Retrovirology - Latest Comments

Emv2

Christine Kozak — 2012-04-05T22:05:03Z

It is clear that the C57BL Emv described in the paper you referenced (Int. J. Cancer 2006 119:1869-1877) is Emv2. The HindIII fragment size in Fig 2 is comparable to that described for Emv2 by Jenkins and colleagues (J Virology 1982, 43:26-36), and, as pointed out earlier (Retrovirology 9:23 and 9:25), BLAST searches with either E-MLV env specific sequences or the full length E-MLV genome identify a single nearly identical sequence in C57BL on distal Chromosome 8, as is also the case with the search you described using the recombinant MelARV. The paper by Lee and colleagues (Retrovirology 8:82, 2011) also noted the high degree of sequence identity between Emv2 and the Emv2-derived MelARV.

My comment emphasized only two factors ¿ Emv2 expression and map location. With regard to the latter, I note that the MelARV paper provided a cytogenetic map location for Emv2 (8qE1), along with more specific information on its chromosomal position relative to a newly acquired insertion, and this position is consistent with the Emv2 map location based on the earlier linkage analyses.

Emv2 and ERVmch8 are one and the same

Christian Lavialle — 2012-03-30T13:52:49Z

Upon reading each of the commentaries published in Retrovirology (9:23, 9:24 and 9:25) on March 22, 2012, about the debated identity of the Emv2 endogenous retrovirus of C57BL6 mice, it appeared to me that none of the authors seemed to be aware of previously published data that I believe contain all the information required to settle the matter.
Indeed, in a paper published in 2006 (Int. J. Cancer 119, 1869-1877), Potlichet, Mangeney and Heidmann had already precisely identified the Emv-2 provirus within the genome of C57BL6 mice by both Southern blotting with an ecotropic MLV-specific probe and nucleotide sequence comparisons. They reported that the melanoma-associated retrovirus (MelARV) expressed in the B16 mouse melanoma is a recombinant between the Emv-2 endogenous retrovirus present in the genome of C57BL6 mice at the position of chromosomal band 8qE1 and a provirus located in the qA3 region of chromosome 5. The authors stated that sequence alignments of MelARV with the endogenous Emv-2 and a chromosome 5 provirus provided evidence that the Emv-2 provirus was the prime contributor to the backbone of MelARV, sharing with it more than 99.8% nucleotide sequence identity, except for two small gag (367 nt) and pol (416 nt) regions that were provided by the chromosome 5 endogenous provirus (see Figure 3 in the reference mentioned above). Using the MelARV sequence (accession number U63133 or DQ366149) as a BLAST query on the mouse genome (NCBI build 37.1), a unique highly homologous region consistent with Emv-2 is found at exactly the same position as that given for the so-called ERVmch8. All the other matching loci have only 90% or less nucleotide similarity. Thus, ERVmch8 and Emv-2 are unambiguously one and the same.

Incorrect interpretation of previously published data in the paper ¿Identification, characterization, and comparative genomic distribution of the HERV-K (HML-2) group of human endogenous retroviruses¿ written by Ravi P Subramanian, Julia H Wildschutte, Crystal Russo and John M Coffin.

Anton Buzdin — 2012-01-17T09:17:27Z

Incorrect interpretation of previously published data in the paper ¿Identification, characterization, and comparative genomic distribution of the HERV-K (HML-2) group of human endogenous retroviruses¿ written by Ravi P Subramanian, Julia H Wildschutte, Crystal Russo and John M Coffin.

Anton A. Buzdin, Eugene D. Sverdlov
Shemyakin-Ovchinnikov Institute of Bioorganic Chemistry, Moscow 117997 Miklukho-Maklaya 16/10

In this paper, Subramanian and colleagues describe the results of their systematic bioinformatics screening of the endogenous retroviral group HERV-K (HML-2) representatives that may exist in human genome both in the form of proviral copies or as the solitary LTR sequences [1]. The paper contains important information on the structural features and genomic distribution of this retroviral group. The authors also provided new estimates of the number of the HERV-K(HML-2) members in human DNA, that are more accurate than several recent estimates.
However, in this publication the authors misinterpret some data published before. For example, the authors state:
¿HML-2 LTRs cluster into one of three subgroups based on phylogeny and shared nucleotide features: LTR5Hs, LTR5A, and LTR5B ..[2].¿
This passage includes citation of our paper published in 2003 describing sequence analysis of the human specific HERV-K (HML-2) inserts, which allowed us to identify a family of HERV-K (HML-2) LTRs that included mostly human specific members (in our paper termed ¿HS family¿). The detailed analysis of this family enabled us to subdivide it into two subfamilies: HS-a and HS-b. The subfamily HS-a included only human specific members, whereas some of the subfamily HS-b members could be also found in the chimpanzee genome. The estimated evolutionary ages for these subfamilies were 5,8 and 10,3 million years, respectively. We published the diagnostic nucleotide positions within the consensus sequence that unambiguously distinguish the HS-a and -b subfamilies. When the paper has been already published, the database of eukaryotic repetitive elements Repbase Update [3] was updated with the consensus sequence of the evolutionary recent fraction of the HERV-K (HML-2) LTRs termed ¿LTR5Hs¿, which had many common features with our ¿HS family¿ consensus sequence, but was not identical to it. Later on, Repbase Update has published two other consensus sequences to characterize evolutionary older HERV-K (HML-2) LTRs, termed LTR5A and LTR5B. Importantly, these consensus sequences built for the evolutionary older LTRs had little in common with our consensus sequences termed HS-a and HS-b created for the youngest subset of the HERV-K (HML-2) LTRs. For example, the consensus sequences from the Repbase Update missed any of the above diagnostic nucleotide substitutions characterizing our HS-a and -b subfamilies.
Unfortunately, Subramanian and colleagues appeared to mix up the two classifications mentioned above (communicated in our paper and in Repbase Update). This mistake, to our mind, led to significant misinterpretation our data in the present manuscript that resulted to the following passages:
¿Overall, our classification largely agrees with the previous report defining the three major subgroups [2]; however, the larger sample size of our data set highlighted inconsistencies in the previous classification system¿In a sequence comparison of 1092 HML-2 LTRs, we successfully identified subgroup-specific features which we used to discriminate the LTR5Hs, LTR5A, and LTR5B elements. We would like to note that previously described sequence polymorphisms [2] were not observed among all sequences in any subgroup, likely due to our large sample size¿ The ages of HML-2 proviruses as a function of LTR subgroup were previously estimated by Buzdin et al. based on the intrabranch divergence between individual elements from the subgroup consensus. Their analysis of ~40 LTRs estimated that the LTR5A and 5B subgroups formed around 5.8 and 10.3 million years ago (mya), respectively [14], with 5A originating from 5B. These are fairly recent estimates for these subgroups, given that most LTR5A and 5B proviruses have shared loci among primates whose divergence from humans substantially predates this timeframe. This underestimation is likely due to faulty molecular clock assumptions as well as the use of relatively few proviruses from early sequence builds¿.Consistent with the LTR-based and internal- based phylogenies (Figures 2, 3, and 4) we found the LTR5A and LTR5B proviruses to have formed earlier, around ~20.1 (± 5.4) mya and ~27.9 (± 12.0) mya, respectively¿.We did not observe the sequence polymorphisms within subgroups of our sequences as previously used to define the groups [2], likely due to our much larger sample set.¿
Subramanian et al. compare the data collected in their study for the groups LTR5A and LTR5B(Repbase Update) with our data on the groups HS-a and HS-b. The disagreement in these datasets, to our mind, is not caused by any kind of methodical faults, but is exclusively due to the structural differences between the LTR5A/B vs HS-a/ b groups.
We believe that in this communication it is extremely important to underline the differnce between the HERV-K (HML-2) LTR classifications published in our paper [2], and those presenting in the Repbase Update database in order to avoid any related confusing situations in the future.

1. Subramanian RP, Wildschutte JH, Russo C, Coffin JM: Identification, characterization, and comparative genomic distribution of the HERV-K (HML-2) group of human endogenous retroviruses. Retrovirology, 8:90.
2. Buzdin A, Ustyugova S, Khodosevich K, Mamedov I, Lebedev Y, Hunsmann G, Sverdlov E: Human-specific subfamilies of HERV-K (HML-2) long terminal repeats: three master genes were active simultaneously during branching of hominoid lineages. Genomics 2003, 81:149-156.
3. Jurka J, Kapitonov VV, Pavlicek A, Klonowski P, Kohany O, Walichiewicz J: Repbase Update, a database of eukaryotic repetitive elements. Cytogenet Genome Res 2005, 110:462-467.

Correction of a typographical error in the abstract

Elisabeth Menu — 2011-08-24T17:16:23Z

In the abstract, the last sentence of the results section should be replaced by : "Using HIV-1 pseudotypes, we found that HIV-1 entry step was inhibited by decidual soluble factors."

Email address update

Mario Chin — 2011-03-11T21:23:07Z

Current corresponding author email address for M.P.S. Chin: mpschin@amvetbio.com

XMRV is inserted into 472 sites in Prostate DNA

Gerwyn Morris — 2011-03-06T14:22:01Z

In his reply to my letters of march 02 and 03 AG Miller referenced the study of Kim and others 2008. On close examination the details of the approach reported by Kim and others are revealing and challenge the conlusions of Garson et al and the accuracy of comments made by AG Miller in his most recent letter. Permission to reproduce the following sections has been granted by the authors.

"We sequenced a total of 508 authentic XMRV integration sites from DU145 cells, and 472 of these sites were mapped to unique locations in the human genome. Integration events were found in all 24 human chromosomes (22 autosomes and the sex chromosomes X and Y) (Fig. 1). The frequencies of integration of XMRV were generally proportional to chromosome size, but the overall frequency of XMRV integration into human chromosomes was different from that of uniformly random integration (P < 0.0001). Notably, chromosomes 1, 17, and 19 were significantly overrepresented (P values of 0.0015, 0.0021, and <0.0001, respectively), while chromosomes 5, 13, and X were significantly underrepresented (P = 0.0081, 0.0099, and 0.0002, respectively). Different integration frequencies among the different human chromosomes have also been observed for other retroviruses Additionally, using the criteria previously defined for integration hot spot which is three or more integrations within a 100-kbp region, we identified four integration hot spots for XMRV "(Table 1).

Thus contrary to the assertions made by Miller XMRV is capable of making three insertions within a 100 kilobase region as noted in the follwing table duplicated from the study;

TABLE 1. Integration hotspots of XMRV
Chromosomal region Integration site positions
7q36.1 151018217 151071783 151075248
8q11.21 49590648 49665683 49668524
19p13.2 7363696 7366399 7366635
20q11.22 33358459 33364040 33370664

The next point is that the integration sites within DU145 DNA were compared to those in DNA obtained directly from tumour tissue .While the regions were the same in 472 instances the precise integration points were all different. The reader is invited to consult figure 1 in the study for a more detailed comparison

"FIG. 1. Positions of XMRV integration sites in the human genome. The human chromosomes are shown numbered. Centromere locations are denoted by chromosomal indentations. Sites of XMRV integration in DU145 cells are indicated as red vertical lines along the top, and XMRV integration sites in prostate cancer tissues are indicated as blue "lollipops" on the bottom. Within each chromosome, the top bar shows the relative densities of RefSeq genes, with higher gene-dense regions shown as a more intense cyan. The second bar shows the chromosome cytobands. The third bar shows the cancer breakpoints, and the frequencies of breakpoints in different chromosomal regions are denoted by different colors (see the key at the bottom right-hand corner). The green shading in the bottom bar denotes the locations of common fragile sites."

Thus if two of a possible 470 integrated XMRV sequences were found in DU145, cells which were identical in terms of nucleotide insertion points, then the experimental environment can be the only possible source.In short the DU145 cells were contaminated by human DNA during the experimental procedure.

The difference in the nucleotide integration pattern produced by the infectious clone VP62 compared to xmrv found inserted in the DNA of prostate tissue isolated directly from patients would also seem worthy of further investigation. Differences in genomic sequences could well explain why thus far PCR systems with cycling conditions adjusted to locate the VP62 clone in vitro have failed to find XMRV in vivo.

reference

Kim S, Kim N, Dong B, Boren D, Lee SA, et al. (2008) Integration site preference of xenotropic murine leukemia virus-related virus, a new human retrovirus associated with prostate cancer. J Virol 82: 9964–9977.

.response rebutting Miller

Gerwyn Morris — 2011-03-05T22:43:35Z

Dr Miller you are commenting on a different study to the one I raised

I am talking about a study by Dong et al 2007 which first detected xmrv in dna taken straight from prostate tissue where they first screened the DU 145 cells to ensure that they did not harbour XMRV.

If you are going to reply to my posts at least have the courtesy of reading them first

You talk of hypothetical situations as though they were fact

As you know the DU 145 cell line environment is one of high oxidative stress and elevated levels of NF-kappa B.you also know that high NF-kappa B instigates the transcription of XMRV and stimulates its replication. If the DUi25 cell line DNA contained integrated XMRV provirus then the cell line wouldexpress XMRV.

Mutations in the cellular and viral DNA would also be expected because of the age of the cell line and the level of oxidative stress involved.None were observed.The most parsimonious hypothesis available which explains all the observations is that human DNA containing integrated XMRV provirus was introduced into the DU125 cells in error.

You speculate about the possibility of contamination without having any empirical evidence whatsoever to base your comments on. It is traditional for scientists to comment on scientific evidence and not to engage in speculation.The abstracts presented at CROI ,for example ,all share the same generic theme.That theme is that XMRV could hypothetically be a contaminant if hypothetical conditions existed. I suggest that many retrovirologists could now benefit from consulting people who are more knowledgeable about study design and adherence to the scientific method Consulting retrovirologists with experience of discovering new human retroviruses would also seem to be adviseable

You saw fit to adopt a condesending attiude in your article and questioned my knowledge.I submit that insulting others has no part to play in anything published in a scientific journal

I suggest that it might benefit you if you spent more time examining the requirements of the scientific method and being knowledgeable enough to at least comment on the correct study.I found.

Response to the comments by Gerwyn Morris posted 02 March and 03 March 2011

A Dusty Miller — 2011-03-04T15:55:36Z

You note that I have not replied to your previous comments. This is because of their voluminous nature, your inability to correctly interpret what is basically a sound article, and because I have other things to do!

In your post of 02 March, the main substantive point questions the proposed direction of contaminant transfer, from the XMRV-infected DU145 cells to the prostate cancer tissue. In the Kim et al. 2008 paper describing these studies, they infected DU145 cells with XMRV at a multiplicity of 0.1 and the cells were grown for 3 days. By this time, it is likely that all of the cells carried at least one integrated copy of XMRV, because XMRV is replication-competent and can readily spread in cultured DU145 prostate cancer cells. In contrast, the number of integrated copies of XMRV in fresh prostate cancer tissue is estimated to be low, on the order of 1 copy per 100 cells, if it is there at all. Therefore, contamination of the prostate tissue DNA with heavily-infected DU145 cell DNA is more likely to result in false positive results than the reverse. Furthermore, the real question here is whether XMRV can be found integrated into prostate cancer cells from a human, not whether the DU145 cells are infected. The finding of identical integration sites in these two samples suggests contamination of the prostate cancer samples with XMRV-infected DU145 DNA or PCR products amplified from this DNA, and threatens the conclusion that XMRV is found integrated into the DNA of prostate cancer tissue from humans.

In your post of 03 March, the main substantive point appears to question how DU145 cells could have been involved in the studies of XMRV integration sites in prostate cancer tissue. The answer is that the analysis of XMRV integration sites in XMRV-infected DU145 cells was performed at the same time and in the same lab as was the analysis of XMRV integration into DNA from human prostate cancer tissue, thus the possibility of cross contamination must be considered. Many of your other questions make no sense to me, and I don’t have the time to delve into your underlying meanings. I suggest you seek further clarification and answers to your questions from knowledgeable colleagues.

Response to the comment by Abigail Smith posted 02 March 2011

A Dusty Miller — 2011-03-04T15:53:26Z

Thanks for the citation (Mitchell et al., PLoS Biology 2:E234, 2004). While these authors did not mention observing any duplicate integrations, they do state in Materials and Methods that only novel integration site sequences were deposited at the NCBI, implying that some sites were not novel and thus were duplicates. Duplicates would not be unexpected in a single experiment, because PCR amplification results in many copies of what might have initially been a unique site, and if one clones and sequences enough sites, duplicates are bound to be found. For example, if 1,000 cells were infected with 10 of their replication-defective HIV vector particles, resulting in 10 integrated proviruses, cloning of 100 integration sites from DNA isolated from these cells would result in detection of multiple identical (duplicate) integration sites. On the other hand, infection of a million cells with 100,000 vector particles followed by cloning of 100 integration sites would be very unlikely to identify identical sites.

Thus, to get at the question of whether retroviruses integrate at identical sites at much higher rates than would be expected by chance, and thus might explain the identical XMRV integration sites identified by Garson et al., one needs to sequence integration sites from at least two independent experiments, and compare integration sites between experiments for duplicate integrations. The data to make this comparison is obviously available, but I’m not aware of anyone having performed this analysis. If someone has done so, I’d appreciate being informed.

This discussion gets at the heart of the contamination issue raised by Garson et al. In the experiments they discuss, 14 XMRV integration sites in prostate cancer tissue are compared to ~500 XMRV integration sites found in acutely-infected DU145 human prostate cancer cells, and two identical (duplicate) integration sites are observed in these theoretically independent experiments. Given the number of possible integration sites in the human genome (~3 billion), cross contamination almost certainly explains the result. That is, unless there are some unusually hot spots for retrovirus integration, which likely would have been identified by now.

the study cannot demonstrate direction of transfer

Gerwyn Morris — 2011-03-03T11:48:16Z

The authors did not follow the scientific method and demonstrate that the DU125 cells did not contain integrated DNA before the start of the experiment. If the cell line was "contaminated" with human XMRV then the results would have been exactly the same

The other problem they have is that none of the studies that identified and isolated XMRV used the the DU145 cell line in any way whatsoever

The experiment that isolated integrated XMRV directly from the prostste tissue of infected patients also did not invole the use of DU145

The question of identical insertion sites is a probabalistic argument and purely determined by sample size and the affinity of a retrovirus for GPG islands

XMRV has been demonstrated a greater affinity for these sites than any known retrovirus.

DU145 cell lines began in 1978 yet Towers asserts that the origin of the XMRV detected in prostae cancer is the "XMRV" in the 22RV1 cell line which began in 1993

Schalberg et al(2009) reported that RT PCR only detected xmrv in samples taken from 6% of prostste cancer suffere while IHC showed 22% of paients to be positive

This is an interesting observation if XMRV is a Mouse contaminant!

I note Miller has not replied to any of the other points I raised