Ti

MaRea Clinb Ins

1. I

tiouplahae

modid

(Fo200hem

Veterinary Microbiology 167 (2013) 403–409

A R

Artic

Rece

Rece

Acce

Keyw

Hem

Bloo

Sero

*

brio

gres

(R. H

037

http

ssue sequestration of ‘Candidatus Mycoplasma turicensis’

rilisa Novacco a,*, Barbara Riond a, Marina L. Meli a, Paula Grest b,gina Hofmann-Lehmann a

ical Laboratory, Vetsuisse Faculty, University of Zurich, Winterthurerstrasse 260, 8057 Zurich, Switzerland

titute of Veterinary Pathology, University of Zurich, Vetsuisse Faculty, Winterthurerstrasse 268, 8057 Zurich, Switzerland

ntroduction

Feline hemoplasmas are the causative agent of infec-s anemia in felids. Currently, at least three hemo-

sma species have been recognized in cats: Mycoplasma

mofelis (M. haemofelis), ‘Candidatus Mycoplasma hae-minutum’ (‘Candidatus M. haemominutum’) and ‘Can-

atus Mycoplasma turicensis’ (‘Candidatus M. turicensis’)ley et al., 1998; Foley and Pedersen, 2001; Willi et al.,5). The pathogenic potential of the different felineoplasma species varies and co-factors, such as

immunosuppression, co-infections with other hemo-plasma species or retroviral infections, may increase theseverity of the disease (Willi et al., 2007). ‘Candidatus M.turicensis’ was first described in a naturally infected catswith severe hemolytic anemia (Willi et al., 2005). It wasdescribed worldwide in wild and domestic cats (Willi et al.,2007). The pathogenic potential of ‘Candidatus M. tur-icensis’ seems to vary, with some isolates inducinghemolytic anemia whereas others result in few clinicalsigns. The peak bacteremia is accompanied with anincrease in the osmotic fragility of red blood cells and adecrease of the PCV (Museux et al., 2009; Willi et al., 2005).The decrease of the PCV is usually proportional to thebacterial blood loads developed during the infection. Nofluctuation of blood loads were detected during the earlyphase of the infection (Museux et al., 2009). After primaryinfection cats may become clinically healthy carriers.

T I C L E I N F O

le history:

ived 27 February 2013

ived in revised form 12 July 2013

pted 13 July 2013

ords:

otropic mycoplasmas

d and tissue loads

logy

A B S T R A C T

‘Candidatus Mycoplasma turicensis’ (‘Candidatus M. turicensis’) is a hemoplasma species

that infects felids. It differs from other feline hemoplasma species due to its particular

infection kinetics and phylogenetic similarity to rodent hemoplasma species. The lower

and shorter bacteremia produced by ‘Candidatus M. turicensis’ suggests a possible tissue

sequestration of the organism. The aim of this study was to explore this possibility. Five

specified-pathogen free cats were subcutaneously inoculated with ‘Candidatus M.

turicensis’ and sacrificed 86 days after inoculation. Thirty-one selected organs were

collected upon necropsy, and samples were analyzed by real-time Taqman1 PCR. The

humoral immune response was monitored by DnaK ELISA. All five cats had detectable

‘Candidatus M. turicensis’ loads in the majority (52–100%) of the tested tissues. High

‘Candidatus M. turicensis’ tissue loads (average 3.46 � 104 copies/10 mg) were detected in

the samples. The presence of the organisms in the tissues could not be explained by the

blood burdens because the blood of four out of five cats tested PCR-negative at the time of

necropsy. This is the first study to describe the distribution of ‘Candidatus M. turicensis’ in

various organs; it also demonstrates that, in contrast to other feline hemoplasma species,

significant sequestration of ‘Candidatus M. turicensis’ occurs in many tissues. These results

represent an important step toward the understanding of the pathogenesis of ‘Candidatus

M. turicensis’.

� 2013 Elsevier B.V. All rights reserved.

Corresponding author. Tel.: +41 44 635 82 79; fax: +41 44 635 89 23.

E-mail addresses: mnovacco@vetclinics.uzh.ch (M. Novacco),

nd@vetclinics.uzh.ch (B. Riond), mmeli@vetclinics.uzh.ch (M.L. Meli),

t@vetpath.uzh.ch (P. Grest), rhofmann@vetclinics.uzh.ch

ofmann-Lehmann).

Contents lists available at ScienceDirect

Veterinary Microbiology

jo u rn al ho m epag e: ww w.els evier .c o m/lo cat e/vetmic

8-1135/$ – see front matter � 2013 Elsevier B.V. All rights reserved.

://dx.doi.org/10.1016/j.vetmic.2013.07.019

M. Novacco et al. / Veterinary Microbiology 167 (2013) 403–409404

In the last few years, the distribution of feline hemoplasmaspecies in different tissues of infected animals has beeninvestigated. Tissue sequestration of the pathogen wouldexplain the chronic nature of hemoplasma infections andthe blood burden fluctuations detectable during M.

haemofelis infection. The presence of M. haemofelis intissues has been previously assessed by sensitive real-timeqPCR (Tasker et al., 2009). Tissue samples from M.

haemofelis-infected cats were collected at different timesduring the infection, and the ratio of hemoplasma bloodloads to tissue loads was calculated. In tissues collectedduring the lowest point of M. haemofelis copy numbercycling, no detectable M. haemofelis copies or only very lowcopy numbers could be identified (Tasker et al., 2009). Forthis reason, the authors concluded that there was nosignificant M. haemofelis sequestration in any of the testedtissues and hypothesized that a better explanation for theblood load fluctuations might be related to the incompleteclearance of the infection by the immune system followedby a rapid replication of the remaining organisms in theblood of infected cats (Tasker et al., 2009). Recently, theratio of hemoplasma blood loads to tissue loads wasassessed in cats experimentally infected with ‘Candidatus

M. haemominutum’ and Feline Leukemia Virus (Wolf-Jackel et al., 2012). ‘Candidatus M. haemominutum’ tissueloads appeared to decrease over time, and most tissuesshowed only few copies; however, the lung, liver, spleenand aorta contained more copies than expected given theblood supply of the tissue. The kinetics of infection of‘Candidatus M. turicensis’ are different than those of M.

haemofelis and ‘Candidatus M. haemominutum’: ‘Candida-

tus M. turicensis’ produces lower blood loads and a shorterduration of bacteremia (Museux et al., 2009; Novacco et al.,2011; Willi et al., 2005). An early tissue sequestration mayexplain these particular kinetics. Therefore, the aim of thepresent study was to investigate ‘Candidatus M. turicensis’tissue loads in infected cats. We analyzed potential tissuesequestration in five experimentally infected cats shortlyafter they had cleared acute bacteremia. Hemoplasmablood and tissue loads were, respectively, measured andcalculated in these cats, as was previously done for otherfeline hemoplasma species (Novacco et al., 2011; Williet al., 2005; Wolf-Jackel et al., 2012). The humoral immuneresponse was monitored by DnaK ELISA to explain thepossible association between tissue sequestration andantibody levels.

2. Materials and methods

2.1. Animals and experimental design

Five specified-pathogen free (SPF) neutered male catswere enrolled in this study (NFS2, NFN1, NFP1, NFQ4 andAKM3). Cats were housed in groups in a confineduniversity facility under ethologically and hygienicallyideal conditions, as described (Geret et al., 2011). All of theexperiments were officially approved by the veterinaryoffice of the canton Zurich (TVB 159/2010). The cats weresubcutaneously inoculated with ‘Candidatus M. turicensis’-positive blood, as previously described (Novacco et al.,

containing 1 � 103 copies of ‘Candidatus M. turicensis’ asdetermined by real-time Taqman1 PCR (Novacco et al.,2012). The acute phase of infection in these cats has beenpreviously described (Novacco et al., 2012). Briefly, all fivecats became bacteremic, and, 31 days post-inoculation,tissue samples were collected by fine needle aspirationfrom the kidney, liver, and salivary glands and bonemarrow aspirates; all tissues samples tested positive byreal-time Taqman1 PCR for ‘Candidatus M. turicensis’(Novacco et al., 2012). The animals were sacrificed 86 dayspost-inoculation at a time point when they were expectedto be ‘Candidatus M. turicensis’ PCR-negative in peripheralblood. Sixteen milliliters of ethylenediaminetetraaceticacid (EDTA)-anticoagulated blood were collected fromeach cat prior to euthanasia. Hematology was performedusing a Sysmex XT-2000iV (Sysmex Corporation, Kobe,Japan) system, which has been used previously for felineblood samples (Weissenbacher et al., 2011). Packed cellvolume (PCV) values of 33–45% (5–95% quantiles of thereference range) were considered to be within thereference range, and anemia was defined as having aPCV <33%. For euthanasia, cats received a light intramus-cular anesthetic injection (20 mg/kg Ketamine and 0.1 mg/kg Midazolam; Dr. E. Graeub AG, Bern, Switzerland)followed by a pentobarbital administration via intravenouscatheter (Esconarkon, 150 mg/kg; Streuli Pharma SA,Uznach, Switzerland).

2.2. Tissue sample collection

The following 31 tissues were collected at necropsy:bone marrow; duodenum; jejunum; mesenteric caudallymph node; ileum; ileum (containing Payer’s Patches);colon; mesenteric cranial lymph node; rectum; liver;spleen; kidney cortex and medulla; urinary bladder;parotid and submandibular salivary gland; submandibularlymph node; tonsil; thyroid; sternal lymph node; myo-cardium; endocardium; aorta; lung; oral mucosa; brain;skeletal muscle, axillary lymph node; cervical lymph node;splenic lymph nodes and hepatic lymph nodes. Hepaticlymph nodes could be found only in one cat (NFQ4), andsplenic lymph nodes were found in 3 out of 4 cats (notavailable in NFP1). For this reason, samples from hepaticand splenic lymph nodes were omitted from the statisticalanalysis. All samples were snap frozen in liquid nitrogenimmediately after collection and stored at �80 8C untilanalysis.

2.3. Nucleic acids isolation

Total nucleic acids (TNA) were purified from 100 mLEDTA-anticoagulated blood using the MagNaPure LC totalnucleic acid isolation kit (Roche Diagnostics, Rotkreuz,Switzerland). In addition, nucleic acids were purified from10 mg of tissue using the DNeasy blood and tissue kit(Qiagen, Hombrechtikon, Switzerland). All TNA sampleswere eluted into 100 mL of elution buffer and stored at�80 8C until use. During all extractions, negative controlsconsisting of 100 mL phosphate-buffered saline wereconcurrently prepared with each batch of samples to

monitor for cross-contamination. 2012). Briefly, each cat received 10 mL of infectious blood

2.4.

tim‘Can

et aonefor

Theasseutmain cresaddalbthesign201andwa

2.5.

peron

thetiss(TatissM.

2.6.

DnprediluperserpreTheposbandesnoiet a

2.7.

PrisDieKru(PD

Frie(PD

anacorcon

M. Novacco et al. / Veterinary Microbiology 167 (2013) 403–409 405

Quantitative real-time TaqMan1 PCR assays

All blood and tissue samples were analyzed by real-e TaqMan1 PCR for the presence and loads ofdidatus M. turicensis’ as previously described (Willil., 2005). The lower limit of detection of the assays is

copy per reaction, which translates into 200 copies/mLblood samples and 20 copies/10 mg for tissue samples.

sensitivity and specificity of the assay was previouslyessed (Willi et al., 2006). The blood samples collected athanasia were tested in nine replicates in order toximize the accuracy of the copy number determinationats with negative or very low hemoplasma loads. This

ulted in a lower limit of detection of 22 copies/mL. Inition, a real-time TaqMan1 PCR amplifying feline

umin (fALB) was used in all tissue samples to test for presence of amplifiable DNA and the absence ofificant PCR inhibitors (Helfer-Hungerbuehler et al.,3). In each PCR run, one positive control (plasmid DNA)

two negative controls consisting of nucleic acid-freeter were included.

Quantification of blood and tissue loads

‘Candidatus M. turicensis’ blood loads were calculated mL of blood. The tissue loads were determined basedthe amount of tissue used for the extraction (�10 mg);

copy numbers expected given the blood supply of theue (66 mL/kg) were estimated, as previously describedsker et al., 2009; Wolf-Jackel et al., 2012). In addition, allue loads were normalized by dividing the ‘Candidatus

turicensis’ copy numbers by the fALB copy numbers.

Quantification of antibodies

Antibody levels were assessed using M. haemofelis

aK enzyme-linked immunosorbent assay (ELISA), asviously described (Wolf-Jackel et al., 2010). A serumtion of 1:100 and 50 ng of the recombinant protein

well were used. Wells containing antigen withoutum samples served as blanks, and wells containing-infection serum samples served as negative controls.

signal-to-noise ratio was calculated by dividing thet-infection absorbance by the pre-infection absor-ce values for each individual cat, as previouslycribed (Wolf-Jackel et al., 2010). An ELISA signal-to-se ratio of �1.5 was considered positive (Wolf-Jackell., 2010).

Statistics

Statistical analyses were performed using the GraphPadm for Windows version 5.0 (GraphPad software, Sango, CA, USA). Non-parametrical tests were used. Theskal–Wallis test (PKW) followed by the Dunn’s post test) was used to compare parameters multiple groups. Thedman’s test (PFriedman), followed by Dunn’s post-test

), was used for longitudinal analysis. Correlationlyses were performed using Spearman’s rank orderrelation coefficient test (rs). P-values <0.05 were

3. Results

3.1. ‘Candidatus M. turicensis’ blood loads in the five cats

Four out of five cats (NFS2, NFN1, NFQ4 and AKM3)tested PCR-negative for ‘Candidatus M. turicensis’ in theperipheral blood at the time of euthanasia in all ninereplicates tested (Fig. 1; Table 1); thus, the ‘Candidatus M.turicensis’ blood loads in these four cats were <22 copies/mL of blood. The time points when the four cats last testedpositive in the blood prior to sacrifice are listed in Table 1.The fifth cat (NFP1) tested ‘Candidatus M. turicensis’ PCR-positive on the day of euthanasia, although at very lowloads (one out of nine replicates tested positive, whichtranslates into 22 copies/mL, Table 1). Cat NFN1 had thelowest peak blood loads during acute infection of all of thecats (Fig. 1).

3.2. ‘Candidatus M. turicensis’ tissue loads in the five cats

Sufficient amounts of DNA and the absence ofsignificant PCR inhibitors were confirmed in all tissuesamples. All cats showed detectable ‘Candidatus M.turicensis’ loads in some of the tested tissues. In one cat(NFS2), all of the analyzed tissues tested (n = 30; 100%)were ‘Candidatus M. turicensis’ PCR-positive. In the fourremaining cats, some of the tissues were ‘Candidatus M.turicensis’ PCR-negative (Table 2): in NFN1, 28/30 tissuestested positive (93%); in AKM3, 26/31 tested positive(87%); in NFP1, 16/29 tested positive (55%); and in NFQ4,16/31 tested positive (52%). Because the blood loads wereundetectable to very low in all five animals, the numberand percentage of positive tissues also represent thenumber and percentage of tissues with ‘Candidatus M.turicensis’ loads higher than expected due to the bloodsupply in these tissues alone. Thus, 52–100% of the tested

Fig. 1. ‘Candidatus M. turicensis’ blood loads in the five infected cats

after inoculation. Cats were monitored for 86 days after inoculation

(day 0). Blood loads are expressed as copies/mL of blood (logarithmic

scale). Four out of five cats were ‘Candidatus M. turicensis’ PCR-negative

on the day of euthanasia; only cat NFP1, was PCR-positive, although at

low level (22 copies/mL). The black arrows indicate the time points of

tissue collection in these cats. The first was performed by fine needle

aspiration and was previously reported (Novacco et al., 2012), and the

nd was performed upon necropsy.

sidered to be significant. seco

M. Novacco et al. / Veterinary Microbiology 167 (2013) 403–409406

tissues had higher loads than expected by the tissue’sblood supply. The overall highest tissue load was5.79 � 105 copies/10 mg tissue and was found in thethyroid of cat NFN1. Tissue loads were significantlydifferent among the cats (PKW< 0.0001). Cats NFS2 andNFN1 had significantly higher ‘Candidatus M. turicensis’tissue loads than cats NFP1 and NFQ4 (PD< 0.0001, Fig. 2).Additionally, cat NFN1 had higher tissue loads than catAKM3 (PD� 0.01, Fig. 2).

3.3. ‘Candidatus M. turicensis’ tissue loads in different organs

No morphological changes were noted in the ana-lyzed tissues upon necropsy. Additionally, no significantalterations were reported after histopathological exam-ination. Most of the tissue samples showed higher‘Candidatus M. turicensis’ loads than expected given theblood supply of the tissue (PFriedman < 0.0001, Fig. 3).Significantly higher tissue loads than expected given the

Table 1

Correlation between median and maximum ‘Candidatus M. turicensis’ tissue loads and other selected parameters. Correlation analyses were performed

using Spearman’s rank order correlation coefficient test (rs).

Selected parameters NFS2 NFN1 NFP1 NFQ4 AKM3 rs

Median tissue loads 2.02 � 103 2.94 � 104 2.00 � 101 2.00 � 101 1.79 � 102

Maximum tissue loads 5.04 � 105 5.79 � 105 9.64 � 103 5.61 � 103 1.65 � 105

Maximum blood loadsa 1.17 � 105 2.10 � 104 8.24 � 104 1.11 � 105 7.34 � 105 ns

Blood loads at time of sacrificea nd nd 22 nd nd ns

Days of peak blood loads 35 28 35 28 28 ns

Days PCR-negative prior to sacrifice 9 3 0 23 3 ns

Antibody levels at day of euthanasiab 3017 3075 1848 5335 1675 ns

Antibodies at the peak levelb 3017 3791 2411 8158 1912 ns

Hematocrit (%) at time of sacrifice 28 34 41 36 25 ns

Leukocytes/mL at time of sacrifice 8100 5400 9300 9800 5900 ns

Lymphocytes/mL at time of sacrifice 4720 3130 3050 5680 1900 ns

Monocytes/mL at time of sacrifice 110 50 200 50 190 ns

Neutrophils/mL at time of sacrifice 2660 1910 5220 3790 3290 ns

Eosinophils/mL at time of sacrifice 630 270 840 280 540 nsa Blood loads are expressed as copies/mL of blood; nd, no detectable copies in the analyzed blood (<22 copies/mL).b Antibody levels are expressed as signal-to-noise ratios. ns, no significant differences were detected between median and maximum tissue loads and the

other selected parameters (P � 0.05).

Table 2

‘Candidatus M. turicensis’ tissue loads in selected tissues (copies per 10 mg of tissue). Tissue loads were determined by real-time Taqman1 PCR and 10 mg of

tissue as starting materials.

Tissue NFS2 NFN1 NFP1 NFQ4 AKM3

Bone marrow 25 116 nd nd nd

Duodenum 5084 378,038 1662 nd 4386

Jejunum 1175 512,179 48 20 974

Mesenteric caudal lymph node 292 4483 nd nd 34

Ileum 983 233,745 76 nd 21

Ileum (with Payer’s Patches) 687 122,531 20 134 56

Colon 6239 200,359 nd 3930 704

Mesenteric cranial lymph node 519 25,162 20 87 136

Rectum 39,092 109,966 nd 1225 1054

Liver 90 172 nd nd nd

Spleen 6513 20 21 1126 92

Kidney cortex 440 nd nd nd 157

Kidney medulla 564 nd nd 624 nd

Urinary bladder 723 1634 nd nd 244

Parotid gland 80,643 283,508 1786 226 165,330

Submandibular salivary gland 102,957 307,784 5211 20 104

Submandibular lymph node 3189 70,877 36 nd 34

Tonsils 156 420,188 51 293 179

Thyroid 504,412 579,365 9642 1862 48,384

Sternal lymph node 1736 20,589 nd 496 191

Myocardium 3960 340 21 nd nd

Endocardium 27,771 29,388 46 nd 320

Aorta 26,477 35,790 123 5605 12,083

Lung 2230 774 nd nd 159

Oral mucosa 388,247 88,931 5423 nd 3666

Brain 58 174 nd nd 245

Skeletal muscle 14,507 41,748 1989 nd 32,344

Axillary lymph node 2015 14,459 nd 97 672

Cervical lymph node 148 1971 nd 212 91

nd, no detectable copies in the analyzed tissues (<20 copies/10 mg).

blosali(PD

in

(PD

Fig.

of eu

scal

25th

‘Can

diffe

two

aste

Fig.

data

wer

M. Novacco et al. / Veterinary Microbiology 167 (2013) 403–409 407

od supply of the tissues were detected in the parotidvary glands (PD� 0.01), thyroid (PD� 0.01) and aorta� 0.01). Significantly higher tissue loads were foundthyroid than in the bone marrow (PD� 0.01), liver� 0.01) and kidney cortex (PD� 0.05).

3.4. Humoral immune response to ‘Candidatus M. turicensis’

All cats developed detectable antibody levels after‘Candidatus M. turicensis’ inoculation (Novacco et al.,2012). Cats became seropositive between 28 and 35 daysafter inoculation and remained seropositive until the endof the study (Novacco et al., 2012). Details on the kinetics ofantibodies were specified elsewhere (Novacco et al., 2012).Antibody levels on the day of euthanasia and at their peakare listed in Table 1.

3.5. Potential association of ‘Candidatus M. turicensis’ tissue

loads with various parameters

No significant correlation was found between themedian or maximum tissue loads of ‘Candidatus M.turicensis’ and any of the tested parameters, includingthe ‘Candidatus M. turicensis’ blood loads at the time ofeuthanasia, the time span between the last ‘Candidatus M.turicensis’ PCR-positive result in blood and euthanasia, theantibody levels at the time of euthanasia, the peakantibody levels, and any of the tested hematologicalparameters (hematocrit, leukocyte, lymphocyte, mono-cyte, neutrophil, eosinophil counts) determined at eutha-nasia (see also Table 1).

4. Discussion

This is the first study to demonstrate the significantsequestration of a feline hemoplasma in a large number of

2. ‘Candidatus M. turicensis’ tissue loads in the five infected cats at day

thanasia. The loads are shown as copies/10 mg of tissue (logarithmic

e). All data are represented as box plots. The boxes extend from the

to the 75th percentile; a horizontal line represents the median.

didatus M. turicensis’ tissue loads were tested for significant

rences between cats by the Kruskal–Wallis test (PKW< 0.0001;

asterisks indicate statistically significant differences PD� 0.01; three

risks PD� 0.001).

3. Comparison of ‘Candidatus M. turicensis’ tissue loads in the analyzed organs. The loads are shown as copies/10 mg of tissue (logarithmic scale). All

are represented as box plots. The boxes extend from the 25th to the 75th percentile; a horizontal line represents the median. The loads in most tissues

e higher than expected given the blood supply to the tissue alone.

M. Novacco et al. / Veterinary Microbiology 167 (2013) 403–409408

tissues in infected cats that had overcome bacteremia.Experimental infections with ‘Candidatus M. turicensis’always resulted in lower blood loads and shorterbacteremia than infections with M. haemofelis and‘Candidatus M. haemominutum’. We had hypothesizedthat the different kinetics of ‘Candidatus M. turicensis’infection could be related to an early tissue sequestration(Novacco et al., 2012). All five cats included in the presentstudy had overcome the acute ‘Candidatus M. turicensis’infection without antibiotic treatment and had very low orundetectable blood loads at the time point of analysis.Remarkably, ‘Candidatus M. turicensis’ was detectable inthe majority of the tissues analyzed. ‘Candidatus M.turicensis’ tissue loads behaved differently from pre-viously reported behaviors of M. haemofelis. High M.

haemofelis tissue loads (up to 3.65 � 106 copies/5 mg) havebeen found in acutely infected cats with high blood loads(2.64 � 109 copies/mL) (Tasker et al., 2009). However, M.

haemofelis tissue loads significantly decreased, and most ofthe tissues tested PCR-negative, when the cats turned PCR-negative in the blood (Tasker et al., 2009). Thus, for M.

haemofelis, the number of the organisms in tissuesappeared to be related to the blood loads. In accordancewith this observation, M. haemofelis was previouslydetected only on the surface of erythrocytes in tissues(Peters et al., 2011), supporting the idea that the tissuelocalization of M. haemofelis is mainly due to the transportof the organisms by erythrocytes in the peripheral blood. Incontrast, in ‘Candidatus M. haemominutum’-infected cats,evidence of sequestration in some tissues (spleen, lung,liver, heart and aorta) has been found (Wolf-Jackel et al.,2012). A previous study showed that the maximumtissue load for ‘Candidatus M. haemominutum’ was1.3 � 104 copies/10 mg in cats that tested PCR-positive(ranging from 1.9 � 105 copies/mL to 1.2 � 106 copies/mLof blood) (Wolf-Jackel et al., 2012). However, all of the catsin that study were sacrificed during bacteremia, and notissue loads were reported in blood PCR-negative cats. Inthe present study, the average ‘Candidatus M. turicensis’tissue copy number was 3.46 � 104 copies/10 mg. Four outof five cats tested ‘Candidatus M. turicensis’ PCR-negativeat the time of necropsy. Only one cat (cat NFP1) tested PCR-positive, although at low levels (22 copies/mL). Thus, bloodloads were not expected to have any significant influenceon the tissue loads. As a result, we were able to concludethat ‘Candidatus M. turicensis’ is significantly sequesteredin various tissues and that the tissue localization persistedafter bacteremia was cleared, in contrast to what had beenreported previously for M. haemofelis (Tasker et al., 2009).Additionally, ‘Candidatus M. turicensis’ tissue loads wereeither as high as or up to 100 times higher than levelsreported for the ‘Candidatus M. haemominutum’ loadsduring bacteremia (Wolf-Jackel et al., 2012).

The highest observed ‘Candidatus M. turicensis’ tissueloads were 5.79 � 105 copies/10 mg in cat NFN1. Remark-ably, this cat had the lowest blood loads during acutebacteremia of all of the cats. We speculated that, at least insome animals, a significant tissue sequestration of theorganisms may lead to a decreased presence of theorganisms in the blood. Thus, lower bacteremia mightbe associated with higher tissue sequestration in

‘Candidatus M. turicensis’-infected cats. However, no suchassociation was found when comparing the peak bloodloads and tissue loads in the five cats; this finding could beascribed to the small number of cats included in the study.Alternatively, the blood load might not be the only factorinfluencing tissue sequestration. To find a potential alter-native explanation for the significantly different ‘Candidatus

M. turicensis’ tissue loads among the five cats, several otherparameters were analyzed for a potential association. Thecat (NFQ4) with lower tissue loads than the other cats (NFS2and NFN1) was the first cat to clear the infection; this cat alsohad the highest peak and final levels of antibodies to thefeline hemoplasma DnaK antigen. However, no overallassociation between tissue loads and any of these threeparameters (time of clearance of the hemoplasma fromblood and antibody levels at any time point) could bedemonstrated; the same was true for all the testedhematological parameters. The lack of a clear associationcould be related to multifactorial influences, but it mightalso be due to the small number of animals in the study. Thenumber of animals was limited for ethical reasons.

‘Candidatus M. turicensis’ tissue sequestration seemedto follow a certain trend during bacteremia and decreasedover time. In this study, we described an early tissuesequestration of ‘Candidatus M. turicensis’ in the five catsenrolled in this study (Novacco et al., 2012); in theprevious study, kidney, liver, salivary gland and bonemarrow samples were collected during the period of peakbacteremia by fine needle aspiration (Novacco et al., 2012;Fig. 1). Interestingly, particularly high loads were found inthe bone marrow of three of the cats that were highlybacteremic at the time (Novacco et al., 2012; Fig. 1), while,at the time of sacrifice when the cats had very low orundetectable bacterial blood loads, the tissue loads in thebone marrow from the same cats were low or undetectable(Table 2). In accordance with these findings, high‘Candidatus M. turicensis’ loads were also reported in thebone marrow (104 copies/mL) of three cats that had beenused to amplify ‘Candidatus M. turicensis’ in vivo and thatwere bacteremic at the time of investigation (106 copies/mL of blood) (Museux et al., 2009).

The tissue sequestration of ‘Candidatus M. turicensis’appears to persist for a prolonged period of time. Thepersistence of ‘Candidatus M. turicensis’ at low levels indifferent tissues was also demonstrated previously inchronically infected animals that had PCR-positive tissuesmany months after the end of bacteremia (Novacco et al.,2011, 2012). Three cats tested PCR-positive in the kidney,liver and bone marrow six to eleven months afterinoculation (Novacco et al., 2011). The copy numbers inthose samples, however, were very low (maximum80 copies/106 cells). It would be of great interest todetermine the exact localization of ‘Candidatus M. tur-icensis’ within the tissue cells. However, the in situ

hybridization technique that was used for the localizationof M. haemofelis in different tissues (Peters et al., 2011)could not be successfully used for a ‘Candidatus M.turicensis’-infected cat in spite of high blood loads(2.6 � 105 copies/mL) (Peters et al., 2011). Further studiesand different approaches are required to address thisquestion.

loain aet aloacolothesaliHowundstuinoeutsubbac

webacprebaccoufousacsignvarmaseqanttheundhostreatheoth

Com

inte

Aut

marevwonecparandthe

Ack

supVet

M. Novacco et al. / Veterinary Microbiology 167 (2013) 403–409 409

We reported higher ‘Candidatus M. turicensis’ tissueds in the thyroid, parotid salivary gland and aorta thanll other analyzed tissues. Peters and colleagues (Petersl., 2011) detected high ‘Candidatus M. turicensis’ tissue

ds in the tonsil, mesenteric lymph nodes, mid-jejunum,n and colonic lymph nodes of one cat by PCR. Overall,

two studies suggest that high loads were present in thevary gland, mesenteric lymph nodes and jejunum.

ever, a comparison of these two studies must beertaken with caution. Only one cat was included in the

dy of Peters and colleagues. Additionally, the cat wasculated intravenously and was still PCR-positive athanasia. In our study, the cats were inoculatedcutaneously and were sacrificed at the end of theteremia.In conclusion, high loads of ‘Candidatus M. turicensis’re found in tissues of infected cats after the clearance ofteremia. The tissue loads were higher than the loadsviously reported in M. haemofelis-infected cats afterteremia. The presence of the organisms in the tissuesld not be explained by the blood loads alone becauser out of five cats tested PCR-negative at the time ofrifice. All of these results support the hypothesis of aificant ‘Candidatus M. turicensis’ sequestration in

ious tissues. This result has direct implications on thenagement and treatment of infected cats. A tissueuestration would explain the decrease of efficacy of theibiotic treatments in obtaining a complete clearance of infection. It would be of great interest to bettererstand the localization of the organisms within thet tissue in order to better target the antibiotictment. Further studies are needed to better clarify

differences between ‘Candidatus M. turicensis’ ander feline hemoplasma species.

peting interests

The authors declare that they have no competingrests.

hor’s contributions

M.N. carried out the in vivo experiments and drafted thenuscript. B.R. was responsible for the SPF cats andised the manuscript. M.M. supported the laboratoryrk and revised the manuscript. P.G. performed theropsy examination and revised the manuscript. R.H.-L.ticipated in the design and coordination of the study

edited the manuscript. All authors read and approved final manuscript.

nowledgments

The laboratory work was performed with logisticport from the Center for Clinical Studies at thesuisse Faculty of the University of Zurich. The authors

express their gratitude to T. Meili and B. Weibel for theirexpert technical assistance during the study. Moreover, theauthors thank the animal caretakers, M. Rios and D.Brasser, for their excellent work with the cats. M.N. is therecipient of a postdoctoral grant by the Novartis Founda-tion, formerly Ciba-Geigy-Jubilee Foundation. The financialsupport for this project was obtained through a researchgrant from the University of Zurich (Stiftung fur wis-senschaftliche Forschung 2009).

References

Foley, J.E., Harrus, S., Poland, A., Chomel, B., Pedersen, N.C., 1998. Mole-cular, clinical, and pathologic comparison of two distinct strains ofHaemobartonella felis in domestic cats. Am. J. Vet. Res. 59, 1581–1588.

Foley, J.E., Pedersen, N.C., 2001. ‘Candidatus Mycoplasma haemominu-tum’, a low-virulence epierythrocytic parasite of cats. Int. J. Syst. Evol.Microbiol. 51, 815–817.

Geret, C., Riond, B., Cattori, V., Meli, M., Hofmann-Lehmann, R., Lutz, H.,2011. Housing and care of laboratory cats: from requirements topractice. Schweiz. Arch. Tierheilkd. 153, 157–164.

Helfer-Hungerbuehler, K., Widmer, S., Hofmann-Lehmann, R., 2013.GAPDH pseudogenes and the quantification of the feline genomicDNA equivalents. Biomed. Res. Int. (submitted).

Museux, K., Boretti, F.S., Willi, B., Riond, B., Hoelzle, K., Hoelzle, L.E.,Wittenbrink, M.M., Tasker, S., Wengi, N., Reusch, C.E., Lutz, H., Hof-mann-Lehmann, R., 2009. In vivo transmission studies of ‘CandidatusMycoplasma turicensis’ in the domestic cat. Vet. Res. 40, 45.

Novacco, M., Boretti, F.S., Franchini, M., Riond, B., Meli, M.L., Hofmann-Lehmann, R., 2012. Protection from reinfection in ‘‘Candidatus Myco-plasma turicensis’’-infected cats and characterization of the immuneresponse. Vet. Res. 43, 82.

Novacco, M., Boretti, F.S., Wolf-Jackel, G.A., Riond, B., Meli, M.L., Willi, B.,Lutz, H., Hofmann-Lehmann, R., 2011. Chronic ‘‘Candidatus Myco-plasma turicensis’’ infection. Vet. Res. 42, 59.

Peters, I.R., Helps, C.R., Willi, B., Hofmann-Lehmann, R., Gruffydd-Jones,T.J., Day, M.J., Tasker, S., 2011. Detection of feline haemoplasmaspecies in experimental infections by in-situ hybridisation. Microb.Pathog. 50, 94–99.

Tasker, S., Peters, I.R., Day, M.J., Willi, B., Hofmann-Lehmann, R., Gruffydd-Jones, T.J., Helps, C.R., 2009. Distribution of Mycoplasma haemofelis inblood and tissues following experimental infection. Microb. Pathog.47, 334–340.

Weissenbacher, S., Riond, B., Hofmann-Lehmann, R., Lutz, H., 2011. Eva-luation of a novel haematology analyser for use with feline blood. Vet.J. 187, 381–387.

Willi, B., Boretti, F.S., Baumgartner, C., Tasker, S., Wenger, B., Cattori, V.,Meli, M.L., Reusch, C.E., Lutz, H., Hofmann-Lehmann, R., 2006. Pre-valence, risk factor analysis, and follow-up of infections caused bythree feline hemoplasma species in cats in Switzerland. J. Clin.Microbiol. 44, 961–969.

Willi, B., Boretti, F.S., Cattori, V., Tasker, S., Meli, M.L., Reusch, C., Lutz, H.,Hofmann-Lehmann, R., 2005. Identification, molecular characteriza-tion, and experimental transmission of a new hemoplasma isolatefrom a cat with hemolytic anemia in Switzerland. J. Clin. Microbiol.43, 2581–2585.

Willi, B., Boretti, F.S., Tasker, S., Meli, M.L., Wengi, N., Reusch, C.E., Lutz, H.,Hofmann-Lehmann, R., 2007. From Haemobartonella to hemoplasma:molecular methods provide new insights. Vet. Microbiol. 125, 197–209.

Wolf-Jackel, G.A., Cattori, V., Geret, C.P., Novacco, M., Meli, M.L., Riond, B.,Boretti, F.S., Lutz, H., Hofmann-Lehmann, R., 2012. Quantification ofthe humoral immune response and hemoplasma blood and tissueloads in cats coinfected with ‘Candidatus Mycoplasma haemominu-tum’ and feline leukemia virus. Microb. Pathog. 53, 74–80.

Wolf-Jackel, G.A., Jackel, C., Museux, K., Hoelzle, K., Tasker, S., Lutz, H.,Hofmann-Lehmann, R., 2010. Identification, characterization, andapplication of a recombinant antigen for the serological investigationof feline hemotropic Mycoplasma infections. Clin. Vaccine Immunol.17, 1917–1925.