We had historical evidence that 3 mass graves excavated in France were used to bury bubonic plague victims. In Vienne, 12 skeletons, including 5 children, buried within the ruins of a Roman temple have been dated from the 7th–9th centuries both by a 5th-century coin and ¹⁴C dating. In Martigues, 205 skeletons buried in 5 trenches were dated from 1720 to 1721 on the basis of coins and detailed parish bills that listed the victims (Figure). In Marseille, 216 skeletons buried in a huge pit dated from a May 1722 epidemic relapse. We previously confirmed the diagnosis of plague at this site (3). Eighteen teeth from 5 skeletons in Vienne, 13 teeth from 5 skeletons in Martigues, and 5 teeth from 3 skeletons in Marseille were processed for the search for Y. pestis DNA in the dental pulp. The teeth were processed according to published criteria for authenticating molecular data in paleomicrobiology (10): 1) there should be no positive control; 2) negative controls, as similar as possible to the ancient specimens, should test negative; 3) a new primer sequence targeting a genome region not previously amplified in the laboratory should be used (suicide PCR); 4) any amplicon should be sequenced; 5) a second amplified and sequenced target should confirm any positive result; and 6) an original sequence that differs from modern homologs should be obtained to exclude contamination.

Accordingly, DNA samples were submitted for suicide-nested PCR conducted by using 1 negative control (18th-century teeth from skeletons of persons without anthropologic and macroscopic evidence of infection) for every 3 specimens. Two microliters (1 μL for nested PCR) DNA were amplified in a 50-μL mixture containing 10 pmol of each primer, 200 μmol/L each deoxyribonucleotide triphosphate (Invitrogen, Cergy-Pontoise, France), 1.5 U Taq polymerase (Invitrogen), and 2.5 μL of a 50-mmol/L solution of MgCl₂ in 1× Taq buffer. Nested PCR aimed to encompass the entire glpD gene incorporated primers: glpD-F1: 5′-GGC TAG CCG CCT CAA CAA AAA CAT-3′ (positions 170080–170103, reference: Y. pestis strain CO92 genome sequence AJ414159.1)/glpD-R2: 5′-GGT GCC AGT TTC AGT AAC AC-3′ (positions 170402–170383) for initial PCR and glpD-F3: 5′-CGC TGT TTC GAA CAT TCA GA-3′ (positions 170230–170249) /glpD-R3: 5′-GGC CAA GGC TTC ACT TAC CA-3′ (positions 170373–170354) for nested PCR. PCRs were performed in a T3 thermocycler (Biolabo, Archamps, France) under the following conditions: an initial 2 min of denaturation at 95°C was followed by 43 cycles (38 cycles for nested PCR) of denaturation for 30 s at 94°C, annealing for 30 s at 58°C, and extension for 90 s at 68°C. The amplification was completed by holding the reaction mixture for 7 min at 68°C. PCR products purified by using a MultiScreen PCR plate (Millipore Corp., Bedford, MA, USA) were sequenced with a DNA sequencing kit (Big Dye Terminator Cycle Sequencing V2.0; PE Biosystem, Courtaboeuf, France) and subjected to electrophoresis with the 3100 Genetic Analyzer (Applied Biosystems, Foster City, CA, USA). The sequences were compared in the GenBank database (www.ncbi.nlm.nih.gov/GenBank) using the multisequence alignment Clustal within the BISANCE environment.

No amplification was observed in 11 negative controls, but 5 of 36 teeth yielded an amplicon of 191-bp length in 2 of 4 persons’ remains from Vienne, 2 of 5 from Martigues, and 1 of 3 from Marseille. Amplicons exhibited 100% sequence similarity with that of the Y. pestis Orientalis glpD gene (GenBank accession nos. AY312359 for tooth 35–0235, Vienne; DQ073797 for tooth SQ401521 and DQ073798 for tooth SQ408113, Martigues; and AY312360 for tooth 25–0225, Marseille) and were characterized by a 93-bp deletion when compared with the glpD gene sequence of the Y. pestis Medievalis biotype (GenBank accession no. AE 013994).

In this study, contamination of the ancient specimens is unlikely because of the extensive precautions we took, including use of the suicide PCR protocol excluding positive controls (4). Accordingly, glpD gene had never been investigated in our laboratory before this study, and negative controls remained negative. The specificity of the amplicons was ensured by complete similarity of experimental sequences with that of the Y. pestis Orientalis glpD gene (8). One site (Marseille, 1722) was previously positive for Y. pestis after sequencing of 2 different targets (chromosome-borne rpob and plasmid-borne pla genes) in other specimens collected in other persons’ remains (3).

These results therefore confirm the detection of Y. pestis–specific DNA in plague patients’ remains from the first and second epidemics (3–6). We observed a 93-bp in-frame deletion within the glpD gene sequences obtained from ancient dental pulp specimens. This deletion has been found only in Orientalis biotype isolates in 2 independent studies comprising a total of 77 and 260 Y. pestis isolates, respectively, of the 4 biotypes (8,9).

After previous demonstration of Y. pestis Orientalis-type multiple spacer type sequences in Justinian and medieval specimens (5), we now have cumulative evidence using 2 different molecular approaches that Y. pestis closely related to the Orientalis biotype was responsible for the 3 historical plague pandemics.