INFECTION AND IMMUNITY, Feb. 2007, p. 634–642 Vol. 75, No. 20019-9567/07/$08.00�0 doi:10.1128/IAI.01228-06Copyright © 2007, American Society for Microbiology. All Rights Reserved.

Infection with Toxoplasma gondii Bradyzoites Has a Diminished Impacton Host Transcript Levels Relative to Tachyzoite Infection�†

A. E. Fouts and J. C. Boothroyd*Department of Microbiology and Immunology, Stanford University School of Medicine, Stanford, California 94305-5124

Received 2 August 2006/Returned for modification 29 August 2006/Accepted 30 October 2006

Toxoplasma gondii, an intracellular pathogen, has the potential to infect nearly every warm-blooded animalbut rarely causes morbidity. The ability for the parasite to convert to the bradyzoite stage and live insideslow-growing cysts that can go unnoticed by the host immune system allows for parasite persistence for the lifeof the infected host. This intracellular survival likely necessitates host cell modulation, and tachyzoites areknown to modify a number of signaling cascades within the host to promote parasite survival. Little is known,however, about how bradyzoites manipulate their host cell. Microarrays were used to profile the host tran-scriptional changes caused by bradyzoite infection and compared to those of tachyzoite-infected and uninfectedhosts cells 2 days postinfection in vitro. Infection resulted in chemokine, cytokine, extracellular matrix, andgrowth factor transcript level changes. A small group of genes were specifically induced by tachyzoite infection,including granulocyte-macrophage colony-stimulating factor, BCL2-related protein A1, and interleukin-24.Bradyzoite infection yielded only about half the changes seen with tachyzoite infection, and those changes thatdid occur were almost all of lower magnitude than those induced by tachyzoites. These results suggest thatbradyzoites lead a more stealthy existence within the infected host cell.

Toxoplasma gondii is an extremely common parasite in hu-mans and animals. Although sexual reproduction of this intra-cellular protozoan takes place only within felines, the interme-diate hosts (many species of mammals and birds) supportasexual reproduction consisting of two stages: tachyzoites andbradyzoites. Tachyzoites replicate rapidly, disseminate throughthe host, and cause tissue damage. Most are then cleared bythe host immune response but not before some have convertedinto the bradyzoite stage. Bradyzoites replicate slowly, form acyst within the host cell, and sustain a chronic infection for thelife of the mammalian host. These bradyzoites latently persistand cause little pathology in a healthy host but, in an immu-nocompromised animal, they can reconvert into the tachyzoitestage and cause potentially fatal encephalitis.

Toxoplasma has a variety of mechanisms to co-opt the hostcell and evade host defenses, thereby promoting intracellularsurvival. In particular, a number of studies indicate thattachyzoites manipulate various signaling pathways within thehost cell. For example, tachyzoite-infected cells have beenshown to be resistant to the induction of apoptosis through thetargeting of multiple, distinct steps (29, 33, 20, 7). Toxoplasmatachyzoites also manipulate host cell NF-�B signaling (32, 28),as well as mitogen-activated protein kinase signaling based onthe fact that tachyzoite-infected macrophages are refractory toadditional stimulation by lipopolysaccharide (26, 23). Recentresearch has also shown that tachyzoite proteins can be in-jected into the host cell upon invasion (19, 21, 30) and that at

least one of these, a protein kinase, can have major effects onhost transcription (30).

To better understand the interaction between parasite andhost, microarray technology has been used by several groupsfor genome-wide analysis of the effects of the intracellulartachyzoite on the host cell transcriptome. Two groups haveshown cell-specific responses to Toxoplasma tachyzoites indendritic cells, macrophages, and retinal vascular endothelialcells (10, 25). Another group compared host gene expression inhuman foreskin fibroblasts (HFFs) infected by Toxoplasmatachyzoites with infection by other pathogens and identifiedtwo genes specifically induced by Toxoplasma (MacMarcks andtransferrin receptor) (17). Further studies have confirmed thatthe parasite-induced increase in host transferrin receptor aidsparasite survival (18). In addition, marked differences betweenearly and later time points following infection with tachyzoiteswere revealed by a time course analysis (3): transcript levelsthat changed at 2 h postinfection (hpi) primarily encoded im-mune response proteins that did not require parasite invasionfor their increase, whereas at 24 hpi there were many addi-tional changes, including increases in the transcript levels ofgenes encoding enzymes in the glycolytic and mevalonate syn-thesis pathways.

In contrast to this large body of data on infection withtachyzoites, relatively little is known about changes mediatedby intracellular bradyzoites. There are many biological differ-ences between tachyzoites and bradyzoites that predict thehost responses to these stages are probably very different. Forexample, a number of studies have revealed developmentallyregulated Toxoplasma genes, including metabolic enzymes, se-creted proteins, and surface proteins (1, 36, 13, 15, 11, 24, 34,31). These differences in gene expression correspond with amuch slower growth rate for bradyzoites and development ofa cyst wall characteristic of this stage, both of which mightcontribute to bradyzoite persistence. Furthermore, unlike

* Corresponding author. Mailing address: Department of Microbi-ology and Immunology, Fairchild Building D305, 300 Pasteur Dr.,Stanford University School of Medicine, Stanford, CA 94305-5124.Phone: (650) 723-7984. Fax: (650) 723-6853. E-mail: john.boothroyd@stanford.edu.

† Supplemental material for this article may be found at http://iai.asm.org/.

� Published ahead of print on 6 November 2006.

634

on March 17, 2021 by guest

http://iai.asm.org/

Dow

nloaded from

on March 17, 2021 by guest

http://iai.asm.org/

Dow

nloaded from

on March 17, 2021 by guest

http://iai.asm.org/

Dow

nloaded from

tachyzoites, which attract a strong proinflammatory response,bradyzoites often persist in the animal without attracting im-mune infiltrates (24). This led us to hypothesize that brady-zoites might produce a unique signature of changes in the hostcell transcriptome.

In the present study, human cDNA microarrays were used toinvestigate whether and how the changes in host gene expres-sion during infection with bradyzoites differ from those duringinfection with tachyzoites. Employing a commonly used methodfor induction of bradyzoite differentiation (high pH and lowserum) (34), parallel cultures of HFFs infected by bradyzoitesand tachyzoites were obtained, and microarrays were used toprofile changes in host gene expression. We observed that,overall, bradyzoite infection caused transcriptional changes oflesser magnitude than those brought about by tachyzoite in-fection but that the two stages have similar effects on hosttranscription.

MATERIALS AND METHODS

Cell culture. Cultures of primary HFFs at passages 9 to 13 were grown toconfluence in 175-cm2 flasks and incubated for 3 to 5 weeks in a humidified,37°C, 5% CO2 incubator. Fresh medium was added to these cultures the daybefore infection. Pru strain parasites expressing green fluorescent protein (GFP)from a bradyzoite-specific promoter (bradyzoite-specific GFP-4, BSG-4) (34)were allowed to lyse their host cells and remain extracellular for approximately24 h before they were washed three times, counted, and added to HFFs at aneffective multiplicity of infection of 1:5. At 4 h postinfection (hpi) the mediumwas removed, and fresh medium of either Dulbecco modified Eagle medium plus10% fetal calf serum (pH 7.5) (“standard”) or RPMI plus 1% fetal calf serum,buffered with 50 mM HEPES to pH 8.2 (“stress”), was added. Flasks withstandard medium were incubated in the conditions described above, whereasflasks with stress medium were incubated as described above but capped tightlywithout added CO2. Tachyzoite-infected cultures were labeled “standard�TZ”;bradyzoite-infected cultures were termed “stress�BZ.” At 44 hpi the cells wereharvested for RNA preparation. Only cultures with infection rates of at least20% of the cells and a bradyzoite conversion of better than 90% were used forRNA isolation. The bradyzoite conversion was assayed by counting all vacuolesin an area and then counting the vacuoles that contained GFP-positive brady-zoites. In the standard�TZ cultures, no conversion to bradyzoites occurredbased on a total absence of detectable GFP expression (data not shown).

cDNA synthesis. Total RNA was extracted by using TRIzol reagent (Gibco-BRL). RNA was assessed for quality using spectrophotometry and gel analysis.mRNA was isolated by using Oligotex mRNA minikit (QIAGEN). cDNA wassynthesized by using Superscript II (Life Technologies) and quantified by mea-suring ethidium fluorescence in electrophoresed samples.

Labeling and microarrays. Second-strand cDNA was labeled by using thedirect incorporation of Cy3 (reference) or Cy5 (sample) dUTPs (Amersham)through random nonomer priming with Klenow enzyme (Gibco-BRL). Type IImicroarray experiments were performed (common reference on all slides) (14)using a reference sample made from a pool of first-strand cDNA from allconditions. Each reference and sample were labeled in parallel and purified byusing YM-30 columns (Amicon) and then simultaneously hybridized to humancDNA microarrays (Stanford Functional Genomics Facility) by using 3.4� SSC(1� SSC is 0.15 M sodium chloride plus 0.015 M sodium citrate), 0.3% sodiumdodecyl sulfate, 20 �g of poly(A) RNA (Sigma), and 2 �g of yeast tRNA (Roche)at 65°C for at least 22 h. Prior to hybridization microarray slides were hydrated,cross-linked, and prehybridized in 25% formamide according to the manufac-turer’s instructions (Corning UltraGAPS-coated slides instruction manual).Slides were rinsed in water, followed by 95% ethanol, and then dried by centri-fugation. Hybridization was followed by washing for 5 min each in 2�, 1�, and0.2� SSC and then drying by centrifugation. Each microarray contains 40,996cDNA spots representing 23,228 unique putative genes (gene identity informa-tion may be found in Fig. 2 and 4).

Data analysis. Slides were scanned by using an Axon Genepix 4000A andgridded by using Genepix 5.1. The data were entered into Stanford MicroarrayDatabase (SMD [genome-www5.stanford.edu]), and two-dimensional spatial lo-cal estimation was used to normalize the spots (span factor of 0.4) (39) to enablecomparison between arrays (raw data will be available from SMD, GEO, and

ArrayExpress repositories). Five arrays per condition were selected to be in-cluded in the analysis based on overall signal-to-noise ratios. These arrays in-clude at least one from each of three biological replicates and an additional twotechnical replicates. Three filters to remove poor quality spots were appliedbefore data were downloaded: the spot had to be flagged as “OK” in Genepix,the regression correlation had to be greater than 0.6, and the channel 2 signalhad to be at least threefold greater than the background. Genes for which thereare high-quality data on 14 of 20 of the arrays (18,425 spots) were entered intoMeV 3.0 (www.tigr.org). The following pairwise combinations of the conditionswere used for two-class unpaired significance analysis of microarrays (SAM)(37) analysis (1% false-positive cutoff): standard versus (standard�TZ), stressversus (stress�BZ), and standard versus stress. Gene Ontology (http://www.geneontology.org/) analysis was performed on the SAM data with the additionalcutoff of requiring a fold change greater than 1.5, either up or down. Theabbreviated gene symbol (e.g., STAT1 [for signal transducer and activator oftranscription 1]) was used to reduce the lists to include one copy of each gene forfurther analysis. Using Onto-Express (http://vortex.cs.wayne.edu/projects.htm[12]), each individual list of significant genes was queried against the entire listof genes present in the input for the SAM analysis. Lists of significant genes fromthe most significant gene ontology (GO) categories were used to query the 18,425spots to obtain data on all replicates of a gene. We used a cutoff of 14 or moregenes in the GO category to consider it for discussion. The replicate experimentswere averaged, entered into MeV 3.1, and clustered with average linkage Eu-clidian parameters for image generation.

ELISA. HFFs were grown in 24-well tissue culture plates. The experimentcommenced exactly as for the cultures used for microarrays, but 2 h before thefinal time point (44 hpi), medium was removed, and cells were washed with warmphosphate-buffered saline. A total of 600 �l of fresh medium of the proper type(standard or stress) without serum was added to the wells. After 2 h, the mediumwas harvested for enzyme-linked immunosorbent assay (ELISA) experiments.ELISAs were performed in duplicates according to manufacturer’s protocol forCCL2 (1:10 dilution of sample; eBioscience, San Diego, CA) and CXCL1 (nodilution of sample; BD-Clontech, Palo Alto, CA).

Semiquantitative PCR. cDNA from microarray experiment 3 (for a descrip-tion of the experiments, see supplemental material) was diluted to 10 ng/�l. Astandard curve was made by using one of the samples at 20 ng/�l and diluting itfurther to 1:10, 1:100, and 1:1,000. PCR was performed with 25 or 50 ng of cDNAand QuantiTect SYBR PCR Mastermix (QIAGEN) according to manufacturer’sinstructions on a Lightcycler 1.2 (Roche). Two replicates of each sample wereperformed, quantitation was performed by using Lightcycler software version 3.5,and abundances were calculated based on the standard curve for each primerpair. Primer pairs were as follows: BCL2A1 (GAAGACGGCATCATTAACTand CCCAGCCTCCGTTTTG), GM-CSF (AGCATGTGAATGCCATC andGTTTCCGGGGTTGGAG), IL-24 (CTCCTTTGCTGGCGAC and GGGCACTCGTGATGTT), ENC1 (GCCGTCGTAGGTATTAGT and ACATCTAGGAACCAGGG), FAS (CCAACCTTAAATCCTGAAACA and GCCAATTACGAAGCAGT), and IFI16 (GTGCCAGCGTAACTCC and CCCGGTATTCCCACTT).

GenBank accession numbers. The accession numbers for the genes identifiedin Fig. 2 and 4 (in the order listed in these figures) are as follows: for Fig. 2A(cholesterol biosynthesis), AA477781, AI214581, N62195, T65790, T56013,N50834, H08205, R06715, AA420437, AA436425, AA779417, N67038, andAI364688; for Fig. 2B (lipid biosynthesis), AI081548, N47716, AA873159,AA991590, W80637, AA464566, AA126676, AI304790, H15842, AA456975,AI924357, AA504461, R25823, T56013, N62195, AA779417, AA216528, N98509,H29214, W48579, R28548, R07295, AA988876, AA972780, N70176, R72174,AA058383, AA401952, AA983633, H19050, AA190764, AI985827, T71976,H15155, N91990, N72215, AA453471, AI261602, T71713, AA457697, N95187,AA443630, AA459293, AA775445, AI393019, and AA693488; for Fig. 2C (ex-tracellular matrix structural constituent), AA293402, AA459308, W90359,AA035657, AA453712, AA455542, AA167222, AA284301, R68634, N73836,W89143, T98611, AA044829, AI679372, AA134757, AA614680, W93067,AA434067, R62612, AI262682, AA953560, N26285, W90740, AA599273,AA430540, AA994760, R52037, AA464042, AI830005, W84711, H77493,AA633747, AA682527, N78828, T99055, AA046525, H99676, AA047208,W93113, AA455157, AA056415, AA418674, AA452840, AI982687, AA490172,N74178, AA857098, N67487, AA488444, AA393766, R77226, R78225,AA150402, AA780815, AA056013, H92621, AI038302, AA177109, AA177011,H17882, AI732248, AA125940, and AA195302; for Fig. 2D (growth factor ac-tivity), W42723, W46900, W46577, AA706226, R50018, N26311, AA011061,AA026118, AI811492, AA401111, AA489383, AA450062, AA292891,AA910443, AW008840, N71102, AA946776, N74623, N76677, H59614, R42395,AA456321, AA128355, AA701502, AA904948, AI083520, AA488780,

VOL. 75, 2007 HOST RESPONSE TO INFECTION BY TOXOPLASMA BRADYZOITES 635

on March 17, 2021 by guest

http://iai.asm.org/

Dow

nloaded from

AA253446, H23457, AA422166, AA234298, R56773, T52484, AA496452,AI054019, T62547, AA431428, AA463224, R70684, R83377, AA293109,H11088, and AA630120; for Fig. 2E (cytokine activity), AA884403, R56773,AA708512, AI075036, AA463224, AA496452, AI054019, AA431428, N26311,AA450062, AA401111, AA281635, AI335002, AI074784, AA995402, R50018,and AA489629; for Fig. 2F (chemokine activity), AI016051, W72294, AA873792,AA780059, AA953842, W42723, W46900, AA935273, AI668847, AA425102,T77816, W69211, AI268937, AI359519, AA486072, and AI298976; for Fig. 4A(tachyzoite-specific decreases), AA971543, AA055520, AA041382, AA142919,AA894927, AI015679, AI375048, AI084504, N92167, R85685, AW028368,AA419176, AA677174, AA486367, AA076085, AA659567, AA459743, N81036,AI659563, H70866, H07920, AA176999, T62048, T69603, AA463224, AA885288,AA058383, T64469, AA971188, AA778276, N31493, N27086, AA229714,N70176, R34682, AA404967, AA079495, AA173408, W44856, AA486836,T71976, R09561, AA411394, AI823924, H50344, AA405488, N30573,AA070357, AA126869, AA479313, AA410636, AI675465, AA912034,AA292213, R64454, AA863403, AA678160, AA676466, AA676804, H13074,AI014468, AA629909, AA953006, AI017846, AA608531, AA190871, AA676405,R92852; for Fig. 4B (tachyzoite-specific increases), AA459263, AI766870,T53705, AA995402, AA449750, AA449720, W96134, W44549, AA293362,H87471, AA281635, N98757, T99302, H55907, AA429661, AA456161, R71691,and T55353; and for Fig. 4C (bradyzoite-specific changes), R62138, AW029226,H72122, N26311, AA126958, AA490996, AA425320, AA450062, N66644,AA044023, AI038014, R63922, AA045792, AA598526, W47003, AA010400,AA293570, AA287732, AA480994, AI361330, AI950601, AI284281, H58872, andAA679565.

RESULTS

Experimental system. To examine bradyzoite-specific effectson host cell transcription, bradyzoite-bearing cells would ide-ally be isolated from an infected animal. Unfortunately, how-ever, this approach is not feasible because of the extremely lownumber of such cells (e.g., an entire mouse brain typicallycontains only ca. 100 to 3,000 cysts). We attempted to use invitro cultures of primary rat astrocytes (neural fibroblasts) butwere unable to obtain consistent results due to limitations oncell viability and the variable purity of the starting material(data not shown). We therefore turned to another primary cellline, HFFs, that has been used extensively for the study ofbradyzoites in vitro. Fibroblasts are involved in the regulationof inflammation, and activated fibroblasts have roles such asproviding costimulatory signaling to leukocytes (38), influenc-ing Th1 or Th2 patterning (22), and downregulating the im-mune response in tissues to limit chronic inflammation (5).These functions of fibroblasts, along with their ease of manip-ulation as primary cell lines, make them a useful and relevantcell type for studying how Toxoplasma manipulates its host cell.

Pru strain parasites have been well characterized regardingtheir ability to convert into bradyzoites in vitro and in vivo. Weused a Pru strain engineered to express GFP by a bradyzoite-specific promoter (LDH2) (34). Parallel cultures were initiatedby adding tachyzoites to HFF monolayers and then, at 4 hpi,the medium was removed, and fresh standard medium wasadded for tachyzoite (“standard�TZ”) conditions or stressmedium was added for bradyzoite (“stress�BZ”) cultures toinduce conversion. The resulting percentage of cells infectedwas examined microscopically and found to be the same forstandard�TZ and stress�BZ cultures (data not shown).Mock-infected cultures were similarly treated to yield unin-fected “standard” and uninfected “stress” cultures. RNA fromall cultures was harvested at 44 hpi, just before tachyzoites lysetheir host cells (bradyzoite cultures are slower to grow and sodo not lyse the host cell for several days, if ever [data not

shown]). In preliminary experiments we determined that theefficiency of bradyzoite conversion decreased dramatically withincreasing multiplicity of infection. The optimum was achievedwhen ca. 20% of host cells were infected, which allowed at least90% bradyzoite conversion (data not shown). The implicationsof this are further discussed below.

To examine host transcription, microarrays were used tocompare mRNA transcript abundance for the four experimen-tal conditions: standard, stress, standard�TZ, and stress�BZcultures. These data are all available in the SMD (2; (http://genome-www5.stanford.edu/microarray). Three separate,pairwise SAM analyses were performed (37) to obtain signif-icance and fold changes for 18,425 spots (9,750 unique genes)on the arrays for these pairs: standard versus standard�TZ,stress versus stress�BZ, and standard versus stress. (A table ofthese results is included as supplemental Fig. 1.) To increaseour confidence in the data, further analysis was performed onlyon genes that were called significant by SAM and changed1.5-fold or more. Due to the low percentage of cells infected(�20%), relatively small changes in transcript levels mightreflect larger changes in the subset of cells that are actuallyinfected. Unfortunately, however, the inherent variabilityacross multiple microarray experiments precluded the inclu-sion of fold changes less than 1.5 in our study and so genes withonly modest changes in infected cells will not be identified inthese analyses. Despite these limitations, 1483(14%) and 692(7%) of unique genes changed significantly in at least one spotas a result of tachyzoite or bradyzoite infection, respectively,whereas just 349 (4%) of the genes’ transcript levels changedsignificantly from stress medium alone (Fig. 1).

Stress conditions induce cholesterol and lipid synthesis. Toreveal the kinds of genes differentially expressed, a gene on-tology-based analysis was used (12). The list of significantgenes from each pairwise SAM output was queried against theentire input list, and P values were obtained for the enrichmentof functional categories of genes. To determine the effects ofthe stress medium on HFFs, the two uninfected conditionswere also compared. The stress conditions alone (i.e., with noinfection) induced primarily genes in the GO categories “cho-lesterol biosynthesis” and “lipid metabolism” suggesting thatthe low serum conditions increased intracellular lipid produc-

FIG. 1. Differential gene expression due to infection and/or culturecondition. Venn diagram of the statistically significant genes in pair-wise comparisons using a SAM 1% false-positive rate and a 1.5-foldchange cutoff. The pairwise comparisons include standard versusstandard�TZ, stress versus stress�BZ, and standard versus stress.

636 FOUTS AND BOOTHROYD INFECT. IMMUN.

on March 17, 2021 by guest

http://iai.asm.org/

Dow

nloaded from

tion (Table 1 and Fig. 2A and B). These results are consistentwith previous microarray studies that examined serum starva-tion in HFFs (8).

Comparing bradyzoite-infected cells to uninfected controlsrevealed a slight, further increase in transcript levels for cho-lesterol biosynthesis genes over and above the effect seen withthe stress conditions alone. An increase was also seen withtachyzoite-infected cells versus uninfected cultures (Table 1and Fig. 2A and B). These data are consistent with previousmicroarray analysis that showed that enzymes in the meval-onate synthesis pathway were upregulated in tachyzoite-in-fected cells at 24 hpi The same upward trend in this latterpathway was evident here, although it failed to achieve statis-tical significance (asterisks in Fig. 2A and B). There is evidencethat this induction occurs only in infected cells (35), and thusthe lesser induction seen here might be due to the lower per-centage of cells infected in the present study.

Infection and stress both cause primarily decreases in ex-pression of genes encoding extracellular matrix proteins andgrowth factors. A primary role of fibroblasts is the secretionand modification of extracellular matrix; thus, it is not surpris-ing that “extracellular matrix structural constituent” was themost significant gene ontology category for all three pairwiseconditions (Table 1). Either stress or infection alone reducedtranscript levels of extracellular matrix constituents, and thecombination of these two conditions appears to cause evengreater decreases in transcript levels (Fig. 2C), suggesting anadditive effect of infection and stress.

The next most highly significant category modulated by bothstress and infection is “growth factor activity” (Table 1). Anumber of fibroblast growth factors were markedly decreasedas a result of all three conditions (Fig. 2D). This is consistentwith the previous finding that pH stress or serum starvationdecreases cell proliferation (6). Our data indicate that infec-tion with Toxoplasma also causes a decrease in gene productsnecessary for the proliferation of fibroblasts although this isexperimentally difficult to examine, given that the cells arephysically lysed by the parasites after about 2 to 3 days ofinfection.

Previous reports, in which transcript levels were examined inHFF cells at 24 hpi, found few transcript levels that weredecreased in infection (17, 3). It was, therefore, not expected tosee the large number of such transcripts, particularly intachyzoite-infected cells at day 2. This is readily explained,however, by the fact that abundance is a function of bothsynthesis and decay and so genes whose transcription is

switched off might not show a significant change in overallabundance until enough time has elapsed for decay of preex-isting message. For the genes described here, this effect mightnot become apparent until 44 h have elapsed. Alternatively,the expression of these genes might not be impacted untilparasite growth has reached a critical point.

Infection by tachyzoites or bradyzoites causes increases inchemokine and cytokine transcript and protein levels. GOcategories highly enriched by tachyzoite and bradyzoite infec-tion, but not stress alone, included “cytokine activity” and“chemokine activity” (Table 1 and Fig. 2E and F). To examinehow changes in transcript levels correlate with protein levels,ELISA experiments were performed with supernatant har-vested from HFFs treated as for the microarray analysis exceptthat at 42 hpi the medium was replaced and harvested 2 h laterfor the measurement of cytokine secretion by ELISA. Weconfirmed that CXCL1 (GRO�) and CCL2 (MCP1) proteinlevels were increased in accordance with their highly inducedtranscript levels (Fig. 3) in tachyzoite and bradyzoite infec-tions. The relative differences in the expression of each of thesemolecules is similar to the transcript level data from the mi-croarrays.

Transcript changes are not specific to bradyzoite infection.To compare host response to bradyzoite and tachyzoite infec-tion, we examined the lists of genes deemed significant by SAMin two pairwise comparisons: standard versus standard�TZand stress versus stress�BZ. The majority (77%) of the genessignificantly affected by bradyzoite infection were also signifi-cant for tachyzoite conditions (Fig. 1). To determine which ofthe remaining 23% were truly bradyzoite specific, the geneswere manually examined according to the following criteria.After obtaining all spotted replicates for each gene consideredsignificant by SAM, we required that two of three of spots showthe same trend and that the ratio of stress�BZ to stress be atleast 1.5-fold the ratio of standard�TZ to standard. The finallist of genes that met these criteria as “bradyzoite specific”consisted of only 14 genes (12 upregulated and 2 downregu-lated), and only two of these genes had fold changes greaterthan two. The 1% false-positive rate of SAM predicts 7 false-positive genes, suggesting that at least half of the 14 genesobserved may not be truly specific to bradyzoite versustachyzoite infection. To explore this further, quantitative PCR(qPCR) was performed on three of the genes specifically in-duced by bradyzoite infection according to the array data. Theresults (see Fig. 5) showed that ENC1 transcript levels changedin the manner predicted by the array data, but IFI16 showed

TABLE 1. Significant GO categories for infection and/or stress

Gene ontology category (no. in category)aP (no. of genes in category)b

Standard�TZ standard Stress�BZ stress Stress standard

Cholesterol biosynthesis (44) 0.66 (2) 0.37 (1) 3.23E-11 (9)Lipid metabolism (126) 0.48 (20) 0.10 (14) 6.04E-04 (14)Extracellular matrix structural constituent (44) <1.00E-12 (26) 3.05E-10 (18) 1.71E-07 (11)Growth factor activity (61) 3.13E-07 (25) 3.48E-05 (15) 3.01E-04 (9)Cytokine activity (39) 4.00E-03 (13) 0.01 (8) 0.08 (3)Chemokine activity (27) 4.00E-03 (10) 0.05 (5) 1.00 (0)

a The category name is followed in parentheses by the number of genes in the analysis that fit into that category.b The corrected P value for a pairwise SAM comparison is followed in parentheses by the number of significant genes in that category. Statistically significant results

are in boldface.

VOL. 75, 2007 HOST RESPONSE TO INFECTION BY TOXOPLASMA BRADYZOITES 637

on March 17, 2021 by guest

http://iai.asm.org/

Dow

nloaded from

FIG. 2. GO categories significantly modulated by infection and/or culture condition. Clusters include genes called significant by SAM in highlysignificant GO categories. Vertical groupings show hierarchical clustering of gene expression in HFF with or without stress and/or parasites 44 hpiThe matrix contains log-transformed medians of normalized ratios, averaged over each condition and converted to a color scale as shown (log2values greater than 3 or less than �3 were assigned as 3 and �3, respectively). Each row represents the color-coded expression of one spotcorresponding to one gene on the arrays. Genes with asterisks are discussed in detail in the text. (A and B) “Cholesterol biosynthesis” and “lipidmetabolism”, two GO categories significantly modulated by stress; (C and D) “extracellular matrix structural constituent” and “growth factoractivity”, the two GO categories most significantly modulated by infection and stress; (E and F) “cytokine activity” and “chemokine activity,” twoGO categories significantly modulated by infection only.

638 FOUTS AND BOOTHROYD INFECT. IMMUN.

on March 17, 2021 by guest

http://iai.asm.org/

Dow

nloaded from

similar induction by both stages, and Fas antigen (FAS) ap-peared to be induced by tachyzoites only. These data indicatethat infection by bradyzoites in vitro causes the induction offew if any host genes distinct from those of tachyzoite infec-tion.

Transcript changes specific to tachyzoite infection. To de-termine whether there were genes induced exclusively bytachyzoite infection, we examined the list of 950 such genesusing the stringent criteria described above. Using these crite-ria, 54 genes (14 up and 40 down) were tachyzoite specific (Fig.4). SAM predicts only 14 false-positive genes in this category,and thus most of the 55 are expected to be real differences. Inaddition, 18 of the 54 had fold changes greater than two,further suggesting that the differences were real. qPCR wasperformed on three genes that the array data indicate areinduced �2-fold in a tachyzoite-specific manner (BCL2-re-lated protein A1 [BCL2A1], granulocyte-macrophage colony-stimulating factor [GM-CSF], and interleukin-24 [IL-24]). Inall three cases, the qPCR data were in agreement with thearray data, indicating that these genes were all specificallyinduced in tachyzoite infection (Fig. 5). The qPCR indicates anunderestimation by the microarrays of the extent of the

tachyzoite induction of these three genes. This is consistentwith a study comparing fold changes measured by qPCR versusmicroarrays that found that microarrays consistently underes-timate the induction relative to qPCR (40).

The GM-CSF induction agrees with the work of others thatfound it to be significantly induced in tachyzoite-infected fi-broblasts (9). This protein promotes granulocyte and macro-phage differentiation and proliferation and would thus in-crease inflammation. This tachyzoite-induced transcript maycontribute to the known phenomenon that tachyzoites causeextensive inflammation, whereas bradyzoites do not.

Antiapoptotic proteins BCL2A1 and TNF receptor-associ-ated factor 1 (TRAF1) are induced by tachyzoites in dendriticcells (10), and these data suggest that they are induced infibroblasts by tachyzoites but not by bradyzoites (Fig. 4 and 5).During the extensive inflammation caused by tachyzoites, theseantiapoptotic genes, as well as IL-24, may help the tachyzoite-infected cell, and thus the tachyzoites inside, survive.

There were 40 genes that were downregulated significantlyin tachyzoite infection, including the “signal transducers andactivators of transcription” STAT1 and STAT3. These proteinshave been well studied in Toxoplasma infection and shown tobe subject to multiple levels of regulation (27, 30, 41). Weexamined the transcript levels for genes whose expression ismediated by activated STAT1 or STAT3 and saw no significantdecrease in infected cells. Consistent with this, we examinedSTAT1 protein levels by Western blot and saw little if anydecrease in the infected cultures (data not shown). Hence, thedecrease in the STAT1/3 transcript levels does not appear toimpact the overall function of the proteins they encode, at leastwithin the time frame (48 h) examined in these experiments.

DISCUSSION

Under the conditions used here, infection by bradyzoitesresults in fewer changes in host gene expression than doesinfection with tachyzoites, and the changes that do occurare essentially a dampened version of those observed withtachyzoites. In addition to the fundamental differences be-tween how bradyzoites and tachyzoites manipulate their hostcell, it is possible that the lesser effects of bradyzoites is aconsequence of the way in which bradyzoites must be gener-ated in vitro. That is, the infections have to be initiated bytachyzoites and the “bradyzoite-inducing” conditions are notapplied until 4 hpi; thus, some host responses might be initi-ated by proteins on the tachyzoite surface and/or tachyzoiteproteins that are injected into the host cell during the invasionprocess itself (19, 21, 30). For example, CCL2 has been re-ported to be induced by a tachyzoite-specific surface antigen,SAG1 (4), and an increase in CCL2 transcripts was clearlyevident in tachyzoite-infected cells and, to a lesser extent,those infected with bradyzoites; both could be a result of theinitial contact with SAG1 on the tachyzoite surface with adiminished effect in the bradyzoite conditions where the SAG1gene is turned off. The large, tachyzoite-specific induction ofgenes such as BCL2A1 and GM-CSF, on the other hand, showsthat not all transcripts induced by tachyzoites are observed inbradyzoite infection. Such instances could be due to specificeffects of the tachyzoites exerted after the initial 4 h of infection,

FIG. 3. Secretion of CCL2 and CXCL1 by uninfected and infectedcells at 44 hpi postinfection. ELISA data representative of four inde-pendent experiments are shown. Tachyzoites were added, and themedium was replaced 4 hpi with normal or stress medium. At 2 hbefore harvest, the medium was removed, cells washed once withphosphate-buffered saline, and fresh medium of the proper type butwithout serum was added. Secretion from the last 2 h before harvest at44 hpi was measured. Error bars reflect the standard deviation of tworeplicate cell culture wells in the experiment. Microarray transcriptlevels are the ratio of sample to reference median intensity averagedover two spots and all five arrays per condition. Error bars reflect thestandard deviation of these data.

VOL. 75, 2007 HOST RESPONSE TO INFECTION BY TOXOPLASMA BRADYZOITES 639

on March 17, 2021 by guest

http://iai.asm.org/

Dow

nloaded from

and/or they could require sustained exposure to tachyzoite-spe-cific molecules.

Two additional factors could contribute to the diminishedmagnitude of host transcript changes in bradyzoite-infectedcells. First, due to the slower growth rate of bradyzoites, thetachyzoite-infected cells contain more parasites at a given time

point, and this higher parasite number could cause larger tran-script changes. Second, the stress conditions used to inducebradyzoite development might have dampened the ability ofthe host cells to respond to the infection.

To determine whether the parasite load within a cell corre-lates to the magnitude of the transcript changes, a slow-grow-

FIG. 4. Clusters of genes that are tachyzoite specific or bradyzoite specific. Clusters include genes called significant by SAM in either tachyzoiteinfection or bradyzoite infection that, upon examination (see the text), met the criteria for being “tachyzoite specific” (A and B) or “bradyzoitespecific” (C). Color-coding is as described for Fig. 2. Genes with asterisks are discussed in detail in the text.

640 FOUTS AND BOOTHROYD INFECT. IMMUN.

on March 17, 2021 by guest

http://iai.asm.org/

Dow

nloaded from

ing tachyzoite mutant ideally would be compared to brady-zoites. The closest to such mutants are those defective incarbamoyl phosphate synthesis, but their defect is so large(complete cessation of growth) that they are not useful forthese purposes (16). Instead, we used ELISA to examine thesecretion of CCL2 and CXCL1 at two different multiplicities ofinfection. Doubling the number of parasites resulted in an�2-fold increase in secretion (data not shown). These resultsare consistent with the larger magnitude of changes seen intachyzoite-infected cells being due in part to the greater num-ber of parasites within the culture, although increasing thepercentage of cells infected is clearly very different from in-creasing the number of parasites per cell.

To determine whether stress conditions suppress the abilityof an infected cell to respond, the secretion of CXCL1 andCCL2 was examined by ELISA using cells that were incubatedfor 44 h with culture supernatant from a tachyzoite-lysed flask.The supernatant stimulated a marked secretion of CXCL1 andCCL2 by HFFs in both standard and stressed conditions, al-though the increases in stressed cells were 2- and 1.5-fold less,respectively, than similarly treated, unstressed cells (data notshown). These results show that stress medium can have asuppressive effect on chemokine secretion by stimulatedcells, but this is not enough to account for the 17- to 20-folddepression in secretion when BCL2A1, granulocyte-macro-phage colony-stimulating factor, and IL-24 in tachyzoite-andbradyzoite-infected cells are compared.

In conclusion, the results presented here show that the ef-fects of maturing bradyzoites on host cells are generally verysimilar to those of tachyzoites. Although we sought to identifybradyzoite-specific changes in host transcript levels, onlytachyzoite-specific changes were identified. The overall lessereffect of bradyzoites on the host cell supports the hypothesisthat infection with bradyzoites is less disruptive than infectionwith tachyzoites. Additional techniques and in vivo work isneeded to study the effects of mature bradyzoites on the hostcell, although the absence of methods to specifically isolatebradyzoite-infected cells from an infected animal, coupled withthe rarity of such cells, makes this a daunting challenge.

ACKNOWLEDGMENTS

This study was supported by grants from the NIH (AI41014) toJ.C.B. and a Stanford Graduate Fellowship and Cell and MolecularTraining Grant fellowship (GM07276) to A.E.F.

We gratefully acknowledge the help of Jeroen Saeij, Michael Cleary,and other members of the Boothroyd lab, as well as Kathleen Rubins,Tammy Doukas, and Lucy Thompson for helpful suggestions andadvice. We also acknowledge the Steinman lab for help with qPCR.

REFERENCES

1. Ajioka, J. W. 1998. Toxoplasma gondii: ESTs and gene discovery. Int. J.Parasitol. 28:1025–1031.

2. Ball, C. A., I. A. Awad, J. Demeter, J. Gollub, J. M. Herbert, T. Hernandez-Boussard, H. Jin, J. C. Matese, M. Nitzberg, F. Wymore, Z. K. Zachariah,P. O. Brown, and G. Sherlock. 2005. The Stanford microarray databaseaccommodates additional microarray platforms and data formats. NucleicAcids Res. 33:D580–D582.

FIG. 5. “Tachyzoite-specific” genes show large induction, but “bradyzoite-specific” genes may be induced by infection by both stages of theparasite. Semiquantitative PCR was performed on three “tachyzoite-specific” and three “bradyzoite-specific” genes. Error bars represent thestandard deviations of two samples. The array data are presented in graphs, with error bars representing the standard deviations of all five arraysfor each condition. GM-CSF, granulocyte-macrophage colony-stimulating factor.

VOL. 75, 2007 HOST RESPONSE TO INFECTION BY TOXOPLASMA BRADYZOITES 641

on March 17, 2021 by guest

http://iai.asm.org/

Dow

nloaded from

3. Blader, I. J., I. D. Manger, and J. C. Boothroyd. 2001. Microarray analysisreveals previously unknown changes in Toxoplasma gondii-infected humancells. J. Biol. Chem. 276:24223–24231.

4. Brenier-Pinchart, M.-P., I. Villena, C. Mercier, F. Durand, J. Simon, M.-F.Cesbron-Delauw, and H. Pelloux. 2006. The Toxoplasma surface proteinSAG1 triggers efficient in vitro secretion of chemokine ligand (ccl2) fromhuman fibroblasts. Microbes Infect. 8:254–261.

5. Buckley, C. D., D. Pilling, J. M. Lord, A. N. Akbar, D. Scheel-Toellner, andM. Salmon. 2001. Fibroblasts regulate the switch from acute resolving tochronic persistent inflammation. Trends Immunol. 22:199–204.

6. Bumke, M. A., D. Neri, and G. Elia. 2003. Modulation of gene expression byextracellular pH variations in human fibroblasts: a transcriptomic and pro-teomic study. Proteomics 3:675–688.

7. Carmen, J. C., L. Hardia, and A. P. Sinai. 2006. Toxoplasma gondii inhibitsultraviolet light-induced apoptosis through multiple interactions with themitochondrion-dependent programmed cell death pathway. Cell. Microbiol.8:301–315.

8. Chang, H. Y., J.-T. Chi, S. Dudoit, C. Bondre, M. van de Rijn, D. Botstein,and P. O. Brown. 2002. Diversity, topographic differentiation, and positionalmemory in human fibroblasts. Proc. Natl. Acad. Sci. USA 99:12877–12882.

9. Channon, J. Y., K. A. Miselis, L. A. Minns, C. Dutta, and L. H. Kasper. 2002.Toxoplasma gondii induces granulocyte colony-stimulating factor and gran-ulocyte-macrophage colony-stimulating factor secretion by human fibro-blasts: implications for neutrophil apoptosis. Infect. Immun. 70:6048–6057.

10. Chaussabel, D., R. T. Semnani, M. A. McDowell, D. Sacks, A. Sherr, andT. B. Nutman. 2003. Unique gene expression profiles of human macrophagesand dendritic cells to phylogenetically distinct parasites. Blood 102:672–681.

11. Cleary, M. D., U. Singh, I. J. Blader, J. L. Brewer, and J. C. Boothroyd. 2002.Toxoplasma gondii asexual development: identification of developmentallyregulated genes and distinct patterns of gene expression. Eukaryot. Cell1:329–340.

12. Draghici, S., P. Khatri, R. P. Martins, G. C. Ostermeier, and S. A. Krawetz.2003. Global functional profiling of gene expression. Genomics 81:98–104.

13. Dzierszinski, F., M. Nishi, L. Ouko, and D. S. Roos. 2004. Dynamics ofToxoplasma gondii differentiation. Eukaryot. Cell 3:992–1003.

14. Eisen, M. B., and P. O. Brown. 1999. DNA arrays for analysis of geneexpression. Methods Enzymol. 303:179–205.

15. Ferguson, D. J. 2004. Use of molecular and ultrastructural markers to eval-uate stage conversion of Toxoplasma gondii in both the intermediate anddefinitive host. Int. J. Parasitol. 34:347–360.

16. Fox, B. A., and D. J. Bzik. 2002. De novo pyrimidine biosynthesis is requiredfor virulence of Toxoplasma gondii. Nature 415:926–929.

17. Gail, M., U. Gross, and W. Bohne. 2001. Transcriptional profile of Toxo-plasma gondii-infected human fibroblasts as revealed by gene-array hybrid-ization. Mol. Genet. Genomics 265:905–912.

18. Gail, M., U. Gross, and W. Bohne. 2004. Transferrin receptor induction inToxoplasma gondii-infected HFF is associated with increased iron-responsiveprotein 1 activity and is mediated by secreted factors. Parasitol. Res. 94:233–239.

19. Gilbert, L. A., S. Ravindran, J. C. Boothroyd, and P. J. Bradley. Toxoplasmagondii targets a protein phosphatase 2C-like protein to the nucleus of in-fected cells. Eukaryot. Cell, in press.

20. Goebel, S., U. Gross, and C. G. Luder. 2001. Inhibition of host cell apoptosisby Toxoplasma gondii is accompanied by reduced activation of the caspasecascade and alterations of poly (ADP-ribose) polymerase expression. J. CellSci. 114(Pt. 19):3495–3505.

21. Hakansson, S., A. J. Charron, and D. Sibley. 2001. Toxoplasma evacuoles: atwo-step process of secretion and fusion forms the parasitophorous vacuole.EMBO. 20:3132–3144.

22. Hogaboam, C. M., M. L. Steinhauser, S. W. Chensue, and S. L. Kunkel.1998. Novel roles for chemokines and fibroblasts in interstitial fibrosis. Kid-ney Int. 54:2152–2159.

23. Kim, L., B. A. Butcher, and E. Y. Denkers. 2004. Toxoplasma gondii inter-

feres with lipopolysaccharide-induced mitogen-activated protein kinase ac-tivation by mechanisms distinct from endotoxin tolerance. J. Immunol. 172:3003–3010.

24. Kim, S. K., and J. C. Boothroyd. 2005. Stage-specific expression of surfaceantigens by Toxoplasma gondii as a mechanism to facilitate parasite persis-tence. J. Immunol. 174:8038–8048.

25. Knight, B. C., C. L. Brunton, N. C. Modi, G. R. Wallace, and M. R. Stanford.2005. The effect of Toxoplasma gondii infection on expression of chemokinesby rat retinal vascular endothelial cells. J. Neuroimmunol. 160:41–47.

26. Lee, C. W., S. Bennouna, and E. Y. Denkers. 2006. Screening for Toxoplasmagondii-regulated transcriptional responses in lipopolysaccharide-activatedmacrophages. Infect. Immun. 74:1916–1923.

27. Luder, C. G., W. Walter, B. Beuerle, M. J. Maeurer, and U. Gross. 2001.Toxoplasma gondii down-regulates MHC class II gene expression and anti-gen presentation by murine macrophages via interference with nuclear trans-location of STAT1 alpha. Eur. J. Immunol. 31:1475–1484.

28. Molestina, R. E., and A. P. Sinai. 2005. Host and parasite-derived IKKactivities direct distinct temporal phases of NF-kB activation and target geneexpression following Toxoplasma gondii infection. J. Cell Sci. 118(Pt. 24):5785–5796.

29. Nash, P. B., M. B. Purner, R. P. Leon, P. Clarke, R. C. Duke, and T. J.Curiel. 1998. Toxoplasma gondii-infected cells are resistant to multiple in-ducers of apoptosis. J. Immunol. 160:1824–1830.

30. Saeij, J. P. J., S. Coller, J. P. Boyle, M. Jerome, M. W. White, and J. C.Boothroyd. Toxoplasma co-opts host gene expression by injection of a poly-morphic kinase homologue. Nature, in press.

31. Schwarz, J. A., A. E. Fouts, C. A. Cummings, D. J. Ferguson, and J. C.Boothroyd. 2005. A novel rhoptry protein in Toxoplasma bradyzoites andmerozoites. Mol. Biochem. Parasitol. 114:159–166.

32. Shapira, S., O. S. Harb, J. Margarit, M. Matrajt, J. Han, A. Hoffmann, B.Freedman, M. J. May, D. S. Roos, and C. A. Hunter. 2005. Initiation andtermination of NF-�B signaling by the intracellular protozoan parasite Toxo-plasma gondii. J. Cell Sci. 118(Pt. 15):p. 3501–3508.

33. Sinai, A. P., T. M. Payne, J. C. Carmen, L. Hardia, S. J. Watson, and R. E.Molestina. 2004. Mechanisms underlying the manipulation of host apoptoticpathways by Toxoplasma gondii. Int. J. Parasitol. 34:381–391.

34. Singh, U., J. L. Brewer, and J. C. Boothroyd. 2002. Genetic analysis oftachyzoite to bradyzoite differentiation mutants in Toxoplasma gondii revealsa hierarchy of gene induction. Mol. Microbiol. 44:721–733.

35. Spear, W., D. Chan, I. Coppens, R. S. Johnson, A. Giaccia, and I. J. Blader.2006. The host cell transcription factor hypoxia-inducible factor 1 is requiredfor Toxoplasma gondii growth and survival at physiological oxygen levels.Cell. Microbiol. 8:339–352.

36. Tomavo, S. 2001. The differential expression of multiple isoenzyme formsduring stage conversion of Toxoplasma gondii: an adaptive developmentalstrategy. Int. J. Parasitol. 31:1023–1031.

37. Tusher, V. G., R. Tibshirani, and G. Chu. 2001. Significance analysis ofmicroarrays applied to the ionizing radiation response. Proc. Natl. Acad. Sci.USA 98:5116–5121.

38. Vaday, G. G., and O. Lider. 2000. Extracellular matrix moieties, cytokines,and enzymes: dynamic effects on immune cell behavior and inflammation.J. Leukoc. Biol. 67:149–159.

39. Yang, Y. H., S. Dudoit, P. Luu, D. M. Lin, V. Peng, J. Ngai, and T. P. Speed.2002. Normalization for cDNA microarray data: a robust composite methodaddressing single and multiple slide systemic variation. Nucleic Acids Res.30:e15.

40. Yuen, T., E. Wurmback, R. L. Pfeffer, B. J. Ebersole, and S. C. Sealfon. 2002.Accuracy and calibration of commercial oligonucleotide and custom cDNAmicroarrays. Nuc. Acids Res. 30:e48.

41. Zimmermann, S., P. J. Murray, K. Heeg, and A. H. Dalpke. 2006. Toxo-plasma gondii contributes to immune evasion in macrophages by blockingIFN- signaling. J. Immunol. 176:1840–1847.

Editor: W. A. Petri, Jr.

642 FOUTS AND BOOTHROYD INFECT. IMMUN.

on March 17, 2021 by guest

http://iai.asm.org/

Dow

nloaded from

INFECTION AND IMMUNITY, June 2007, p. 3209 Vol. 75, No. 60019-9567/07/$08.00�0 doi:10.1128/IAI.00462-07

AUTHOR’S CORRECTION

Infection with Toxoplasma gondii Bradyzoites Has a Diminished Impact on HostTranscript Levels Relative to Tachyzoite Infection

A. E. Fouts and J. C. BoothroydDepartment of Microbiology and Immunology, Stanford University School of Medicine, Stanford, California 94305-5124

Volume 75, no. 2, p. 634–642, 2007. Supplemental material: In supplemental file 1, “Supplementary Figure 1. List and log2 foldchanges of genes . . . ” should read “Supplementary Figure 1. List and linear fold changes of genes . . . ” and in the boxheads ofcolumns 7, 9, and 11 “ . . . Log2 Fold Changes” should read “ . . . Fold Changes”. Revised supplemental material is posted athttp://iai.asm.org/cgi/content/full/75/2/634/DC1.

3209