Screening for 15 pathogenic viruses in human cell lines registered at the JCRB Cell Bank: characterization of in vitro human cells by viral infection

Human cell lines have been used in a variety of research fields as an in vitro model. These cells are all derived from human tissue samples, thus there is a possibility of virus infection. Virus tests are routinely performed in clinical practice, but are limited in cell lines. In this study, we investigated 15 kinds of viruses in 844 human cell lines registered at the Japanese Collection of Research Bioresources (JCRB) Cell Bank. Our real-time PCR analysis revealed that six viruses, EBV, HTLV-1, HBV, B19V, HHV-6 and HHV-7, were detected in 43 cell lines. Of them, 20 cell lines were transformed by intentional infection in vitro with EBV or HTLV-1. Viruses in the other 23 cell lines and one EBV transformed cell line are derived from an in vivo infection, including five de novo identifications of EBV, B19V or HHV-7 carriers. Among them, 17 cell lines were established from patients diagnosed with virus-associated diseases. However, the other seven cell lines originated from in vivo cells unrelated to disease or cellular tropism. Our approach to screen for a set of 15 viruses in each cell line has worked efficiently to identify these rare cases. Virus tests in cell lines contribute not only to safety assessments but also to investigation of in vivo viral infection which can be a characteristic feature of cell lines.

lines carrying the same name appear with different registration numbers. These cell lines were deposited by various depositors through different routes, and would normally have the same characteristics, but there is the possibility of subtle strain differences.

Microbiological tests
The residual cells from every freezing lot were inoculated into two tubes and one agar plate (NB medium-nutrient broth with 2% yeast extract, thioglycollate medium, 1-blood agar plate) for a sterility test. These tubes and plates were incubated at 35°C for 2 weeks to examine bacterial and fungal contamination.

DNA and RNA extraction
Total DNA and RNA samples were extracted from each cell line with the AllPrep DNA/RNA Mini kit (QIAGEN, Germany, 80204). Some DNA was prepared by a conventional protocol using phenolchloroform. When extracting RNA for copy number titration, DNA contamination was eliminated by digestion with RNase-free DNase I (QIAGEN, Germany, 79254) at room temperature.

Detection of viral DNA with real-time PCR
The viral DNA test was targeted to nine and four kinds of DNA and RNA viruses, respectively (table 1) and performed using customized 96-well plates inoculated with primers and probes in trehalose solution [17]. This format included two sets of multiplex

Reverse Transcriptase Assay
Some cell lines were transfected with the luciferase expressing gene using retrovirus or lentivirus vectors.
To examine lentivirus replication, the cultured medium of these cell lines was centrifuged at 250g for 10 min at 4°C to remove cell debris. The supernatant was concentrated by centrifugation (22 000g) for 2 h at 4°C, then the pellets were analysed by the Reverse Transcriptase Assay (Roche, Germany, 11468120910) as described by the manufacturer.

Results
Our virus test detected six viruses, EBV, HTLV-1, HBV, B19V, HHV-6 and HHV-7, in 43 cell lines (table 2). One of them, AT(L)7KY, was positive for both EBV and HHV-7. Six cell lines are sublines, corresponding to three individuals, owing to duplicated registrations, different names or misidentification. The other nine viruses were not found in any of the 844 cell lines. Whole results are listed in the electronic supplementary material, table S1 in which positive data are shown by Ct value or DNA copy number. All cell lines were negative for bacterial microorganisms.

EBV
The most common virus we detected in human cell lines was EBV, found in 30 cell lines, all of which are derived from blood cells (table 3.1). Two of them, P32/ISH and ITSM, have not been reported to be EBV positive and have been identified in this study. EBV is closely associated with endemic Burkitt lymphoma, but this is low in Japanese cases [18]. Six of nine Burkitt lymphoma cell lines, Daudi, HS-Sultan, Namalwa, NC-37, P32/ISH and RAJI, were positive for EBV in contrast to three negative cell lines, KHM-10B, Minami-2 and Ramos (RA1). NC-37 is a misidentified cell line and has been revealed as the derivative of RAJI. Except for P32/ISH, the other five positive cell lines different from Japanese origin showed a high copy number of EBV. P32/ISH, as described previously [19], has an immature B-cell phenotype on the basis of immunological surface marker analysis. EBV nuclear antigen and terminal deoxynucleotidyl transferase activity were negative, which would be related to low copy number of EBV. The other eight EBV-positive cell lines were established from different haematological diseases, leukaemia, multiple myeloma, lymphoma, pseudomyxoma peritonei or severe chronic active EBV infection. One of them, KAI3, is non-B-cell and has been reported as an NK-like cell line [20]. RPMI 1788 was not associated with any disease and was established from normal cells, which would be EBV positive in vivo.
Compared with these 14 EBV-positive cell lines originated from in vivo infection, 16 lymphoblastoid cell lines were established from in vitro transformation of peripheral blood mononuclear cells using EBV. They were obtained from patients with retinoblastoma and their parents, or patients with a rare disease such as Bloom syndrome, ataxia telangiectasia or xeroderma pigmentosum. Different from these cell lines     . related to inherited diseases, TK6(IVGT) is characterized by being heterozygous for thymidine kinase and has been used for in vitro mutagenesis assays [21]. EBV-transformed cell lines generally proliferate well, and high copy number of EBV was detected.

HTLV-1
HTLV-1 was detected in six cell lines, one lung cancer and four leucocytes (table 3.2). KHM-3S was established from pleural effusion of a patient with small cell lung cancer and showed negative for leucocyte common antigen [22], indicating its unique tropism. MT-1 was derived from a patient with adult T-cell leukaemia (ATL), which shows strong association with HTLV-1 [23]. These two cell lines originated from an in vivo infection. MT-2, MT-3 and MT-4 were established by co-culture with lymphoid cells from ATL patients and were infected with HTLV-1 in vitro [24]. The viral RNA was detected from these six cell lines and the DNA showed a high copy number. Five types of HTLV-1 proviral sequences were detected in 11 different sites of the MT-2 genome, indicating the integration [25]. It is reported that these MT-1, MT-2, MT-3 and MT-4 cell lines are virus-producing cell lines [26].

HBV
HBV was detected from three hepatoma cell lines, huH-1, JHH-7 and PLC/PRF/5 (table 3.3). PLC/PRF/5 is also known as Alexander cells, registered as IFO50069. The viral DNA copy number varied between these cell lines. It is reported that huH-1 and PLC/PRF/5 carry hepatitis B surface antigen (HBsAg) [27].
Although the target region of our real-time PCR is the S gene coding for HBsAg of the virus envelope, the antigen was not detected in JHH-7 [28]. Two integration sites composed of incomplete HBV genome are identified in the JHH-7 genome [28]. Other hepatoma cell lines without HBsAg, JHH-1, -2, -4, -5 and -6, were negative for HBV [28], indicating that JHH-7 exhibits a distinctive feature in the HBVpositive cells.

FPC5JTO
skin fibroblast of familial polyposis coli patient JCRB0318

XP2SA
skin fibroblast of xeroderma pigmentosum (variant type) patient JCRB0305  Six cell lines are sublines consisting of three different in vivo origins. a Identical origin because of misidentification in NC- 37. b Different registration numbers due to independent depositions. c PLC/PRF/5 has also been used as Alexander cells.

HHV-6
HHV-6 was detected in a human vascular endothelial cell line, HUV-EC-C, derived from normal umbilical cord tissue (table 3.5). We have previously reported that chromosomally integrated HHV-6B has been identified at the distal end of chromosome 9q in this particular cell line [30]. The copy number is calculated to be one copy per cell, indicating that the integration could have occurred in the germ line. The viral integration in the HUV-EC-C cell line would be caused by in vivo infection. This is distinct from other HHV-6-positive cell lines generated by in vitro transfection [31].

HHV-7
HHV-7 was detected in one cell line, AT(L)7KY, which was not previously reported to be positive. This cell line is a B-cell line transformed by EBV, indicating that both EBV and HHV-7 can be detected in this cell line (table 3.6). It was established from peripheral blood mononuclear cells of a patient with ataxia telangiectasia, which is not associated with HHV-7 infection. Different from HHV-7-infected cell lines transduced with an adenovirus vector [32], AT(L)7KY is, to the best of our knowledge, the first report of an HHV-7-positive cell line likely to be originated from in vivo infection which remains detectable in cell culture.

HIV-1 and HIV-2
A very small copy number of DNA fragments targeting the gag gene of HIV-1 and LTR of HIV-2 were detected in some of the luciferase-expressing cell lines (electronic supplementary material, table S4). However, these viral RNAs were not detected in these cell lines, and reverse transcriptase activity was below the detection limit. The absence of the RNA and reverse transcriptase in these cell lines confirmed that the cell lines are negative for HIV-1 and HIV-2. The luciferase gene was introduced into these cell lines by the lentiviral vector [33], which may be associated with the detection by PCR.
Because HIV-1 and HIV-2 are highly pathogenic and must be handled under biosafety level 3 (BSL3), such cell lines are not received and distributed from our JCRB Cell Bank. All cell lines examined in this study are free of HIV-1 and HIV-2, which minimizes safety concerns over the use of our cell lines.

Discussion
Virus infection is one of the major public health concerns and is related to disease pathogenesis [34]. Virus tests have been established for clinical samples in diagnostic laboratories [9][10][11]. These tests are also routinely performed on donated blood to avoid infections through blood products. Blood samples are examined mainly by targeting antigens or antibodies detected in plasma. Positive results from these indirect methods identify individuals carrying a viral infection, but do not specify which cells carry the virus. Different from those on in vivo samples, our virus tests for cell lines detect viral DNA or RNA in the cytoplasm or the nuclei. The presence of these viral components in cells can be a molecular characterization of the cell line.
Our previous study demonstrated that chromosomally integrated HHV-6B has been identified in a human vascular endothelial cell line, HUV-EC-C (IFO50271) [15]. The analysis of the HUV-EC-C cells reveals that integration of HHV-6B is stable during long-term culture of HUV-EC-C, despite genomic instability in the host cell. A complete HHV-6 genome can be present as the form of integration into the host germ line genome [35], leading to the unexpected detection in different cell types of tropism observed as the HUV-EC-C cell line. It was reported that a human B cell line, RPMI 1788, was established from the peripheral blood of a normal donor and labelled as normal cells [36], however this cell line was found to be EBV positive in another study [37]. This suggests that lymphocytes had been infected with EBV in vivo, leading to establishment of the immortalized cell line. These examples show that further investigation of virus-positive cell lines could lead to an understanding of viral dynamics and that such cell lines can serve as invaluable research materials. Virus tests contribute to the characterization of cell lines and elucidation of cellular origins. Because each cell line is derived from a specific tissue and established from a certain cell population in the sample, cell lines that show negative on virus tests are not necessarily derived from uninfected individuals.
Cells in vitro always have a risk of viral infection in the culture environment because, by themselves, they lack a defence system against infection. Assessment of virus contamination in cells is undertaken for the purpose of risk management [38]. When cell lines are used as biological products in the biotechnology industry, quality control is strictly maintained during production [39]. Examination of the parainfluenza virus has been performed in 123 human cell lines without any such infection being detected, assuring that those lines are free from the infection [40]. Another study of a screen for murine leukaemia viruses in 577 human cell lines reported that 17 of these cell lines produced active retroviruses [41]. This implies that human cell lines have a risk of viral contamination through transplantation of cells into mice, which is distinguished from in vivo infection characterizing cell lines. The use of human cell lines is accompanied by a potential biohazard, because virus tests are limited to certain species and do not cover all pathogens. It is recommended that human cell lines are primarily handled at a biosafety level 2 laboratory, regardless of results from virus tests. Especially, cell lines which produce virus particles need careful handling and their application should be restricted. Each virus has distinct host cell tropisms, implying that negative results would be expected from the majority of cell lines. Because it is not clear that viral tropism is identical between in vivo and in vitro environments, we applied the whole set of virus tests to all cell types. Apart from the cell lines transformed using EBV or HTLV-1 by in vitro infection, 22 cell lines had been infected with virus in vivo before their establishment. Of them, two cell lines are inconsistent with the cellular tropism, showing unexpected results.
It is reported that induced pluripotent stem cells can be established from EBV-transformed lymphoblastoid cell lines, and that the reprogrammed cells have been shown to be EBV free [42,43]. This suggests that a change in cell type could lead to elimination of viruses infecting cell lines. These viruses might present in the form of episomes in cells, which are different from integration into host genomes. Differences in the form of viruses in cells would be related to the amount of virus present during cell culture. In this report, human papillomavirus (HPV) is excluded because it has diverse genotypes related to malignant potential, which we will analyse in a future study. Analyses of RNA sequencing data led to identification of EBV, HHV-8 and HTLV-1 in lymphoma cell lines [44]. DNA sequence data from a large number of human cancer cell lines have been accumulated in the Cancer Cell Line Encyclopedia (CCLE) [45] and the Catalogue of Somatic Mutations in Cancer (COSMIC) [46]. Virus-positive cell lines will be identified from these databases, and bioinformatics approaches will help to investigate viral integration sites and genotypes.
Ethics. This work is based on the analysis of publicly available materials, which are registered with the JCRB Cell Bank.
This work did not involve direct access to any human samples and is not subject to human ethics review.
Data accessibility. The datasets supporting this article have been uploaded as the electronic supplementary tables.