of January 14, 2019.This information is current as

to VaccinationLinkedMechanism of Neonatal Pancytopenia

Alloantibodies against MHC Class I: A Novel

SchelcherLacroux, Cyrielle Franchi, Odile Burlet-Schiltz and FrançoisPichereaux, Cécile Caubet, Catherine Trumel, Caroline Gilles Foucras, Fabien Corbière, Christian Tasca, Carole

http://www.jimmunol.org/content/187/12/6564doi: 10.4049/jimmunol.1102533November 2011;

2011; 187:6564-6570; Prepublished online 14J Immunol 

Referenceshttp://www.jimmunol.org/content/187/12/6564.full#ref-list-1

, 9 of which you can access for free at: cites 44 articlesThis article

        average*  

4 weeks from acceptance to publicationFast Publication! •    

Every submission reviewed by practicing scientistsNo Triage! •    

from submission to initial decisionRapid Reviews! 30 days* •    

Submit online. ?The JIWhy

Subscriptionhttp://jimmunol.org/subscription

is online at: The Journal of ImmunologyInformation about subscribing to

Permissionshttp://www.aai.org/About/Publications/JI/copyright.htmlSubmit copyright permission requests at:

Email Alertshttp://jimmunol.org/alertsReceive free email-alerts when new articles cite this article. Sign up at:

Print ISSN: 0022-1767 Online ISSN: 1550-6606. Immunologists, Inc. All rights reserved.Copyright © 2011 by The American Association of1451 Rockville Pike, Suite 650, Rockville, MD 20852The American Association of Immunologists, Inc.,

is published twice each month byThe Journal of Immunology

by guest on January 14, 2019http://w

ww

.jimm

unol.org/D

ownloaded from

by guest on January 14, 2019

http://ww

w.jim

munol.org/

Dow

nloaded from

The Journal of Immunology

Alloantibodies against MHC Class I: A Novel Mechanism ofNeonatal Pancytopenia Linked to Vaccination

Gilles Foucras,*,† Fabien Corbiere,*,† Christian Tasca,*,† Carole Pichereaux,‡,x,{

Cecile Caubet,*,† Catherine Trumel,* Caroline Lacroux,*,† Cyrielle Franchi,*

Odile Burlet-Schiltz,‡,x and Francois Schelcher*,†

Fetal/neonatal alloimmune thrombocytopenia is a frequent disease in humans where alloantibodies against platelet Ags lead to plate-

let destruction and hemorrhage. Although a role in the disease for Abs against MHC has been suspected, this has not been formally

demonstrated. Since 2007, a hemorrhagic syndrome due to thrombocytopenia and designated as bovine neonatal pancytopenia

(BNP) has been recognized in calves in several European countries. An inactivated antiviral vaccine is strongly suspected to be

involved in this syndrome because of its highly frequent use in the dams of affected calves. In this study, we show that BNP is

an alloimmune disease, as we reproduced the signs by transferring serum Abs from vaccinated BNP dams into healthy neonatal

calves. Ab specificity was strongly associated with the presence of allogeneic MHC class I Abs in the dams. MHC class I staining was

also observed on Madin–Darby bovine kidney cells, a cell line related to the one used to produce the vaccine Ag. Our report

emphatically demonstrates that alloimmunization against MHC class I is associated with a substantial risk of developing cytopenia-

associated syndromes in neonates when a cell line of the same species is used to produce an inactivated vaccine injected into the

mother. The Journal of Immunology, 2011, 187: 6564–6570.

Since 2007, an emergent life-threatening disease of neonatalcalves has been reported in several European countries (1–4) and has been designated as bovine neonatal pancyto-

penia (BNP). The disease is characterized by severe thrombocy-topenia and leukopenia—essentially a strong neutropenia anda mild lymphopenia—in 2- to 3-wk-old calves and in most casesleads to hemorrhages and a fatal issue. Bone marrow is affectedas a profound medullar aplasia is identified in most clinical cases(2). As no infectious agent could be isolated until now and despitethe early suspicion of a circovirus cause (2) that was not furtherconfirmed (5), other hypotheses have emerged.BNP has been produced experimentally in some calves through

ingestion of particular colostrums (6, 7), indicating that immuneeffectors might be involved. Colostrum has a high concentrationof IgG (between 50 and 200 g/l) and provides the calf with its solesource of passive immunity in the first hours of life. Indeed, due

to the syndesmo-chorial placentation, there is no transplacentalpassage of Abs during pregnancy in bovines. Indeed, BNP isclinically very similar to human neonatal thrombocytopenia andgranulocytopenia with an alloimmune cause (8). Recent reportsindicating that alloantibodies can be detected in the dams of af-fected calves support this hypothesis (9). However, Ab specificityand cause of BNP remained hidden until now. Recently, a descrip-tive epidemiological study strongly suggested that the alloimmuneresponse developed in BNP dams after vaccination with an inac-tivated bovine viral diarrhea (BVD) vaccine (10). Inasmuch, it wasshown that BNP-dam sera recognize the kidney cell line used toproduce the vaccine Ag (11).The aim of this work was to assess the role of Abs in the de-

velopment of BNP and to identify the specificity of the allore-sponse to clarify the cause of the disease.In this study, we provide substantial evidences that the

alloimmune response is directed against MHC class I Ags andis probably responsible for the pancytopenia observed in bovineneonates after the dam received one or several injections of aninactivated vaccine.

Materials and MethodsSera, animals, and BNP experimental model

Sera were collected from dams of confirmed BNP calves in Pregsure (PfizerAnimal Health)-vaccinated herds. IgG was precipitated using a final con-centration of 40% ammonium sulfate as previously described (12) and thensolubilized in NaCl 0.9% before extensive dialysis to remove ammoniumsalt. IgG solutions were 0.2-mm-filter sterilized and stored at 280˚C untilinjection.

Neonatal calves without any parental link were collected from herdswithout BNP history. They were separated from their dams immediatelyafter delivery and before colostrum intake. Each calf was injected i.v. with2 l of a solution containing ∼100–150 g total IgG. Calves were clinicallyexamined and sampled for the analysis of hematological and biochemicalparameters several times during the first 24 h after IgG injection and thenevery 2–3 d. After IgG injection, total plasma protein concentrations in-creased from 39.2 6 6.1 g/l to 56.3 6 3.1 g/l, corresponding approxi-mately to an increase of 15 g/l in the globulin fraction. This increase is

*Universite de Toulouse, Institut National Polytechnique de Toulouse, Ecole Natio-nale Veterinaire de Toulouse, F-31076 Toulouse, France; †Institut National de laRecherche Agronomique, Unite Mixte de Recherche 1225, Interactions Hotes-Agents Pathogenes, F-31076 Toulouse, France; ‡Centre National de la RechercheScientifique, Institut de Pharmacologie et de Biologie Structurale, F-31077 Toulouse,France; xUniversite de Toulouse, Universite Paul Sabatier, Institut de Pharmacologieet de Biologie Structurale, F-31077 Toulouse, France; and {Centre National de laRecherche Scientifique, Federation de Recherche 3450, Agrobiosciences, Interac-tions et Biodiversite, F-31326 Castanet Tolosan, France

Received for publication September 1, 2011. Accepted for publication October 4,2011.

This work was supported in part by Pfizer Animal Health (to G.F.), by grants fromthe Region Midi-Pyrenees, and by Fond Europeen de Developpement Regional.

Address correspondence and reprint requests to Prof. Gilles Foucras, Unite Mixte deRecherche 1225, Interactions Hotes-Agents Pathogenes, 23 Chemin des Capelles, BP87614, 31076 Toulouse Cedex 03, France. E-mail address: g.foucras@envt.fr

Abbreviations used in this article: BNP, bovine neonatal pancytopenia; BoLA, bovineleukocyte Ag; BVD, bovine viral diarrhea; FDR, false discovery rate; FNAIT, fetal/neonatal alloimmune thrombocytopenia; b2m, b2-microglobulin; MDBK, Madin–Darby bovine kidney; nano-LC-MS/MS, nano-liquid chromatography and nano-spray ionization mass spectrometry; PBLk, peripheral blood leukocyte; siRNA, smallinterfering RNA.

Copyright� 2011 by TheAmericanAssociation of Immunologists, Inc. 0022-1767/11/$16.00

www.jimmunol.org/cgi/doi/10.4049/jimmunol.1102533

by guest on January 14, 2019http://w

ww

.jimm

unol.org/D

ownloaded from

consistent with a colostrum intake of 10% of body weight. Bone marrowaspirations were made at the sternum before IgG injection and after 10 d,and smears stained with May–Grunwald–Giemsa were prepared for cy-tological examination.

All experiments were done in accordance with local and French regu-lations. The experimental protocol was approved by Comite d’Ethique deMidi-Pyrenees under the agreement number MP/03/25/06/10.

Immunocytochemistry, leukocyte Ab detection by flowcytometry, and ELISA

Peripheral blood leukocytes (PBLks) were prepared from experimentalcalves and recovered BNP-calves using EDTA-anticoagulated bloodtreated with ammonium chloride to lyse RBCs. PBLks were incubatedwith a titration of pathogenic or control sera, and IgG binding was revealedusing DyLight 488-labeled F(ab9)2 goat anti-bovine IgG (Jackson Im-munoResearch). Dead cells were gated out using 7-aminoactinomycin D(BD Biosciences), and a minimum of 20,000 events were acquired on aFACSCalibur (BD Biosciences) prior to analysis with FlowJo software(Tree Star).

For the competitive assay, cells (0.5 3 106), either PBLks or Madin–Darby bovine kidney (MDBK) cells, were incubated with a dilution ofserum prior to staining with DyLight 488-labeled anti-bovine IgG (JacksonImmunoResearch) and allophycocyanin-labeled anti-MHC class I W6/32mAb (BioLegend). After extensive washing, at least 20,000 events wereacquired, and data were analyzed with FlowJo software.

For bovine leukocyte Ag (BoLA) MHC class I quantification in bio-logical products, a commercial ELISA was used following the manu-facturer’s recommendations (Cusabio Biotech). The range of BoLA MHCdetection is 0.04 to 10 ng/ml.

MHC class I down-expression

Knockdown of MHC class I was carried out with a small interfering RNA(siRNA) targeting bovine b2-microglobulin (b2m) or a control siRNA(Eurogentec) at a final concentration of 10 nM using ICAFectine442siRNA transfection reagent (Eurogentec).

Immunoprecipitation and visualization of precipitated proteinsby SDS-PAGE and blotting

PBLks were adjusted to 2 3 107 to 3 3 107 cells/ml and were surfacelabeled with EZ-link Sulfo-NHS-LC-LC-Biotin (GE Healthcare) accord-ing to the manufacturer’s instructions. Biotinylated cells were incubatedwith a 1:20 dilution of BNP-IgG, a negative serum, or PBS (45 min, onice). After extensive washing to remove unbound and nonspecificallybound Abs, cells were solubilized with lysis buffer (50 mM Tris-HCl,pH 8, 150 mM NaCl, 5 mM EDTA, 1% Triton X-100, 1% deoxycholate,0.1% SDS, 1 mM PefaBlock SC, and 0.5 mg/ml leupeptin) on ice for 20 min.After centrifugation (15,000 3 g; 20 min), the supernatant was incubatedwith goat anti-bovine IgG (SouthernBiotech) coupled to Adembeads ProtAG (Ademtech). After washing five times with lysis buffer, precipitatedproteins were eluted twice with the elution buffer provided with the beads.

Immunoprecipitates were boiled in sample buffer for 5 min and thensubjected to SDS-PAGE on NuPAGE 4–12% Bis-Tris gel (Invitrogen).Proteins were transferred onto nitrocellulose membrane (Whatman) for 90min at 100 V using Tris/glycine buffer. After transfer, the membrane wasfirst blocked with 5% skim milk and 0.1% Tween 20 in TBS and thenincubated with HRP-conjugated streptavidin (KPL). Signals were devel-oped using the Immun-Star Western C kit reagent (Bio-Rad) according tothe manufacturer’s instructions. Alternatively, blotting for MHC class Iwas performed with a BoLA class I-specific IL-A88 mAb, a kind gift ofS.A. Ellis (Institute for Animal Health, Compton, U.K.), and anti-mouseIgG-HRP for revelation.

Nano-liquid chromatography and nanospray ionization massspectrometry analysis

After Coomassie blue staining of the SDS-PAGE gel, bands correspondingto those detected on the Western blots were excised and digested bythe addition of 30 ml of a solution of modified trypsin in 25 mM NH4HCO3

(20 ng/ml, sequence grade; Promega). The mixture was incubated at 37˚Covernight. The resulting peptide mixtures were analyzed by nano-liquidchromatography and nanospray ionization mass spectrometry (nano-LC-MS/MS) using an Ultimate3000 system (Dionex) coupled to an LTQ-Orbitrap Velos mass spectrometer (Thermo Fisher Scientific). Fivemicroliters of each sample were loaded on a C18 precolumn (300-mminner diameter 3 5 mm; Dionex) at 20 ml/min in 5% acetonitrile, 0.05%trifluoroacetic acid. After 5 min of desalting, the precolumn was switched

online with the analytical C18 column (75-mm inner diameter 3 15 cm;packed in-house) equilibrated in 95% solvent A (5% acetonitrile, 0.2%formic acid) and 5% solvent B (80% acetonitrile, 0.2% formic acid).Peptides were eluted using a 5–50% gradient of solvent B during 80 min ata flow rate of 300 nl/min. The LTQ-Orbitrap was operated in data-dependent acquisition mode with the Xcalibur software. Survey scan massspectrometry spectra were acquired in the Orbitrap on the 300–2000 m/zrange with the resolution set to a value of 60,000. The five most intenseions per survey scan were selected for CID fragmentation, and theresulting fragments were analyzed in the linear ion trap (LTQ). Dynamicexclusion was used within 60 s to prevent repetitive selection of the samepeptide ion for mass spectrometry analysis.

Database searching and data analysis

The Mascot Daemon software (version 2.3.2; Matrix Science, London,U.K.) was used to perform database searches in batch mode with all theraw files acquired on each sample. To extract automatically peak lists fromXcalibur raw files, the Extract_msn.exe macro provided with Xcalibur(version 2.0 SR2; Thermo Fisher Scientific) was used through the MascotDaemon interface. The following parameters were set to create the peaklists: parent ions in the mass range 400–4500, no grouping of MS/MS scans,and threshold at 1000. A peak list was created for each analyzed frac-tion (i.e., gel slice), and individual Mascot searches were performed foreach fraction. Data were searched against all entries in the SwissProtTrembl_20100907 protein database (12,155,553 sequences; 3,930,097,083residues) Mammalia (338,129 sequences). Oxidation of methionine andcarbamidomethylation of cysteine were set as variable modifications for allMascot searches. Specificity of trypsin digestion was set for cleavage afterLys or Arg except before Pro, and one missed trypsin cleavage site wasallowed. The mass tolerances in mass spectrometry and MS/MS were set to5 ppm and 0.8 Da, respectively, and the instrument setting was specified as“ESI-Trap.” Mascot results were parsed with software developed in-house,Mascot File Parsing and Quantification, version 4.0. Protein hits wereautomatically validated if they satisfied one of the following criteria:identification with at least one top-ranking peptide of a minimal length of 8aa and with a Mascot score higher than the identity threshold at p = 0.001(99.9% probability); or identification with at least two top-ranking peptideseach of a minimal length of 8 aa and with a Mascot score higher than theidentity threshold at p = 0.05 (95% probability). To calculate the falsediscovery rate (FDR), the search was performed using the “decoy” optionin Mascot and using the same criteria in Mascot File Parsing and Quan-tification to validate decoy and target hits. The FDR was calculated at theprotein level [FDR = (100 3 the number of validated decoy hits)/(thenumber of validated target hits + number of validated decoy hits)] and,using the specified validation criteria, it ranked between 0 and 0.5% for allthe samples analyzed with an average value of 0.4%.

ResultsPurified IgG from BNP dams induces thrombocytopenia andleukopenia in neonatal calves

To assess whether BNP is mediated by maternal Abs, we i.v.injected six colostrum-deprived newborn calves with one of fourpools of purified IgG prepared from cows that had been vaccinatedwith the same inactivated BVD vaccine. The pools were from fourdifferent herds of three different breeds, and cows (n = 16) had allgiven birth to calves that developed clinically confirmed cases ofBNP. Two calves of six died 2 d after IgG injection, after experi-encing intense respiratory distress and severe hemorrhages (datanot shown). In the four remaining calves, blood platelet (Fig. 1A)and leukocyte counts (Fig. 1B) were significantly diminished after10 d (Mann–Whitney U test, n = 4, p , 0.01). In contrast, anothercalf that received IgG from a nonvaccinated control cow remainedhealthy without any blood or medullar depletion. Hematologicalchanges resembled that of two other BNP cases that had receivedcolostrums from two of the BNP dams at birth (Fig. 1A, 1B). After10 d, bloody feces were observed in three of the four survivingBNP-IgG–transferred calves, and petechiae on mucous mem-branes were observed in all four animals, both of which areindications of clotting disorders (data not shown). A decrease ofthe cellularity was observed on aspirated bone marrow at day 10where it was significantly lower than that on day 0, the day of IgG

The Journal of Immunology 6565

by guest on January 14, 2019http://w

ww

.jimm

unol.org/D

ownloaded from

injection (day 0 grade 3.3 6 0.5 versus day 10 grade 1 6 0.8 ona scale of 4, n = 4, Mann–Whitney U test, p , 0.01). The pro-portions of most cell types were reduced, with the exceptions ofeosinophils, lymphocytes, and macrophages (Fig. 1C). After 15and 17 d respectively, two of the four remaining calves were re-cumbent and were euthanized for ethical reasons. Necropsiesrevealed hemorrhages throughout the bodies of both calves in theviscera, joints, and muscles. Microscopic examinations of theirbone marrow revealed massive cell depletions with no visiblemegakaryocytes (Fig. 1D), which is consistent with previouslydescribed cases of spontaneous BNP (2). The other two calvesrecovered spontaneously from the disease and developed normallyafterward. Their WBC and platelet counts (Fig 1A, 1B) and bonemarrow composition progressively returned to normal within 1mo. The clinical and biological signs of the IgG-injected animalsthat manifested disease were the same as those reported in natural(1, 2) and colostrum-induced BNP cases (6, 7). Our data un-equivocally show that IgG collected from BNP dams can inducede novo BNP in unrelated newborn calves.

IgG recognizes MHC class I on blood leukocytes and platelets

Soon after the transfusion of BNP-dam IgG or after the ingestionof colostrum from BNP dams, flow cytometry analysis of calfblood cells using DyLight 488-labeled (Fab9)2–anti-bovine IgG

revealed the presence of bound Abs on high proportions of calfgranulocytes, monocytes, and lymphocytes (Fig. 2A). We alsofound that BNP-dam IgG stained blood and bone marrow cellsfrom three recovered cases of spontaneous BNP. IgG binding wasalso demonstrated on platelets (data not shown). Conversely, whenBNP-dam blood cells were assayed, no staining was detected.Altogether, our data indicate that IgG from BNP dams recognizesone or several blood cell surface Ag(s), as has been previouslysuggested by others in studies of sera from BNP dams (9) or ofBNP induced in calves by administration of colostrum from BNPdams (6). Thus, BNP is an alloimmune response in which cowsproduce Abs against one or several determinant(s), hypotheticallyof paternal origin, widely expressed on calf blood cells and pla-telets.Having characterized a set of pathogenic IgG batches, and

further to define the specificity of the IgG associated with BNPdevelopment in neonatal calves, we investigated the Ag presenton the PBLks of five healed calves (two experimental and threespontaneous cases). To immunoprecipitate the Ag(s), we incubatedbiotinylated PBLks with dilutions of pathogenic sera or BNP-IgGand released the Ag–Ab complexes by lysing the cells. We pre-cipitated IgG-bound proteins from the supernatant using magneticbeads coated with goat anti-bovine IgG Abs. Next, the precipitatedproteins were separated by SDS-PAGE and visualized by silver

FIGURE 1. Injection of IgG from BNP dams into newborn calves induces pancytopenia and bone marrow depletion. Colostrum-deprived newborn calves

were injected i.v. with IgG prepared from the sera of a group of four BVD-vaccinated BNP dams (n = 4; s) or with control serum (n = 1; d). Two other

calves received colostrum from two different BNP dams (:). A and B, Numbers of platelets (A) and WBCs (B) were recorded every 2 to 3 d after

transfusion. C, Bone marrow aspiration at the sternum was done before transfusion (day 0) and 10 d afterward for analysis of bone marrow cellularity and

composition. Bone marrow cell numbers were significantly reduced, and the proportions of several cell lineages were substantially altered (n = 4, Mann–

Whitney U test, **p , 0.01). D, At necropsy, histological examination of bone marrow from the sternum indicated severe cell depletion in the two calves

that died 15 and 17 d, respectively, after transfusion (hemalaun-eosin stain, original magnification 3100). BC, band cell; EA, acidophilic erythroblast; EB,

basophilic erythroblast; EP, polychromatophilic erythroblast; GNB, basophil; GNE, eosinophil; GNN, neutrophil; L, lymphocyte; M, myelocyte; MB,

myeloblast; MM, metamyelocyte; MP, macrophage; PE, proerythroblast; PM, promyelocyte.

6566 ANTI-MHC CLASS I-MEDIATED NEONATAL PANCYTOPENIA

by guest on January 14, 2019http://w

ww

.jimm

unol.org/D

ownloaded from

staining (Fig. 2B). Multiple bands were visible on the gel, but wedetected no clear, condition-specific differences. By contrast, afterthe immunoprecipitated proteins were transferred onto a nitro-cellulose membrane and the biotinylated surface proteins wererevealed using peroxidase-conjugated streptavidin and chemilu-minescence, a prominent band of ∼40–45 kDa (Fig. 2C) was ob-served in the lane containing proteins precipitated by BNP-IgG.In control immunoprecipitations using PBS or a nonpathogenicserum, no such signal was detected.To identify the putative BNP-IgG Ag(s), the proteins in the gel

regions containing the 40- to 45-kDa band identified throughblotting were digested in the gel with trypsin, and the resultingpeptides were analyzed using online coupling of nano-LC-MS/MS.A search against the mammalian SwissProt database led to theidentification of an average of ∼30 proteins in each of the samplesanalyzed including control samples. The proteins precipitated bypathogenic sera were identified and compared with those precip-itated by different combinations of sera/cell sources and controlsamples (nonpathogenic serum and PBS). This analysis unam-biguously identified BoLA MHC class I protein as the Ag rec-ognized by the pathogenic sera, as at least two peptides matched

the extracellular domain of this protein for four of the five path-ogenic sera analyzed (Table I). The identity of the band wasfurther confirmed using IL-A88, an anti-BoLA class I monomor-phic mAb, which identified the expected 45-kDa protein band onan immunoblot (Fig. 2D). Thus, bovine MHC class I protein wasrecognized as a target of BNP-inducing sera.To confirm further the specificity of the Ab response on a larger

number of samples, PBLks from one BNP-afflicted calf werepreincubated with either pooled BNP-IgG or individual sera from12 BNP dams and 12 non-BNP cows from within the same BVD-vaccinated herds before staining with W6/32, a monomorphicHLA-(A, B, and C)-specific mAb that recognizes MHC class Imolecules in several mammalian species including the bovines.Preincubation with pooled BNP-IgG and sera from individual BNPdams partially abolished subsequent staining by W6/32 (Fig. 3A,3B), indicating that Abs in the BNP-IgG and 12 sera from BNPdams competed with W6/32 for a closely related epitope on theBoLA class I molecules. Incubation with non-BNP sera fromvaccinated individual cows (Fig. 3B) or with an irrelevant mAbagainst CD45 did not elicit similar reductions in W6/32 binding(data not shown). This strongly supports the view that these

FIGURE 2. BNP-inducing sera recognize BoLA MHC class I determinant(s). A, The presence of surface IgG on PBLks from a calf before or 5 h after

injection with BNP-IgG was revealed using a DyLight 488-labeled F(ab9)2 secondary Ab and flow cytometry analysis. Data are representative of two IgG-

injected calves and one calf that received colostrum. B and C, Surface biotin-labeled PBLks from a healed BNP calf were incubated with a dilution of BNP-

IgG. After cell lysis, membrane Ag–IgG complexes were captured with magnetic beads coated with goat anti-bovine IgG Abs, and the proteins were

separated by gel electrophoresis. Total proteins were visualized by silver staining the gel (B), and surface proteins were revealed using streptavidin–

peroxidase and chemiluminescence after transfer onto a nitrocellulose membrane (C). D, Identity of the Ag was further confirmed through BoLA class

I-specific IL-A88 mAb immunoblotting. These results are representative of at least five different experiments.

Table I. Identification of bovine MHC class I protein as recognized by pathogenic sera

ProteinAccessionNumbera

TheoreticalMolecularMass (Da)

SequenceCoverage

(%) Peptide Sequenceb m/zChargeState

ExperimentalMass(Da)

PeptideScorec

SampleNumber

MHC class I Ag Q3YJF9 40,623 8 57RFDSDAPNPR65 509.7336 2 1017.4526 24 1509.7328 1017.4510 38 2

133RGTTQYGYGDR142 582.2610 2 1162.5043 14 1582.2593 1162.5040 42 2

208AHVTHHPISER218 642.3336 2 1282.6526 30 1642.3337 1282.6528 45 2

Q3YJG8 or Q3YJG9 40,794 9 43YLEVGYVDDTQFVR56 852.4163 2 1702.8180 60 3

143DYIALNEELR152 618.3160 2 1234.6174 37 3

208AHVTHHPISER218 642.3342 2 1282.6538 36 3Q3YFH3 39,751 5 137DYLALNEDLR146 611.3072 2 1220.5998 27 4

202AHVTHHPISER212 642.3344 2 1282.6542 36 4

aFrom SwissProt Trembl database.bThe amino acid position in the protein sequence is indicated.cThe peptide score is given by the Mascot algorithm according to the parameters set for database search as described in Materials and Methods.

The Journal of Immunology 6567

by guest on January 14, 2019http://w

ww

.jimm

unol.org/D

ownloaded from

pathogenic sera contained Abs directed against one or severalepitopes on BoLA class I molecules.

Alloantibodies recognize MHC class I on the vaccine parentalcell line

To look for the presence of MHC class I in the vaccine and in-vestigate the hypothesis that the Ab response follows vaccina-tion, the antigenic fraction of the vaccine was isolated using 10%butanol as previously described (13). MHC class I was detectedon an immunoblot by the IL-A88 mAb (Fig. 4), indicating thatthe vaccine actually contains BoLA class I. Using a commercialELISA, the BoLA MHC concentration was estimated to be around0.5–1 ng/ml in the aqueous phase recovered after butanol treat-ment, indicating that the vaccine may serve as a source of allo-antigen for mounting an Ab response against MHC Ags. VaccineAg is produced on a cell lined derived from MDBK cells. Weobserved that not only did BNP-IgG stain the MDBK cells (Fig.5A), but also that staining with BNP-IgG blocked subsequentMHC class I staining on this cell line (Fig. 5B). Specificity wasfurther confirmed by immunoblotting of MDBK cell lysate withthe four pathogenic BNP-IgG, as performed previously for the

blood leukocytes (Fig. 5C). Indeed, when MHC class I expressionon MDBK cells was partially extinguished using RNA interfer-ence targeting b2m-coding mRNA, bovine IgG staining withpooled or individual BNP-sera and MHC class I staining weresimilarly reduced (Fig. 5D). This again supports the assertion thatthe vast majority of the allogeneic Abs in BNP-dam sera arespecific for BoLA class I–b2m complexes.

DiscussionIn this study, we provide several compelling lines of evidencethat anti-MHC class I Abs are sufficient to induce alloimmunethrombocytopenia and leukopenia in neonates. In humans, MHCclass I alloantibodies in individual cases of fetal/neonatal alloim-mune thrombocytopenia (FNAIT) have been reported a numberof times during the past decades (14–25). In addition to throm-bocytopenia, the presence of an associated neutropenia is noticedin a few cases (16–28). These Abs are sometimes detected at thesame time as other HPA Abs commonly involved in FNAIT (27–29). Although HLA class I Abs are commonly encountered inthe mothers of FNAIT cases (30), their role as the main triggerremains controversial (17, 31, 32).Several conclusions can be drawn from our study. First, as no

other specificity in common among pathogenic sera could yet beidentified, an alloresponse against MHC class I is probably thecorrect explanation for induction of thrombocytopenia and leu-kopenia in the neonates. However, the pathophysiological mech-anism remains unclear. Whether the peripheral consumption of thecalf recipient platelets and cells, or the central depletion of stemcells, or both mechanisms are responsible for the development ofthe disease has to be determined. Whereas MHC class I moleculesare widely expressed in the body, the strong impact of MHCalloantibodies on the hematopoietic system, and notably the bonemarrow, is perplexing. Higher expression of MHC class I mole-cules on hematopoietic cells is one possibility. After injectioninto pregnant mice, paternal MHC class I Abs were transferred tothe fetus and accumulated predominantly in the blood, thymus,and liver (33). Moreover, in our situation, all cell types inside thehematopoietic system are not equally affected as some lympho-cytes seem to be less sensitive to the depletion and still remain,whereas granulocytes, with a lower intensity of IgG staining, arefirst to disappear during the course of the leukopenia. Heteroge-neous expression of bovine MHC class I, as recently shown insome primates (34) conversely to humans, would explain thisobservation.Although all calves developed a thrombocytopenia upon BNP-

IgG injection, the disease ending varied among cases in our model.This is also in accordance with previous reports where BNP wasinduced by ingestion of colostrum (6, 7). In the field, the variabilityof the Ab concentration in the colostrum, the quantity of ingestedcolostrum, and the rate of absorption through the digestive tractmay modulate the total amount of acquired alloantibodies andexpression of the disease. In our study, the BNP-IgG solutionswere shown to have very similar titers of alloantibodies whenassayed against the MDBK cell line (data not shown), and i.v.injection of IgG diminished the variability of IgG transfer com-pared with that of colostrum intake. Thus, several other factors areplaying a role to modulate incidence and severity of the disease.In humans, a correlation between HPA1a Ab titers in the motherand severity of the thrombocytopenia at birth has been reported,despite several analytical variation factors (35–37). FcRn, theneonatal IgG receptor, has been shown to modulate FNAIT ina mouse model (38). Other genetically determined factors mayalso exist. It is worth noting that some haplotypes of b2m, whichis involved in the assembly of both FcRn and MHC class I mole-

FIGURE 3. BNP-IgG and sera from BNP dams compete for MHC class

I staining. Design of the competitive Ab staining assay: PBLks from

a healed BNP calf were stained with allophycocyanin-labeled anti-MHC

class I mAb W6/32 after preincubation with BNP-IgG or control IgG. A,

Competition for MHC class I binding was detected through extinction of

the W6/32 signal. B, Results of the competitive binding assay for BoLA

class I are presented for control IgG (n = 5), BNP-IgG (n = 4), sera from

BNP dams (n = 12), and sera from non-BNP cows (n = 12) from three

BVD-vaccinated herds (n = 24, Student t test).

FIGURE 4. BoLA class I Ag is present within a commercial vaccine

suspected to be responsible for the alloimmune response in cattle. Vaccine

emulsion was broken down into oil and water phases using 10% butanol as

indicated in Materials and Methods. The total protein contents of the su-

pernatant and the protein pellet were assayed using SDS-PAGE and silver

staining or by immunoblotting with IL-A88 mAb as previously. Data were

reproduced at least three times.

6568 ANTI-MHC CLASS I-MEDIATED NEONATAL PANCYTOPENIA

by guest on January 14, 2019http://w

ww

.jimm

unol.org/D

ownloaded from

cules, are associated with failure of passive transfer in neonatalcalves (39). Whether there is a concentration threshold of MHCclass I alloantibodies that can be tolerated by the neonate needsalso to be further investigated.Furthermore, not only the amount of alloantibodies transferred

to the calf but also the allospecificity is expected to control se-verity of the disease. Partial BNP phenotype may be due to the loweravidity of alloantibodies for the MHC molecules expressed bythe calf. Because MHC class I genes have the most polymorphicalleles within the mammalian genome and combinatorial haplo-types are numerous, both recognition and affinity may vary. Byproviding plasma IgG concentration in the range of that permittedby colostrum intake in bovines, we were able to induce throm-bocytopenia after all attempts. This high frequency of BNP de-velopment contrasts with previous reports (6, 7) and might also beexplained by the fact that we used pooled sera from four cattledespite individual sources of Abs from either colostrum or serumand by our possibility to increase the diversity range of IgGspecificities. The same phenomenon may have heightened thenumber of spontaneous cases in regions or herds where the dis-tribution of pooled colostrum to calves is a common practice.Altogether, the frequency of the disease remains relatively low,

as indicated by the results of epidemiological studies; the estimatedoverall prevalence is probably lower than 0.3% and is usually notmore than 10% within a herd (Ref. 11 and G. Foucras, F. Corbiere,and F. Schelcher, unpublished observations). Another consequenceis the lower than expected prediction of BNP occurrence in theoffspring of MHC Ab-positive dams due to a combination of fac-tors that are difficult to evaluate a priori.The second finding of our study is that the specificity supports the

hypothesis of an alloimmunization linked to the vaccination withan inactivated viral vaccine. In humans, allo-MHC immunization

is generally observed after the first pregnancy, transfusion of non-cell-depleted blood products, or transplantation. To our knowledge,the development of an alloimmune response after injection of avaccine produced on a human cell line has never been reported,although never truly investigated. Similarly, all products containingallogeneic cell culture constituents could be considered as possiblyimmunogenic. The origin of the alloresponse questions the pe-culiarities of this vaccine, as both Ag and adjuvant may be in-volved. The fact that we were able to detect MHC Ags within theantigenic fraction of the vaccine may be sufficient to prime analloresponse in MHC-discordant recipients. Moreover, the adju-vant, by inducing a strong Ab response compared with that of othercommercial vaccines against BVD virus (11, 40), may favor thedevelopment of high titers of alloreactive Abs in some animalshighly responsive to immunization. Because of the large vari-ability of the MHC in mammals, the frequency of the allores-ponders is expected to be relatively low, which fits with the lowoccurrence of the disease as reported by epidemiological studies.In the current case, if the Ab response was induced by injection

of an inactivated BVD vaccine containing bovine MHC class Iproteins, the genetic background of the dam and that of the neo-nate and of the father, which share the recognized epitopes, aredetermining factors in occurrence of the disease, as described inhumans for FNAIT. Even the reported presentation of fetal pro-teins to the dam by placental leakage could be discussed as a causeof this syndrome for rare cases of the disease in which no BVDvaccination took place (41). Later, this may also enhance the Abresponse once priming has occurred.Third, the approach we developed for the identification of the

allospecificity was reported by others during the time of our study(42–44). Our results are one more example that shows the powerof detection, no matter whether it is an alloimmune or an auto-

FIGURE 5. Alloreactive sera recognize MHC class

expressed on MDBK cell line. A, MDBK cells, the

parental cell line of the one used to produce the vac-

cine Ag, were incubated with BNP-IgG, and binding

was revealed using DyLight488-(Fab9)2 anti-bovine

IgG prior to flow cytometry analysis (gray, control

serum; black, BNP-IgG; dotted histogram, secondary

Ab only). B, Competitive assay for MHC class I

binding indicated that MHC class I Ags expressed on

MDBK cells are recognized by BNP-IgGs (dotted

line, no staining; plain line, PBS; gray, control serum;

black, BNP-IgG). C, MDBK cell lysate proteins were

separated using SDS-PAGE and blotted with a 1:2000

dilution of BNP-IgG and HRP-labeled anti-bovine

IgG or IL-A88 mAb and HRP–anti-mouse IgG, re-

spectively, to confirm BoLA MHC class I specificity.

Lanes 1–4 correspond to the different BNP-IgGs used

to produce experimental BNP cases. D, MDBK cells

were transfected with b2m-specific siRNA. After 12 h,

cells were stained with anti-MHC class I W6/32 or

with BNP-IgG and DL488–anti-bovine IgG as previ-

ously. BNP-IgG staining and MHC class I Ab staining

were diminished by similar proportions, indicating

that BoLA class I represents most of the specificity

present in these BNP-IgG. Data are representative of

three experiments performed.

The Journal of Immunology 6569

by guest on January 14, 2019http://w

ww

.jimm

unol.org/D

ownloaded from

immune disease, for the discovery of Ab specificities. Despitea relatively large number of proteins that are retained by immu-nocapture in one sample, comparison of several samples enablesidentification of the critical Ag. Moreover, prestaining of cellsurface proteins with biotin facilitates further isolation of mem-brane proteins.At the time when the strategies to reduce alloimmunization are

investigated in humans (45), we have identified the possibility thatvaccination with inactivated vaccines containing low amounts ofallogeneic proteins may induce the development of an immuneresponse against MHC class I that may later be detrimental to thefetus or the neonate. The possibility that this situation alreadyexists in humans, or will arise in the future, should be considered.

AcknowledgmentsWe are grateful to Dr. S.A. Ellis (Institute for Animal Health, Compton,

U.K.) for the kind gift of IL-A88 mAb.

DisclosuresThe authors have no financial conflicts of interest.

References1. Pardon, B., L. Steukers, J. Dierick, R. Ducatelle, V. Saey, S. Maes,

G. Vercauteren, K. De Clercq, J. Callens, K. De Bleecker, and P. Deprez. 2010.Haemorrhagic diathesis in neonatal calves: an emerging syndrome in Europe.Transbound Emerg Dis 57: 135–146.

2. Kappe, E. C., M. Y. Halami, B. Schade, M. Alex, D. Hoffmann, A. Gangl,K. Meyer, W. Dekant, B.-A. Schwarz, R. Johne, et al. 2010. Bone marrow de-pletion with haemorrhagic diathesis in calves in Germany: characterization ofthe disease and preliminary investigations on its aetiology. Berl. Munch. Tier-arztl. Wochenschr. 123: 31–41.

3. Bell, C. R., P. R. Scott, N. D. Sargison, D. J. Wilson, L. Morrison, F. Howie,K. Willoughby, and C. D. Penny. 2010. Idiopathic bovine neonatal pancytopeniain a Scottish beef herd. Vet. Rec. 167: 938–940.

4. Sanchez-Miguel, C., M. McElroy, and E. Walsh. 2010. Bovine neonatal pan-cytopenia in calves in Ireland. Vet. Rec. 166: 664.

5. Willoughby, K., J. Gilray, M. Maley, A. Dastjerdi, F. Steinbach, M. Banks,S. Scholes, F. Howie, A. Holliman, P. Baird, and J. McKillen. 2010. Lack ofevidence for circovirus involvement in bovine neonatal pancytopenia. Vet. Rec.166: 436–437.

6. Bridger, P. S., R. Bauerfeind, L. Wenzel, N. Bauer, C. Menge, H.-J. Thiel,M. Reinacher, and K. Doll. 2011. Detection of colostrum-derived alloantibodiesin calves with bovine neonatal pancytopenia. Vet. Immunol. Immunopathol. 141:1–10.

7. Friedrich, A., M. Buttner, G. Rademacher, W. Klee, B. K. Weber, M. Muller,A. Carlin, A. Assad, A. Hafner-Marx, and C. M. Sauter-Louis. 2011. Ingestion ofcolostrum from specific cows induces bovine neonatal pancytopenia (BNP) insome calves. BMC Vet. Res. 7: 10.

8. Berkowitz, R. L., J. B. Bussel, and J. G. McFarland. 2006. Alloimmunethrombocytopenia: state of the art 2006. Am. J. Obstet. Gynecol. 195: 907–913.

9. Pardon, B., E. Stuyven, S. Stuyvaert, M. Hostens, J. Dewulf, B. M. Goddeeris,E. Cox, and P. Deprez. 2011. Sera from dams of calves with bovine neonatalpancytopenia contain alloimmune antibodies directed against calf leukocytes.Vet. Immunol. Immunopathol. 141: 293–300.

10. European Medicines Agency. Committee for Medicinal Products for VeterinaryUse. Available at: http://www.ema.europa.eu/docs/en_GB/document_library/Press_release/2010/07/WC500094734.pdf. Accessed June 15, 2011

11. Bastian, M., M. Holsteg, H. Hanke-Robinson, K. Duchow, and K. Cussler. 2011.Bovine Neonatal Pancytopenia: is this alloimmune syndrome caused by vaccine-induced alloreactive antibodies? Vaccine 29: 5267–5275.

12. Hebert, G. A. 1974. Ammonium sulfate fractionation of sera: mouse, hamster,guinea pig, monkey, chimpanzee, swine, chicken, and cattle. Appl. Microbiol.27: 389–393.

13. Miles, A. P., H. A. McClellan, K. M. Rausch, D. Zhu, M. D. Whitmore, S. Singh,L. B. Martin, Y. Wu, B. K. Giersing, A. W. Stowers, et al. 2005. Montanide ISA720 vaccines: quality control of emulsions, stability of formulated antigens, andcomparative immunogenicity of vaccine formulations. Vaccine 23: 2530–2539.

14. Sharon, R., and A. Amar. 1981. Maternal anti-HLA antibodies and neonatalthrombocytopenia. Lancet 1: 1313.

15. Chow, M. P., K. J. Sun, C. H. Yung, H. Y. Hu, J. L. Tzeng, and T. D. Lee. 1992.Neonatal alloimmune thrombocytopenia due to HLA-A2 antibody. Acta Hae-matol. 87: 153–155.

16. Starcevic, M., M. Tomicic, M. Malenica, and V. Zah-Matakovic. 2010. Neonatalalloimmune thrombocytopenia caused by anti-HLA-A24 alloantibodies. ActaPaediatr. 99: 630–632.

17. King, K. E., K. J. Kao, P. F. Bray, J. F. Casella, K. Blakemore, N. A. Callan,S. D. Kennedy, and T. S. Kickler. 1996. The role of HLA antibodies in neonatalthrombocytopenia: a prospective study. Tissue Antigens 47: 206–211.

18. Tanaka, T., N. Umesaki, J. Nishio, K. Maeda, T. Kawamura, N. Araki, andS. Ogita. 2000. Neonatal thrombocytopenia induced by maternal anti-HLAantibodies: a potential side effect of allogenic leukocyte immunization for un-explained recurrent aborters. J. Reprod. Immunol. 46: 51–57.

19. De Tar, M. W., E. Klohe, A. Grosset, and T. Rau. 2002. Neonatal alloimmunethrombocytopenia with HLA alloimmunization: case report with immunohe-matologic and placental findings. Pediatr. Dev. Pathol. 5: 200–205.

20. Grainger, J. D., G. Morrell, J. Yates, and D. Deleacy. 2002. Neonatal alloimmunethrombocytopenia with significant HLA antibodies. Arch. Dis. Child. FetalNeonatal Ed. 86: F200–F201.

21. Han, K.-S., E.-Y. Song, and M.-H. Park. 2002. Neonatal alloimmune thrombo-cytonpenia due to HLA antibodies. Int. J. Hematol. 76(Suppl 1): 361–363.

22. Saito, S., M. Ota, Y. Komatsu, S. Ota, S. Aoki, K. Koike, I. Tokunaga, T. Tsuno,G. Tsuruta, T. Kubo, and H. Fukushima. 2003. Serologic analysis of three casesof neonatal alloimmune thrombocytopenia associated with HLA antibodies.Transfusion 43: 908–917.

23. Tomicic, M., M. Starcevic, J. Bux, V. Zach, Z. Hundric-Haspl, V. Drazic, andB. Grahovac. 2003. Severe neonatal neutropenia due to anti-human leucocyteantigen B49 alloimmunization only: a case report. Transfus. Med. 13: 233–237.

24. Moncharmont, P., V. Dubois, C. Obegi, M. Vignal, Y. Merieux, L. Gebuhrer, andD. Rigal. 2004. HLA antibodies and neonatal alloimmune thrombocytopenia.Acta Haematol. 111: 215–220.

25. Thude, H., U. Schorner, C. Helfricht, M. Loth, B. Maak, and D. Barz. 2006.Neonatal alloimmune thrombocytopenia caused by human leucocyte antigen-B27 antibody. Transfus. Med. 16: 143–149.

26. Hagimoto, R., K. Koike, K. Sakashita, T. Ishida, Y. Nakazawa, Y. Kurokawa,T. Kamijo, S. Saito, A. Hiraoka, M. Kobayashi, and A. Komiyama. 2001. Apossible role for maternal HLA antibody in a case of alloimmune neonatalneutropenia. Transfusion 41: 615–620.

27. Marın, L., A. Torıo, M. Muro, R. Fernandez-Parra, A. Minguela, V. Bosch,M. R. Alvarez-Lopez, and A. M. Garcıa-Alonso. 2005. Alloimmune neonatalneutropenia and thrombocytopenia associated with maternal anti HNA-1a, HPA-3b and HLA antibodies. Pediatr. Allergy Immunol. 16: 279–282.

28. Gramatges, M. M., P. Fani, K. Nadeau, S. Pereira, and M. R. Jeng. 2009.Neonatal alloimmune thrombocytopenia and neutropenia associated with ma-ternal human leukocyte antigen antibodies. Pediatr. Blood Cancer 53: 97–99.

29. Panzer, S., W. R. Mayr, and B. Eichelberger. 2005. Light chain phenotypes ofHLA antibodies in cases with suspected neonatal alloimmune thrombocytopenia.Vox Sang. 89: 261–264.

30. Davoren, A., B. R. Curtis, R. H. Aster, and J. G. McFarland. 2004. Humanplatelet antigen-specific alloantibodies implicated in 1162 cases of neonatalalloimmune thrombocytopenia. Transfusion 44: 1220–1225.

31. Taaning, E. 2000. HLA antibodies and fetomaternal alloimmune thrombocyto-penia: myth or meaningful? Transfus. Med. Rev. 14: 275–280.

32. Bussel, J. B., and A. Primiani. 2008. Fetal and neonatal alloimmune thrombo-cytopenia: progress and ongoing debates. Blood Rev. 22: 33–52.

33. Adeniyi-Jones, S. C., and K. Ozato. 1987. Transfer of antibodies directed topaternal major histocompatibility class I antigens from pregnant mice to thedeveloping fetus. J. Immunol. 138: 1408–1415.

34. Greene, J. M., R. W. Wiseman, S. M. Lank, B. N. Bimber, J. A. Karl, B. J. Burwitz,J. J. Lhost, O. E. Hawkins, K. J. Kunstman, K. W. Broman, et al. 2011. DifferentialMHC class I expression in distinct leukocyte subsets. BMC Immunol. 12: 39.

35. Bessos, H., M. K. Killie, M. Matviyenko, A. Husebekk, and S. J. Urbaniak.2008. Direct comparison between two quantitative assays in the measurement ofmaternal anti-HPA-1a antibody in neonatal alloimmune thrombocytopenia(NAIT). Transfus. Apheresis Sci. 39: 221–227.

36. Killie, M. K., A. Husebekk, J. Kjeldsen-Kragh, and B. Skogen. 2008. A pro-spective study of maternal anti-HPA 1a antibody level as a potential predictor ofalloimmune thrombocytopenia in the newborn. Haematologica 93: 870–877.

37. Bessos, H., M. K. Killie, J. Seghatchian, B. Skogen, and S. J. Urbaniak. 2009.The relationship of anti-HPA-1a amount to severity of neonatal alloimmunethrombocytopenia - Where does it stand? Transfus. Apheresis Sci. 40: 75–78.

38. Chen, P., C. Li, S. Lang, G. Zhu, A. Reheman, C. M. Spring, J. Freedman, andH. Ni. 2010. Animal model of fetal and neonatal immune thrombocytopenia: roleof neonatal Fc receptor in the pathogenesis and therapy. Blood 116: 3660–3668.

39. Clawson, M. L., M. P. Heaton, C. G. Chitko-McKown, J. M. Fox, T. P. L. Smith,W. M. Snelling, J. W. Keele, and W. W. Laegreid. 2004. Beta-2-microglobulinhaplotypes in U.S. beef cattle and association with failure of passive transfer innewborn calves. Mamm. Genome 15: 227–236.

40. Raue, R., S. S. Harmeyer, and I. A. Nanjiani. 2011. Antibody responses toinactivated vaccines and natural infection in cattle using bovine viral diarrhoeavirus ELISA kits: assessment of potential to differentiate infected and vaccinatedanimals. Vet. J. 187: 330–334.

41. Newman, M. J., and H. C. Hines. 1980. Stimulation of maternal anti-lymphocyteantibodies by first gestation bovine fetuses. J. Reprod. Fertil. 60: 237–241.

42. Heinold, A., B. Kuehl, G. Brenner-Weiss, G. Opelz, and T. H. Tran. 2010. Se-quential analysis by immunoprecipitation-MALDI-TOF: a novel method for de-tection and identification of alloantibody specificities.Hum. Immunol. 71: 462–467.

43. Greinacher, A., J. Wesche, E. Hammer, B. Furll, U. Volker, A. Reil, and J. Bux.2010. Characterization of the human neutrophil alloantigen-3a. Nat. Med. 16: 45–48.

44. Littleton, E., M. Dreger, J. Palace, and A. Vincent. 2009. Immunocapture andidentification of cell membrane protein antigenic targets of serum autoanti-bodies. Mol. Cell. Proteomics 8: 1688–1696.

45. Zimring, J. C., L. Welniak, J. W. Semple, P. M. Ness, S. J. Slichter,S. L. Spitalnik; NHLBI Alloimmunization Working Group. 2011. Currentproblems and future directions of transfusion-induced alloimmunization: sum-mary of an NHLBI working group. Transfusion 51: 435–441.

6570 ANTI-MHC CLASS I-MEDIATED NEONATAL PANCYTOPENIA

by guest on January 14, 2019http://w

ww

.jimm

unol.org/D

ownloaded from