HLA-DR Genotyping and Mitochondrial DNA Analysis Reveal the Presence of Family Burials in a Fourth Century Romano-British Christian Cemetery

In Colchester, Britain's oldest recorded town, during the Roman period there were areas which were clearly used solely as cemeteries. One of the most significant is at Butt Road, which includes a late Roman probable Christian cemetery with an associated building, apparently a church, that overlies and developed from a pagan inhumation cemetery. DNA was extracted from the long bones (femurs) of 29 individuals, mostly from a large complex of burials centered on two timber vaults. These were thought to comprise a number of family groupings, deduced from osteological analysis, stratigraphical and other considerations. The use of a modified version of the silica-based purification method recovered nanogram quantities of DNA/gram of bone. Two-stage amplification, incorporating primer-extension preamplification-polymerase chain reaction, permitted simultaneous amplification of both mitochondrial and nuclear DNA. Sequence-specific oligonucleotide probes yielded human leukocyte antigen (HLA)-DR typing of seven samples, with four revealing the infrequent HLA-DR10 genotype. Examination of the control region of mitochondrial DNA (mtDNA) by direct sequencing revealed polymorphisms yet to be reported in the modern population. HLA-DRB typing and mtDNA analysis affirmatively supported kinship among some, if not all, individuals in the “vault complex” and demonstrate a continental European origin of the individuals investigated.


42
Several Period 2 burial groups were located within a complex of graves centred around two Group C is a postulated family group comprised of G298, G299, G369, G390 and G433. This 52 group is first identified in Period 1 Phase 3. Two Group C graves (G298, G390) were placed 53 in the group by virtue of their grave goods and two more (G369, G433) owing to their 54 position relative to the first two. In addition, G295 and G320 may also belong to Group C 55 based on their deposits. However, their alignments slightly differ.

57
The three Group C graves G299, G298 and G369 lay head to foot in a row which 58 corresponded roughly to the line of CF59 projected westwards towards CF32. The two 59 sampled Group C graves, G369 (young adult female) and G298 (adult, sex uncertain), cut 60 G283 (neonate), and thus either may be related to that burial. It is also possible that G283 is 61 related to G285 (male, age uncertain) which lay adjacent to it in the south, or to G392 62 (neonate), just over 1 m to the east. The grouping G663/G667/G674 is interpreted as a multiple burial, as corroborated by clear 154 direct stratigraphical relationships, which also suggest possible family relationships. The 155 three burials were deposited within a large isolated Period 2 grave pit, HF53, in which G663 156 was the latest burial, emphasizing their association.

162
The grave of an elderly male. Isolated burial.

166
A femur excavated during the study from a Roman site at Colchester was also analyzed. The 167 bone was dug out from the ground and placed into an UV irradiated autoclave bag. Only C.P.

168
Voong had contact with the femur. Therefore the only opportunity and source of 169 contaminating the sample with contemporary human DNA would be from him. This made 170 the process of authenticating subsequent DNA sequences less complicated.

182
The silica-based extraction technique involved vigorous washing of the silica pellet, causing 183 some DNA loss. To investigate the efficiency of this method and to determine the amount of 184 DNA recovered, known amounts of DNA (0.2 -1.0 μg of 1kb ladder) were added to the DNA 185 extractions. A known quantity (10 μg) of the 1kb molecular weight ladder DNA was added to 186 2 ml of extraction buffer and incubated at 50C for 1 h in an orbital shaker. A negative 187 control, containing everything except the DNA, was performed in parallel. The tubes were 188 then centrifuged for 10 min at 12,000g. The supernatant was transferred to 1.5 ml screw on 189 cap micro-centrifuge tubes with 30 μl of silica suspension and incubated for a further 45 min 190 at room temperature on a tube rotator. PicoGreen was used to determine how much DNA was 191 recovered in each extraction. The effect of decalcification on DNA loss was also determined through a UV analysis using 205 the dye PicoGreen. Sample G375 was selected, as this sample was well preserved and 206 therefore more likely to contain DNA. The sample was decalcified and extracted as above, 207 the EDTA supernatant was retained and the DNA eluted once with 150 μl Purite water. A 100 208 μl aliquot of the DNA extract, EDTA supernatant and EDTA was added to a PicoGreen 209 working solution and the fluorescence taken between 500-600 nm.

211
The agarose gel revealed smearing that was concentrated in the high molecular weight range.

212
The smearing was more evident in non-decalcified samples than in the decalcified samples.

213
Thus omitting the decalcification step does appear to make some improvement. However, the observed was not blue/green, but white. Although the agarose gels indicate that 220 decalcification does make a difference, it is only slight.

222
The UV absorbance results for decalcified and non-decalcified samples were very similar.

223
The purity of the DNA extracts, judging from the 260/280 nm values, again indicated that the 224 extracts were not pure and contained microbial contaminants. Therefore, this was not an 225 accurate estimation for aDNA concentration.

269
The purpose of using P1 nuclease was to determine whether the high fluorescence observed 270 in the EDTA supernatant was due to DNA. However, since EDTA inhibits the activity of the To demonstrate that it was DNA that caused the fluorescence in the extracts, the P1 nuclease 279 was added to an aliquot of the non-decalcified sample. After incubation at 37C for 10 min, 280 the extract was mixed with PicoGreen and the fluorescence taken.

282
The mass spectrum showed that peak fluorescence intensity was lower for the decalcified 283 sample (110 nm) than for the non-decalcified sample (165 nm). This indicates DNA is lost 284 through decalcification. The spectra showed the fluorescence was reduced compared to the 285 extract not treated with nuclease. Although the use of nuclease has not allowed us to 286 determine directly whether DNA caused the extremely high fluorescence observed in the 287 EDTA supernatant or not, it has allowed us to establish beyond doubt that aDNA is being 288 recovered, since as much as half of the DNA appears to be lost through decalcification. The 289 amount of DNA lost during decalcification was not determined for every femur sampled. Collagen has been identified as a major inhibitor of PCR. This protein constitutes around 296 90% of the organic fraction of bone (1/3 by weight), the remaining 10% (2/3 by weight) 297 consisting mainly of water and inorganic material, most of which is calcium phosphate. To

306
After collagenase treatment, the extract was eluted once with 150 μl Purite water. The 307 modified extract was added to PEP-PCR with 0.2, 0.6, 1.0, 1.25 and 1.5 μl of Taq. An aliquot 308 of the PEP-PCR products was added to specific mitochondrial PCR with primers L16055 and 309 H16142 flanking a 125-bp region of hypervariable segment I of the mitochondrial control 310 region. Both PEP and PCR products were mixed with loading buffer and electrophoresed 311 through a 3% agarose gel pre-stained with ethidium bromide. The extract was also used to 312 spike control PCR.

314
Removing collagen through enzymatic hydrolysis greatly reduced inhibition. However, 315 inhibition could not be fully prevented. This suggested that it was not inhibitors that were 316 disrupting amplification, but rather that it was the damaged and modified aDNA that did not 317 allow amplification. This, however, was found not to be the case, since small volumes (μl) of 318 aDNA extracts added to PCR mixture with abundant undamaged templates were shown to 319 inhibit and reduce amplification. Through modification of the standard PCR, including the 320 use of wax-mediated hot start and touchdown PCR, specific amplification products of low 321 target copy templates were obtained. By diluting the extracts, using more Taq and adding

330
As a further control, standard and modified extracts (5-15 μl) were added to the control PCR 331 mixture with mitochondrial primers. Amplification was observed in all PCR reactions spiked 332 with the modified extracts. However, for PCR reactions spiked with standard extracts, 333 amplification, despite being extremely weak, was detected for the reaction with 5μl of ancient 334 extract added but negative for 10 and 15 μl. The template used for the control PCR reaction 335 was from Raji cells, extracted with the same method as that used for extracting DNA from the 336 bones.

355
In PCR-SSOP typing, the PCR product is fixed on a nylon membrane and the probes are then attached to the membrane, the membranes were exposed to UV light for 1 min.

369
The membranes were first washed (non-stringent 5x SSC, 0.1% SDS) at room temperature 370 for 5 min, and then washed (stringent 1x SSC, 0.1% SDS) at a specific temperature for 30 371 min. After the stringent wash, the nylon membranes were washed twice in buffer (1.0M Urea, 372 0.1M NaCl, 5% Triton X-100, 1% Dextran Sulphate) for 1 min at room temperature before 373 being submerged in 25 ml of the same buffer containing horse radish peroxidase (HRP).

375
Hybridization of the probe with the amplified DNA was detected using chemiluminescence.

376
Equal volumes of the detection reagents containing luminol and H 2 O 2 were mixed and 377 applied to the membrane. After incubation for 2 min, the membranes were wrapped in Saran Wrap. The exposure of the detection reagents to HRP produces a blue light that is detected on 379 X-ray film. The membranes were exposed to X-ray film for 5 min. After removing the first 380 film, a second exposure was taken immediately; the exposure time for second was dependent 381 on the signal intensity obtained with the first.

Supplementary information tables 384
Archaeological context

Phylotree generation 503
Methods 504 The mtDNA sequencing data and SNP data were used to generate proper HSD files using the 505 mtDNAProfiler server (http://mtprofiler.yonsei.ac.kr/). The haplogroups were automatically assigned 506 using Phylotree data by the HaploGrep v.2.0 server (http://haplogrep.uibk.ac.at). Haplogroup 507 classification was based on Kulczynski distance. For each burial specimen and the two controls a 508 lineage graphical representation of the haplogroup classification per sample was generated. populations, with a >2% frequency distribution of mtDNA J within Europe. Subgroup J2a is 516 homogenously spread in Europe, but absent in the nations around the Caucasus. It is not known to be 517 found elsewhere. The time of origin of the younger branches of mtHG J for J2a1a1a is 7,591.6 ± 518 2,889.6 (between 4,700 and 10,500 before present (ybp). 519 Haplogroup H1+16239 (G361): H1 encompasses an important fraction of Western European 520 mtDNA lineages, reaching its local peak among contemporary Basques (27.8%). It also occurs at high 521 frequencies elsewhere in the Iberian Peninsula, as well as in the Maghreb (Tamazgha). The 522 haplogroup frequency is above 10% in many other parts of Europe (France, Sardinia, parts of the 523 British Isles, Alps, large portions of Eastern Europe. 524

Results 525
The haplogroup H2a2a1 is the most common and is represented by the Control mtDNA and in 526 specimens from G370. It is a subtype of the "Helena" clan haplogroup (Sykes, 2007: "Saxons, 527 Vikings and Celts: The Genetic Roots of Britain and Ireland"). H2a2 is found throughout Europe (ref: possible Roman descent of the burials. 536 The haplogroup J2a1a1a, represented in the specimens from G298 and G346, is most prevalent in 537 Wales, Scotland and Ireland (https://www.eupedia.com/genetics/britain_ireland_dna.shtml), with J2a1 538 being found mostly in western, central and northern Europe, particularly around the Alps.