Supporting InformationLo et al. 10.1073/pnas.1006901107SI Materials and MethodsPreparation of Blood Samples. For each patient, three tubes (∼5mL/tube of sodium citrate tubes or EDTA tubes) of blood weresent with an ice-pack using overnight shipment to Lo’s laboratoryat the Armed Forces Institute of Pathology, Walter Reed ArmyMedical Center, Washington, DC. The samples of plasma andperipheral blood mononuclear cells (PBMCs) were preparedimmediately from the whole blood received. Plasma sampleswere prepared from whole blood by centrifugation at ∼900 × gfor 15 min at 4–6 °C, and PBMCs were isolated from whole bloodusing standard gradient centrifugation protocol for purificationon a Histopaque 1077 (Sigma). Aliquots of chronic fatigue syn-drome (CFS) patients’ cell-free plasma, PBMC and whole bloodsamples were prepared and stored at −80 °C. The PBMC sam-ples similarly prepared from 44 healthy blood donors were col-lected under National Institutes of Health (NIH) exempt statusin the Department of Transfusion Medicine, NIH, from 2003 to2006. The cells were stored at −190 °C until they were revived forDNA isolation and processed in parallel with the controls.

DNA Isolation. The frozen whole-blood samples and PBMCs keptat−80 °C were used for DNA preparation in the study. DNAfrom PBMCs (∼107 cells) was extracted using a DNeasy bloodand tissue kit (QIAGEN) or the conventional phenol/chloroformextraction method. To prepare DNA from whole-blood samples,200 μL of samples were used and DNA was extracted using theDNeasy blood and tissue kit. The extracted DNA was dissolvedin TE buffer and stored at −80 °C in small aliquots. DNAsamples were quantified using a NanoDrop 2000 spectropho-tometer (Thermo Scientific).

RNA Isolation and RT–PCR. To prepare RNA from the frozenplasma sample, 0.5 mL of the frozen plasma was thawed andsubjected to ultracentrifugation at 195,500 × g on an OptimaMax-XP ultracentrifuge (Beckman Coulter) for 4 h at 4 °C. Thepellet was dissolved in 250 μL of PBS, and RNA was extractedusing TRIzol reagent (Invitrogen) according to the manu-facturer’s protocol. RNA pellet was dissolved in 10 μL of dieth-ylpyrocarbonate-treated water and immediately used for cDNAsynthesis with SuperScript II RT (Invitrogen) in a total reactionvolume of 20 μL. To perform nested PCR, 5 μL of the preparedcDNA was used for the first round of PCR. The PCR productswere then processed according to the protocol for nested PCRspecific for the xenotropic murine leukemia virus-related virus(XMRV)/murine leukemia virus (MLV) gag gene sequences asdescribed below.

XMRV/MLV gag Nested PCR. To prevent potential laboratory DNAcontamination, all PCR studies were performed using completelyseparate reagents in each experiment and set up inside a PCRhood in a separate room free of any amplicons and plasmid DNA.Negative controls were also included in every experiment. Thefollowing primers were used: 419F (5′-ATCAGTTAACCTAC-CCGAGTCGGAC-3′) and 1154R (5′-GCCGCCTCTTCTTC-ATTGTTCTC-3′) in the first round of PCR; GAG-I-F (5′-TCTCGAGATCATGGGACAGA-3′) and GAG-I-R primer (5′-AGAGGGTAAGGGCAGGGTAA-3′) or NP116 (5′-CATGG-GACAGACCGTAACTACC-3′) and NP117 (5′-GCAGATCG-GGACGGAGGTTG-3′) in the second round of the nested PCR.For the first round PCR, the amplification reaction was per-

formed as follows: total reaction volume, 20 μL; template con-centration, 30–50 ng of total cellular DNA/reaction, 1× Taq

buffer, 2.5 mM MgCl2, 0.2 mM dNTP, 0.25 pmol/μL of 419Fprimer, 0.25 pmol/μL of 1154R primer, and 0.5 units of PlatinumTaq polymerase (Invitrogen). The cycles were 4 min at 94 °C (1min at 94 °C, 1 min at 57 °C, 1 min at 72 °C) × 40 cycles and 10min at 72 °C.For the second round PCR, the amplification reaction was

performed as follows: total reaction volume, 20 μL; 2 μL of round 1PCR product, [1× Taq buffer, 2.5 mMMgCl2, 0.2 mM dNTP, 0.25pmol/μL of NP116 primer or GAG-I-F primer, 0.25 pmol/μL ofNP117 primer or GAG-I-R primer, and 0.5 units of Platinum Taqpolymerase (Invitrogen)]. The cycles were 4 min at 94 °C (1 min94 °C, 1 min 57 °C, 1 min 72 °C) × 45 cycles and 10 min 72 °C.To verify the PCR product, 5 μL of sample from each PCR

round was separated on 2% agarose gel. The first-round PCR isintended to amplify a target fragment of ∼730 bp from the gaggene and the second-round PCR to result in a product of 413 bpusing the GAG-I-F and GAG-I-R primer set or a product of 380bp using the NP116 and NP117 primer set. Both variants of thePCR products from the second-round PCR were used to screenall of the DNA and cDNA samples. A plasmid (pNP101-730)with a cloned DNA fragment of ∼730 nt of a gag gene fragmentof endogenous MLV amplified from mouse DNA was used asa positive control in all PCR reactions. PCR products from thefirst and second rounds of nested PCR specific for the XMRV/MLV gag region that were close to the expected sizes were re-trieved and purified from the gel using QIAEX II gel extractionkit (QIAGEN) and sequenced by conventional Sanger DNAsequencing in the Food and Drug Administration core facility(Bethesda, MD).

PCR Sensitivity and Specificity Testing. To evaluate the sensitivity ofPCR testing, a DNA fragment of ∼730 nt of PCR product wasamplified from a mouse genomic DNA sample using 419F/1154Rprimer set and cloned into pGEM-T-Easy vector (Promega) byTA cloning according to themanufacturer’s protocol to obtain thepNP101-730 plasmid. The sequence of the target mouse retrovi-rus gag gene cloned into the plasmid was confirmed by DNA se-quencing. The resulting pNP101-730 plasmid was used for thePCR sensitivity testing. Serial 10-fold plasmid dilutions withknown plasmid copy number were spiked into 30 ng of humanDNA. The detection threshold for the first-round PCR using theprimer set 419F/1154R targeting the gag gene and producinga 730-bp product was 100 copies in the background of 30 ng hu-man DNA per reaction. The detection threshold for the nestedPCR using either the primer set GAG-I-F/GAG-I-R (intended tobe more XMRV specific) or the primer set NP116/NP117 (highlyconserved sequences for XMRV and MLV) in the second stagewas five gene copies per reaction.DNA polymerase enzymes from different sources including Ap-

plied Biosystems (AmpliTaq-gold), USB (HotStart-IT FideliTaq),and Qiagen (QIA HotStartTaq Plus) were carefully assessed andcompared for sensitivity and specificity using the plasmid-clonedMLV gag gene of different dilutions spiked into negative blooddonor DNA. Platinum Taq polymerase from Invitrogen gave us thebest results. None of more than 300 negative controls run in parallelin multiple assays as well as in the routine enzyme quality controlstudies has ever produced a positive gag gene amplicon with any ofthe three gag gene-specific primer sets.

Phylogenetic Analysis. The viral gag gene sequences of mouseendogenous retrovirus (mERV) chromosome 8 (AC163617; nt85328–86073), mERV chromosome 11 (AL731805; nt 59192–

Lo et al. www.pnas.org/cgi/content/short/1006901107 1 of 7

59937), mERV chromosome 7 (AC127565; nt 71563–72307),mERV chromosome 3 (AC161425; nt 200890–201634), mERVchromosome 10 (AC119828; nt 58801–59580), mERV chromo-some 5 (AC114666; nt 101281–102120), polytropic mERV-clone5 (FJ544576; nt 971–1717), and polytropic MLV (AF490352, nt1022–1768) were selected from the National Center for Bio-technology Information (NCBI) nonredundant database byBLAST querying with the gag gene sequences obtained from thepositive PCR products (746 nt after the first round of nestedPCR) as the most closely related sequences. Sequences of thecorresponding 730-nt region of the gag genes from XMRV-VP35(DQ241301; nt 424–1154), XMRV-VP62 (DQ399707; nt 424–1154), XMRV-VP42 (DQ241302; nt 424–1154), XMRV-WPI-1178 (GQ497343; nt 384–1114), and XMRV-WPI-1106(GQ497344; nt 384–1114) were obtained from GenBank and alsoincluded for the sequence alignment and phylogenetic analysis inFig. S1 and Fig. 3A. In addition, the gag gene sequences of poly-tropic mERV clone 51 (FJ544578; nt 639–948), murine AIDS-related virus (S80082V; 1060–1390), and mERV chromosome 9(AC121813; nt 45132–44802) were similarly chosen from theNCBI nonredundant database by BLAST querying with the se-quence of PCR product obtained from BD-28 patient’s DNA(∼380 nt after the second round of nested PCR) as themost closelyrelated sequences for analysis (shown in Fig. S2 and Fig. 3B).To classifymERVMLVsequences obtained fromGenBank, we

used Vector NTI software (www.invitrogen.com/bioinformatics)based on the predicted reactivity with the specific probes for theenv region: JS-4, JS-5, and JS-6 from an early publication (1). Wealso used BLAT search (http://genome.ucsc.edu) to refer mERVelements according to the standard nomenclature described ina more recent publication (2). Such classification of MLVs is re-flected in Figs. 3 and 4 and in Figs. S1 and S2 as polytropic (Pmv),modified polytropic (Mpmv), or xenotropic (Xmv).For the gag protein partial sequence alignment in Fig. S3, we

included themost relevantMLVs such as exogenous ecotropicMLV:Cas-Br-E (X57540), Friend (Z11128), Moloney (J02255),Rauscher (U94692) RadLV (K03363), SL3-3, tumor derived AKR(AF169256), Graffi GV-1.2 (AB187565), SRS-19.6 (AF019230);amphotropic: 1313 A, wild mouse, California (AF411814); xeno-tropic: XMRVs (DQ241301, DQ399707, DQ241302, GQ497343,GQ497344); endogenous ecotropic: Akv, AKR mice (J01998),HEMV, Mus spicelegus (AY818896), PSR3 (M87550); polytropic:MCF1233, isolated from C57BL mice containing Akv-type viruses(U13766); endogenous nonecotropic: (Mpmv1) mERV chromo-some 7 (AC127565; nt 71563–72307), (Mpmv8) mERV chromo-some 11 (AL731805; nt 59192–59937), and (Mpmv11) mERVchromosome 12 (AC153658; nt 93404–92660) were selected fromthe NCBI database. The protein sequences were generated frommERVs, CFS types 1, 2, and 3 and BD-22 nucleotide sequences bytranslation starting from the AUG codon using Vector NTI soft-ware (www.invitrogen.com/bioinformatics) and were included inthe alignment for phylogenetic analysis.

Amplification of MLV-Like Virus env Gene Sequences. PBMC DNA(35 ng/each reaction) originating from the 37 CFS patients and 44healthy blood donors was also tested by PCR targeting the MLV-related viral env gene. A semi-nested PCR was used to amplifythe MLV-related viral env gene by producing a target ampliconof ∼252 bp. The following primers were used: first round—5922F(5′-GCTAATGCTACCTCCCTCCTGG-3′) and 6273R (5′-GG-

AGCCCACTGAGGAATCAAAACAGG-3′) (3); second round—5922F and 6173R (5′-CTGTCCAGTGGTCTCACATC-3′)(in-house design). A nested PCR was used to amplify the MLV-related viral env gene by producing a target amplicon of ∼218 bp,using primers 5922F and 6273R in the first round of amplifica-tion and primers 5942F (5′-GGGGACGATGACAGACAC-TTTCC-3′) and 6159R (5′-CACATCCCCATTTGCCACAGT-AG-3′) (4) in the second round of amplification. The nested andsemi-nested PCR conditions for the amplification of env genesequences were the same as those described for the amplificationof gag gene sequences. The PCR products with their sizes closeto the target amplicons were sequenced for confirmation.In the sequence alignment and the construction of the phylo-

genetic tree, the following sequences were selected from theNCBInonredundantdatabase closehomologybyBLASTqueryingwith the env gene sequences obtained from a CFS patient (206 nt)and a blood donor (240 nt): Friend spleen focus-forming virusisolate derived from mouse erythroleukemia cells (MEL179)(FJ556973), Friend mink cell focus-forming virus (DQ199949),mERV-chromosome 7-Mpmv1, mERV-chromosome 3-Mpmv10,mERV-chromosome 5-Pmv11, mERV-clone 51(P), and XMRVisolates VP-62, VP-42, and WPI-1106 (see the accession numbersfrom previous gag gene alignments). Alignment of env partialsequences was performed using Clustal W2 and the same pa-rameters as described for the gag gene sequence alignment. Pri-mers were excluded from the sequences in the phylogeneticanalysis.

Mouse Mitochondrial DNA Assay. The complete mitochondrialDNA (mtDNA) sequences of humans andmice were downloadedfromGenBank and aligned using Clustal W2. Sequence alignmentrevealed that 439 bp of the 3′ end of mouse mtDNA (beyond15,862 bp, according to the coordinates of the BALB/c mouse;accession no. AJ512208) was not present in human mtDNA.Primer sets were designed for a semi-nested mouse-specificmtDNA PCR based on the sequence in this region of mousemtDNA using Primer-Blast from NCBI. The external PCR pri-mers were designated as mt15982F (5′-AGACGCACCTAC-GGTGAAGA-3′) and mt16267R (5′-AGAGTTTTGGTTCAC-GGAACATGA-3′), which produce a predicted amplicon of 286bp. The internal primers of the semi-nested PCR were desig-nated as mt16115F (5′-TGCCAAACCCCAAAAACACT-3′) andmt16267R, which would produce a predicted amplicon of 153 bpfrom mouse mtDNA. PCR system and setup were the same as forthe gag gene nested PCR study.Comparison of the sensitivity in amplifying mouse DNA by the

semi-nested PCR targeting mouse-specific mtDNA and by thenested PCR targeting the MLV-like virus gag gene. Serial dilu-tions (from 40 pg to 2.5 fg) of mouse spleen DNA were spikedinto 35 ng of total human PBMC DNA and compared in parallelfor the mouse DNA detection sensitivity of the two PCR assays.All of the CFS patients’ and healthy blood donors’ samples thatwere positive for MLV-like gag gene sequences were tested forthe presence of mouse DNA contamination using the semi-nested PCR assay targeting mouse-specific mtDNA. Serial di-lutions (from 50 fg to 1 fg) of mouse DNA were spiked into 35 ngof human DNA and were used as the controls to verify the assaysensitivity in the detection of mouse DNA. Multiple reactionswith 35 ng of human DNA without spiking with any mouse DNAwere included and tested in parallel as the negative controls.

1. Coffin JM, Stoye JP, Frankel WN (1989) Genetics of endogenous murine leukemiaviruses. Ann NY Acad Sci 567:39–49.

2. Jern P, Stoye JP, Coffin JM (2007) Role of APOBEC3 in genetic diversity amongendogenous murine leukemia viruses. PLoS Genet 3:2014–2022.

3. Lombardi VC, et al. (2009) Detection of an infectious retrovirus, XMRV, in blood cells ofpatients with chronic fatigue syndrome. Science 326:585–589.

4. Hong S, et al. (2009) Fibrils of prostatic acid phosphatase fragments boost infectionswith XMRV (xenotropic murine leukemia virus-related virus), a human retrovirusassociated with prostate cancer. J Virol 83:6995–7003.

Lo et al. www.pnas.org/cgi/content/short/1006901107 2 of 7

Fig. S1. (Continued)

Lo et al. www.pnas.org/cgi/content/short/1006901107 3 of 7

Fig. S1. Multiple sequence alignment (MSA) of 746 nt of the MLV-related virus gag gene nucleotide sequences amplified from 21 CFS patient samples and oneblood donor (BD22) using primer set 419F/1154R, which includes closely related MLVs and XMRVs. There are three different sequences (CFS types 1–3)identified among the 21 CFS patient samples that were PCR positive for the MLV-like virus gag gene. The gag gene sequence found in BD22 is also differentfrom those of CFS types 1–3 and XMRVs. The alignment was conducted with ClustalW2. The GenBank accession numbers for all of the sequences aligned aredescribed in SI Materials and Methods. The names of the corresponding MLV genetic elements present in the mouse genome according to a standard no-menclature (2) are listed in parentheses. The regions of all of the primer sets used in the nested PCR study are marked by arrows. A unique 15-nt deletion in a 5′gag leader sequence found in all XMRVs examined is indicated by dashes. Dots denote identical nucleotides. (P), polytropic.

Lo et al. www.pnas.org/cgi/content/short/1006901107 4 of 7

Fig. S2. MSA of 380-nt MLV-related virus gag gene nucleotide sequences amplified from blood samples of 21 CFS patients and 3 normal blood donors (BD22,BD26, and BD28) using the primer set NP116/NP117, which includes closely related MLVs and XMRVs. Origins of the sequences used for the alignment aredescribed in Fig. 3 and inMaterials and Methods. The names of the corresponding MLV genetic elements present in the mouse genome according to a standardnomenclature (2) are listed in parentheses. The regions of the primers used in the nested PCR are marked by arrows. A specific 21-nt deletion of the gag genenucleotides found in PBMC DNA of BD28 and in polytropic mERV-clone 51 is indicated by dashes. Dots denote identical nucleotides. (P), polytropic.

Lo et al. www.pnas.org/cgi/content/short/1006901107 5 of 7

Fig. S3. Protein sequence alignment of CFS types 1, 2, and 3 and BD-22 and MLVs gag protein sequences. Gag protein sequences starting from the AUGinitiation codon are aligned with those of relevant endogenous as well as exogenous MLVs. Sequences were translated using Vector NTI and aligned withClustalW2. The Gag protein sequences of other MLVs were selected from the NCBI database (SI Materials and Methods for the accession numbers) and used forprotein sequence alignment and phylogenetic analysis. Sequences of MLVs are referred to as polytropic (P), ecotropic (E), amphotropic (A), or modifiedpolytropic (Mpmv). The deletions are indicated by dashes. Dots denote identical amino acids.

Lo et al. www.pnas.org/cgi/content/short/1006901107 6 of 7

A. B.5922F

6173R

5942F

6159R

5922F

6173R

5942F

6159R

0.50.5

Fig. S4. (A) MSA of MLV-related virus env gene nucleotide sequences amplified from one CFS patient (170 nt shown in red) and one blood donor (BD-26) (240nt shown in blue) by a nested and a semi-nested PCR, respectively, as well as those of XMRVs and closely related MLVs with the highest sequence homology inthe BLAST search. The alignment was conducted with ClustalW2. The GenBank accession numbers for all of the sequences aligned are described in SI Materialsand Methods. The regions of primer sets used in the PCR study are marked by arrows. Dots denote identical nt. A deletion of 12 nt identified in env genesequences amplified from both the CFS patient and the blood donor, in comparison with those of XMRVs, is indicated by dashes. The missing sequences at bothends of the CFS-env sequence are also shown in dashes. (B) Phylogenetic tree was generated in ClustalW2 program using the neighbor-joining method basedon 170- to 240-nt env nucleotide sequences of the corresponding MSA.

Lo et al. www.pnas.org/cgi/content/short/1006901107 7 of 7