Abstract
Animal donors such as pigs could provide an alternative source of organs for transplantation. However, the promise of xenotransplantation is offset by the possible public health risk of a cross-species infection1, 2. All pigs contain several copies of porcine endogenous retroviruses (PERV)3, 4, and at least three variants of PERV can infect human cell lines in vitro in co-culture, infectivity and pseudotyping experiments3, 5, 6, 7. Thus, if xenotransplantation of pig tissues results in PERV viral replication, there is a risk of spreading and adaptation of this retrovirus to the human host. C-type retroviruses related to PERV are associated with malignancies of haematopoietic lineage cells in their natural hosts8. Here we show that pig pancreatic islets produce PERV and can infect human cells in culture. After transplantation into NOD/SCID (non-obese diabetic, severe combined immunodeficiency) mice, we detect ongoing viral expression and several tissue compartments become infected. This is the first evidence that PERV is transcriptionally active and infectious cross-species in vivo after transplantation of pig tissues. These results show that a concern for PERV infection risk associated with pig islet xenotransplantation in immunosuppressed human patients may be justified.
Juvenile-onset diabetes mellitus is a major health problem and exogenous insulin therapy is only partially successful in preventing its many complications. Although islet transplantation holds great promise for a cure, the number of potential human pancreas donors are extremely unlikely to provide enough islet tissue to treat the millions of patients worldwide. The xenotransplantation of pig islets is therefore being actively investigated as a potential alternative clinical therapy. It is still uncertain whether cellular transplantation has the same infectious risk profile as vascularized organ xenotransplantation. To develop a small-animal model for islet xenotransplantation suitable for studying the risk of PERV infection in immunosuppressed human patients, we chose the NOD/SCID mouse strain. This strain can be successfully engrafted with human haematopoietic cells and tissues9, 10, 11, 12 as well as pig islets which function to reverse chemically induced diabetes (unpublished data).
We first determined the potential of adult pig islets to produce infectious PERV using in vitro co-culture assays. Cultured pig islets express PERV mRNA based on quantitative RT–PCR (polymerase chain reaction after reverse transcription of RNA) for the spliced product of PERV env sequence (data not shown). This result is consistent with preliminary data that reverse transcriptase (RT) activity was detected in cultures of fetal pig islets13. Next, neo-resistant U293 cells were co-cultured with pig islets for eight weeks. The neomycin analogue, G418, was added after seven days to eliminate the pig cells. The use of neomycin selection has the advantage of avoiding irradiation, a procedure that might activate retroviral transcription. As shown in Fig. 1a, the U293 cells were clearly positive for PERV pol DNA at 18 and 28 days of co-culture. However, no pig-specific cytochrome c sequence could be detected by Southern blotting of PCR products after 11 days of G418 selection (a total of 18 days of culture), indicating that all the pig cells were eliminated. Infection was confirmed by quantitative PCR (Fig. 1b) demonstrating detectable PERV-specific pol sequence and undetectable pig centromeric sequence. We demonstrated that U293 cells infected by pig islet-derived PERV secrete infectious virions by showing reverse transcriptase activity and infectivity of these culture supernatants (Fig. 1c). Therefore, purified pig islets release infectious PERV virions and productively infect the human epithelial cell line, U293. U293 is a well established target cell line for retrovirus infection; however, we and others6 have also infected mouse cell lines with PERV in culture.
Figure 1: Pig islets release infectious PERV virions and productively infect human U293 cells.
a, U293/neo cells co-cultured with pig islets or PK15 became clearly positive for PERV pol whereas no pig-specific cytochrome c sequence could be detected after G418 selection (one of four similar experiments is shown). b, The number of PERV pol and pig centromeric DNA copies per 100 ng DNA from the same cells after 44 days of culture was quantified (n = 2). c, Cells incubated with conditioned media from PERV-infected cells clearly became positive, demonstrating that U293 co-cultured with islets are productively infected. RT enzyme activity was tested at week 6 of culture. ND, not done.
High resolution image and legend (59K)Pig islets transplanted under the kidney capsules of immunodeficient mice were explanted 7–8 weeks later and analysed by confocal immunohistochemistry for expression of PERV p30 Gag protein. Several cells within the islet grafts were stained positive for PERV p30 (Fig. 2c). This excludes the possibility that p30 expression is restricted to a small population of passenger leukocytes or islet dendritic cells. Moreover, insulin staining (Fig. 2a) did not specifically co-localize with p30 staining. Staining was characteristically in a granular pattern and essentially identical to the staining that we have documented in U293 and primary human bone marrow endothelial cells infected with PERV in co-cultures. At various time points after transplantation (18 to 56 days), islets under the kidney capsule or in three-dimensional collagen templates coated with an arginine, glycine and aspartic acid (RGD) peptide placed in perispinous muscle were analysed for the presence of spliced PERV env mRNA using quantitative RT–PCR (Fig. 2d). PERV mRNA was detected in islet grafts in four independent experiments. Inflammatory cytokines and local tissue ischaemia associated with surgical injury might increase the patient's exposure to PERV by activating viral replication shortly after transplantation. We determined PERV mRNA expression in the islets before transplantation and from islets explanted from the same animals 3, 7 and 14 days after transplantation. As shown in Fig. 2e, there is a greater than tenfold increase in subgenomic PERV mRNA expression from day 0 to day 7. Transcriptional activation of PERV genes thus occurs in vivo after pig islet transplantation.
Figure 2: PERV expression in pig islets after transplantation.
a, c, Immunohistochemical staining for PERV Gag protein p30
(c) and insulin (a) in pig islets transplanted under the mouse
kidney capsule. A number of clearly p30-positive islet cells are identified.
b, The rabbit antisera control. Magnification, 
400. d, Spliced
mRNA for PERV env was determined by quantitative RT–PCR and detected
in islet grafts after transplantation in NOD/SCID mice. N/A, not applicable.
e, PERV env mRNA expression in islet grafts was significantly increased
at day 7 (asterisk, P < 0.015, Wilcoxon) compared to day 0 and day
3 after transplantation. Shown are the means 
s.e.m. (n = 9)
of three independent experiments.
Retrospective analysis of patients in clinical trials exposed in vivo or extracorporeally to pig cells has demonstrated long-term survival or microchimaerism of pig cells in some patients. Chimaerism after pig islet transplantation was detected by PCR for mitochondrial DNA in the peripheral blood at six months in four of ten patients13. Chimaerism was detected by PCR for pig mitochondrial and centromeric sequences in 23 of 100 patients after extracorporeal splenic perfusion, persisting for up to 8.5 years in one patient14. We determined the extent of chimaerism in NOD/SCID mice transplanted with pig islets. In 17 of 25 mice (68%), one or more tissues other than the transplanted islets tested positive for PERV pol and centromeric DNA (Table 1). In the chimaeric animals, 30% of tested tissues contained pig sequences. Thus, our results show the migration of pig cells after islet transplantation into several tissue compartments increasing the range of tissue exposure to potential PERV infection far beyond the immediate transplantation site.
It is vital to determine whether pig islets release infectious virus after
transplantation. Every pig cell contains multiple copies of PERV proviral
DNA3, 4 and centromeric sequence repeats. Our approach to distinguish
pig cell chimaerism from a true infection of mouse tissues is based on the
premise that the ratio between PERV proviral DNA and pig centromeric sequences
(P/C ratio) in any given pig cell or pig tissue is fixed. We
tested five different islet preparations from outbred Landrace pigs used for
these transplants to obtain a mean P/C ratio of 0.050
0.009 (s.d.). We also tested liver tissues from 43 Landrace pigs (P/
C ratio = 0.042 0.041) and peripheral blood cells from 9 Yorkshire/Hampshire
pigs (P/C ratio = 0.050 0.033) to obtain a mean
P/C ratio for control pig tissues of 0.043 0.040 (s.d.).
Thus, there is consistency between the P/C values calculated
for islets, liver and peripheral blood cells supporting the premise that the
ratio between these two genes is fixed. If the P/C ratio is
fixed, then a significant increase in the P/C ratio in chimaeric
tissues can only be achieved by a relative increase in PERV DNA consistent
with infection of the mouse tissue. To test the sensitivity of this approach
we calculated P/C ratios by mixing 20 pig cells with 10
7 human cells (1 pig cell per 5 105 human cells)
and spiking the mixture with plasmid PERV DNA from 0 to 3 10
4 molecules. The P/C ratio was found to be significantly
different when the equivalent of more than 1,000 human cells (>0.01%) are
infected by a single copy of PERV.
Table 2 shows the actual copy numbers of PERV and
pig centromeric DNA in all 13 tissues from 8 animals in which a P/
C ratio could be calculated. In 9 of the 17 chimaeric animals shown in Table 1, an accurate P/C ratio could not be
confidently determined, despite positive tests for PERV and pig centromeric
DNA, because the PERV values were too low (1–10 copies per 3.3 
g
DNA). The remaining 8 of 25 total animals that demonstrated no evidence of
pig cell chimaerism are not shown and were all negative for PERV. The results
demonstrate PERV infection in one or more tissues in the 8 chimaeric animals
(32%) based on significantly higher P/C ratios when compared
against the ratios of the 5 pig islet preparations used in these transplantation
studies (P = 0.002, Wilcoxon) or 52 control pig tissues (P < 0.001).
A possible complication for our model is that mice are known to contain xenotropic
endogenous retroviruses15, 16. Thus, it is possible that endogenous
murine retroviruses could infect transplanted pig cells and then rescue or
pseudotype PERV to allow infection of mouse cells. In addition, recombination
of PERV and endogenous murine retroviruses could have resulted in a recombinant
PERV virus with altered tropism.
Here we show that pig islets release infectious PERV virus which can productively infect human target cells in vitro. After islet transplantation we demonstrate active proviral DNA transcription by measuring subgenomic (spliced) mRNA for the PERV env gene and expression of p30 Gag by islet cells. Moreover, we demonstrate PERV infection of mouse cells in multiple tissue compartments. We also demonstrate an extensive chimaerism of pig cells in these animals after transplantation, indicating that the host's exposure to PERV infection will extend well outside the transplantation site. In fact, we have only found infection in compartments chimaeric for pig cells, suggesting that cell–cell contact may be one important mechanism for spreading infection.
It is interesting to consider the recent evidence that PERV infection was not demonstrated in humans exposed to pig cells13, 14, 17. First, PERV infection in these human studies was only determined with peripheral blood lymphocytes and serum whereas our studies involved several tissue compartments in the mouse. There is evidence suggesting that primary human peripheral blood cells cannot be productively infected with PERV in vitro3, 18. Furthermore, comparisons of host ranges for PERV consistently demonstrate increased susceptibility for human epithelial cell lines as compared to both T and B cell lines6. Secondly, most patients in the clinical series were not immunosuppressed at the time of pig cell exposure, and the majority of immunosuppressed recipients of islets did not demonstrate long-term survival and function of the grafts. Moreover, the retrospective nature of these studies did not allow consecutive cases to be analysed at set time points post-exposure so the possibility of missing evidence of infection in the peripheral blood cannot be excluded. For example, the 14 patients with pig islet transplants had early post-transplant sera tested for antibodies but peripheral blood lymphocytes were only tested at 4 to 7 years afterwards13. Nor can we exclude the possibilities that mouse tissue is more susceptible to infection than human or that pseudotyping of PERV with mouse endogenous retrovirus did not increase the host range in our model.
This study provides evidence of cross-species transmission of PERV
in vivo following pig tissue transplantation. We have created a model
to study PERV infection in an immunodeficient mouse that is permissive for
transplantation of both human and pig tissues. A human patient is unlikely
to have an immunodeficiency of equivalent magnitude and the mice do not have
natural immunity against galactose 
(1-3)-galactose terminal sugars3, 19. But pig cell chimaerism was detected in human patients, including
23 patients that were not immunosuppressed14. Moreover, successful
xenotransplantation will involve overcoming the natural immune barrier by
clinical strategies, such as engineering pigs with complementary regulatory
proteins or alterations in enzymatic pathways responsible for sugar modification
of cell-surface molecules which will also eliminate this primary immune barrier
to PERV infection19, 20. Thus, we conclude that successful pig
islet xenotransplantation to human patients may result in long-term exposure
to replication competent endogenous retrovirus. Our model supports the concern
that such an exposure will be associated with PERV infection of cells in multiple
tissue compartments.
Methods
Pig NOD/SCID chimaeras
NODlt/szSCID mice (L. D. Shultz) are maintained under pathogen-free conditions. 2,000–8,000 pig islet equivalents (IEQ) were transplanted under the kidney capsule21 or perispinous muscles in three-dimensional collagen templates coated with RGD peptide analogues. These templates were developed with Integra Life Sciences (San Diego) to protect islet structure and enhance angiogenesis. In some experiments (Fig. 2e) multiple templates (8,000 IEQ each) were placed in the perispinous muscles and removed at 3, 7 and 14 days after transplantation for quantification of PERV mRNA.
Cell lines, pig islets and cell co-cultures
U293 (human
kidney epithelial, #CRL-1573) and PK15 (pig kidney, #CCL-33) cell lines were
obtained from American Type Culture Collection. U293 transfected with a neomycin
resistance gene (U293/neo) were selected with 800 g ml
-1 G418 (Calbiochem). U293 (2 104) were co-cultured
with either PK15 (2 104) or pig islets (2,000 IEQ)
for 8 weeks. Infections of U293 by supernatants were done with conditioned
media from PERV-infected U293 (passage 10 or later), added in a 2:1 volume
with fresh media every 3 days for 12 days after which cells were cultured
for 4 weeks. Pig islets were purified from female Landrace pigs. Freshly harvested
pancreata underwent intraductal distention with solution containing 0.27–0.33%
of collagenase (Serva). Islets were isolated within one hour by a continuous
digestion–filtration device and purified on a discontinuous gradient
(Biochrom)22.
Immunohistochemistry
Detection of p30 PERV-expression was done with rabbit polyclonal anti-recombinant p30-specific antisera and donkey Fab2 FITC anti-rabbit IgG (Jackson Labs). Similar results were obtained with rabbit anti-p30 peptide-specific antisera. Insulin was stained with sheep anti-human insulin (The Binding Site) followed by donkey Fab2 LSRC anti-sheep IgG (Jackson Labs). Sections were blocked with donkey serum and secondary reagents were pre-absorbed to eliminate cross-species reactivity. Sections were analysed with a Zeiss microscope equipped with a scanning laser confocal attachment (MRC 1024, BioRad).
RT assay, PCR, RT–PCR and Southern blotting
RT activity was measured in cell supernatants using a Mn2+-dependent
RT assay (Cavidi HS-kit). Supernatants after 7 days of culture were diluted
1:15 and 1:30 and incubated overnight with an immobilized template/primer
and BrdUTP. RT activity was measured by BrdUMP incorporation using a specific
alkaline phosphatase-conjugated antibody. Results were compared to MMuLV rRT
and expressed as U ml-1. Values less than two
times the background (<30 U ml-1) were
below the limit of detection.
PERV sequences were detected by PCR with conserved pol sequence-derived
probes14: 5'-AGCTCCGGGAGGCCTACTC-3' and 5'-ACAGCCGTTGGTGTGGTCA-3'.
Pig-specific cytochrome c DNA was detected with 5'-CATTCTACGAGGTCTGTTCCG-3'
and 5'-GCCTATTCATCCACGTAGGC-3' probes (GenBank accession no. U18827).
Cycling parameters were 45 s at 64 °C, 60 s at 72 °C
and 30 s at 94 °C for 35 cycles. A pol gene-containing plasmid
was used as a positive control for PERV, and PK15 DNA as a positive control
for cytochrome c. Detection of PCR products was done by Southern or
slot blotting (Schleicher & Schuell) of DNA using digoxigenin-labelled
probes (full-length PERV pol and pig cytochrome c PCR products)
and rabbit anti-digoxigenin alkaline phosphatase (Boerhinger Mannheim). Quantitation
of PERV sequences using fluorescence-based, real-time PCR (ABI/Perkin-Elmer)
was performed in 100 l containing 5U Taq polymerase, 1U UNG,
300 nM of each primer and 100 nM of probe 5'-FAM-CCACCGTGCAGGAAACCTCGAGACT-TAMRA-3'.
Cycling parameters were 2 min at 50 °C, 10 min at 95 °C
and then 60 cycles of 15 s at 95 °C and 1.5 min at 60 °C.
The limit of detection for this assay is one PERV copy in 3.3 g
genomic DNA (500,000 cells); the sensitivity of the assay was defined as 10
copies per 500,000 cells, which gave >99.99% confidence of detecting 
10
copies14.
DNA samples were tested for the presence of porcine centromeric sequences
using primers 5'-TAGCCATGCTGCATGTAATGC-3', 5'-GGAGCGTGGCCCAAT-3'
and probe 5'-FAM-ATGCTGCATGGAATGCACTACCTTCAA-TAMRA-3'14.
Less than 10 copies in 3.3 g DNA was considered negative. The
number of PERV and centromeric sequence repeats in 43 pig liver samples was
67 7.2 (s.e.m.) and 3,694 688 per cell, respectively. Because
the number of PERV copies per cell did not conform to a gaussian distribution,
the P/C ratios were determined for each of the 43 control samples
individually. This approach yielded a mean P/C ratio of 0.042
rather than the value of 0.018, which is the simple ratio of the values
for the average PERV versus centromeric copy numbers (67/3700). Significance
of detected differences was tested by Wilcoxon Two Sample test (http://fonsg3.let.uva.nl/Service/Statistics.html).
We determined the level of a spliced form of the PERV env mRNA using primers to a conserved region. Total RNA was isolated using RNAzol B (Tel-Test) and reverse transcribed by priming with oligo d(T)16 and 75U MuLV RT (10 min at 25 °C followed by 15 min at 42 °C, then 5 min at 99 °C). cDNA was amplified using primers and probe specific for PERV env: 5'-GTTTGCATCAAGACCGCTTCT-3' (300 nM) and 5'-AGCCGTTGGTGTGGTCAAAA-3' (300 nM) and 5'-FAM-CCACCGTGCAGGAAACCTCGAGACT-TAMRA-3' (75 nM) (GenBank accession no. AF038601). Cycling parameters were 2 min at 50 °C, 10 min at 95 °C and then 50 cycles of 15 s at 95 °C and 1 min at 60 °C. A standard curve of 10 2 to 106 copies of plasmid containing spliced env cDNA allowed quantitation of starting cDNA concentrations.
