, Dorsch M, KazdaJ, WoltersJ, StackebrandtE. of 16. DewhirstFE, PasterBJ, Olsen I, FraserGJ. Phylogenyof 54 representa- Phylogeny members of the Int J tive strainsof species in the family Pasteurellaceaeas determinedby rapidly growing genus Mvcobacteriumz. Syst Bacteriol1992;42:337-43. comparisonof 16S rRNA sequences.J Bacteriol1992; 174:2002-13. 34. DewhirstFE, PasterBJ, Olsen I, FraserGJ. of the Pasteurel- 17. BrosiusJ, Palmer ML, Kennedy PJ, Noller HF. Complete nucleotide Phylogeny laceae as determined of 16S ribosomal ribonucleic sequence of a 16S ribosomalRNA gene from Escherichiacoli. Proc by comparison Natl Acad Sci USA 1978;75:4801-5. acid sequences. Int J Med Microbiol Virol Parasitol Infect Dis 18. Guttell RR, WeiserB, Woese CR, Noller HF. Comparativeanatomy of 1993;279:35-44. 16-S-like ribosomal RNA. Prog Nucleic Acid Res Mol Biol 35. CarbonP, EhresmannC, EhresmannB, Ebel JP. The complete nucleo- 1985;32:155-216. tide sequence of the ribosomal 16-S RNA from Escherichiacoli. Ex- 19. Fox GE, WisotzkeyJD, JurtshukP Jr. How close is close: 16S rRNA perimental details and cistron heterogeneities. Eur J Biochem sequence identity may not be sufficientto guaranteespecies identity. 1979; 100:399-410. Int J Syst Bacteriol1992;42:166-70. 36. Ruano G, Kidd KK. Coupledamplification and sequencingofgenomic 20. Russell DS, Fildes P. A case of endocarditisin man associated with DNA. Proc Natl Acad Sci USA 1991;88:2815-9. Bacillus parainfluenzae, Rivers, 1922. J Pathol Bacteriol 37. TracyTE, MulcahyLS. A simplemethod for directautomated sequenc- 1928:31:651-6. ing of PCR fragments.Biotechniques 1991; 11:68-75.