The attenuated Towne strain of human cytomegalovirus may revert to both endothelial cell tropism and leuko- (neutrophiland monocyte-) tropism in vitro

Transcription

1 Journal of General Virology (2002), 83, Printed in Great Britain... The attenuated Towne strain of human cytomegalovirus may revert to both endothelial cell tropism and leuko- (neutrophiland monocyte-) tropism in vitro Giuseppe Gerna, Elena Percivalle, Antonella Sarasini, Fausto Baldanti and M. Grazia Revello Servizio di Virologia, IRCCS Policlinico San Matteo, Pavia, Italy The Towne strain of human cytomegalovirus (HCMV), originally recovered from the urine of a congenitally infected newborn, was attenuated through 125 passages in human embryonic lung fibroblast cell cultures. Although reliable markers of attenuation were not identified, the virus was shown to be attenuated by inoculation of both healthy human volunteers and immunocompromised patients. More recently, Towne (like other laboratory-adapted strains) was shown not to have two biological properties typical of recent clinical isolates: endothelial cell tropism and polymorphonuclear leukocyte tropism. These markers of attenuation are lost by all clinical isolates on extensive propagation in cell cultures and are apparently associated with one another. Here, we show that Towne may reacquire both endothelial cell tropism and leuko- (polymorphonuclear- and monocyte-) tropism on adaptation to growth in endothelial cell cultures. However, reversion to endothelial cell tropism is dissociated from reversion to leukotropism, since the latter was reacquired passages later. Thus, these two biological properties, which were considered to be encoded by the same viral gene(s), appear to be distinct. Both restriction fragment length polymorphism and Southern blot analysis demonstrated the identity of the attenuated and endothelial cell tropic variants of Towne, thus suggesting that only minor variations (mutations) of the viral genome may be responsible for loss or reacquisition of the two biological properties. Viral genes involved in endothelial cell tropism and leukotropism remain to be identified. However, reversion of attenuated strains to pathogenicity in vivo cannot be excluded a priori. Introduction The Towne strain of human cytomegalovirus (HCMV) was initially isolated in 1970 in WI-38 cell cultures from the urine of a 2-month-old infant with microcephaly and hepatosplenomegaly (Plotkin et al., 1975). The first 10 passages in WI-38 were accomplished using cell-associated virus, while the following 28 passages were performed using cell-free virus released into the tissue culture supernatant fluid. Subsequently, virus was propagated until passage 125 by inoculating onto WI-38 cell monolayers cell-free virus suspensions obtained by pooling at each passage virus released from sonicated cells with the relevant infected cell supernatant. Potential markers of attenuation were identified as follows: (i) the ratio of virus released into the cell culture medium to the number of infected cells, ranging from 0 1 to1 0 infectious virus per infected cell for high-passage strains to for low-passage fresh Author for correspondence: Giuseppe Gerna. Fax g.gerna smatteo.pv.it isolates; (ii) thermostability for 24 h at 4 C for high-passage strains, in contrast to fresh isolates which are highly unstable at that temperature; (iii) resistance to trypsin treatment for highpassage strains in contrast to sensitivity of low-passage viruses (Plotkin et al., 1975). However, due to the lack of an animal model, these hypothetical in vitro correlates of in vivo attenuation could only be tested in humans. Indeed, that the Towne strain was truly attenuated was shown in volunteers, in whom the virus subcutaneous inoculation did not cause systemic symptoms, virus excretion or clinical laboratory abnormalities (Plotkin et al., 1976, 1989; Fleisher et al., 1982; Quinnan et al., 1984). In contrast, a low-passage virus named Toledo, isolated from the urine of a congenitally infected infant from Toledo (NJ, USA), when inoculated subcutaneously to seronegative individuals as cell-free virus, following five cell-to-cell passages in MRC-5 human fibroblasts, caused a mild mononucleosis syndrome associated with atypical lymphocytosis, increase in liver enzyme levels and virus shedding (Quinnan et al., 1984; Plotkin et al., 1989). Recently the Towne strain, like other high-passage strains SGM BJJD

2 G. Gerna and others such as AD169 or Davis, has been shown to lack two biological properties which have been found to be shared by all of the recent clinical HCMV isolates thus far tested (more than 100 strains), i.e. endothelial cell (EC) tropism and leukotropism (Revello et al., 1998, 2001; Gerna et al., 2000, 2002a). Endothelial cell tropism is the ability of an HCMV strain to infect and to grow in endothelial cells and, in particular, in human umbilical vein endothelial cells (HUVEC), while leukotropism is the ability of an HCMV strain to enter peripheral blood leukocytes and to replicate either abortively, as in polymorphonuclear leukocytes (PMNL) or productively, as in monocytes macrophages, according to several reports (Lathey & Spector, 1991; Ibanez et al., 1991; Taylor-Wiedeman et al., 1991, 1994; So derberg et al., 1993; So derberg-naucle r et al., 1997, 2001; Fish et al., 1995, 1996; Gerna et al., 2000). In addition, we have shown that all clinical HCMV isolates lose both biological properties after extensive propagation in human fibroblast cell cultures, thus documenting that modifications of biological properties of laboratory-adapted HCMV strains had occurred during propagation in cell cultures (Revello et al., 2001). Loss of endothelial cell tropism and loss of leukotropism have been proposed as potential in vitro correlates of a true in vivo attenuation in terms of pathogenicity (Gerna et al., 2002a). Reversion of high-passage (laboratory-adapted) HCMV strains to endothelial cell tropism and leukotropism has never been fully investigated. In the present report, we describe the adaptation to growth in HUVEC of two different preparations of the Towne strain, which reacquired leukotropism more than 20 passages after initial adaptation to growth in HUVEC. Thus, endothelial cell tropism and leukotropism, which have been so far found to be apparently associated with one another, are two distinct biological properties. Methods Cell cultures. Human embryonic lung fibroblast (HELF) cell cultures were derived from a cell strain developed in the laboratory in 1980 and were used at passages HUVEC were obtained by trypsin treatment of umbilical cord veins and used at passages 2 5 as reported (Revello et al., 1998). Human umbilical artery endothelial cells (HUAEC) were obtained similarly by trypsin treatment of arteries of the same umbilical cords and used within the same range of passages. All EC preparations were tested for HCMV DNA by nested PCR, as reported (Gerna et al., 1998b). Virus strain. The HCMV isolate VR6110, originally recovered from the blood of a patient with AIDS, was previously adapted to growth in HUVEC and was initially used as a reference HUVEC-tropic and PMNL-tropic low-passage strain (Revello et al., 1998). Subsequently, a new low-passage HCMV strain, originally recovered from cervical secretions of an immunocompetent pregnant woman (VR1814), was used routinely as a reference HUVEC-tropic PMNL-tropic HCMV strain (Revello et al., 2001). The three laboratory-adapted HCMV strains Towne (originally obtained from E. Go nczo l, Wistar Institute, PA, USA), AD169 and Davis (both obtained from ATCC) were used as reference high-passage HCMV strains grown in human fibroblasts and lacking both HUVEC tropism and PMNL tropism (Revello et al., 1998, 2001). Two preparations of the Towne strain were used for adaptation of the virus to growth in HUVEC, both at passages between 129 and 134 in human fibroblasts: one obtained from E. Go nczo l (Wistar) on MRC-5, and the other one from RIT (Genval, Belgium). The former was inoculated onto HUVEC as cell-associated virus, the latter as cell-free virus. Finally, the Toledo strain (currently referred to as the reference wild-type or lowpassage strain), also kindly provided by E. Go nczo l (Wistar), and shown recently to lack both HUVEC tropism and PMNL tropism (Gerna et al., 2002a), was tested as a reference strain. Assay for HUVEC tropism. Two protocols were used to test for endothelial cell tropism (Gerna et al., 2002b). The first was relevant to transmission of HELF-associated virus to HUVEC (Revello et al., 2001). HELF were infected with Towne Wistar (Towne-W) at an m.o.i. of 1 5. Following a 7 day incubation at 37 C, infected HELF showing 100% CPE were trypsinized and inoculated at a ratio of 1:3 onto confluent monolayers of uninfected HUVEC grown in 24-well plates. After a further 7 days of incubation, infected HUVEC were trypsinized and mixed at a ratio of 1:2 with uninfected HUVEC. This procedure was repeated weekly until passage 6, when cells were sonicated and cell-free virus propagated. Virus growth in HUVEC was checked 7 days p.i., at each passage, by immunofluorescence using monoclonal antibodies to the major immediate-early (IE) protein p72 or glycoprotein B (gb) (Gerna et al., 1990). Monoclonal antibodies to gb were kindly provided by Lenore Pereira (UCSF, CA, USA). The second protocol was relevant to infection of HUVEC with cell-free Towne RIT (Towne-RIT) virus preparation grown in HELF at an m.o.i. of 5. Weekly passages of infected HUVEC followed the same protocol. Virus growth was checked 7 days p.i. by immunofluorescence using monoclonal antibodies as reported above. The degree of infection was determined subjectively by light microscopy following counterstaining with % Evans blue. Assays for PMNL and monocyte tropism. Concentrated PMNL preparations from healthy blood donors were cocultured for 3 h with infected HUVEC h p.i. (Revello et al., 1998). Following coculture, to separate PMNL from infected HUVEC detached from the growth surface, cell suspensions were placed for 3 h at 37 C in the upper compartment of a cell culture device separated by a transwell filter (5 µm pore size, Costar) from the lower compartment containing 10 M N- formyl-met-leu-phe-ala (FMLP, Sigma) according to a reported procedure (Revello et al., 1998; Gerna et al., 1998a). In these conditions, PMNL migrate to the lower compartment, thus separating from contaminating HUVEC and reaching a level of purity comparable to that of fluorescence-activated cell sorting (Revello et al., 1998). In parallel, Ficoll-prepared peripheral blood mononuclear cell (PBMC) suspensions were separated in the monocyte subpopulation by purification through a Percoll gradient. Following overnight coculture according to the above protocol, monocytes were separated from infected HUVEC by the same procedure used for PMNL, except for a higher concentration of FMLP (10 ) in the lower compartment of the transwell device. PMNL and monocyte suspensions were then tested for presence of pp65 in cytospin preparations of 2 10 cells, according to a procedure originally reported for monitoring of HCMV antigenaemia in blood of immunocompromised patients (Gerna et al., 1992) and more recently standardized (Gerna et al., 1998a). Cytospin preparations of either PMNL or monocytes were examined for pp65 antigen following immunofluorescent staining with a pool of three monoclonal antibodies reactive to different epitopes of HCMV pp65 (Gerna et al., 1992). In addition, monocyte preparations were examined for p72 and gb. BJJE

3 Reversion of HCMV Towne to leukotropism Restriction fragment length polymorphism (RFLP) analysis. Genomic regions UL54, UL97 and UL123 of the HCMV strains examined were amplified by PCR using the pairs of primers previously reported (Revello et al., 2001). PCR products were then cleaved using two to four of the following endonucleases: HaeIII, MspI, HinP1I, AluI and BstUI (New England Biolabs). RFLP patterns were compared by agarose gel electrophoresis. Southern blot analysis. The genomes of Towne-W variants and Toledo were digested with EcoRI, HindIII and BamHI, blotted onto nylon membranes (Boehringer Mannheim) and hybridized using a set of cosmid probes from the VR6110 genome library spanning almost (94 98%) the entire virus genome. The VR6110 cosmid library was constructed as previously reported (Revello et al., 2001). Growth of Towne strain in human umbilical artery endothelial cells (HUAEC). HUAEC were prepared from the same umbilical cords used for HUVEC preparations. However, the cell yield for HUAEC was markedly lower compared to HUVEC. Cells could be propagated and used within 4 5 passages. Virus growth in HUAEC was monitored, as reported for HUVEC, by looking for CPE and immunostaining cell cultures with monoclonal antibodies to both p72 and gb. Results Growth of the Towne strain in HUVEC with dissociation of EC tropism and leukotropism Thus far, the following findings have been repeatedly reported: (i) all recent clinical HCMV isolates are leukotropic in HELF, where they lose this property following several tens of passages (Revello et al., 2001); (ii) all HCMV strains thus far tested which are leukotropic in HELF, are able to grow in HUVEC either using HELF-associated or HELF-free virus or leukocyte-associated virus as inoculum (Gerna et al., 2002b); (iii) all HCMV strains adapted to growth in HUVEC are also leukotropic in HUVEC and do not appear to lose this property as in HELF; (iv) following loss of leukotropism during propagation in HELF, as shown to occur for parental strains, plaque isolates and laboratory-adapted HCMV strains, adaptation to growth in HUVEC no longer seemed to occur (Revello et al., 2001). Surprisingly, when we tried to make the Towne strain grow in HUVEC, attempts were successful either when using the HELF-associated or the HELF-free virus preparation (Table 1 and Fig. 1). Using the cell-associated Towne-W preparation originally obtained from Wistar, the adaptation process to HUVEC was slow and the number of infected cells decreased after the first passage until passage 6, when cells were sonicated and cell-free virus propagated (Table 1). Afterwards, the aliquot of infected cells was about 1% at passage 10, 10% at passage 20 and 50% after passage 25. PMNL-tropism was negative until passage 20, and became positive at a low level at passage 25 when about 50% of HUVEC were infected, reaching medium and high levels at passage 30 and 35 in HUVEC when infected Table 1. Sequential reversion to HUVEC tropism and leukotropism of two HELF-derived virus preparations of the Towne strain lacking both biological properties N, negative; ND, not done; L, low ( cells) number of pp65-positive leukocytes; M, medium (11 100) number; H, high ( 100) number. HUVEC, human umbilical vein endothelial cells; PMNL, polymorphonuclear leukocytes; HELF, human embryonic lung fibroblast cell strain established in the laboratory in Sequential Leukotropism passages on % HCMV-infected Virus preparation HUVEC HUVEC PMNL Monocytes Towne-W (Wistar) HUVEC N ND MRC-5 2 HELF N ND (cell-associated virus) N ND N ND N ND L ND M ND H ND Towne-RIT HELF 12 HUVEC N N (cell-free virus) N N N N N N N N N N H H H H BJJF

4 G. Gerna and others Fig. 1. Adaptation to growth in HUVEC of the cell-free Towne-RIT preparation. (A) Passage 1, rare foci of virus infection. (B, C) passages 2 and 3 showing progressive increase in the number of infected cells. (D, E) Passages 4 and 5, showing a trend toward generalized spreading of the infection. (F) Passage 8, at a lower magnification (150 vs 600 of panels A E), showing dissemination of the infection to the entire cell monolayer. Indirect immunofluorescent staining at 7 days p.i. using a pool of monoclonal antibodies to immediate-early protein p72 and gb (Gerna et al., 1990). cells were 60% and 80%, respectively. It is important to recall that we could determine that, when using a recent clinical isolate, even in the presence of a HUVEC monolayer infected less than 1 %, a fair number of pp65-positive PMNL following coculture was detected (data not reported). Monocyte-tropism was not tested during this experiment. Fig. 2. In vitro generated pp65-positive (A, B) polymorphonuclear leukocytes and (C F) monocytes following coculture with HUVEC infected with (A, C) a recent clinical isolate or (B, D F) the Towne-RIT virus strain. Immunofluorescent staining with a pool of (A D) pp65-specific (Gerna et al., 1992), (E) p72-specific and (F) gb-specific monoclonal antibodies. (A D) 600 ; (E, F) When using the cell-free Towne-RIT virus preparation, the number of infected HUVEC, which was very low upon first passage, rapidly increased till reaching about 30% and 50% of the cell monolayer at passage 4 and 5, respectively. The great majority of HUVEC were infected after passage 10. However, until at least passage 20, no PMNL nor monocyte was found to be pp65-positive after coculture, thus documenting a striking dissociation between EC tropism and leukotropism (Table 1). The rapid process of adaptation of Towne-RIT to BJJG

5 Reversion of HCMV Towne to leukotropism A Pol 1-2 Haelll Mspl Hinp1l MW MW B Pol 3-4 Haelll Mspl Hinp1l MW MW Fig. 3. Adaptation to growth in HUAEC of the cell-free Towne-RIT preparation. (A) Passage 1, single infected cells. (B) Passage 3, multiple single infected cells. (C) Passage 5, several cells are infected throughout the cell monolayer.(d) Passage 8, at a lower magnification (150 vs 600 of panels A C), showing spread of the infection to the majority of the cell monolayer. Indirect immunofluorescence staining using a pool of monoclonal antibodies to immediate-early protein p72 and gb. growth in HUVEC during the very first passages is documented in Fig. 1, where a large number of HUVEC are already productively infected in panel (F) corresponding to passage 8. Starting at passage 25, coculture of Towne-RIT-infected HUVEC with either PMNL or monocytes yielded a high number of pp65-positive cells of either leukocyte subpopulation (Fig. 2), while the relative number of positive cells of the two subpopulations was comparable, suggesting that the efficiency of transmission to PMNL and monocyte was similar. In addition, following overnight coculture with infected HUVEC, monocytes were positive for IE protein p72 in the nucleus, while showing granules of gb in the cytoplasm. To verify whether the Towne strain was able to grow in HUAEC as well as in HUVEC, cell-free Towne-RIT virus preparation was inoculated onto arterial EC from umbilical cord arteries as well (Fig. 3). The adaptation process occurred similarly to that observed in HUVEC, although at passage 8 the number of infected cells appeared to be slightly (10 20%) lower (Fig. 3). Fig. 4. RFLP analysis of two HCMV DNA polymerase fragments, (A) Pol1 Pol2 (nt ) and (B) Pol3 Pol4 (nt ), digested with HaeIII, MspI and HinP1I endonucleases. RFLP profiles are identical for Towne-W F ( ), Towne-W E ( ), Towne-RIT F (Tow- RIT F ) and Towne-RIT E (Tow-RIT E ) in all combinations, while Toledo-W F ( ) and are differentiated from Towne variants in (A) by Hinp1I digestion and in B are differentiated from each other and from Towne variants by HaeIII and Hinp1I digestion. Genetic identity of the Towne strain grown in HELF (Towne F ) and in HUVEC (Towne E ) as determined by RFLP and Southern blot analysis RFLP analysis of both attempts at adaptation to growth in HUVEC of the two different preparations of the Towne strain showed the apparent identity of the two parental variants grown in HELF (Towne-W F and Towne-RIT F ) and the two subsequent EC tropic variants, Towne-W E and Towne-RIT E, the former negative and the latter positive for leukotropism (Fig. 4A, B). On the basis of previously acquired experience, we believe that RFLP analysis of three different genomic regions with multiple restriction enzymes may be sufficient to consider the different Towne preparations examined as truly identical. BJJH

6 G. Gerna and others A Kb EcoRI BamHI EcoRI BamHI EcoRI BamHI EcoRI BamHI EcoRI BamHI EcoRI BamHI EcoRI BamHI B Kb UL ULb J US ULb ULb Fig. 5. Southern blot analysis of Towne variants. (A) BamHI and EcoRI digestion profiles of Toledo-W F ( ), Towne-W F ( ) and Towne-W E ( ) DNA following hybridization with a cosmid set spanning the entire HCMV genome. (B) Schematic of HCMV genome and position of cosmid probes (derived from cloning of the entire genome of a clinical isolate, VR6110) used in the Southern blot analysis (Revello et al., 2001). This identity was confirmed to a larger extent by Southern blot analysis, spanning the entire Towne genome (Fig. 5A, B). These findings show unequivocally that: (i) no contamination with other virus strains was responsible for the acquisition of the new biological properties; (ii) only minor mutations in the viral genome were likely to be responsible for the reversion to biological properties commonly shared by recent clinical isolates; (iii) this is the first reported case of a laboratory strain reverting to potential markers of in vivo pathogenicity. Discussion All recent clinical HCMV isolates thus far tested have been shown to be leukotropic in HELF, i.e. to be able to be transferred to leukocytes following coculture with infected cells (Revello et al., 1998; Gerna et al., 2000). This has been repeatedly shown for PMNL, but recently transfer of virus and viral products has also been shown to occur from infected HELF to monocytes (unpublished data). As shown here, monocytes, following overnight coculture with Towne-infected HUVEC, may harbour p72 in the nucleus, besides pp65, and granules of gb in the cytoplasm. While nuclear p72 may represent early stages of virus replication, gb must be considered as passively transferred from infected HUVEC, possibly according to a mechanism already described for transfer of viral and cellular material from infected cells to PMNL (Gerna et al., 2000). Thus far, leukotropism has been found to be consistently associated with EC tropism or HUVEC tropism (Revello et al., 2001). Although adaptation to growth in HUVEC could be achieved by propagating in EC either cell-associated or cell-free virus from infected HELF, the best approach to adaptation to growth in HUVEC was to inoculate onto EC PBL carrying infectious virus either when virus was recovered directly from blood or when virus recovered from clinical samples other than blood was first transferred from infected HELF to PBL from healthy blood donors and then inoculated onto HUVEC (Gerna et al., 2002b). The latter procedure was considered to select virus variants able to grow in EC within a heterogeneous viral population, in which the relative number of EC-tropic variants was low or very low when the clinical isolate was derived from a clinical source other than blood, while it was more abundant when virus was isolated directly from blood (Gerna et al., 2002b). On the other hand, all viral variants selected in vitro after extensive propagation in HELF and shown to be unable to infect leukocytes were also shown to be unable to grow in BJJI

7 Reversion of HCMV Towne to leukotropism HUVEC (Revello et al., 2001). According to these findings, laboratory-adapted HCMV strains such as Towne were considered to lack both biological properties, EC tropism and leukotropism. Unexpectedly, the results of the present study indicate that two different preparations of the Towne strain, both following extensive propagation in HELF, were able to grow readily in HUVEC after a few passages, thus suggesting that EC-tropic viral variants are still represented in the Towne strain preparations after more than 100 passages in human fibroblasts. The apparent identity at the molecular level (as shown by both RFLP and Southern blot analysis) of the Towne strain extensively propagated in human fibroblasts and shown to have lost leukotropism and the Towne strain grown in HUVEC and shown to have reacquired leukotropism, indicated that only minor genomic mutations should be responsible for the reversion of a laboratory attenuated virus to biological properties commonly shared by recent clinical isolates. To explain this reversion, the hypothesis of selection in vitro of viral variants previously represented in a minor proportion of the population appears more plausible than that of the sudden emergence of a new mutation rapidly representing the majority of the viral population. However, the result which was most unexpected in this study was the dissociation (observed for the first time) of EC tropism and leukotropism. In fact, the two properties were found to be clearly distinct during the reversion process, with leukotropism being reacquired a number of passages after adaptation of Towne to growth in HUVEC. This dissociation was observed during adaptation of both Towne preparations, suggesting that different viral gene functions might be involved in the expression of the two biological properties. However, adaptation to growth in HUVEC appears to precede leukotropism, thus suggesting a link between the two properties. It is well known that, when administered to humans, the Towne vaccine has never been found to be pathogenic nor to undergo latency (Plotkin et al., 1976, 1989; Quinnan et al., 1984). However, if leukotropism and EC tropism, as recently suggested (Gerna et al., 2002a), are in vitro correlates of in vivo pathogenicity, data herein reported do not seem to exclude the risk of reversion of the Towne strain to pathogenicity when administered as a vaccine to humans. In fact, replication in EC during primary infection of the immunocompetent or during both primary and reactivated infections of the immunocompromised host may be one of the first steps preceding virus dissemination in vivo through interaction of EC with circulating leukocytes, both PMNL and monocytes (Grundy et al., 1998; Revello et al., 1998; Gerna et al., 2000). Recently the Toledo strain, which is currently considered the prototype of wild-type HCMV strains, has been unexpectedly reported to lack both leukotropism and EC tropism in three different virus preparations from three different sources (Gerna et al., 2002a), in contrast to the large number of recent clinical isolates tested and sharing consistently both properties (Revello et al., 2001). However, low-passage Toledo has been actually shown to be pathogenic when inoculated in seronegative immunocompetent individuals (Quinnan et al., 1984; Plotkin et al., 1989). Whether Toledo strain lacking leukotropism and EC tropism is a true low-passage virus is uncertain at this time. Furthermore, results consistently obtained with the other virus strains representing true recent clinical isolates suggest that it is likely that Toledo strain preparations now circulating within laboratories most likely are no longer low-passage viruses. On this basis, we were interested in analysing at a molecular level the Toledo strain which has recently been reported to be adapted to growth in HUVEC (Grundy et al., 1998). We used both RFLP and Southern blot analysis to demonstrate that the Toledo strain grown in HUVEC was not the true Toledo strain, but instead was likely to be an entirely different HCMV strain, a new virus strain resulting from recombination of Toledo with a new HCMV strain (F. Baldanti, M. G. Revello, E. Percivalle, N. Labo & G. Gerna, unpublished). Thus, these data further confirm that Toledo strain is not per se able to grow in HUVEC, as shown in our laboratory by consistently unsuccessful adaptation attempts. In conclusion, we believe that leukotropism and EC tropism, which are lost following extensive propagation in HELF of laboratory-adapted HCMV strains, such as the Towne vaccine strain, may be reacquired if EC-tropic viral variants are still represented in HELF-derived viral populations. However, reversion to the ability to grow in HUVEC is not simultaneously accompanied by reversion to leukotropism, which is regained a series of passages afterwards. These findings suggest that a vaccine strain such as Towne may reacquire biological markers that are typical of wild-type HCMV strains. This work was partially supported by Ministero della Salute, Istituto Superiore di Sanita, Programma Nazionale di Ricerca sull AIDS, Ricerca Finalizzata IRCCS Policlinico San Matteo (grant 820RFM99 01) and IRCCS L. Spallanzani (grant ICS120.5 RF00 124), and Ricerca Corrente IRCCS Policlinico San Matteo (grant 80206). The authors thank Linda D Arrigo for revision of the English, and the Personnel of the Obstetrics and Gynecology ward of the Clinic Citta di Pavia, together with Nazarena Labo and Daniele Lilleri, for providing umbilical cords. In addition, the authors are indebted to the entire technical personnel of Servizio di Virologia, and to Daniela Sartori for typing the manuscript. References Fish, K. N., Depto, A. S., Moses, A. V., Britt, W. & Nelson, J. A. (1995). Growth kinetics of human cytomegalovirus are altered in monocytederived macrophages. Journal of Virology 69, Fish, K. N., Britt, W. & Nelson, J. A. (1996). A novel mechanism for persistence of human cytomegalovirus in macrophages. Journal of Virology 70, Fleisher, G. R., Starr, S. E., Friedman, H. M. & Plotkin, S. A. (1982). Vaccination of pediatric nurses with live attenuated cytomegalovirus. American Journal of Diseases of Children 136, BJJJ

8 G. Gerna and others Gerna, G., Revello, M. G., Percivalle, E., Zavattoni, M., Parea, M. & Battaglia, M. (1990). Quantification of human cytomegalovirus viremia by using monoclonal antibodies to different viral proteins. Journal of Clinical Microbiology 28, Gerna, G., Revello, M. G., Percivalle, E. & Morini, F. (1992). Comparison of different immunostaining techniques and monoclonal antibodies to the lower matrix phosphoprotein (pp65) for optimal quantitation of human cytomegalovirus antigenemia. Journal of Clinical Microbiology 30, Gerna, G., Percivalle, E., Torsellini, M. & Revello, M. G. (1998a). Standardization of the human cytomegalovirus antigenemia assay by means of in vitro-generated pp65-positive peripheral blood polymorphonuclear leukocytes. Journal of Clinical Microbiology 36, Gerna, G., Zavattoni, M., Baldanti, F., Sarasini, A., Chezzi, L., Grossi, P. & Revello, M. G. (1998b). Human cytomegalovirus (HCMV) leukodnaemia correlates more closely with clinical symptoms than antigenemia and viremia in heart and heart-lung transplant recipients with primary HCMV infection. Transplantation 65, Gerna, G., Percivalle, E., Baldanti, F., Sozzani, S., Lanzarini, P., Genini, E., Lilleri, D. & Revello, M. G. (2000). Human cytomegalovirus replicates abortively in polymorphonuclear leukocytes after transfer from infected endothelial cells via transient microfusion events. Journal of Virology 74, Gerna, G., Percivalle, E., Baldanti, F. & Revello, M. G. (2002a). Lack of transmission to polymorphonuclear leukocytes and human umbilical vein endothelial cells as a marker of attenuation of human cytomegalovirus. Journal of Medical Virology 66, Gerna, G., Percivalle, E., Sarasini, A. & Revello, M. G. (2002b). Human cytomegalovirus and human umbilical vein endothelial cells: restriction of primary isolation to blood samples and susceptibilities of clinical isolates from other sources to adaptation. Journal of Clinical Microbiology 40, Grundy, J. E., Lawson, K. M., MacCormac, L. P., Fletcher, J. M. & Yong, K. L. (1998). Cytomegalovirus-infected endothelial cells recruit neutrophils by the secretion of C-X-C chemokines and transmit virus by direct neutrophil-endothelial cell contact and during neutrophil transendothelial migration. Journal of Infectious Diseases 177, Ibanez, C. E., Schrier, R., Ghazal, P., Wiley, C. & Nelson, J. A. (1991). Human cytomegalovirus productively infects primary differentiated macrophages. Journal of Virology 65, Lathey, J. L. & Spector, S. A. (1991). Unrestricted replication of human cytomegalovirus in hydrocortisone-treated macrophages. Journal of Virology 65, Plotkin, S. A., Furukawa, T., Zygraich, N. & Huygelen, C. (1975). Candidate cytomegalovirus strain for human vaccination. Infection and Immunity 12, Plotkin, S. A., Farquhar, J. & Hornberger, E. (1976). Clinical trials of immunization with the Towne 125 strain of human cytomegalovirus. Journal of Infectious Diseases 134, Plotkin, S. A., Starr, S. E., Friedman, H. M., Go nczo l, E. & Weibel, R. E. (1989). Protective effects of Towne cytomegalovirus vaccine against low-passage cytomegalovirus administered as a challenge. Journal of Infectious Diseases 159, Quinnan, G. V., Jr, Delery, M., Rook, A. H., Frederick, W. R., Epstein, J. S., Manischewitz, J. F., Jackson, L., Ramsey, K. M., Mittal, K., Plotkin, S. A. & Hilleman, M. R. (1984). Comparative virulence and immunogenicity of the Towne strain and a nonattenuated strain of cytomegalovirus. Annals of Internal Medicine 101, Revello, M. G., Percivalle, E., Arbustini, E., Pardi, R., Sozzani, S. & Gerna, G. (1998). In vitro generation of human cytomegalovirus pp65 antigenemia, viremia, and leukodnaemia. Journal of Clinical Investigation 101, Revello, M. G., Baldanti, F., Percivalle, E., Sarasini, A., De Giuli, L., Genini, E., Lilleri, D., Labo, N. & Gerna, G. (2001). In vitro selection of human cytomegalovirus variants unable to transfer virus and virus products from infected cells to polymorphonuclear leukocytes and to grow in endothelial cells. Journal of General Virology 82, So derberg, C., Larsson, S., Bergstedt-Lindqvist, S. & Mo ller, E. (1993). Definition of a subset of human peripheral blood mononuclear cells that are permissive to human cytomegalovirus infection. Journal of Virology 67, So derberg-naucle r, C., Fish, K. N. & Nelson, J. A. (1997). Reactivation of latent human cytomegalovirus by allogeneic stimulation of blood cells from healthy donors. Cell 91, So derberg-naucle r, C., Streblow, D. N., Fish, K. N., Allan-Yorke, J., Smith, P. P. & Nelson, J. A. (2001). Reactivation of latent human cytomegalovirus in CD14+ monocytes is differentiation dependent. Journal of Virology 75, Taylor-Wiedeman, J., Sissons, J. G., Borysiewicz, L. K. & Sinclair, J. H. (1991). Monocytes are a major site of persistence of human cytomegalovirus in peripheral blood mononuclear cells. Journal of General Virology 72, Taylor-Wiedeman, J., Sissons, P. & Sinclair, J. (1994). Induction of endogenous human cytomegalovirus gene expression after differentiation of monocytes from healthy carriers. Journal of Virology 68, Received 15 February 2002; Accepted 19 April 2002 CAAA