High diversity of the immune repertoire in humanized NOD.SCID.γc−/− mice


  • Gilles Marodon,

    Corresponding author
    1. Université Pierre et Marie Curie - Paris 6; Centre National de la Recherche Scientifique UMR 7211, Paris, France
    2. Institut National de la Santé et de la Recherche Médicale U959, Hôpital de la Pitié Salpêtrière, Paris, France
    • UPMC/CNRS UMR 7211/INSERM U959, 83 Bd de l'Hôpital, F-75651 Paris cedex 13, France Fax: +33-14-217-7462
    Search for more papers by this author
  • Delphine Desjardins,

    1. Université Pierre et Marie Curie - Paris 6; Centre National de la Recherche Scientifique UMR 7211, Paris, France
    2. Institut National de la Santé et de la Recherche Médicale U959, Hôpital de la Pitié Salpêtrière, Paris, France
    Search for more papers by this author
  • Laetitia Mercey,

    1. Université Pierre et Marie Curie - Paris 6; Centre National de la Recherche Scientifique UMR 7211, Paris, France
    2. Institut National de la Santé et de la Recherche Médicale U959, Hôpital de la Pitié Salpêtrière, Paris, France
    Search for more papers by this author
  • Claude Baillou,

    1. Université Pierre et Marie Curie - Paris 6; Centre National de la Recherche Scientifique UMR 7211, Paris, France
    2. Institut National de la Santé et de la Recherche Médicale U959, Hôpital de la Pitié Salpêtrière, Paris, France
    Search for more papers by this author
  • Pierric Parent,

    1. Université Pierre et Marie Curie - Paris 6; Centre National de la Recherche Scientifique UMR 7211, Paris, France
    2. Institut National de la Santé et de la Recherche Médicale U959, Hôpital de la Pitié Salpêtrière, Paris, France
    Search for more papers by this author
  • Manuarii Manuel,

    1. ImmunID Technologies, CEA/iRTSV, Grenoble, France
    2. INSERM U590, Centre Léon Bérard, Université Lyon, Lyon, France
    Search for more papers by this author
  • Christophe Caux,

    1. INSERM U590, Centre Léon Bérard, Université Lyon, Lyon, France
    Search for more papers by this author
  • Bertrand Bellier,

    1. Université Pierre et Marie Curie - Paris 6; Centre National de la Recherche Scientifique UMR 7211, Paris, France
    2. Institut National de la Santé et de la Recherche Médicale U959, Hôpital de la Pitié Salpêtrière, Paris, France
    Search for more papers by this author
  • Nicolas Pasqual,

    1. ImmunID Technologies, CEA/iRTSV, Grenoble, France
    Search for more papers by this author
  • David Klatzmann

    1. Université Pierre et Marie Curie - Paris 6; Centre National de la Recherche Scientifique UMR 7211, Paris, France
    2. Institut National de la Santé et de la Recherche Médicale U959, Hôpital de la Pitié Salpêtrière, Paris, France
    Search for more papers by this author


The diversity of the human immune repertoire and how it relates to a functional immune response has not yet been studied in detail in humanized NOD.SCID.γc−/− immunodeficient mice. Here, we used a multiplex PCR on genomic DNA to quantify the combinatorial diversity of all possible V–J rearrangements at the TCR-β chain and heavy chain Ig locus. We first show that the combinatorial diversity of the TCR-β chain generated in the thymus was well preserved in the periphery, suggesting that human T cells were not vastly activated in mice, in agreement with phenotypic studies. We then show that the combinatorial diversity in NOD.SCID.γc−/− mice reached 100% of human reference samples for both the TCR and the heavy chain of Ig. To document the functionality of this repertoire, we show that a detectable but weak HLA-restricted cellular immune response could be elicited in reconstituted mice after immunization with an adenoviral vector expressing HCV envelope glycoproteins. Altogether, our results suggest that humanized mice express a diversified repertoire and are able to mount antigen-specific immune responses.


The recent description of long-term human T-cell reconstitution from CD34+ precursors in immunodeficient mice 1, 2 opens new possibilities for investigations of the biology of the human immune system (reviewed in 3). “Humanized” mice appear to be a simple, affordable and robust model of the human immune system to develop new therapeutic strategies, such as vaccines or cell and gene therapies. Recent data showed productive infection with HIV in the humanized RAG-2−/−.γc−/−4–7 and NOD.SCID.γc−/− (NSG) 8 mice models, in which high viral replication and CD4+ T-cell depletion were observed. These models could thus be used to test new therapeutic strategies against HIV, as recently shown with a CCR5 antagonist in mice transplanted with human PBMC 9.

Establishment of a proper diversity in the immune system is crucial to mount an efficient yet not self-destructive adaptive immune response 10. For example, constriction of the immune repertoire has been associated with progression to AIDS 11. The diversity of the T-cell repertoire is therefore one of the key parameter that needs to be addressed to evaluate the relevance of humanized mice. A mouse model expressing a limited set of human TCR or Ig would have limited usefulness to assess new therapeutic strategies, such as vaccines. Earlier studies reported the presence of several variable elements of the β chain TCR (TRBV) families using flow cytometry 2, 12, 13, but this kind of analysis is far from being exhaustive. Diversity can also be assessed by the size of CDR3 lengths 14. The resulting distributions are often evaluated visually and the repertoire is qualified as “diverse” if CDR3 lengths grossly follow a Gaussian distribution. Previous analysis using CDR3 spectratyping revealed major distortions in the human T-cell repertoire generated in immunodeficient mice compared with regular human PBMC 15, casting doubts on the “normality” of the T-cell repertoire generated in humanized mice. However, a major disadvantage of standard qualitative CDR3 spectratyping is that it only reveals the presence of a particular CDR3 length within a given TRBV family, without any information on the frequency of this “clone” within the entirety of the repertoire. Furthermore, information regarding the variable α chain elements of the TCR (TRAV) using CDR3 spectratyping is also very difficult to obtain.

Recent studies on the ability of human cells generated in mice from CD34+ precursors to mount an antigen-specific immune response led to conflicting results. Although earlier studies showed that T cells were able to proliferate ex vivo against EBV-transformed autologous B cells in humanized RAG-2−/−.γc−/− mice 2, other studies reported limited T- and B-cell responses against HIV in the same model 4. Another report in NSG mice described high levels of HIV-specific antibodies in the serum of HIV-infected mice compared with controls, suggesting the existence of a functional B- and T-cell response in this model 8. Various immunization protocols applied to both murine strains led to the detection of a specific IgG response or antigen-specific cytotoxicity 6, 16, suggesting that humanized mice are useful to assess vaccine efficacy. Noticeably IFN-γ production, a well recognized marker for a functional cellular immune response, has only been observed in NOD/SCID mice humanized with bone marrow and liver transplants (BLT) following infection with a live EBV virus or after injection with a superantigen 17. A T-dependent humoral immune response has also been recently described in BLT mice following immunization with a hapten-carrier antigen 18. Furthermore, a weak immune response was elicited in NSG injected with antibodies targeting the EBNA-1 antigen to DC 19. To date however, there is no evidence for the ability of humanized mice to mount an antigen-specific cellular immune response against an antigen expressed by a potential vaccine. This aspect is the key for the future application of humanized mice to pre-clinical testing of potential vaccines.

In this study we determined the quality of the immune repertoire in humanized NSG mice and analyzed how it could correlate with an immune response elicited by a non-replicative genetic vaccine that target a limited portion of the repertoire. To overcome the limitations in CDR3 spectratyping, we turned to a recently described multiplexed PCR technique 20 that focuses on combinatorial rather than clonal diversity.


We first aimed at quantifying human chimerism in our NSG colony following injection of purified CD34+ cells from umbilical cord blood. In total, 16–22 wk after injection, almost all of the mice had human cells in their thymus, mesenteric LN and spleen. More than 90% of the cells in the lymphocyte gate were of human origin in the thymus or LN (Fig. 1A). The T-cell differentiation process in the thymus was indeed well preserved with normal proportions of CD4+CD8+ (DP) and SP cells (Supporting Information Fig. 1A). Although chimerism was lower in the spleen, this organ contained the majority of human cells (Fig. 1B) and some male mice presented enlarged spleen. Detailed kinetic studies in the blood revealed that human CD45+CD3+ cells appeared from 8 wk on and steadily increased afterwards (Supporting Information Fig. 1B), the rest of the human cells being mostly CD19+ B cells that followed the reverse pattern, i.e. appeared first and steadily diminished thereafter (Supporting Information Fig. 1B). Among CD3+ T cells, CD8+ represented close to 60% of all CD3+ T cells in blood at 10 wk indicating that these T cells appeared first in the periphery (Fig 1D). Their proportions in the blood declined in parallel to increased proportions of CD4+ T cells (Fig. 1C), showing that neither CD4+CD8+ nor CD4CD8 T cells aberrantly developed. A high proportion of “naïve” CD8+CD45RA+CCR7+ was detected among CD45+CD3+CD19 cells of the spleen and the mesenteric LN (Table 1 and Supporting Information Fig. 2). The proportions of naïve CD4+CD45RA+CCR7+ cells were much lower, and elevated frequencies of effector memory CD45RACCR7 T cells were found in some mice (Table 1), consistent with their expansion (Fig. 1C). These results indicated that human CD4+ T cells were much more activated in the murine environment than CD8+ T cells. Interestingly, we also detected CD4+CD25+Foxp3+ cells in the thymus and the spleen of NSG mice, indicating the presence of T cells with a regulatory phenotype (Fig. 2A). Furthermore, we identified a population of large HLA-DRloCD123+CD303+ cells that have a phenotype resembling that of plasmacytoïd DC (Fig. 2B), essential to initiate immune responses 21.

Figure 1.

Percentages of hCD45+ cells and total numbers of cells in reconstituted NSG mice. (A) Lymphoid-sized cells were selected based on forward-size scatter /side-size scatter profiles from the indicated organs of reconstituted NSG mice and the frequencies of human CD45+ cells within the lymphoid cell population was measured by flow cytometry. (B) Absolute numbers of cells in the indicated organs were determined by trypan blue exclusion. Analysis was performed in 16-22 wk-old mice. (C) Kinetics of CD4+ and (D) CD8+ T-cell reconstitution in NSG mice. Percentages of CD4+ and CD8+ T cells within the CD45+CD3+CD19 cell population in the blood of reconstituted NSG animals over time. (A–D) Results shown are from two separate experiments and are representative of five independent experiments. Each dot represents data from a single mouse. The mean values are indicated by a horizontal bar.

Figure 2.

(A) CD4+CD25+Foxp3+ T-cells generation in NSG mice. CD25 and Foxp3 expression in CD4+SP cells in the thymus and the spleen in a reconstituted NSG mouse. Similar results were obtained in three other mice. Numbers indicate the percentage of each population in each quadrant. (B) Phenotypic characterization of plasmacytoid DC in NSG mice. A first gate on large cells based on forward-size scatter and side-size scatter was established in total splenocytes of humanized mice (left panel, inset top right hand corner). A second gate around HLA-DRloLin (CD3CD19CD14CD56) cells was set (boxed area, left hand side, containing 8.4% of the total population) and the proportions of CD303 and CD123 cells were determined within this gate (right panel). Shown is a representative staining of one out of three mice examined.

Table 1. Seven-color phenotypic analysis of T cells in spleen and mesenteric LN of humanized NSG micea)
 SpleenMesenteric LN
  • a)

    a) Splenocytes and mesenteric LN from 12–20 wk-old NSG mice were stained with fluorescent mAb against human CD45, CD3, CD19, CD4, CD8, CD45RA and CCR7 and analyzed by flow cytometry. Shown are the frequencies of cells of the indicated phenotype within CD45+CD19CD3+CD4+ or CD45+CD19CD3+CD8+ cells±SD determined in six mice in three independent experiments. A representative analysis is shown in Supporting Information Fig. 2.


We then used multiplex PCR to determine the combinatorial diversity of the T-cell repertoire in humanized mice. Four series of experiments were performed using eight different cord blood samples. A total of 22 reconstituted mice were analyzed. Six samples of PBMC from healthy subjects and one thymus from a child were used as references in the study. Each signal obtained for a particular V–J rearrangement was represented by a spike, on a 3-D plot (Fig. 3A). It was already visible at this stage that the repertoire of a reconstituted mouse displayed numerous spikes over the landscape, indicating a wide usage of various V and J elements. In order to evaluate the combinatorial diversity of the immune T-cell repertoire in the thymus, we calculated a combinatorial diversity index for each analyzed thymic sample (Fig. 3B). The combinatorial diversity index represents the percentage of detected rearrangements over the total number of theoretical rearrangements (i.e. 276 for the TRBV chain and 168 for the TRAV chain). Our data revealed that the combinatorial diversity at the TRBV and TRAV loci in the thymus was similar to the reference for four out of six mice (Fig. 3B), showing that the human repertoire generated in the thymus did not suffer major alterations. In order to compare the TRBV repertoire in the thymus and in the periphery of humanized NSG mice, we plotted the mean value of PCR signals detected in the thymus of six individual mice against the one detected in sorted CD45+CD3+ T cells from the spleen. As shown in Fig. 3C, there was a statistically significant correlation between the two, indicating that on average the human repertoire generated in the thymus of NSG mice was well preserved once exported to the periphery.

Figure 3.

(A) Constel'ID analysis of TCR-β chain repertoire diversity. Comparison of PCR signals obtained by multiplex PCR on a human sample of reference (PBMC, left panel) and on a splenic sample of a reconstituted mouse (NSG, right panel). The signal corresponding to each TRBV–TRBJ rearrangement is represented on a 3-D plot by a spike. (B) High level of combinatorial diversity of the T-cell repertoire generated in the thymus of NSG mice. Results show the percentage of detected rearrangements over the total number of possible rearrangements for the α (white bars) and β (black bars) chain of the TCR in the thymus of six NSG mice (C2, C3, C4. C6, M and 5) each reconstituted with CD34+ cells from one of three different donors (CB7, CB8 and CB11) as compared with a reference (ref i.e. thymocytes of a child). (C) Correlation analysis of the mean PCR signals corresponding to single TRBV–TRBJ in six thymuses and six spleen samples. The coefficient of determination (r2) and the p-value calculated by the Prism software are shown.

We next wanted to compare the TRBV repertoire generated in NSG mice and in humans. For this, we compared the mean combinatorial diversity index of NSG mice with six healthy subjects and 26 cancer patients. Cancer patients with lymphocyte counts below 700 cells/μL were used as a reference for a lymphopenic condition in humans. Cancer patients with lymphocyte counts above 1500 cells/μL and healthy subjects were used as non-lymphopenic controls. Indeed, non-lymphopenic cancer patients presented a significant reduction in the diversity of their repertoire compared with healthy subjects (Fig. 4A). The mean combinatorial diversity index at the TRBV locus in NSG mice encompassed 71 and 101% of the mean detected in healthy subjects or in non-lymphopenic cancer patients, respectively. Moreover, the diversity index observed in NSG mice was 145% of the one observed in lymphopenic cancer patients (Fig. 4A). A similar analysis of the combinatorial diversity at the TRAV locus was also performed (although data on cancer patients were not collected in this case). Similar to the TRBV chain, the mean diversity index of the TRAV chain in NSG mice recapitulated 80% of the reference. In addition, the diversity and the composition of the Ig heavy chain repertoire were very similar in human samples and in sorted CD45+CD3 B cells based on the analysis of all 48 possible VH–JH rearrangements (Fig. 4B). Altogether, these results suggest a high diversity of the human immune repertoire generated in NSG mice.

Figure 4.

(A) Combinatorial diversity of the T-cell repertoire in humans and in NSG mice. Percentages of detected rearrangements in the 276 β (left panel) and 168 α (right panel) chain rearrangements in 22 spleen samples of NSG mice, six human PBMC as a reference (PBMC, healthy) and two groups of patients (LP, NLP) suffering a breast cancer relapse. The first group of patients (LP, n=12) had a lymphocyte count below 700 lymphocytes/μL and was considered lymphopenic; the second group (NLP, n=14) had a lymphocyte count above 1500 lymphocytes/μL and was considered non-lymphopenic. Statistical analysis was performed using t-test from the Prism software. Indicated are the p-value calculated by the software (NS, non-significant; **p<0.005; ***p<0.001;). Each dot represent a single sample from the indicated groups. The mean value for each group is indicated by the horizontal bar. (B) Combinatorial diversity of the heavy chain Ig repertoire in NSG mice and correlation with human samples. Percentages of detected rearrangements at the IgH locus in ten NSG mice and in three human samples of reference (PBMC) over the 48 VH–JH theoretical rearrangements (left panel). Correlation analysis of the mean PCR signals obtained for each rearrangement in three PBMC samples and those obtained for each rearrangement in ten humanized NSG mice (right panel). The coefficient of determination (r2) and the p-value calculated by the Prism software are shown.

In an attempt to correlate the quality of the immune repertoire in NSG mice and the immune response, we assessed the ability of reconstituted animals to mount an antigenic-specific immune response. We deliberatly choose a setting where a limited portion of the repertoire would be targeted. We immunized reconstituted NSG mice with a recombinant adenoviral vector encoding the E1 and E2 envelope glycoproteins of HCV (AdHCV), or an adenoviral vector encoding GFP (AdGFP) as a control. As shown in Fig. 5A, there was a tenfold increase in the mean number of HCV-specific IFN-γ-producing cells in AdHCV- compared with AdGFP-immunized mice. This specific response was however modest and detected in only 30% of AdHCV-immunized mice. In two mice that did not respond, a higher specific response was observed when mesenteric LN cells were added to the culture (Fig. 5B). More importantly, the HCV-specific response was still observed when sorted human CD45+ cells were used (Fig. 5C). Indeed, a higher proportion of responding mice (80%) was detected after sorting out murine cells. These data suggested that the HCV-specific response was HLA-restricted rather than elicited by murine MHC molecules. In addition, we detected neutralizing antibodies against the adenoviral vector in some but not all immunized mice (Fig. 5D). This neutralization correlated with the detection of anti-adenoviral antibodies of the IgM isotype (data not shown). This result was consistent with a weak in vitro cellular immune response against adenoviral antigens (data not shown). To conclude, our data suggest that a specific albeit low-grade cellular immune response is elicited in humanized NSG mice immunized with a recombinant adenoviral vaccine encoding HCV antigens.

Figure 5.

(A) HCV-specific cellular immune response after a single immunization in NSG mice. IFN-γ production assessed by ELISPOT from splenocytes of NSG mice immunized with either a single dose of 108 pfu of Ad5 defective vector either encoding GFP (AdGFP, n=5) or the E1/E2 glycoproteins of HCV (AdHCV, n=10). Unsorted splenocytes from each immunization protocol were stimulated with a VSV-G pseudotyped lentiviral vector encoding the E1/E2 glycoproteins of HCV or with an empty lentiviral vector 9–11 days later. Represented is the difference in the number of SFU per million cells (ΔSFU) in the two stimulation conditions, hence there are negative values for some mice. Results shown were obtained in five independent experiments. Each dot is a single mouse. The horizontal bar indicates the mean value for this particular group (B) Cellular immune response in splenocytes and gangliocytes from HCV-vaccinated humanized mice. IFN-γ production by splenocytes alone (spleno.) or a three to one mixture of splenocytes and mesenteric LN cells (spleno+LN) of AdHCV and AdGFP immunized mice upon restimulation with lentiviral vectors as described in (A) in two mice tested out of the ten represented in (A) in two independent experiments. (C) Cellular immune response in human cells from HCV-vaccinated mice. IFN-γ production by sorted human CD45+ cells of AdHCV and AdGFP immunized mice upon restimulation with lentiviral vectors as described in (A) in five mice tested out of the ten represented in (A) in three independent experiments. (D) Kinetics of appearance of adenovirus neutralizing antibodies in serum of three different mice (S2, S3 and S4) after two intra muscular injections of 108 pfu AdHCV defective vector starting 30 days after the first immunization (days post-boost). Percentage of neutralization was determined at 1/100 dilution of serum. Similar results were obtained at lower dilutions although the background tended to increase. Serum from non-immunized humanized mice is represented by the dotted line on the graph. Percentage of neutralization was calculated as described in the Materials and methods.


In this study, we report that the combinatorial diversity of the NSG mice repertoire recapitulates up to 100% of the one detected in humans. We also observed that the diversity originated in NSG mice was clearly superior to lymphopenic cancer patients. Thus,T cells in the periphery of NSG mice did not originate solely from lymphopenia-induced proliferation, in which case a constriction of their repertoire would have been expected 22, 23. That human T cells originated from spontaneous post-thymic expansion rather than homeostatic proliferation (driven by self-antigens) is also supported by our observations that the diversity of the peripheral repertoire in NSG mice was not vastly different from the one observed in the thymus and that a high proportion of naïve T cells were observed in the periphery. The high combinatorial diversity of the human T-cell repertoire detected in the murine thymus also suggests that the recombination “machinery” at the origin of V(D)J rearrangements was fully functional in this organ. Previous report showed that IL-7 was able to enhance human T-cell reconstitution in NSG mice 15, suggesting that murine IL-7 produced by thymic epithelial cells participate to the development of immature human progenitors in the murine thymus 24. Whether IL-7 also contributes to the diversity of the TRBV–TRAV repertoire, as recently demonstrated in humans 25, remains a matter for future studies. It is also unclear whether human HLA molecules expressed in the thymus (presumably by progenitors or resident DC) played a role in the generation of repertoire diversity. It has been shown that human T cells generated in BALB/c RAG-2−/−.γc−/− mice are tolerant to BALB/c but not to C57Bl/6 background and also less reactive against donor-derived DC than against allogeneic DC 2. We thus speculate that the humanized mice repertoire was selected on both human and murine MHC molecules recognition. Reconstitution experiments in either murine MHC-KO humanized mice or in mice transgenic for human HLA molecules will clarify this point.

On the other hand, we observed that male mice invariably fell ill 20–24 wk after reconstitution and suffered from various symptoms such as weight loss, anemia or skin afflictions. These symptoms correlated with expansion of activated CD4+CD3+ T cells in peripheral lymphoid organs (data not shown), a phenotype remminiscent of a strong homeostatic proliferation. Adverse effects following immune reconstitution in NOG mice have also been reported 8 and lymphopenia in NOD mice is known to be associated with autoimmunity 26. Noticeably, activated CD4+ and CD8+ T cells could be detected up to one year after reconstitution in female mice (data not shown) with no apparent signs of any disease. Currently, we do not hold a satisfactory explanation for these clinical symptoms in males. It may be related to our relatively “standard” breeding conditions. It has also been shown that estrogens have a strong impact on regulatory T cells and notably foxp3 expression 27. Moreover, regulatory T cells are thought to control xeno-reactivity in RAG2−/−.γc−/− mice 28. Therefore, one possible explanation would be that male mice develop a low-grade chronic xeno-reactivity as they would carry less efficient regulatory T cells. Recent reports in RAG2−/−.γc−/− mice showed that CD4+CD25+Foxp3+ human cells were indeed true regulatory T cells able to suppress CD3/CD28 as well as SEB-mediated proliferation 29, 30. Adding to the imperfections of the model, we observed that all NSG mice examined so far carried a single mesenteric LN with no other peripheral LN, even after immunization. Although not stated in previous studies, this observation seems widely shared. We have performed a thorough examination of the literature and noticed that the mesenteric LN is the only LN reported so far in various models of humanized mice. However, Sun et al. 31 reported the presence of peripheral LN in BLT models. The lack of LN in γc-deficient mice may be related to poor differentiation of lymphoid-tissue inducer cells in this strain, which were recently identified as NK cells producing IL-22 and IL-17 in humans 32. Improvements in reconstitution protocols are still the focus of numerous studies 33.

We then showed that the diversity generated in mice translated into a weak functional immune response in a limited number of mice using a potential HCV vaccine. We showed however that the low frequency of NSG mice responding to the vaccine more likely reflect the poor sensitivity of the assay using total splenocytes than a poor functionality of the repertoire. Indeed, we could detect a specific immune response in 80% of the immunized NSG mice if human cells were isolated before the assay. Also, the E1/E2 antigens of HCV probably target a limited number of clones within the repertoire. This assumption is supported by the low frequency of IFN-γ secreting cells in our ELISPOT assay (around 15 cells/million human splenocytes), albeit lower than the one elicited in wild type BALB/c mice that were tested in the same conditions (80 cells/million) 34. It is possible that missing specificities in the repertoire of humanized mice prevented an efficient response against E1 and E2 of HCV. The immune response elicited by the same vector in humans is however unknown, making the relative quantification of the response difficult. Moreover, low-grade E1 and E2 immune responses have been reported in acute and chronic HCV patients 35, 36. Besides the demonstration that sorting human cells is key to detect rare antigen-specific T cells in humanized mice, our results suggest that the major restriction elements of the HCV-specific response could be HLA molecules. However, these results do not rule out a possible role for murine antigen presenting cells in vivo, as proposed earlier in RAG2−/−γc−/− mice 2. Our data revealed the presence in humanized NSG mice of cells with a plasmacytoid DC phenotype and it is possible that these cells may have primed T cells in vivo. Further studies are needed to determine the implications of human versus murine MHC molecules and of human plasmacytoid DC in the immune response elicited in humanized animals. We and others have been unable to detect antigen-specific IgG antibodies in humanized mice 4, 19. A robust IgG secretion has however been reported in immunized NSG mice and also in non-immunized RAG-2−/−.γc−/− mice 2, 37. Therefore, there is still some uncertainty relative to T–B cooperation in humanized mice.

In summary, we have shown a high combinatorial diversity of the T-cell and B-cell repertoires in humanized NSG mice that can translate into a weak but specific cellular immune response after immunization with a genetic vaccine. Therefore, our results support the usefulness of humanized mice for the evaluation of some aspects of vaccine development.

Materials and methods


NOD.Cg-Prkdcscid Il2rgtm1Wjl/SzJ (NSG) (stock ≠ 005557, The Jackson Laboratory) were bred in our own animal facilities in SPF conditions (accreditation number from the Veterinary services: A75-13-10) with an enriched fat regime and addition of Bactrim in drinking water every other week. The colony was regularly checked for γ-c deficiency by PCR according to the Jackson Laboratory protocol. The Regional Ethical Committee on Animal Experimentation has approved the project the October 17, 2006. 24–48 h-old NSG neonate mice were irradiated at 1 Gy and 105 to 3×105 purified human CD34+ cells were directly injected in the liver using a 300 μL insulin syringe (Terumo, Tokyo, Japan) to generate reconstituted NSG mice.

CD34+cell purification

Umbilical cord blood was obtained from regular birth at the maternity of the Hospital La Pitié-Salpêtrière after informed consent of the mother. Mononuclear cells were enriched on a Ficoll (Sigma-Aldrich, St Quentin, France) gradient. Cells were resuspended at 5.108 cells/mL and 10% of the volume of CD34 CliniMacs purification kit (Miltenyi Biotech, Paris, France) was added for a 30 min incubation with gentle agitation. Cells were then adjusted to a concentration of up to 2.108 cells/mL for immunomagnetic sorting on two consecutive LS and MS columns (Miltenyi Biotech).

Antibodies and cell sorting

For flow cytometry analysis, the following mAb were used: for phenotypic characterization; CD45-allophycocyanin (APC) (Becton Dickinson, Le Pont de Claix, France), CD45-PB (Pacific Blue) (Biolegend, San Diego, CA, USA); CCR7-PE (Becton Dickinson); CD45RA-FITC (Fluoro-isothiocyanate) (Becton Dickinson); CD8-peridinin-chlorophyll-protein (Becton Dickinson), CD8-APC-Cy7 (Becton Dickinson) or CD8-PE-Cy7 (Becton Dickinson); CD4-Alexa700 (Biolegend); CD3-FITC (Becton Dickinson), CD3-PE-Cy7 (eBioscience, San Diego, CA, USA), or CD3-PE-Cy5 (Dako), or CD3-peridinin-chlorophyll-protein-Cy5.5 (Becton Dickinson), or CD3-APC (Becton Dickinson); CD19-PB (Biolegend); Foxp3-Alexa488 (eBioscience); for phenotypic characterization of DC; CD3, CD56, CD14, CD19, labeled with PE (Becton Dickinson) to exclude Lin+cells; HLA-DR-Alexa700 (Biolegend); CD303-APC (Miltenyi); CD123-PE-Cy7 (Biolegend). Up to 106 cells were stained with optimal quantity of directly labeled mAb at room temperature for 20 min under gentle shaking with 100 μL of the mix containing all of the mAb of interest diluted in sterile PBS SVF 3%. The compensation matrix was set up using CompBeads (Becton Dickinson) complexed to each mAb present in the mix. Data were acquired on a LSRII flow cytometer equipped with solid-state coherent Sapphire™ (488 nm, 20 mW) and Vioflame™ (405 nm, 25 mW) lasers and with a Helium-Neon gas JDS Uniphase 1334 P (633 nm, 17 mW) laser using DiVA software (Becton Dickinson) and analyzed with FlowJo (TreeStar, Arshland, Oregon) with in some instances, slight modifications of the original matrix. Total cells from the spleen were stained with optimal amount of CD45-APC (Becton Dickinson) and CD3-FITC (Becton Dickinson) and sorted on a FACSAria (Becton Dickinson) with purity greater than 90%.

Multiplex PCR assay

Genomic DNA was extracted using standard techniques from sorted cells in NSG mice and from total peripheral blood cells in cancer patients after informed consent. Thymic sample from a child undergoing corrective cardiovascular surgery has been obtained in collaboration with the surgical team of Pr Alain Serraf at the Centre Chirurgical Marie-Lannelongue (CCML, Service de Chirurgie Thoracique, Le Plessis Robinson, France), after informed consent of parents. Multiplex PCR was performed as described 15 using an upstream primer specific of all functional members of a given V family and a downstream primer specific of a given J segment. This assay allows the simultaneous detection of several V–J rearrangements in the same reaction. By using this technique, it is possible to detect 276 different TRBV–TRBJ rearrangements and 168 TRAV–TRAJ rearrangements covering, respectively, 100 and 25% of the possible TRBV and TRAV combinatorial rearrangements. Similarly, it is possible to detect 48 different IgHV–IgHJ rearrangements covering 100% of the possible rearrangements at both loci. All V–J1, J2, J3, J4, Jn products were separated as a function of their size with a maximum amplicon size of ∼5 kb. PCR signals were detected and analyzed using the Constel'ID software developed by ImmunID Technologies (France). It is here important to mention that the multiplex PCR strategy could potentially amplify a DNA fragment corresponding to a rearrangement involving a Vx+1 gene located 3′ of the targeted Vx gene. However, these side products were recognized and excluded from the final results by the Constel'ID software. PCR products were generated using 1.3 unit/reaction of Expand High Fidelity PCR system (Roche Diagnostics, Meylan, France). The cycling conditions were 5 min at 94°C, 26 cycles of 1 min at 94°C, 1 min at 58°C, 6 min at 72°C, and one cycle of 10 min at 72°C. In order to perform semi-quantitative analysis, PCR reactions were stopped at the exponential step of the PCR. In order to normalize DNA quantity in each reaction the actin gene was amplified in the same PCR run. DNA isolated from kidney and/or HEK cell lines, in which no rearrangement takes place, was used in a negative PCR control. PCR products were separated on a 1% agarose gel, directly stained with SyberGreen I and quantified using a CCD camera equipped with BIO-1D (Vilbert Lourmart, France) and Constel'ID softwares for further analytical studies.

Lentiviral construction, production and titration

To generate HCV-recombinant or control VSV-G pseudotyped lentiviral particles, 293 T cells seeded at 3×106 cells in 10 cm plates were cultivated in DMEM medium (Invitrogen, France) supplemented with 2 mM L-glutamine, 100 U/mL of penicillin and streptomycin (Invitrogen) and 10% heat-inactivated fetal calf serum. The next day, cells were transfected with the lentiviral Gag-Pol pCMVR8.91 (10 μg), the VSV-G envelope pMΔG (5 μg) with or without the packaging-competent E1/E2 HCV glycoproteins-expressing plasmid (10 μg) using a calcium phosphate transfection protocol, as described 20. An HIV p24-specific ELISA assay (HIV-1 p24 antigen ELISA Kit, Zeptometrix, NY, USA) was used to determinate the p24 concentrations of the lentiviral vector stocks after ultrafiltration concentration (Centricon, Millipore, St Quentin, France).

Immunizations and ELISPOT assays

Mice were immunized intramuscularly with recombinant adenovirus (ΔE1/ΔE3) expressing the E1 and E2 HCV envelope glycoproteins derived from the 1a HCV genotype or expressing the GFP protein (kind gift of Dr. S. Kochanek, Division of Gene Therapy, University of Ulm, Germany). The cellular immune response against E1 and E2 glycoproteins was evaluated 9–12 days later by using the ELISPOT IFN-γ assay (Mabtech, Nacka Strand, Sweden), according to the manufacturer's instructions. Cells were added in triplicate wells (3.105 cells/well or 4.105 cells/well when gangliocytes were added) in RPMI supplemented with 2 mM L-glutamine, 100 U/mL of penicillin and streptomycin (Invitrogen) and 1% heat-inactivated mice serum and incubated for 20 h at 37 C in 5% CO2 in presence of 20 ng p24 per well of HCV-expressing or control lentiviral particles (Supporting Information Fig. 3). Genome-free lentiviral particles pseudotyped with the VSV-G envelope were used as a control for HCV-specific restimulation. After revelation, spots were counted using an ELISPOT reader (AID Reader System ELR03, Autoimmun Diagnostika, Germany). Due to variations in the background among various mice in independent experiments, we calculated a difference in the number of spots (ΔSFU) between the HCV-stimulated and the VSV-G-stimulated conditions in all of our experiments.

Neutralization assay and ELISA

TE671 cells were seeded (1x105/well) in a 24-well plate and incubated overnight at 37°C 5% CO2. Heat inactivated sera were diluted into DMEM+2% FBS (1:20–1:100), and pre-incubated for 1 h at 37°C with a rAd5-GFP (multiplicity of infection=2) in DMEM+2% FBS. The mixture of rAd5-GFP+ serum was added to the plate and incubated for 1 h 30 min at 37°C. Then, cells were washed in PBS 1× and incubated in fresh medium for 48 h at 37°C. The percentage of GFP-expressing cells was determined by FACS analysis (FACScalibur, Becton Dickinson) and the percentage of neutralization was calculated as follows: [1%–(of GFP-expressing cells/percentage of GFP−expressing cells in controls without serum)]×100. Isotype of anti-adenoviral antibodies were determined using ELISA kits (IBL, Hamburg, Germany), according to the manufacturer's instructions.

Statistical analysis

Two-tailed unpaired t-test with 95% confidence intervals and correlation statistical analyses were performed using Prism 4.0c for Macintosh (GraphPad Software, San Diego, CA, USA). The mean values were considered statistically different when the p-value was below 0.05.


This work received financial supports from the Agence Nationale de la Recherche contre le SIDA (ANRS), the Association Francaise contre les Myopathies (AFM) to GM and the FP6 European Compuvac and Clinigene grants to D. K. The authors would like to thank Pr F. Lemoine, Dr. A. Six, Dr. B. Salomon, Dr. J. Cohen (CNRS UMR 7211/INSERM U959) and Pr G. Gorochov (INSERM UMR-S 945, Hopital de la Pitié Salpétrière, Paris, France) for helpful suggestions and critical reading of the manuscript; Dr. S. Cohen-Kaminsky (CNRS UMR 8162, Université Paris Sud, Le Plessis Robinson, France) for the kind gift of the human thymic sample and Dr. S. Kochanek (University of Ulm, Germany) for the kind gift of adenoviral vectors. In addition, we thank B. Bault (CNRS UMR 7211, INSERM U959), N. Plantier, S. Perez, C. Alberti-Segui (ImmunID Tech.) and G. Boisserie (Service de radiothérapie, Hopital de la Pitié Salpétrière, Paris, France) for their help in various aspects of this work.

Conflict of interest: The authors declare no financial or commercial conflict of interest.