1985 Coronaviruslike Particles in Human Gastrointestinal Disease

Coronaviruslike Particles inHuman Gastrointestinal Disease

Epidemiologic, Clinical, and Laboratory Observations

Mary Lund Mortensen, MD; C. George Ray, MD; Claire M. Payne, PhD;Allan D. Friedman, MD; Linda L. Minnich, MS; Celine Rousseau, MD

\s=b\Coronaviruslike particles (CVLPs)were visualized by direct electron micro-scopy (EM) of diarrheal stools in 49 of126 infants and children between 1month and 12 years ofage during a three\x=req-\year observation period. The clinical andepidemiologic features of these patientswere analyzed and compared with pa-tients with diarrhea who were sheddingrotaviruses, or whose stools were nega-tive for enteric viruses by EM. Seasonaland age distributions for CVLP sheddingwere similar to those for rotaviruses (inmost cases less than 1 year of age; peakmonths were September through Janu-ary), as were the symptoms and mediandurations of illness. Prospective studiesof three subsequent patients suggestthat the duration of shedding in acuteillness is five to at least 25 days. Multipleattempts to cultivate the CVLPs wereunsuccessful. In addition, partial pu-rification of CVLPs from stool spec-imens was performed, and immunologicanalysis by immunoelectron micros-copy and radial immunodiffusionshowed no antigenic relatedness to pro-totype human (OC43 and 229E) or animal(bovine and canine) coronaviruses.These findings suggest that CVLPs maybe an important cause of acute gastroin-testinal illness in infancy, and may repre-sent a virus antigenically unrelated toknown human and animal coronavi-ruses.

(AJDC 1985;139:928-934)

From the Departments of Pediatrics (Drs Mor-tensen, Ray, Friedman, and Rousseau) andPathology (Drs Ray and Payne and Ms Minnich),University of Arizona College of Medicine, Tuc-son.

Reprint requests to Department of Pathology,Arizona Health Sciences Center, 1501 N Camp-bell Ave, Tucson, AZ 85724 (Dr Ray).

Coronaviruses are medium-sizedRNA-containing viruses that have

characteristic club-shaped projectionsor peplomers on their surface.1,2 Someserotypes have been implicated ascausative agents of enteric disease inanimals and man.3"5 Bovine coronavi¬ruses are a cause of enteritis in neo¬

natal calves and have been associatedwith necrotizing enterocolitis in theseanimals.3 Coronaviruslike particles(CVLPs) in fecal samples have beeninconsistently associated with diar¬rhea and other gastrointestinal symp¬toms in humans. Mathan et al6 were

first to report CVLPs seen in stoolsamples from children and adults withtropical sprue. In the same year,CVLPs were associated with an out¬break of explosive diarrhea in youngadults.7 Subsequent reports have im¬plicated CVLPs in outbreaks of diar¬rhea or necrotizing enterocolitis inneonates.8"10 The relationship betweenCVLPs in fecal samples and illness,however, remains controversial, sincesome workers have found CVLPs instools of controls as well as patientswith gastrointestinal disease. Most ofthese latter reports were of children oradults in underdeveloped countries.11,12

Uncertainty about the associationbetween CVLPs and human gastroin¬testinal disease may be due in part tothe lack of detailed epidemiologie in¬formation, such as symptoms in rela¬tion to duration of particle shedding infeces and characterization of at-riskpopulations. Such data are difficult togather because electron microscopy(EM) is the only available means foridentifying CVLPs in fecal samples.

Furthermore, identification of CVLPsrequires a vigilant and experiencedelectron microscopist to distinguishthe particles from other membraneprofiles in the preparation.13 With fewexceptions,14,15 attempts to cultivateCVLPs in organ or tissue culture, or inmycoplasma media, have been unsuc¬

cessful, thereby hampering efforts todevelop an immunologic or other rapiddiagnostic assay.9,12,16

At our institution, an outbreak ofgastrointestinal illness among pa¬tients in the intensive care nursery9and sporadic cases of diarrhea associ¬ated with CVLPs among older infantsand children have suggested thatCVLPs may be an enteric pathogen. Inthis report, we describe clinical andepidemiologie characteristics of symp¬tomatic infants and children over 1month of age with CVLPs, the dura¬tion of CVLP shedding, attempts tocultivate and purify CVLPs, and testsof antigenic relatedness betweenCVLPs and prototype human and ani¬mal coronaviruses.

PATIENTS AND METHODSEpidemiologie and

Clinical Observations

From May 1979 through April 1982, stoolsamples from 126 ill infants and childrenbetween 1 month and 12 years of age, ofwhom all but one were hospitalized, were

studied by EM. Forty-nine of these sampleswere found to contain CVLPs. Of the re¬

mainder, 12 had rotavirus, three had ade-novirus, six had Norwalk-like virus, one

had astrovirus, and 55 were negative byEM. We analyzed the characteristics of thegroup with CVLPs and compared them

Downloaded From: http://archpedi.jamanetwork.com/ by a Western University User on 06/06/2015

with the rotavirus group and EM-negativegroup. The charts of these 126 patientswere reviewed for pertinent clinical andlaboratory data. Epidemiologie data in¬cluded age, sex, seasonal distribution, andduration of symptoms. Clinical symptomsand signs included diarrhea; vomiting;mucousy, green, or bloody stools; fever;and nongastrointestinal symptoms. Labo¬ratory data included: peripheral whiteblood cell counts, hematocrit, differential,platelet counts, stool culture, stool ex¬amination for ova and parasites by lightmicroscopy, viruses by EM, abdominalroentgenograms, and miscellaneous stud¬ies associated with any nongastrointestinaldiagnoses. Statistical comparisons were

made with Student's t test.

Duration of CVLP SheddingSubsequent to the retrospective review,

during June 1982 through March 1983, anyhospitalized nonneonate with CVLPs iden¬tified by EM in a fecal sample was eligible.Bacterial pathogens were not identified infecal samples of these subjects. Twice orthree times weekly, stool samples were

collected and prepared for EM (describedbelow). A subject was considered to haveceased shedding CVLPs when particleswere not observed in two consecutivesamples.

Laboratory InvestigationsVirus Detection.—Electron microscopy

of stool specimens was done by a previouslydescribed method.9 Stool diluted with dis¬tilled water was vortexed and centrifugedat 3,500 rpm for ten minutes, then thesupernatant was removed and centrifugedagain at 3,000 rpm for ten minutes in aclinical centrifuge to remove debris. Super¬natant, placed on 300-mesh copper gridspreviously covered with a carbon-coatedpolyvinyl formal (Formvar) layer, was air-dried, negatively stained with 2% phos-photungstic acid (pH, 6.5), and then di¬rectly examined under an EM. Thepleomorphic appearance of characteristicCVLPs is shown in Fig 1. These particlesare morphologically distinct from respira¬tory coronavirus in that CVLPs have a

flexible-appearing fringe compared withthe more rigid-appearing fringe thatcharacterizes coronaviruses. Coronavirus¬like particles also have more closely spacedpeplomers compared with typical coro¬naviruses.

Cultivation Attempts.—Five to 20 stoolsamples that were positive for CVLPs byEM and five stools negative by EM were

used in each of various culture attempts.These were processed according to stan¬dard methods17 from either fresh sam-

Fig 1.—Representative coronaviruslike particles in negatively stained stool specimen frompatient with gastroenteritis. Eight distinct pleomorphic, fringed, viral-like particles areshown. Fringe surrounding individual viral particles varies in length and appears "flexible."Although ends of some filaments appear "bulbous," others may simply represent end-onviews of the flexible filaments. Among viral-like particles are two nonspecific membranelikeprofiles with thickened electron-lucent areas (phosphotungstic acid, 154,000).

pies or ones that had been kept frozen at-60 °C. Aliquote (0.1 mL) were inoculatedundiluted and diluted 1:10 onto the follow¬ing cell cultures: a rhabdomyosarcoma line(RD), human embryonic intestine (intes¬tine 407), a human rectal tumor line (HRT18), a heteroploid lung line, human fetallung (Flow 2,000), primary cynomolgusmonkey kidney, and four strains of humanembryonic tonsil fibroblasts developed inour laboratory. In addition, six continuouscell lines obtained from the American TypeCulture Collection (Vero, BHK21, LLC-MK2, Madin-Darby canine kidney, Buffalogreen monkey, and SIRC) were used. Allspecimens were observed for cytopathiceffect and passaged on days 5 to 7 for five to15 times.

Processed specimens were also dilutedwith equal volumes of trypsin at concentra¬tions of 250, 25, 2.5, and 0.25 µ-g/mL,incubated at 4 °C, 25 °C, and 35 °C for 30minutes, and inoculated onto intestine 407and two lines of human embryonic tonsilfibroblasts. These were passaged five timeson days 6 and 7 with and without trypsintreatment.

Cultures were also tested for the pres¬ence of hemagglutinin at 4 °C, 25 °C, and35 °C with human O, mouse, and guinea pigerythrocytes.

Mycoplasma culture attempts includedinoculation of SP4 broth and agar plateswith 10% horse serum. The plates wereincubated in 5% carbon dioxide at 35 °C,

and observed for colonies for ten days.Broth cultures were observed for 30 daysfor any changes in pH.

Attempt to Artifactually ProduceCVLP.—Mucosa was teased from a portionof human jejunum obtained at autopsy froma patient without gastrointestinal symp¬toms. The jejunal contents were examinedby EM to determine the absence of CVLPs,thereby excluding the possibility of sub¬clinical infection. A 4-mL aliquot of homog¬enized jejunum was incubated at 31 °C witheither 1 mL of 0.25% trypsin or physiologicsaline solution (control) for ten or 120 min¬utes. At the end of incubation, 3 mL of 5%bovine serum albumin was added. Spec¬imens were immediately prepared for EM.In addition a sample of untreated homoge¬nized and jejunal mucosa was examinedby EM.

Antigen Purification.—To obtain a suf¬ficient quantity for purification, approx¬imately 25 CVLP-positive fecal specimenswere pooled. Ultrapure ammonium chlo¬ride was added to the pooled specimen toproduce a 60% saturated solution. The mix¬ture was stirred at 4 °C for three hours andthen centrifuged at 10,000 rpm (7,800 g) for20 minutes. The supernatant was dialyzedovernight at 4 °C against phosphate-buf¬fered saline (pH, 7.4). In the final stage thedialysate was centrifuged at 90,000 g fortwo hours, resulting in supernatant andpellet. Prior to purification and followingeach step in the procedure, an aliquot of the


Age, mo

Fig 2.—Age distribution of patients. Solid bars indicate coronaviruslikeparticles; shaded bars, rotavirus; and open bars, electron microscopy-negative.

pooled specimen was examined by EM toverify the presence of CVLPs.

Mouse Inoculation.—Suckling whiteSwiss mice (24 hours old) received 0.1 mL ofpellet antigen by the intraperitoneal (n = 4)or intragastric (n = 3) route. During thenext seven days, the animals were observedfor signs of illness (diarrhea, ascites) ordeath. On day 7, one mouse (intragastricinjection) was killed and the gastric con¬tents and large intestine were homogenizedand centrifuged at 3,500 rpm (1,564 g). Thepellet was resuspended in approximately0.5 mL of sterile, distilled water and pre¬pared for EM.

Antisera.—Porcine anti-bovine entericcoronavirus (anti-BEC) and guinea piganti-canine coronavirus (anti-CCV) serawere contributed privately and Guinea piganti-229E serum and murine anti-OC43 as-citic fluid were obtained from the Centersfor Disease Control, Atlanta. Human anti-CVLP serum was obtained from a labora¬tory technologist who had the onset ofCVLP-associated acute diarrhea within 24hours after the accidental aerosolization ofCVLP-positive fecal samples. Pre-illnessserum was that obtained approximatelytwo years before the episode and convales¬cent serum was obtained two weeks afterthe onset of illness. Paired serum sampleswere also obtained from one infant studiedfor duration of CVLP fecal excretion; how¬ever, the earliest available serum was ob¬tained approximately ten days after theonset of diarrhea ("acute"), and a laterpostillness serum specimen was collected30 days after onset.

Immunodiffusion Studies.—Periph-

o

_

MonthFig 3.—Monthly distribution of electron microscopic stool examinations over three-yearperiod. Solid bars indicate coronaviruslike particles; shaded bars, rotavirus; and open bars,electron microscopy-negative.

eral wells of four radial immunodiffusionplates were each filled with 0.1 mL of thefollowing: normal goat serum (two controlwells), CVLP-positive fecal samples (threewells), and rotavirus-positive fecal sample(one well). The central well of each platewas filled with one of the following sera:

anti-BEC, anti-CCV, anti-OC43, or

anti-229E. After incubation at 4 °C theplates were examined at 24 and 48 hours.The study was performed on two occasions,using the same fecal samples and goatserum.

Immuno-EM.—A fecal sample contain-

ing numerous intact, nonaggregatedCVLPs was selected. In addition to anti-sera against the four prototype coronavi¬ruses, rabbit anti-herpes simplex (anti-serum control) and normal goat sera(serum control) were used. Bovine serumalbumin (0.22%) was used to make serialtwofold dilutions of antisera and goat se¬

rum, from 1:20 through 1:320.A 0.05-mL aliquot of antiserum (or se¬

rum) was pipetted into the U-shaped well ofa microtiter plate. Next, 0.05 mL of stoolextract was added to each well and the platewas gently agitated. Following a 45-minute


Clinical and Laboratory Characteristics of the Three Study Groups*Group

CVLP Rotavirus EM-NegativeSymptoms, No. positive/No. reported (%)

Diarrhea 46/49 (94) 12/12 (100) 50/55 (89)Fever (>38°C) 31/49 (63) 7/12 (58) 35/55 (63)Vomiting 25/49 (51) 8/12 (67) 36/55 (64)

Stool characteristics,No. positive/No. reported (%)

Occult blood 7/38 (18) 0/12 (0) 12/30 (40)Watery 25/38 (66) 11/12 (92) 43/55 (77)Mucoid 12/38 (32) 1/12 (8) 13/43 (24)Green 10/38 (26) 5/12 (42) 18/55 (32)

Duration of symptoms, daysMean 18 12 12Median 7 5 9

Range 2-60 2-90 1-60Peripheral blood counts, mean

WBCs/cumm 14,200 10,600 12,600PMNs/Cumm 5,274 4,322 4,361Hematocrit 35.8 32.7 34.7Platelets/cu mm 449,000 412,000 466,000

*CVLP indicates coronaviruslike particle;PMNs, polymorphonucleocytes.

EM, electron microscopy; WBCs, white blood cells; and

incubation at 37 °C, an EM grid was dippedinto each well, air-dried, and processed as

previously described.The electron microscopist examining and

scoring immune aggregates on the gridswas blinded to the specific antiserum used,and to the sequences of antisera and dilu¬tions. For each microtiter well, ten fieldsper grid were examined at 4,800 magni¬fication and scored for clumping of envel¬oped particles (0 for no immune aggregatesto 4 for extensive aggregation). The mean ofthe ten fields was used for comparisonamong the different antisera.

RESULTSEpidemiologie and

Clinical Observations

Epidemiology.—Of the 49 patientswith CVLPs in their stools, 59% were

male, 88% were 2 years old or less, and71% were 1 year old or less. The medianage was 7 months (Fig 2). Althoughsporadic cases occurred throughoutthe year, the majority were detected inthe fall and early winter (Fig 3).Thirty-eight (78%) of the cases oc¬curred between September and theend of January.

Nine patients (18%) with CVLP alsohad at least one other identifiable stoolpathogen. These included Salmonella(four patients), Shigella (three pa¬tients), Campylobacter (one patient)

and an enterovirus (one patient).Nine of the 12 patients with rota¬

virus were male. Ages ranged from 2months to 12 months, with a median of7 months. Eleven of these patientswere ill in the fall and early winter.Three patients with rotavirus infectionalso had other pathogens concurrentlyidentified in their stools: two patientshad Norwalk-like particles and onehad Shigella.

Sixty-four percent of the patientswith EM-negative stools were male,with ages ranging from five weeks to 11years. The median age was 7 months.The seasonal distribution for thisgroup is shown in Fig 2. No otherpathogens were detected in these pa¬tients.

Clinical Findings.—The clinicalcharacteristics of patients in each ofthe three groups (CVLP-positive,rotavirus-positive, and EM-negative)are summarized in the Table.

The median duration of symptoms inall patients with CVLP was seven

days, compared with five and nine daysin the rotavirus- and EM-negativegroups, respectively. These differ¬ences were not statistically signifi¬cant.

The most common gastrointestinalsymptoms described in the CVLP

group included diarrhea (94%), fever(63%), and vomiting (51%). The pres¬ence of these symptoms was not signif¬icantly influenced by the presence ofanother pathogen in addition toCVLPs. The occurrence of thesesymptoms in the other two groups wassimilar to the CVLP group. Therewere no statistically significant differ¬ences. Stools were commonly de¬scribed as watery (66%), green (26%),and mucoid (32%) in the CVLP group,and occult blood was detected in thestools of seven patients (18%). Nogrossly bloody stools were reported.Of these stool characteristics, only oc¬

cult blood in the stool was more fre¬quent in the EM-negative group thanthe CVLP group (P<.05). None of therotavirus group had occult blood. How¬ever, only slightly more than one half ofthe patients with stools negative byEM had stools tested for blood; stoolguaiac tests were not done on one

fourth of the CVLP group. Greenstools were more common in the rota¬virus group and mucus was more com¬mon in the CVLP group, but thesedifferences were not statistically sig¬nificant. Comparisons of peripheralblood cell counts (Table) showed no

significant distinguishing characteris¬tics among the three groups. Statis¬tical evaluation was also done that ex¬

cluded all patients with more than one

intestinal pathogen. This did notchange the analysis.

Duration of Shedding.—Duringthe fall of 1982, three hospitalized in¬fants with CVLPs identified in fecalspecimens were serially studied.

A 1-month-old boy was found to haveCVLPs after 13 days of vomiting anddiarrhea. An extensive workup ruledout bacterial pathogens (Salmonella,Shigella, Yersinia enterocolitica,Campylobacter jejuni, and toxin-pro¬ducing Clostridium difficile), intes¬tinal parasites, immunoglobulin defi¬ciency, and metabolic disease. TheCVLPs were detected in fecal samplesfor an additional five days, and thedisappearence of the particles coin¬cided with the resolution of diarrhea.

A 2-month-old boy with chronic lungdisease had fever with a temperatureof 39 °C and loose stools, but no othersymptoms of illness. Bacterial culturesof blood, urine, and cerebrospinal fluid


Fig 4.—Composite electron micrographs comparing typical coronaviruslike particles withjejunal homogenates (phosphotungstic acid, 154,000). Top left, Typical coronaviruslikeparticle from patient with gastrointestinal illness. Note particle's irregular shape, thick fringe,and lack of stain penetration into particle interior. Top right, Membrane profile from jejunalhomogenate incubated at 37 °C (ten minutes) without trypsin. Note distinct surface fringeand penetration of negative stain (electron-dense areas) into particle. Bottom left, Mem¬brane profile from jejunal homogenate incubated at 37 °C (ten minutes) with trypsin. Noteirregular, jagged contour and penetration of negative stain (large electron-dense areas) intoparticle interior. Bottom right, Smooth-surfaced membrane profile from jejunal homogenateincubated at 37 °C (ten minutes) with trypsin.

were sterile, as were viral cultures ofcerebrospinal fluid and nasopharyn¬geal and rectal swabs. There were

CVLPs in fecal samples collected overthe next 25 days. The patient was

intermittently febrile for seven days.Although his stools were noted to besomewhat loose and occasionally wa¬

tery, he continued to tolerate gavagefeedings of breast milk without diffi¬culty.

A 3-month-old girl had watery diar¬rhea and required intravenous hydra¬tion and parenteral alimentation. Thissecretory diarrhea subsided only afterall oral feedings were discontinued at14 days after onset. No enteric patho¬gens were isolated from the feces, butCVLPs were seen in a sample submit¬ted 19 days after the onset of diarrheaand CVLPs were shed in the feces foran additional four days.

Laboratory Investigations

Cultivation Attempts.—All at¬tempts to cultivate the CVLPs failed.Aside from some irregularly observed,nonspecific cytopathic effects thatcould not be serially passaged or con¬

firmed by EM of the culture superna¬tant or a cell pellet, the only isolateswere two polioviruses of presumedvaccine origin and one adenovirus. Nodetectable hemagglutinin was found,and all mycoplasma cultures were neg¬ative.

Mouse Inoculation.—None of thesuckling mice became ill or died duringthe seven days of observation. NoCVLPs were noted in the intestinalhomogenate of a mouse killed on theseventh day after intragastric inocula¬tion.

Attempts to Produce CVLPs Ar-tifactually.—The jejunal contentswere negative for CVLPs by EM. Acharacteristic CVLP (Fig 4, top left)with a surface fringe approximately20 nm thick was used to compare withthe homogenized jejunum (Fig 4, topright, bottom left and right). A non-

trypsinized jejunal homogenate incu¬bated at 37 °C for ten minutes disclosedthe presence of membrane profiles,some of which displayed a surfacefringe approximately 10 nm thick(Fig 4, top right). The individual fila¬ments that made up the fringe were ofa width (1.9 to 3.2 nm) similar to thatassociated with characteristic CVLPs(Figs 1 and 4, top left). The CVLPshad, in addition, bulbous tips, some ofwhich measured up to 5 nm in diame¬ter. Jejunal homogenates trypsinizedfor ten minutes showed either mem¬

brane profiles with a jagged irregularsurface fringe (Fig 4, bottom left)or smooth-surfaced membrane pro¬files with no fringe (Fig 4, bottomright). Homogenates exposed to tryp¬sin for two hours contained a sparsenumber ofsmooth-surfaced membraneprofiles. In addition to the filamentcharacteristics, the negative-contraststaining features of typical CVLPsdiffered from the membrane profilesfound in the homogenate. The CVLPsappear to exclude the negativestain from the interior of the mem¬

brane profiles (Figs 1 and 4, topleft), whereas the fringed membrane


profiles from the homogenizedjejunum appear to allow the stainto penetrate (Fig 4, top right andbottom left). Despite an exhaustivesearch, no typical CVLPs were identi¬fied in any of the homogenate prepara¬tions.

Antigen Purification.—Absence ofrotavirus antigen in the pooled stoolspecimen was confirmed using a com¬mercial enzyme-linked immunosor¬bent assay (ELISA) kit. The partiallypurified supernatant fraction con¬tained greater than ten CVLPs per300-mesh grid square. Some of theparticles had characteristic fringes.No flagella and few nonspecific mem¬

brane profiles were seen. The pelletfraction contained greater than 100CVLPs per 300-mesh grid square. Nocontaminants were seen and the parti¬cles did not appear to be clumped. Thefractions were used without success inattempts (not detailed herein) to de¬velop specific antisera in rabbits toapply ELISA methods for CVLP anti¬gen detection and characterization.

Immunodiffusion.—There was noevidence of reaction between the pro¬totype antisera (BEC, CCV, 229E,and OC43) or rotavirus antiserum andCVLP antigens in the stool samples.Similarly, the goat serum control didnot react with any of the stool anti¬gens.

Immuno-EM.—For all anticorona-virus sera tested, aggregates of par¬ticle envelopes appeared at randomserum dilutions. These aggregatesconsisted of both fringed and non-

fringed membrane profiles, as previ¬ously described and illustrated.9 Noconsistent pattern of enhanced aggre¬gation with changing serum dilutionwas discernible for any of these anti-sera. These aggregates were as likelyto be seen in control wells (anti-herpessimplex virus serum) as in a well con¬

taining prototype antiserum of thesame dilution.

COMMENTA variety of coronaviruses or "coro¬

naviruslike" agents have been de¬tected in mammalian and avian spe¬cies. Langpap et al3 demonstrated thepresence of coronaviruses in the stoolsof calves by negative-staining EM andby fluorescent antibody examination

of intestine and colon tissue sections.The calves ranged in age from 1 day to 3months. Green, mucoid, fecal castswere seen in the colon at necropsy andthe histologie characteristics were

compatible with necrotizing enteroco¬litis.3 This bovine enteric coronavirushas been shown to possess membraneantigens that cross-react with humanrespiratory coronavirus OC43.18

The association of CVLPs with gas¬trointestinal symptoms in humans hasbeen reported recently, and their rolein such disease is less clear. Identical-appearing particles have also beendemonstrated in the stools of asympto¬matic persons, making it more difficultto prove a causal relationship.5,11,12

Caul and Clarke15 have reported suc¬cessful in vitro propagation of CVLPsfrom one patient. More recently,CVLPs were associated with gastro¬enteritis in children and adults resid¬ing in England and France, two ofwhom also had a hemolytic-uremicsyndrome19; Gema et al20 found CVLPsby EM in 34 (16.3%) of 208 Italianinfants and children with acute gastro¬enteritis and only three (1.6%) of 182age-matched control subjects.20

Evidence has been presented to sup¬port an association between neonataldisease and the presence of CVLPs instool. We have previously described anoutbreak of gastrointestinal disease ina neonatal intensive care unit, and a

subsequent 40-week survey demon¬strated a statistically significant asso¬ciation between the presence of CVLPand gastrointestinal symptoms, in¬cluding blood in stools, watery stools,gastric retention, bilious gastric aspi¬rates, and abdominal distension.9 MostCVLPs were identified during the falland early winter.

Chany et al10 have reported a clusterof cases of neonatal necrotizing entero¬colitis associated with CVLPs in thestool. The association occurred in bothpremature and full-term infants. Ahigh incidence of CVLPs was alsofound in the staff members caring forthese infants.

We have characterized the gastroen¬teritis associated with CVLPs amonghospitalized patients beyond the neo¬

natal age group. Over a three-yearperiod, the incidence of CVLPs infeces was greatest in fall and early

winter. Diarrhea was the most com¬mon symptom (94% of patients), butvomiting and fever were frequentlypresent (63% and 51%, respectively).Stools were often characterized as be¬ing watery, green, and mucus-contain¬ing.

Seasonal distribution was similar forrotavirus and CVLP-associated gas¬troenteritis, with the peak incidencefor both occurring in fall and earlywinter. The age distribution for bothagents was similar, with the majorityof patients being less than 3 years old.The clinical manifestations and dura¬tion of illness associated with bothagents were not significantly differ¬ent; however, occult blood was more

frequent in the stools of CVLP-in-fected than rotavirus-infected pa¬tients. Smears and Gram stains ofacute-phase fecal specimens were notregularly recorded, so data were notavailable to determine if the heme-positive stools reflect a greater entero-invasiveness of CVLPs than rotavirus.

In samples of nonhuman primatefeces, CVLPs were rarely found innonweaned compared with weanedsimians (1/35 vs 23/34, respectively).21The protective effect of nursing mayextend to humans; premature infantsreceiving intensive care are oftenweaned, and seem to be at risk forsevere CVLP-associated disease.9,10The relationship between gastrointes¬tinal illness and weaning status of in¬fants in our retrospective study wasnot examined because of incompleteinformation about whether infantswere formula- or breast-fed, or if re¬cent changes in feeding practice hadbeen made prior to the illness.

The three patients prospectivelystudied were not sufficient to deter¬mine specific correlation between fecalCVLP excretion and duration or sever¬

ity of illness; it is apparent that a

larger number of patients will need tobe similarly studied to determinethese relationships.

Because some investigators havequestioned the accuracy of CVLP di¬agnosis by EM,13,22 we attempted topropagate the agent in cell culture andsuckling mice, and to develop an

ELISA for detecting fecal CVLPs. Asin other studies, we were unable tocultivate the agent in a variety of cell


lines or by using mycoplasma media.We were unsuccessful in producinggastrointestinal symptoms by inocu¬lating suckling mice with the purifiedCVLP antigen. Because of the smallamount ofantigen available, we did notorally infect the animals. In the future,this and perhaps other routes shouldbe considered.

To determine if CVLPs might be anartifact of enzyme action on the muco¬sal lining of the intestine, we treated a

jejunal homogenate with trypsin. Ourinability to produce any particles re¬

sembling CVLPs suggests that theagent is not artifactual. The fact thatCVLPs can also be observed in thestools of symptomatic adults precludesthe possibility that CVLPs are ar¬tifacts produced only in the intestinalcontents of children infected with an as

yet undiagnosable virus. The defini¬tive proof of their viral nature maycome from the demonstration of nu¬

cleic acid in these particles and their

ultimate propagation.We were unable to purify fecal

CVLPs sufficiently to determinewhether the particles are antigeni-cally unrelated to prototype coronavi¬ruses (OC43, 229E, CCV, or BEC).However, when antisera to prototypecoronaviruses were used for immuneEM, only spontaneous, random aggre¬gation was seen. Similarly, immunodif-fusion results indicated a lack of anti¬genic sharing between CVLPs and theprototype coronaviruses. Gema et al20suggested antigenic relatedness be¬tween their patients' CVLPs and OC43coronavirus, but others have sug¬gested that CVLPs are antigenicallyunique. Beards et al19 showed anti¬genic unrelatedness of their humansubjects' CVLPs to prototype corona¬

viruses, and suggested that these mayrepresent a new viral genus, related tothe Breda and Berne viruses of calvesand horses, respectively.19

The lack of consistent results using

convalescent sera in immune EM ex¬

periments may reflect the presence ofa dominant local immune response insome or most patients. Clarification ofthis will require further investigationofhumoral and mucosal immune mech¬anisms, perhaps utilizing methodsother than immuno-EM.

In conclusion, we have described theepidemiology and clinical features ofCVLP-associated gastroenteritis inhospitalized nonneonates. Using sev¬eral techniques, we were unable eitherto cultivate or to purify adequately theagent. We believe that these particlesare of potential importance in humandisease, and do not merely representfecal or mucosal artifacts. However,better definition of their role in diseasemay require other methods of in vitrocultivation and antigen purification.

Brian A. David, PhD, provided the HRT 18,Alan Liss, PhD, Binghamton, NY, provided theSP4 broth and agar plates, and E. A. Carbrey,VMD, MS, Ames, Iowa, provided the anti-CCVand anti-BEC sera.

References1. Tyrrell DAJ, Bynoe ML: Cultivation of a

novel type of common-cold virus in organ culture.Br Med J 1965;1:1467-1470.

2. McIntosh K, Dees JH, Becker WB, et al:Recovery in tracheal organ cultures of novelviruses from patients with respiratory disease.Proc Nati Acad Sci USA 1967;57:933-940.

3. LangpapJJ, Bergeland ME, Reed DE: Cor-onaviral enteritis of young calves: Virologic andpathologic findings in naturally occurring infec-tions. Am J Vet Res 1979;40:1475-1478.

4. Appel MJ, Cooper BJ, Greisen H, et al:Canine viral enteritis: Status report on corona\x=req-\and parvo-like viral enteritis. Cornell Vet 1979;69:123-133.

5. Clarke SKR, Caul EO, Egglestone SI: Thehuman enteric coronaviruses. Postgrad Med J1979;55:135-142.

6. Mathan M, Mathan VI, Swaminathan SP,et al: Pleomorphic virus-like particles in humanfaeces. Lancet 1975;1:1068-1069.

7. Caul EO, Paver WK, Clarke SKR: Coro-navirus particles in faeces from patients withgastroenteritis. Lancet 1975;1:1192.

8. SeigelJD, Luby JP, Laptook AR, et al:Identification of coronavirus (CRNV) inapre-

mature nursery during an outbreak ofnecrotizingenterocolitis (NEC) and diarrhea (D). PediatrRes 1983;17(suppl):181A.

9. Vaucher YE, Ray CG, Minnich LL, et al:Pleomorphic, enveloped, virus-like particles as-sociated with gastrointestinal illness in neonates.J Infect Dis 1982;145:27-36.

10. Chany C, Moscovi O, Lebon P, et al: Associ-ation of coronavirus infection with neonatal nec-

rotizing enterocolitis. Pediatrics 1982;69:209-214.11. Marshall JA, Birch CJ, Williamson HG,

et al: Coronavirus-like particles and other agentsin the faeces of children in Efate, Vanuatu. J TropMed Hyg 1982;85:213-215.

12. Puel JML, Orillac MS, Bauriaud RM, et al:Occurrence of viruses in human stools in theAhaggar (Algeria). J Hyg 1982;89:171-174.

13. Dourmashkin RR, Davies HA, Smith H,et al: Are coronavirus-like particles seen in diar-rhea stools really viruses? Lancet 1980;2:971-972.

14. Caul EO, Egglestone SI: Further studieson human enteric coronaviruses. Arch Virol 1977;54:107-117.

15. Caul EO, Clarke SKR: Coronavirus propa-gated from patient with non-bacterial gastroen-teritis. Lancet 1975;2:953-954.

16. Baker SJ, Mathan M, Mathan VI, et al:Chronic enterocyte infection with coronavirus:One possible cause of the syndrome of tropicalsprue? Dig Dis Sci 1982;27:1039-1043.

17. Lennette EH, Schmidt NJ: DiagnosticProcedures for Viral, Rickettsial and Chla-mydial Infections, ed 5. New York, AmericanPublic Health Association, 1979.

18. Gerna G, Battaglia M, Cereda PM, et al:Reactivity of human coronavirus OC43 and neo-natal calf diarrhea coronavirus membrane-asso-ciated antigens. J Gen Virol 1982;60:385-390.

19. Beards GM, Hall C, Green J, et al: Anenveloped virus in stools of children and adultswith gastroenteritis that resembles the Bredavirus of calves. Lancet 1984;1:1050-1052.

20. Gerna G, Passerani N, Cereda PM, et al:Antigenic relatedness of human enteric coro-navirus strains to human coronavirus OC43: Apreliminary report. J Infect Dis 1984;150:618-619.

21. Smith GC, Lester TL, Heberling RL, et al:Coronavirus-like particles in nonhuman primatefeces. Arch Virol 1982;72:105-111.

22. Caul EO, Ashley CR, Egglestone SI: Rec-ognition of human enteric coronaviruses by elec-tron microscopy. Med Lab Sci 1977;34:259-263.

Librarian's Note.—See also: Rettig PJ, Altshuler GP: Fatal gastroenteritis associatedwith coronaviruslike particles (AJDC 1985;139:245-248).


1985 Coronaviruslike Particles in Human Gastrointestinal Disease

Documents