| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Regular Articles |
Molecular Medicine Program, Mayo Clinic, Rochester, Minnesota 55905
 |
ABSTRACT |
|---|
 Top ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
|
INTRODUCTION |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Replicating attenuated MV for the therapy of lymphoid and other malignancies has recently been investigated, with encouraging results (3, 4, 5) . We showed that locally or systemically injected MV resulted in regression of large established lymphoma xenografts in SCID mice (3) . Viral replication within the tumor was confirmed, however, viral RNA and protein expression was distributed in discrete areas within the tumor. Despite this, we frequently observed complete regression of MV-treated tumors, suggesting that host response to MV-infected tumor cells contributed to the therapeutic effect. We observed in sections of residual tumor a considerable inflammatory response, associated with large necrotic areas of tumor. This observation, together with the published data regarding the antitumor properties of neutrophils, led us to hypothesize that neutrophils are important contributors to the oncolytic effect of MV in this setting and to investigate this hypothesis further.
We reasoned that the local expression of the immunostimulatory cytokine GM-CSF, which potentiates many neutrophil functions, may synergize with MV infection and improve the oncolytic effect. The pleotropic actions of GM-CSF on neutrophils include enhancement of antimicrobial activity against intracellular pathogens via stimulation of phagocytosis, oxidative metabolism, lysozyme release, and complement recruitment (6) . GM-CSF is known to increase neutrophil transmigration by augmenting chemotaxis and enhancing adhesion to vascular endothelium (7) . GM-CSF also enhances antimicrobial activity of monocyte macrophages and enhances NK cell function (8 , 9) .
We, thus, exploited the capacity of MV to express exogenous proteins from ATU and generated MVs expressing murine GM-CSF. To express this protein at different levels, we took advantage of the MV transcription gradient, whereby the level of gene transcription and protein expression is determined by the position of the foreign gene in the MV genome (10) . The replication and toxicity of these recombinant viruses was characterized in a murine model permissive for MV replication by virtue of transgenic expression of the MV receptor human CD46. Oncolytic activity against human Raji tumor xenografts was then tested in SCID mice. Our studies confirmed the hypothesis that therapy with replicating MV stimulates a neutrophil antitumor response, which can be cytokine-enhanced to improve the oncolytic capacity of MV.
|
MATERIALS AND METHODS |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Virus Rescue.
The viruses were rescued by Lipofectamine (Invitrogen, Grand Island, NY)-mediated cotransfection of 4 µg of full-length p+MVmGM-CSF cDNA, along with a set of three plasmids from which the MV polymerase complex was expressed (1 µg each of plasmids encoding MV-N and MV-P and 0.5 µg of plasmid encoding MV-L) into 293 cells that had been infected for 1 h by modified Vaccinia Ankara encoding T7 polymerase (12)
. After 72 h, the 293 rescue cells were overlaid onto Vero cells in 10-cm dishes. Individual syncytia were picked from the dishes and transferred to fresh Vero cells in 35-mm wells. When a cytopathic effect was evident in the wells, the virus was transferred to Vero cells in T75 flasks and subsequently propagated in bulk preparations.
Viruses and Virus Production.
MVs were inoculated onto 106 Vero (African green monkey) cells in T75 tissue-culture flasks at a MOI of 0.01 in 2 ml of Optimem (Life Technologies, Inc., Rockville, MD) at 37°C for 2 h. The viral inoculum was removed and replaced by normal medium. The cultures were then observed until maximal syncytial formation had occurred, at which point the cells were harvested in 2 ml of Optimem and the virus was released by two cycles of freeze-thawing. Virus was titrated on Vero cells, and the 50% tissue-culture infectious dose was calculated according to the method of Kärber (13)
.
Cells.
Vero cells (ATCC CCL-81; American Type Culture Collection, Manassas, VA) were grown in DMEM supplemented with 5% FCS. Raji cells (ATCC CCL-86) were grown in RPMI supplemented with 10% FCS (Life Technologies, Inc.). The 293 cells were grown in DMEM supplemented with 10% FCS.
Animals.
Four-week-old CB17 SCID mice (Harlan Sprague Dawley, Indianapolis, IN) were housed in a barrier facility and cared for according to standards set by the Institutional Animal Care and Use Committee. Mice received s.c. injections in the flank region with 107 viable Raji tumor cells. For MV therapy, after the tumors reached a volume of approximately 0.4 cm3 , they were injected daily with MV or MVmGM-CSF/P in a total volume of 100 µl for 10 days. As controls, tumors were injected daily with the same volume of UV-inactivated virus. Tumor measurements were made daily in two diameters, and the tumor volume was calculated according to the formula V = a2 b/2 where a is the shortest and b the longest diameter. Mice whose tumors reached a volume of 2.5 cm3 or had begun to invade surrounding tissues were euthanized. The experimental protocol was approved by the Mayo Clinic Institutional Animal Care and Use Committee.
mGM-CSF ELISA.
mGM-CSF was quantified by ELISA kit (R&D Systems, Minneapolis, MN) according to the manufacturer’s instructions.
Immunohistochemistry.
Residual tumors were carefully removed, and a portion was fixed in 10% neutral buffered formalin. After paraffin embedding, the tissue was cut in 0.5-µm sections. CD18 staining was performed on paraffin-embedded tissue sections after deparaffinization and antigen retrieval by steaming in citrate (pH 6.0). Tissue sections were incubated with biotinylated antimouse CD18 (C71/16) or the isotype control at a concentration of 10 µg/ml for 1 h, followed by horseradish peroxidase-streptavidin conjugate. The labeled antigen signal was amplified using Tyramide Signal Amplification (Molecular Probes, Eugene, OR), followed by another incubation with HRP-streptavidin. The peroxidase label was developed using the chromogen 3-amino-9-ethylcarbazole (DAKO Corp., Carpinteria, CA).
Flow Cytometric Analysis of Tumor Cells.
Tumors were dissected into small pieces and suspended in a digestion mixture consisting of 0.4 mg/ml collagenase, 0.4 mg/ml DNase, 0.2 mg/ml CaCl2, and 0.2 mg/ml MgCl2 in PBS for 45 min at 37°C, with occasional agitation. The tumor digest was filtered through one layer of gauze, then centrifuged for 5 min at 1000 rpm. RBCs were lysed with NH4Cl lysing solution. The digest was centrifuged again, and the pellet was resuspended in PBS with 2% BSA. Viable cells were counted. Three aliquots of 106 cells each were stained to detect neutrophils, NK cells, and macrophages. For neutrophil detection, a biotinylated rat antimouse antibody, Ly6G (RB6–8C5), at 5 µg/ml, followed by streptavidin-FITC at 2.5 µg/ml was used. For NK cell detection, a phycoerythrin-conjugated rat antimouse pan NK cell antibody (DX5) was used. For macrophage detection, 2 µg/ml phycoerythrin-conjugated rat antimouse Mac-3 (M3/84; after a 5-min incubation with Fc block-rat antimouse CD16/CD32 at 10 µg/ml) was used. Appropriate isotype controls were used for all antibodies. All antibodies were obtained from PharMingen (San Jose, CA).
Neutrophil Chemotaxis Assay.
Neutrophils isolated from whole blood by centrifugation on Ficoll-Paque PLUS (Amersham Biosciences, Uppsala, Sweden) were incubated with Calcein AM (Molecular Probes, Eugene, OR) at a final concentration of 5 µm in PBS/0.5% BSA for 30 min at 37°C. Neutrophils were washed, resuspended in fresh PBS/BSA at a concentration of 4 x 106/ml to 2.5 x 106, and then loaded into the top chamber of a ChemoTx System chemotaxis system, (Neuro Probe Inc., Gaithersburg, MD). Normal Raji cells or Raji cells infected with MV at MOI 0.1, 1 or 10 for 24 h were washed free of virus and resuspended in fresh RPMI at a concentration of 1 x 106/ml. These cells were loaded into the bottom chamber of the chemotaxis system containing a polycarbonate filter with a pore size of 8 µm. The system was incubated for 90 min at 37°C, 5% CO2. Normal RPMI media or 100 nM FMLP (Sigma Chemical Co., St. Louis, MO) were used as negative and positive controls, respectively. After incubation, the neutrophils remaining on the top of the membrane were washed off. The chemotaxis system was centrifuged for 10 min at 450 rpm to collect any neutrophils that had partially migrated through the membrane. Fluorescence of neutrophils that had migrated to the bottom well of the chemotaxis plate was quantified at absorbance 485/535 using a Spectrofluor Plus Fluorescence Microplate Reader (Tecan U.S. Inc., Research Triangle Park, NC).
|
RESULTS |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
. The growth kinetics of the mGM-CSF-expressing viruses did not differ significantly from that of the parent virus, MVNSe, except for a slight delay in replication of the virus expressing mGM-CSF before the N gene (data not shown). Fig. 1, B and C
, shows that, after infection of 106 cells at a MOI of 0.1, mGM-CSF in nanograms per milliliter quantities was detected in the cell supernatant of both Vero and Raji cells. Thus, mGM-CSF can be expressed from ATU at various positions in the genome of an attenuated MV without compromising viral replication.
|
/ß IFN receptor. Attenuated MV replicates and disseminates systemically in these mice after intranasal and i.p. administration (14, 15, 16)
and is ultimately self-limited. We wished to determine whether, after systemic administration, GM-CSF would be detectable in the circulation, whether any effects on the peripheral blood cells would be observed, and whether infection with the GM-CSF-expressing MVs resulted in any symptoms or pathology related to the expression of this cytokine in a replicating virus. Therefore, CD46GeInfarko mice received intraperitoneal injections of 106 pfu of each of the three mGM-CSF-expressing viruses i.p. (n = 10 per group). White cell counts with differentials, serum mGM-CSF levels, and anti-MV antibody response were serially monitored and compared with those of control mice that received injections of MVNSe. In an additional experiment, a mouse was sacrificed at various time points during the experiment, so that relevant organs could be sectioned and examined.
The general health of the mice was closely observed. None of the mice suffered any apparent toxicity. Fig. 2A
shows the levels of mGM-CSF measured in the serum at four time points over a 28-day period after infection. The highest serum level of mGM-CSF was detected in mice that received injections of MV GM-CSF N, lower levels in mice receiving injections of MV GM-CSF P, and no elevations in serum GM-CSF beyond background levels were detected in mice that received injections of MV GM-CSF H. This is consistent with expected levels of protein production generated by the MV transcription gradient. Serum mGM-CSF was not detectable at any time later than the 3-day time point, consistent with previous studies, in which the peak of virus replication in this model occurs at 3 days after infection (13)
.
|
. In particular, as shown in Fig. 2C
, no elevation in neutrophil counts were noted. An antibody response, measured by ELISA, was observed to each of the injected viruses at days 14 and 28 after infection. There was no significant difference in the magnitude of this response between each of the GM-CSF-expressing viruses and the MVNSe control (data not shown). H&E staining of sections of various organs, in particular, the spleens, did not differ between MV-GM-CSF-injected mice and MVNSe-injected controls (data not shown). Thus, administration of MV GM-CSF N and MV GM-CSF P results in transient elevations in serum GM-CSF levels but does not significantly affect the peripheral blood counts nor result in any overt toxicity in CD46 transgenic mice. A typical antibody response is generated against the GM-CSF-expressing MVs, and these viruses are eliminated with similar kinetics to the control MVNSe virus.
MV-producing mGM-CSF Has Enhanced Oncolytic Activity in a SCID/Raji Model of Human Lymphoma.
We chose MV GM-CSF P, expressing GM-CSF at intermediate levels, to investigate the antitumor efficacy of MV GM-CSF in a model of human lymphoma in SCID mice. Raji tumors were seeded and allowed to grow to a volume of 0.2–0.4 cm3
. Groups of mice (n = 10/group) then received injections intratumorally, daily for 10 days, of MV GM-CSF P, MVNSe, or UV-irradiated MV control.
Fig. 3A
shows the change in tumor volume over time. There was a significant difference in tumor progression between the three groups. In the tumors injected with UV-inactivated MV, all of the animals required euthanization by day 15. Consistent with our previous experience, injection of MVNSe resulted in a significant slowing of tumor growth compared with control. Moreover, after the injection of MV GM-CSF, there was an additional improvement in antitumor efficacy that was especially pronounced at later time points. Histological examination of sections of residual Raji tumors revealed a significant infiltrate in tumors treated with MVNSe and MV GM-CSF P, which stained positively with an anti-CD18 antibody. This was not present in control tumors treated with UV-irradiated MV, suggesting that leukocytes were involved in the antitumor response. Photomicrographs of representative tumor sections from mice that received injections of MVNSe and UV-irradiated MV are shown in Fig. 3, B and C
.
|
shows mean (+/-SE) murine CD45-positive (CD45+) cells as a percentage of total tumor cells, for tumors injected with MV GM-CSF, MVNSe, or recombinant GM-CSF (n = 10/group). Infiltrating CD45+ cells were seen in all groups. However, there was a very significant difference in the magnitude of the infiltrate between each group. Whereas in tumors injected with recombinant GM-CSF alone, the mean CD45+ infiltrate less than 5% of total cells, those injected with MVNSe contained a mean of 16% CD45+ cells and those with MV GM-CSF a mean of 34% CD45+ cells. This suggested that MVNSe in itself stimulated a considerable host leukocyte response to the tumor. Furthermore, the magnitude of the infiltrate elicited by MV GM-CSF was greater than that stimulated by GM-CSF and MVNSe added, raising the possibility of a synergistic interaction between the two.
|
as the mean (+/-SE) percentage of total cells detected by Ly6.1 for each group. We did not detect any macrophages or NK cells within the tumors. Murine neutrophils were detected in each of the groups. In a result that was consistent with the previous experiment, the tumors injected with MV GM-CSF P contained the greatest percentage (26%) of neutrophils; once again, there was a larger percentage of neutrophils in the MV GM-CSF P-injected tumors than in both the MVNSe (16%) and untreated tumors (4%) added together, again suggesting a possible synergistic interaction between MV and the virally expressed GM-CSF in stimulating a host response. Interestingly, the untreated tumors in this experiment contained a similar percentage of neutrophils to those tumors that, in the previous experiment, had been injected with recombinant GM-CSF. Again, this suggests synergistic local interaction.
To verify that the neutrophil response to the tumors did not result merely from increased numbers of neutrophils in the peripheral blood, we performed leukocyte counts with differentials on all mice, shown in Fig. 4, C and D
. There was no difference in the mean peripheral blood counts and differential counts between mice with MVNSe-treated tumors and those with untreated tumors. Mice with MV GM-CSF-treated tumors had slightly but significantly lower peripheral blood leukocyte counts than the other two groups and, in addition, a slight but significant difference in the percentage of neutrophils. This finding possibly resulted from large numbers of neutrophils having been recruited to the tumor site.
We measured serum GM-CSF levels in all of the mice at the same time point. GM-CSF was detected only in the serum of mice whose tumors were injected with MV GM-CSF P and not in controls. The mean GM-CSF level after six injections of virus was 13.2 pg/ml (range, 0–33), a level similar to that detected on day 3 after infection of the CD46GeInfarko mice with one dose of MV GM-CSF P.
Taken together, these data support the hypothesis that therapy with replicating MV stimulates a neutrophil antitumor response. The hypothesis is further supported by the fact that this response can be cytokine enhanced, with a concomitant improvement in the oncolytic capacity of MV.
Chemotaxis of Human Neutrophils to MV-infected Raji Cells.
We set out to determine whether neutrophils were specifically attracted to MV-infected Raji cells. Chemotaxis assays were performed in triplicate, using isolated human peripheral blood neutrophils from five different individuals. Fig. 5
shows the mean fluorescence of the migrated neutrophils (+/-SE). Medium alone was used as a negative control, and FMLP was used as a positive control. In all individuals, neutrophils migrated toward Raji cells but MV infection of the cells did not increase the migration. Thus, we did not find evidence that MV infection of Raji tumor cells directly stimulated neutrophil chemotaxis.
|
|
DISCUSSION |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
The augmentation of the oncolytic effect ascribed to MV GM-CSF was more pronounced in the last 2 weeks of the experiment (Fig. 3A)
, again suggesting a cellular host-response effect as opposed to direct viral oncolysis.
An antitumor cytotoxicity of neutrophils has been described previously. In vitro and in vivo cytotoxicity of neutrophils has been demonstrated against various murine tumor cells, in particular, after activation of the neutrophils with substances derived from microorganisms [e.g., a fungal derivative, ß1–3-D-glucan (17) , or a streptococcal preparation, OK-432 (18 , 19) ]. It is, therefore, of interest that a neutrophil response in this model of lymphoma is enhanced by a replicating virus that has oncolytic activity. Our studies indicate that virus replication is necessary to achieve this effect, because there was no augmentation of antitumor activity when the control, UV-irradiated MV was used.
We showed, in vitro, that chemotaxis of human neutrophils was stimulated by the Raji tumor cells themselves, but MV infection of the cells did not increase it. Thus, the mechanism by which MV replication within the Raji tumors in vivo stimulates neutrophil infiltration remains unclear. It has been demonstrated that "tumor take" after injection of murine adenocarcinoma can be controlled by neutrophils and that the tumor rejection was associated with neutrophil chemotaxis in vivo after G-CSF-producing tumor cells were injected (20) . However, a recent study in which CD95 (Fas) ligand-expressing tumors were rejected by antitumor T cell receptor transgenic, perforin knockout mice strongly suggested that neutrophils were responsible for the tumor rejection, albeit that no chemotactic activity of CD95 ligand was demonstrated, nor could antitumor cytotoxicity of the neutrophils be shown in vitro (21) .
A number of mechanisms are proposed in the literature for direct killing of tumor cells by neutrophils, including release of hydrogen peroxide (22) , the nitric oxide pathway (23) , and neutrophil-mediated inhibition of glutamine uptake by tumor cells (24) . It is of interest that glutamine seems to play a particularly important role as an energy source for cells of lymphoid origin (25) . Determination of the mechanistic aspects of enhanced tumor cell killing by neutrophils are of interest to us in future studies.
It is important to recall that our studies have been performed in a SCID mouse model, allowing for the study of the role of neutrophils in the cytolytic response in relative isolation. However, the lack of T cells as antitumor immune effector cells may have exaggerated the role of neutrophils in this model system. Nonetheless, our work clearly demonstrates that cytolysis and tumor regression after infection with an oncolytic virus is, at least in part, attributable to a neutrophil inflammatory response. This conclusion is supported by the fact that oncolytic response is enhanced by viral expression of GM-CSF that correlates with a considerable increase in neutrophil number within the tumors. We have also demonstrated that attenuated replicating MVs expressing GM-CSF are not toxic in CD46 transgenic mice. Thus, in the presence of a fully intact immune system, other activities of GM-CSF, shown to be the most potent cytokine to induce durable antitumor immunity when secreted by tumor cells (26) , may provide additional antitumor effects.
Our studies in CD46 transgenic mice, used as a preclinical model for toxicity studies, demonstrate that replicating attenuated MVs expressing murine GM-CSF are safe after systemic administration. This virus has provided further insights into the mechanisms of MV-mediated oncolysis. MV expressing human GM-CSF may be more effective than parental MV as a replicating vector for the therapy of human cancer.
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
1 To whom requests for reprints should be addressed, at The Royal Free and University College Hospital Medical School, Royal Free Campus, Rowland Hill Street, London NW3 2PF, United Kingdom. Phone: 44-0207-830-2833; Fax: 44-0207-830-2092; E-mail: a.fielding{at}rfc.ucl.ac.uk 
2 The abbreviations used are: SCID, severe combined immune deficient; MV, measles virus; GM-CSF, granulocyte macrophage colony-stimulating factor; mGM-CSF, murine GM-CSF; NK, natural killer; ATU, additional transcription unit(s); MOI, multiplicity of infection; pfu, plaque-forming unit(s); FMLP, N-formyl-methionyl-leucyl-phenylalanine.
3 L. Hangartner. Development of measles virus as a vector: Expression of green fluorescent protein from different loci (thesis). Zurich, Switzerland: University of Zurich, 1997.
Received 5/ 9/03. Revised 7/ 8/03. Accepted 7/21/03.
|
REFERENCES |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  A. Gauvrit, S. Brandler, C. Sapede-Peroz, N. Boisgerault, F. Tangy, and M. Gregoire Measles Virus Induces Oncolysis of Mesothelioma Cells and Allows Dendritic Cells to Cross-Prime Tumor-Specific CD8 Response Cancer Res., June 15, 2008; 68(12): 4882 - 4892. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 F. Errington, L. Steele, R. Prestwich, K. J. Harrington, H. S. Pandha, L. Vidal, J. de Bono, P. Selby, M. Coffey, R. Vile, et al. Reovirus Activates Human Dendritic Cells to Promote Innate Antitumor Immunity J. Immunol., May 1, 2008; 180(9): 6018 - 6026. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
C. Allen, S. Vongpunsawad, T. Nakamura, C. D. James, M. Schroeder, R. Cattaneo, C. Giannini, J. Krempski, K.-W. Peng, J. M. Goble, et al. Retargeted Oncolytic Measles Strains Entering via the EGFRvIII Receptor Maintain Significant Antitumor Activity against Gliomas with Increased Tumor Specificity Cancer Res., December 15, 2006; 66(24): 11840 - 11850. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 H. D. Lum, I. N. Buhtoiarov, B. E. Schmidt, G. Berke, D. M. Paulnock, P. M. Sondel, and A. L. Rakhmilevich In vivo CD40 ligation can induce T cell-independent antitumor effects that involve macrophages J. Leukoc. Biol., June 1, 2006; 79(6): 1181 - 1192. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 G. G. Welstead, C. Iorio, R. Draker, J. Bayani, J. Squire, S. Vongpunsawad, R. Cattaneo, and C. D. Richardson Measles virus replication in lymphatic cells and organs of CD150 (SLAM) transgenic mice PNAS, November 8, 2005; 102(45): 16415 - 16420. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
L. Gillet, B. Dewals, F. Farnir, L. de Leval, and A. Vanderplasschen Bovine Herpesvirus 4 Induces Apoptosis of Human Carcinoma Cell Lines In vitro and In vivo Cancer Res., October 15, 2005; 65(20): 9463 - 9472. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
B. D. Anderson, T. Nakamura, S. J. Russell, and K.-W. Peng High CD46 Receptor Density Determines Preferential Killing of Tumor Cells by Oncolytic Measles Virus Cancer Res., July 15, 2004; 64(14): 4919 - 4926. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Cancer Research | Clinical Cancer Research |
| Cancer Epidemiology Biomarkers & Prevention | Molecular Cancer Therapeutics |
| Molecular Cancer Research | Cancer Prevention Research |
| Cancer Prevention Journals Portal | Cancer Reviews Online |
| Annual Meeting Education Book | Meeting Abstracts Online |
, MVNSe; cross X, MV GM-CSF N; 
, MV GM-CSF P; 
, MV GM-CSF H.
), or MV GM-CSF P (MV GM-CSF, 
). B and C, CD18 antibody staining of sections of residual tumor from mice treated with MVNSe (B) or UV-irradiated MV (C).
), MVNSe (
), or recombinant mGM-CSF control (
, MV-infected Raji cells.