Construction of a Gammaretrovirus with a Novel Tropism and Wild-Type Replication Kinetics Capable of Using Human APJ as Entry Receptor

Abstract

We have constructed a replication-competent gammaretrovirus (SL3-AP) capable of using the human G-protein-coupled receptor hAPJ as its entry receptor. The envelope protein of the virus was made by insertion of the 13-amino-acid peptide ligand for hAPJ, flanked by linker sequences, into one of the variable loops of the receptor binding domain of SL3-2, a murine leukemia virus (MLV) that uses the xenotropic-polytropic virus receptor Xpr1 and which has a host range limited to murine cells. This envelope protein can utilize hAPJ as well as murine Xpr1 for entry into host cells with equal efficiencies. In addition, the SL3-AP virus replicates in cells expressing either of its receptors, hAPJ and murine Xpr1, and causes resistance to superinfection and downregulation of hAPJ in infected cells. Thus, SL3-AP is the first example of a retargeted replication-competent retrovirus, with replication characteristics and receptor interference properties similar to those of natural isolates.

INTRODUCTION

Enveloped viruses gain access to the cellular replication machinery of their target cells through fusion of their viral and cellular membranes. In retroviruses, this process is mediated by the viral envelope protein, a single gene product encoded by the viral genome. It is cleaved into two subunits, which remain associated through noncovalent interactions or in some cases through a disulfide bridge. The larger subunit, known as SU (surface subunit), is mostly responsible for binding to the cellular receptor(s), whereas the C-terminally encoded smaller subunit, TM (transmembrane subunit), contains the fusion machinery that ultimately fuses the viral and cellular membranes.

The fusion process involves complex conformational changes in TM, including formation of an elongated triple helix which inserts a fusion peptide into the target membrane and the subsequent pulling of the membranes together. The high activation energy of bringing the two charged hydrophilic surfaces of the membranes close together is overcome by the potential energy stored in the TM subunit. On the surface of the virus, TM is in a metastable conformation, which is kinetically arrested through association with the larger SU. Receptor binding triggers SU dissociation and allows TM to begin the stepwise transformation into its stable conformation. The released energy is used to overcome the activation energy of the membrane fusion (reviewed in references 10 and 12).

Receptor binding is a key step in membrane fusion mediated by the envelope protein and one of the most important determinants of viral tropism. Different closely related viruses use different cellular proteins as entry receptors (3, 18, 24). This suggests that it is theoretically feasible to change the tropism of a virus by redirecting the affinity of its envelope protein to a specific cellular protein, although in practice this has been proven very difficult. The ecotropic murine leukemia viruses (MLVs) have been extensively used for retargeting experiments, since their limited tropism to rodent cells makes them very suitable experimental systems (2, 29). Several attempts to retarget ecotropic MLV through insertion of single-chain antibodies into SU have failed mainly because the resulting envelopes cannot induce the fusion of membranes after binding to the new receptors. In other cases, insertion of peptide ligands into SU has been used to confer new tropism, usually resulting in very inefficient retargeted envelope proteins (5, 13, 17, 21, 31). However, two cases of efficient targeting have been reported, both based on ecotropic MLVs. In one case, CXCR4 was the targeted receptor, and a titer similar to those of wild-type (wt) viruses has been achieved in one specific cell line expressing the CXCR4 receptor, while in other receptor-expressing cell lines the infection efficiency was 100-fold lower (25). In the other case, wild-type-like efficiency of infection through the somatostatin receptor was achieved at the cost of the ability to use the ecotropic receptor (19). In a third case, selection of a library based on the feline leukemia virus (FeLV) envelope protein yielded a chimeric envelope that uses HuPAR-1 as entry receptor, which is also used by porcine endogenous retrovirus A (PERV-A). Interestingly, this chimeric FeLV envelope protein has also lost its ability to use its native receptor for entry to feline cells (22, 23).

We have previously shown that insertion of apelin, a small peptide ligand for the G-protein-coupled receptor (GPCR) APJ, into one of the variable loops of the receptor binding domain (RBD) of the ecotropic Moloney MLV results in a modest affinity for APJ when flanked by flexible linkers (5). APJ was chosen as an experimental target for several reasons. For one, it functions as a coreceptor for HIV (11) and has a physiological peptide ligand, apelin, the C-terminal 13 amino acids of which bind to APJ with high affinity (28). Interestingly, the binding of apelin and that of HIV envelope map to the same area of the receptor (33).

Here, we use the SL3-2 envelope protein as a novel scaffold for construction of a retargeted envelope protein. This envelope protein uses the xenotropic-polytropic receptor (Xpr1) for entry (9, 27, 32) but, unlike close relatives, is unable to infect nonrodent cells, providing the same benefits as the ecotropic envelope proteins for retargeting experiments (4). We find that this new chimeric envelope protein is able to mediate entry through APJ very efficiently. Interestingly, the retargeted envelope protein still retains its ability to infect murine cells through Xpr1 and thus represents an example of an envelope protein that uses two different receptors with equal efficiency. Furthermore, we show that a complete virus genome (SL3-AP) containing this chimeric envelope protein is able to replicate in nonmurine cells in an APJ-dependent manner and that the infected cells show downregulation of surface APJ. Thus, SL3-AP has the same infection characteristics that would be expected from a wild-type isolate and to our knowledge represents the first replicating virus with an artificially determined tropism.

MATERIALS AND METHODS

Envelope design and titer measurements.

The envelope constructs were cloned in a bicistronic vector described elsewhere (7) using standard cloning techniques. The sequences around the insertion sites are as follows (apelin-13 and the linker sequences are underlined): AP@86 (DWDSGGSGQRPRLSHKGPMPFSGGSGGLG), AP@155 (GNTQGIYQCSGGSGQRPRLSHKGPMPFSGGSGCGPCY), and AP@165 (YDSSGGSGQRPRLSHKGPMPFSGGSGSSS).

The resulting vectors were transfected into HEK293T cells together with a gag-pol expression plasmid, and the resulting virions were used to transduce CeB semipackaging cell lines, which subsequently produced virions containing vectors bearing both neo and env genes (7). The virions were transferred to target cells in serial 10-fold dilutions. After 24 h, the cells were subjected to selection in medium containing 600 μg/ml of G418 until colonies appeared, which were counted and used to calculate the titers in CFU. All titer measurements were reproduced in at least three independent experiments.

Cell cultures.

NIH 3T3 and the semipackaging cell line (CeB) (7) were grown in Dulbecco's modified Eagle's medium with Glutamax-1 and 10% (vol/vol) newborn calf serum. HEK293T cells were grown in the same medium containing 10% fetal calf serum. D17 and D17 APJ (5) cells were grown in minimal essential medium alpha (MEM-alpha) supplemented with Glutamax-1 and 10% fetal calf serum. All growth media contained 100 U of penicillin/ml and 100 μg of streptomycin/ml. All cells were incubated at 37°C in 90% relative humidity and 5.0% CO₂.

Flow cytometry analysis.

The cells were released from growth plates, washed using phosphate-buffered saline (PBS) containing 2% serum, and incubated with the primary antibody (200 μl of supernatant from the 83A25 hybridoma cell line [14] for envelope labeling or 40 μl of anti-hAPJ antibody [MAB856; R&D Systems] for APJ labeling) at 4°C for 45 min, followed by two washes and 45 min of incubation with the secondary antibody (5 μl of 1:40-diluted goat anti-rat Ig R-phycoerythrin [PE] conjugate [Southern Biotechnology Associates, Inc.] for envelope labeling and 4 μl fluorescein isothiocyanate [FITC] goat anti-mouse IgG/IgM [BD Pharmingen] for APJ labeling). The samples were washed in PBS containing 2% serum, suspended in a 2% formalin solution, and analyzed on a Beckman Coulter Gallios flow cytometer.

Construction of the replication-competent vector SL3-AP.

The replication-competent virus was constructed as described before (6). Briefly, the virus was constructed by cloning SL3-2 AP@165 envelope proteins into the SL3-3 MLV using standard cloning techniques. The resulting viral genome was transfected into HEK293T cells, and 48 h posttransfection, supernatants were transferred to NIH 3T3 cells in order to establish a virus-producing culture. Successful infection was confirmed by flow cytometry analysis. Supernatants from the infected NIH 3T3 cells were used to infect D17 or D17 APJ cells.

RESULTS

Design of the retargeted SL3-2 envelope protein.

Since the SL3-2 envelope protein is related to that of feline leukemia virus B (FeLV-B), we used the crystal structure of the latter (8) to identify potential insertion sites for apelin. As seen in Fig. 1, alignment of the receptor binding domains (RBDs) from these two envelope proteins shows that the main differences are found in two clusters, which correspond to variable regions A and B (VRA and VRB, respectively) identified previously (4). Based on this information, three constructs were made by insertion of the apelin-13 sequence flanked by two linker sequences at positions 86, 155, and 165 of the SL3-2 envelope protein. Linker sequences were included based on our previous finding that they enhance binding affinity of the chimeric ecotropic envelopes to APJ (5). Position 86 corresponds to the VRA loop, which in ecotropic envelopes contains the main determinant of receptor binding (3, 20). Positions 155 and 165 are both found in the VRB loop. Position 155 was chosen based on the realization that FeLV sequence contains an extra disulfide bonded loop compared to SL3-2. The rationale was that the introduction of a new disulfide bonded loop containing apelin would be tolerated by the overall structure. Position 165 was chosen since in the structure of the FeLV RBD, it corresponds to the tip of the large VRB loop. This was expected to expose the ligand most efficiently to the environment and thus to interaction with APJ.

Fig 1.

(A) Crystal structure of the receptor binding domain (RBD) of FeLV-B strain 1LCS (8). The insertion sites for apelin are marked on the structure because of its similarity to the RBD of SL3-2. (B) Alignment of the receptor binding domain sequences of SL3-2 and FeLV-B; the insertion sites for apelin are indicated by boxes. (C) The sequences corresponding to the boxes depicted in panel B in the three apelin/SL3-2 chimeric constructs are shown. The apelin cassette is shown in blue whereas the FeLV-B sequences are underlined; the black sequences are of SL3-2 origin. The insertion sites for apelin are marked in panels A and B.

Activity of the chimeric envelope proteins.

In order to test the activity of the chimeric envelopes, viral particles containing the chimeric envelope and a transfer vector expressing the neo gene were produced as described earlier (7). Briefly, the bicistronic vectors expressing the chimeric envelope proteins were cotransfected with gag-pol plasmids into 293T cells. The resulting virions were used to infect NIH 3T3-based semipackaging cells that stably express gag-pol. The infected semipackaging cells were selected with G418. The resulting cell population stably produces virions containing the chimeric envelopes and capable of a single round of infection, which are used to perform the subsequent experiments. Serial dilutions of the virions produced by semipackaging cells were used to transduce murine NIH 3T3 cells, canine D17 cells, and D17 cells expressing either APJ or the irrelevant CXCR4 receptor. Selection of the transduced cells with G418 resulted in the emergence of transduced colonies used to calculate the titers (Fig. 2).

Fig 2.

Titer of the chimeric SL3-2 envelope proteins. The figure shows results representative of several independent experiments. Note the logarithmic scale.

As seen in Fig. 2, the AP@86 envelope protein could not produce measurable titers on either cell line, suggesting misfolding of the protein. Interestingly, both envelopes with insertions in the VRB loop were able to infect murine cells as well as D17 cells expressing APJ, with AP@165 showing titers on the same level as those of the wild-type SL3-2. The background titers on D17 or D17 CXCR4 cells were several orders of magnitude lower. Therefore, the infection of the D17 APJ cells must be through interaction of the virus with APJ. Since AP@165 had the highest titer on D17 APJ, it was chosen for the subsequent experiments.

While it is evident that AP@165 utilizes APJ as an entry receptor, it had retained its ability to infect murine cells as well. Although it seemed more probable that this happens through interaction with Xpr1, it could not be ruled out that the infection is through a putative APJ receptor on the NIH 3T3 cells. In order to clarify the precise receptor usage profile of the chimeric envelope protein, we examined whether blocking of either Xpr1 or APJ affects the titers (Fig. 3).

Fig 3.

The titer of virions containing AP@165 envelope. Blocking of either Xpr1 through interference with MCF247 or APJ through incubation with apelin reduces the titers. Note the logarithmic scale.

Xpr1 usage was examined through interference with MCF247; infected NIH 3T3 cells express this polytropic envelope protein on their surface, which will bind to Xpr1 and thus prevent interaction with incoming viruses (6). The titer on NIH 3T3 cells infected with MCF247 was around 5 orders of magnitude lower than that on uninfected NIH 3T3 cells, confirming that the AP@165 envelope uses Xpr1 as an entry receptor on murine cells.

APJ was blocked through competitive binding to apelin-13 peptide at 6 μM. All APJ molecules are expected to bind apelin-13 at this ligand concentration, which is several orders of magnitude higher than the K_d (dissociation constant) value (16). On D17 APJ cells, infectivity of this chimeric protein was around a factor of 100 lower in the presence of the apelin peptide, confirming that infection of D17 APJ cells was dependent on interaction with the APJ receptor. These results show that AP@165 can use both APJ and Xpr1 as entry receptors with equal efficiency.

Replication-competent MLV with APJ tropism.

The efficiency of AP@165 in using a heterologous protein as entry receptor prompted us to investigate whether a replication-competent MLV containing this envelope protein will be able to replicate in a cell culture in an APJ-dependent manner. The virus was designed in the context of an N-tropic SL3-3 backbone as described previously (6). The CA of SL3-3 is restricted by human TRIM5alpha, which enhances the safety profile of this chimeric virus (1). Virions were produced in NIH 3T3 cells by transfection and were subsequently used to infect both D17 and D17 APJ cells. Three weeks later, we measured the surface expression of SL3-AP envelope protein using flow cytometry as an indicator for presence of the virus, since infected cells express the viral proteins, including Env, on their surface. As seen in Fig. 4A, D17 APJ cells showed high levels of envelope protein, suggesting that the virus has replicated in these cells, while D17 cells without APJ were not infected. Control viruses containing the SL3-2 GI envelope protein (which can use Xpr1 for entry in both human, mink, and murine cells) (4, 6) were unable to infect the D17 APJ cell line (Fig. 4B). Both viruses were able to infect murine NIH 3T3 cells (Fig. 4C and D). Thus, the SL3-AP is a gammaretrovirus which can replicate using APJ as its entry receptor. This infection seems to be mediated by cell-free virions since supernatant transfer from an infected D17 APJ culture to an uninfected one results in complete infection (data not shown).

Fig 4.

Infected cells stably express viral proteins, including the envelope protein. Thus, replication of a virus results in gradual growth in the number of Env-expressing cells in a culture. To confirm viral replication, expression levels of the envelope protein were determined by flow cytometry. (A) The SL3-AP virus is able to replicate in D17 only if APJ is expressed in these cells. (B) A replication-competent virus expressing a variant of the SL3-2 envelope protein with extended tropism is not able to replicate in D17 APJ cells. (C and D) Both viruses are able to replicate in murine NIH 3T3 cells.

Wild-type MLV envelope proteins are known to be able to bind and block access to their respective receptors when expressed from inside a cell. Since SL3-AP uses two different receptors, we examined whether this virus shows any difference in the efficiency of receptor interference with regard to either Xpr1 or hAPJ. This was done by measuring the titers of either wild-type SL3-2 or AP@165 containing vectors on murine NIH 3T3 or canine D17 APJ cells infected with replication-competent SL3-AP. We had previously shown that the SL3-2 envelope shows suboptimal receptor interference (6). As evident in Fig. 5, SL3-AP also showed similar incomplete interference with infection through Xpr1. Interestingly, the interference with infection through hAPJ was at 100%, suggesting a strong interaction between the chimeric envelope protein of SL3-AP and hAPJ.

Fig 5.

Receptor interference illustrated by reduced titers on SL3-AP-infected cells (which express the chimeric envelope protein SL3-2 AP@165 blocking the APJ receptor).

Surface expression of APJ is affected by SL3-AP infection.

At least in some cases, arrest of envelope receptor complexes in the Golgi complex and the consequent lack of expression of the receptor on the cell surface are observed upon viral infection (30). The strong interference with infection through hAPJ exhibited by SL3-AP might be due to similar reasons. To investigate whether AP@165 (the envelope protein of SL3-AP) affects surface expression of APJ in a similar way, expression levels of APJ on infected versus noninfected D17 APJ cells were determined by flow cytometry. As can be seen in Fig. 6, infection of D17 APJ cells with SL3-AP severely reduces the labeling of APJ by the antibody. Since preincubation of D17 APJ cells with apelin peptide does not affect labeling by the anti-APJ antibody (data not shown), this suggests that this chimeric envelope protein is capable of interaction and downregulation of the surface expression of its receptor.

Fig 6.

Flow cytometry analysis of the surface expression of APJ. D17 APJ cells infected with SL3-AP virus show severely reduced surface binding of anti-APJ.

DISCUSSION

We have constructed a gammaretrovirus envelope protein that is capable of utilizing the G-protein-coupled receptor APJ for entry with an efficiency comparable to that of the wild-type isolates. This was achieved by insertion of apelin-13 in one of the variable regions of the receptor binding domain of the SL3-2 envelope protein. A similar approach based on the envelope protein of Moloney MLV (AP@79SGGSG) had resulted in a construct able to use APJ as entry receptor, albeit at much lower efficiency (5). Ecotropic envelopes have generally been used as experimental systems for retargeting efforts since their limited tropism (capable of infecting only murine cells) makes them adaptable to such experiments. However, most attempts to produce an envelope protein capable of utilizing a heterologous protein as entry receptor either have been unsuccessful or have resulted in tropism change with severely reduced infection efficiencies (5, 13, 17, 21, 31). Only in two cases has efficient retargeting been achieved. Retargeting toward the somatostatin receptor resulted in the loss of the affinity for the wild-type mCAT-1 receptor, while using CXCR4 as the entry receptor was efficient only in a specific cell line (19, 25).

SL3-2 envelope protein, because of its limited and ecotropic-like tropism, offers another candidate for a retargeting scaffold. This envelope belongs to the xenotropic-polytropic group but can infect only murine cells (4). Using it as a scaffold, we have produced a retargeted envelope protein that can infect through a heterologous receptor as well as its natural surface receptors. This envelope protein can be incorporated into a virus genome, and the resulting virus (SL3-AP) can both replicate in cells expressing hAPJ and interfere with superinfection through and downregulate surface expression of this receptor. Thus, this new retrovirus shows the replication characteristics that would be expected from a wild-type isolate: specific tropism dependent on a surface protein receptor.

One interesting aspect of the SL3-2 AP@165 envelope protein is its dual tropism. It shows interference with the polytropic MLV MCF247 strain, proving that this chimeric envelope does not lose its ability to use Xpr1. This resembles the receptor usage by the gammaretrovirus 10A1, which uses the related phosphate transporters Pit-1 and Pit-2 as receptors (26). However, in the case of Xpr1 and hAPJ, no obvious relationship is evident.

In the presence of 6 μM apelin-13 peptide, the infectivity of this virus was reduced by 2 orders of magnitude, presumably as a result of APJ internalization and a reduced steady-state surface level of APJ as well as blocking of surface APJ by the peptide. However, the concentration of peptide used in these experiments was much higher than the K_d value for its binding to APJ, which is in nanomolar range (16). Therefore, all available APJ molecules on the cells would be expected to bind to the peptide, resulting in complete blocking of infection by the virions. The relatively high titers despite this blocking might be explained by the fact that SL3-2 virions seem to be able to bind to nonpermissive cells (6). If this is true for D17 cells, the presence of the envelope proteins close to the cell membranes would have resulted in very high local concentrations of the envelope protein and potentially contributed to competition with the peptide molecules for binding to APJ.

SL3-AP shows receptor interference with infection through both its receptors. While interference through Xpr1 is incomplete and reminiscent of wild-type SL3-2 (6), this virus is able to block superinfection through hAPJ completely. Residual surface expression of hAPJ remains in cells infected with SL3-AP, suggesting that SL3-2 AP@165 envelope binds and blocks hAPJ on the cell surface as well.

In contrast to the Moloney MLV-based AP@79SGGSG (5), which had apelin inserted in the VRA of the receptor binding domain, insertion of apelin into the VRA loop of SL3-2 envelope resulted in a nonfunctional protein, whereas VRB could tolerate insertions at two different sites. Interestingly, comparison of the sequence and crystal structures for the ecotropic Friend MLV (15) receptor binding domain with those of the (SL3-2-like) FeLV-B (8) shows a considerable difference in the relative sizes of the VRA and VRB in these two envelope proteins. While VRA is much larger than VRB in ecotropic envelopes, the two variable regions are similar in size in the FeLV-B envelope. However, in both cases the combined sizes of the two variable regions are roughly the same, resulting in RBDs with very similar three-dimensional structures (Fig. 7). The VRA in ecotropic envelopes constitutes a surface which has been shown to contain the binding site for the ecotropic receptor (3, 20). Based on the structure of the FeLV-B receptor binding domain, the corresponding surface in SL3-2 is expected to be formed by both VRA and VRB. The insertion site at position 165 falls onto one end of this surface (Fig. 1A). Since this insertion does not remove the ability of the AP@165 envelope protein to use its natural receptor, while insertion at position 86 completely inactivates the envelope function, the binding site for Xpr1 might be expected to be found on the other end of this surface, mostly in VRA and VR3. This is supported by our previous finding that two amino acid alterations at VR3 can enable SL3-2 to use Xpr1 in other species (4) (Fig. 7). This degree of redundancy in the receptor binding surface of gammaretroviral envelope proteins might be an evolutionary advantage in that “dual-tropic” viruses such as SL3-AP might constitute the transitional forms in evolution of viruses with new tropisms. Thus, it might be expected that SL3-AP might lose its affinity for Xpr1 upon successive passages in APJ-expressing cells.

Fig 7.

Crystal structures of FeLV-B strain 1LCS (8) (left) and Friend MLV strain 1AOL (15) (right) receptor binding domains.

The choice of receptor for retargeting of retroviruses does not seem to be trivial. Here, apelin/APJ was chosen as the ligand/receptor pair because of the high binding affinity, the small size of apelin, and the fact that APJ is one of the coreceptors of HIV. Our efforts to design a retargeted SL3-2 envelope against other surface molecules, including other G-proteins and monoamine transporters, have not been successful (S. Bahrami and F. S. Pedersen, unpublished results), resulting in envelope proteins which, while keeping their ability to infect through Xpr1, show little or no infectivity through the target receptor. In these cases, the peptide ligands might have been presented on the surface of the envelope protein in such a way that their interaction with the receptor has been inhibited. Another intriguing possibility is that surface proteins must fulfill as-yet-undefined criteria for being able to mediate entry of a retrovirus. APJ is one of the coreceptors of HIV (11), and the other reported cases of efficient retargeting have been toward CXCR4 (25), another of the HIV coreceptors; somatostatin (19); and HuPAR-1 (22, 23), all of them belonging to the GPCR family of proteins. The design of retargeted envelope proteins does not seem to be a case of simplified generic design but requires a case-by-case development.