Distant lymph nodes serve as pools of Th1 cells induced by neonatal BCG vaccination for the prevention of asthma in mice

Authors


  • Edited by: Angela Haczku

Correspondence

Dr Hua-Hao Shen, Department of Respiratory and Critical Care Medicine, Second Affiliated Hospital, Zhejiang University School of Medicine, Hangzhou 310009, China.

Tel.: (86) 571 87783729

Fax: (86) 571 87783729

E-mail: hh_shen@yahoo.com.cn

Abstract

Background

Neonatal Bacillus Calmette–Guérin (BCG) vaccination induces vigorous T-helper type 1 (Th1) responses and inhibits allergy-related airway dysfunction, but the exact mechanisms remain unclear. The objective of this study was to address where the Th1 cells induced by neonatal BCG vaccination are generated and stored, and how they are recruited into the inflamed airway for the prevention of allergen-induced airway inflammation.

Methods

We vaccinated neonatal C57BL/6 mice with BCG in a mouse model of asthma and analyzed the expression and function of Th1 cells in vivo and in vitro.

Results

BCG vaccination–induced Th1 cells in the local inguinal lymph nodes (ILN) migrated into the lungs upon inhaled ovalbumin (OVA) challenge in OVA-sensitized mice. These CD4+ T cells in the ILN exhibited potentials of activation, proliferation and cytokine secretion and expressed high levels of CXCR3. Adoptive transfer of CD4+ T cells from BCG-treated ILN significantly decreased allergic airway responses. In addition, the protective effect of BCG vaccination against allergic airway inflammation was lost upon the excision of the ILN.

Conclusions

These data demonstrate that ILN serves as a ‘weapon’ pool of Th1 cells following BCG vaccination, and these cells are ready for the migration into the inflamed lungs upon the allergen challenge, thereby inhibiting allergen-induced airway disorder.

Abbreviations
BCG

Bacillus Calmette–Guérin

ILN

Inguinal lymph nodes

OVA

Ovalbumin

CFSE

5-(and-6)-carboxyfluorescein diacetate, succinimidyl ester

MLN

Mediastinal lymph nodes

Allergic asthma is an immunodysregulatory disease that is mediated by excessive T-helper type 2 (Th2) immune responses to allergens in the lung [1-4]. In contrast, antigen-induced Th1 responses and associated IFN-γ production can down-regulate Th2 responses, thereby inhibiting allergen-induced airway inflammation [5, 6].

Vaccination with Bacillus Calmette-Guérin (BCG) induces strong Th1 responses and Th1 cells infiltrating into the lung and then modulates the development of asthma in both rodents and humans [7-14]. The accumulation of IFN-γ-expressing T cells may result from the migration of activated T cells from circulation and/or the rapid expansion of residual memory T cells in the lung [15, 16]. Notably, BCG vaccination or chronic infection of mycobacterium tuberculosis (M. tb) usually fails to establish an effective memory T-cell immunity due to the long-term presence of sustainable antigen [17, 18]. Furthermore, effector T cells can migrate into nonlymph target tissues during the process of inflammation [19, 20]. Therefore, it is likely that BCG-induced Th1 cells behave like these aforementioned effector T cells. However, where the IFN-γ-expressing Th1 cells induced by BCG vaccination are generated and stored and how these cells are recruited into the inflamed airway following OVA challenge have not been clarified.

In this study, we tested the hypothesis that some nickels might exist in the body where the enriched BCG-specific Th1 cells were stored and ready for migration into the inflamed lung to protect against allergic airway inflammation. We found inguinal lymph nodes (ILN), not the spleen or mediastinal lymph nodes (MLN), served as a pool of Th1 cells that were then recruited into the lung and ameliorated OVA-induced airway inflammation.

Materials and methods

Mice

C57BL/6 mice were from the Experimental Animal Center of Zhejiang University and housed in a specific pathogen-free facility. Both male and female neonates were used, and the experimental protocols were approved by the Ethical Committee for Animal Studies at Zhejiang University, China.

BCG vaccination and experimental protocol for allergen-induced asthmatic model

The details are presented in Supplement S1.

Analysis of BALF

The details are presented in Supplement S1.

Lymphocyte isolation from the lung, spleen, and lymph nodes

The details are presented in Supplement S1.

Preparation of lung homogenate

The details are presented in Supplement S1.

Histological analysis of the lungs

The details are presented in Supplement S1.

Real-time PCR

The details are presented in Supplement S1.

Cell culture, activation, and proliferation assays

The details are presented in Supplement S1.

Analysis of cytokines

The details are presented in Supplement S1.

Flow cytometric analysis

The details are presented in Supplement S1.

CD4+ T-cell isolation

The details are presented in Supplement S1.

Adoptive transfer assay

Neonatal C57BL/6 mice were vaccinated with BCG or injected with saline. These mice at 12 weeks of age were killed and their CD4+ T cells from the ILN or spleen were isolated as donor cells, respectively. The purified CD4+ T cells from the ILN or spleen were transferred intravenously into individual C57BL/6 recipients (2 × 106/mouse) that had been sensitized i.p. with OVA/alum; 24 h later, the recipients were challenged with 1.0% OVA for three consecutive days. After the last OVA challenge, airway inflammation of individual mice was measured. To characterize the migration of CD4+ T cells in vivo, the purified CD4+ T cells from the ILN of BCG-vaccinated or control mice were labeled with 0.2-μm 5-(and-6)-carboxyfluorescein diacetate, succinimidyl ester (CFSE) for 5 mins in vitro and washed; the CFSE-labeled CD4+ T cells (2 × 106/mouse) were used for the adoptive transfer assay, as described above. Forty-eight hours after the last OVA challenge, the frequency of ILN CD4+ T cells in the lungs was analyzed by flow cytometry.

ILN resection

Neonatal C57BL/6 mice were vaccinated with BCG or injected with saline. The ILN from the mice was excised surgically 10 days before OVA challenge (termed BCG/ILN excision/OVA and saline/ILN excision/OVA), and the ILN from the mice without BCG vaccination and OVA challenge was also excised as the control (termed saline/ILN excision/saline). The wound was usually healed within 7 days after the operation.

Chemotaxis assay

The details are presented in Supplement S1.

Statistical analysis

Data are expressed as the mean ± SEM. Data were analyzed for their normality by the Kolmogorov–Smirnov test. Differences in normally distributed data among the groups were analyzed by one-way analysis of variance (one-way anova) or by an unpaired Student's t-test using SPSS 13.0 software. The Kruskal–Wallis H-test was used for non-normal data. A P value of ≤0.05 was considered statistically significant.

Results

Neonatal BCG vaccination enhances Th1 cells, but not Th17 or Foxp3+CD4+ Treg recruitment in the lung after OVA challenge

Similar to our previous study [13], neonatal vaccination with BCG inhibited airway inflammation and mucus production (Fig. S1). Next, the impact of BCG vaccination on the infiltration of T cells in the lung following OVA challenge was examined. Mice were injected with saline or BCG and sensitized/challenged with saline or OVA (termed saline/saline, BCG/saline, saline/OVA, and BCG/OVA). The frequency of IFN-γ+CD4+, but not IFN-γ+CD8+, IL-17A+, and Foxp3+CD4+, T cells in the lung of BCG/OVA mice was significantly higher than that in the BCG/saline and saline/saline mice (6.15 ± 0.42% vs 1.47 ± 0.14% and 6.15 ± 0.42% vs 1.24 ± 0.07%, respectively, both P < 0.001) (Fig. 1).

Figure 1.

An increased frequency of Th1 cells in the lung of BCG-vaccinated mice upon OVA challenge. Groups of neonates were injected with saline or BCG and sensitized/challenged with saline or OVA later (termed saline/saline, saline/OVA, BCG/saline, and BCG/OVA). Forty-eight hours after the last OVA challenge, their lung tissues were dissected out for further analysis. (A–C) Representative FCM plots (A), quantitative analysis (B) of CD3+IFN-γ+ T cells, and representative FCM plots of CD3+IL17+ T cells (C) in the lung. Lymphocytes were isolated from the lungs of different groups of mice and stimulated with PMA/ionomycin in vitro for 5 h. The cells were then stained with PE-Cy5-CD3, FITC-CD8, and PE-IFN-γ or PE-IL-17A and gated on CD3 for FACS analysis. CD3+CD8- T cells were considered as CD4+ T cells in this experiment. (D) Representative FCM plots of Foxp3+CD4+ T cells in the lung. Data are expressed as the mean ± SEM of individual groups (n = 6–8 mice per group) from two separate experiments. *P < 0.05; ***P < 0.001.

OVA challenge promotes the migration of BCG vaccination–induced Th1 cells in the ILN

We next sought to examine the origin of the increased Th1 cells in the lung of BCG/OVA mice. The frequency of splenic IFN-γ+CD4+, but not IFN-γ+CD8+, T cells in the BCG/OVA mice was almost 3-fold higher than that in other groups of mice (all P < 0.001) (Fig. 2A), but MLN IFN-γ+CD4+ T cells in the OVA-challenged mice was significantly lower than that in the nonchallenged mice (Fig. 2B). These findings suggest that OVA challenge–induced Th2 responses in MLN counteracted the development of Th1 cells [21, 22]. The percentage of ILN IFN-γ+CD4+, but not IFN-γ+CD8+, T cells in the ILN of BCG/saline mice was significantly higher than that in BCG/OVA mice (0.69 ± 0.14% vs 0.36 ± 0.05%, P < 0.01) (Fig. 2C). The ILN size, but not the spleen, from the BCG-vaccinated mice was significantly larger than that from the saline mice (Fig. 3A), accompanied with significantly increased total numbers of ILN lymphocytes (data not shown). Of note, the number of IFN-γ-expressing CD4+ T cells in the BCG-vaccinated ILN was reduced by 77.15% after OVA challenge (164.61 ± 16.21 × 103 vs 37.61 ± 9.52 × 103, P < 0.001) (Fig. 3B). These results suggested that OVA challenge triggered a migration of Th1 cells out of the ILN.

Figure 2.

IFN-γ-expressing T cells in different organs. Newborn mice were injected with saline or BCG and sensitized/challenged with saline or OVA later (termed saline/saline, saline/OVA, BCG/saline, and BCG/OVA). Forty-eight hours after the last OVA challenge, their spleen, MLN, and ILN were dissected out for further analysis. Lymphocytes were prepared and stimulated with PMA/ionomycin for 5 h. The cells were then stained with PE-Cy5-CD3, FITC-CD8, and PE-IFN-γ and gated on CD3 for FACS analysis. (A–C) IFN-γ-expressing CD3+CD8 (CD4+ T cells) and CD3+CD8+ T cells in the spleen (A), MLN (B), and ILN (C). Data are expressed as the mean ± SEM of individual groups (n = 6–8 mice per group) from two separate experiments, *P < 0.05; **P < 0.01; ***P < 0.001.

Figure 3.

The size of ILN and the number of IFN-γ-expressing CD4+ T cells in the ILN of BCG-vaccinated mice. Newborn mice were injected with saline or BCG and sensitized/challenged with saline or OVA later (termed saline/saline, saline/OVA, BCG/saline, and BCG/OVA). Forty-eight hours after the last OVA challenge, their spleens and ILN were dissected out for further analysis. (A) Morphologies of the ILN and spleen. (B) The number of IFN-γ-expressing CD4+ T cells in the ILN. Lymphocytes were prepared and stimulated with PMA/ionomycin for 5 h. The cells were then stained with PE-Cy5-CD3, FITC-CD8, and PE-IFN-γ and gated on CD3 for FACS analysis. CD3+CD8 T cells were considered as CD4+ T cells in this experiment. The number of IFN-γ+CD4+ T cells in the ILN = the frequency of IFN-γ+CD4+ T cells in the ILN× total cell counts in the ILN. Data are expressed as the mean ± SEM of individual groups (n = 6–8 mice per group) from two separate experiments, ***P < 0.001.

Phenotype and function of CD4+ T cells in the ILN of BCG-vaccinated mice

CD4+ T cells can be divided into naive (CD44lowCD62high Tnaive), central memory (CD44highCD62high Tcm), and effector/memory T cells (CD44highCD62Llow Tef/em) [23-25]. In comparison with the other groups, increased frequency of ILN Tef/em, but decreased Tcm cells were detected in the BCG-vaccinated mice (Fig. 4A and 4B). Following anti-CD3e/CD28 stimulation, significantly higher frequency of ILN CD25+CD4+ T cells was observed in the BCG-vaccinated mice (Fig. 4C and D). The CFSE-labeled ILN CD4+ T cells from the BCG-vaccinated mice displayed more divisions than controls (Fig. 4E), accompanied by significantly higher level of IFN-γ production (1526.36 ± 27.13 vs 477.96 ± 6.73 pg/ml, P < 0.001) (Fig. 4F). Therefore, these BCG vaccination–induced CD4+ T cells in the ILN exhibited high potentials of activation, proliferation, and IFN-γ secretion in vitro.

Figure 4.

Phenotype and functional analysis of CD4+ T cells in the ILN. Neonatal mice were vaccinated with saline or BCG. At 12 weeks of age, their ILNs were dissected and their lymphocytes were isolated for further analysis. (A and B) Representative FCM plots (A) and quantitative analysis (B) of naive CD4+ T cells (CD44lowCD62Lhigh, Tnaive), effector/effector memory CD4+ T cells (CD44highCD62Llow, Tef/em) and central memory CD4+ T cells (CD44highCD62Lhigh) in the ILN. Lymphocytes from the ILNs were stained with antibodies against CD4, CD62L, and CD44 and gated on CD4 for FACS analysis. Data are shown as the mean ± SEM (n = 3–5 mice per group, *P < 0.05; **P < 0.01). (C and D) FCM plots of CD25+CD4+ T cells in the ILN without in vitro stimulation (C) or stimulated with anti-CD3ε/CD28 in vitro for the indicated periods (D). (E) The proliferation of CD4+ T cells in the ILN was determined by CFSE dilution assay. These results in (C), (D), and (E) present one of the two independent experiments with similar results (n = 3–5 per group for each experiment). (F) The concentrations of IFN-γ in the supernatants of cultured lymphocytes from the ILN were detected by ELISA. Data are shown as the mean ± SEM of individual groups (n = 3–5 mice per group) from two separate experiments; ***P < 0.001.

Distant ILN Th1 cells have potential to migrate into inflamed lungs and directly inhibit airway inflammation

To test the function, ILN or splenic CD4+ T cells from the BCG-vaccinated or saline-injected mice or control PBS were transfused into OVA-sensitized mice (termed BCG-ILN/OVA, saline-ILN/OVA, BCG-spleen/OVA, saline-spleen/OVA, or PBS/OVA). The BCG-ILN/OVA mice displayed less eosinophil infiltrates, pulmonary inflammation, and mucus production, but higher IFN-γ in BALF and greater number of CFSE+CD4+ T cells in the lung than the other groups (Fig. 5). However, adoptive transfer with the same number of splenic CD4+ T cells failed to modulate airway inflammation and mucus production (data not shown). As expected, the protection of ILN CD4+ T cells was abrogated by presurgical removal of the ILN from the BCG-vaccinated mice following OVA challenge (Fig. S2), indicating the importance of ILN CD4+ T cells.

Figure 5.

Adoptive transfer of CD4+ T cells from the ILN of BCG-vaccinated mice attenuates allergic airway inflammation. (A–D) CD4+ T cells were isolated from the ILNs of the BCG-vaccinated or saline-injected control mice at 12 weeks of age and intravenously transferred into the recipient mice (2 × 106/mouse) that had been sensitized with OVA. One day later, the recipients were challenged with OVA for three consecutive days. Forty-eight hours after the last OVA challenge, the recipients were killed. (A) The numbers of total and differentiated cell counts in BALF. Eos: eosinophils, Mac: macrophages, Neut: neutrophils, and Ly: lymphocytes. Data are expressed as the mean ± SEM of individual groups (n = 8–10 mice per group). *P < 0.05; **P < 0.01. (B) Representative photomicrographs of lung inflammation (H&E staining, upper panels) and mucus secretion (PAS staining, lower panels). (Bar = 100 μm). (C) Quantification of mucus production. PAS score was described in Supplement S1. (n = 8–10 mice per group). *P < 0.05. (D) The concentrations of IFN-γ in BALF were detected by ELISA. Data are expressed as the mean ± SEM of individual groups (n = 8–10 mice per group). *P < 0.05. (E) Quantitative analysis of CFSE+CD4+ T cells in the lung. CD4+ T cells were purified from the ILN of BCG-vaccinated or control mice and labeled with 0.2-μm CFSE. After CFSE labeling, these CD4+ T cells (2 × 106/mouse) were used for the adoptive transfer assay, as described above. The number of CFSE+CD4+ T cells in the lung = the frequency of CFSE+CD4+ T cells in the lung × total cell counts in the lung. The control group was considered as the basal level (designated 1), and the other groups were expressed as fold changes compared with that in the control. The fold change values are shown as the mean ± SEM of individual groups (n = 8–10 mice for per group), *P < 0.05.

High levels of CXCR3 expression on CD4+ T cells in the ILN from the BCG-vaccinated mice and high levels of CXCL9 expression in the OVA-challenged lungs are associated with the potential of CD4+ T-cell migration

The relative levels of CXCR3, but not CCR5 and CXCR6, mRNA transcripts in the ILN of BCG-vaccinated mice were 3-fold higher than those in the control mice (Fig. 6A). The relative levels of CXCL9, CXCL10, and CXCL11 mRNA transcripts in the lung from the OVA-challenged mice were significantly higher than those in the control mice (Fig. 6B). In addition, BCG vaccination increased the CXCR3 protein expression on CD4+ T cells in the ILN compared with that in the controls (14.84 ± 1.49% vs 7.84 ± 1.12%, P < 0.05) (Fig. 6C). The concentrations of CXCL9 and CXCL10, but not CXCL11, in the lung of BCG/OVA mice were significantly higher than those in the saline/saline mice (Fig. 6D). Furthermore, dramatically increased frequency of migrated ILN CD4+ T cells, in response to CXCL9, was detected in the BCG-vaccinated mice (10.83 ± 2.15% vs 2.42 ± 1.59%, P < 0.05) (Fig. 6G).

Figure 6.

High levels of CXCL9 in the lung of asthmatic mice facilitate the migration of CD4+ T cells from the ILN of BCG-vaccinated mice via high expression of CXCR3. (A and B) The relative levels of CXCR3, CCR5, and CXCR6 mRNA transcripts in the ILN (A) and CXCL9, CXCL10, and CXCL11 in the lung (B) were determined by real-time PCR. Data are expressed as the mean ± SEM of individual groups of mice from two independent experiments. (n = 6–8 mice per group, *P<0.05; **P<0.01). (C) Representative FCM plots (left panels) and the quantified analysis (right panels) of CXCR3+CD4+ T cells in the ILN from mice vaccinated with or without BCG. The cells were gated on CD4 for FACS analysis. Data are expressed as the mean ± SEM of individual groups (n = 3–4 mice per group) from two independent experiments, *P < 0.05. (D–F) The concentrations of CXCL9 (D), CXCL10 (E), and CXCL11 (F) in the lung homogenates were tested by ELISA. Data are expressed as the mean ± SEM of individual groups of mice (n = 6–8 mice per group, *P < 0.05). (G) The migration of CD4+ T cells in the ILN of BCG-vaccinated and saline-injected mice in response to chemokine CXCL9 in vitro. CD4+ T cells were isolated and their chemotactic responses to CXCL9 were analyzed by transwell assay as described in Supplement S1. Data are expressed as the mean% ± SEM of individual groups of mice from two independent experiments. (n = 4–6 mice per group, *P < 0.05).

Discussion

In this study, the ILN was identified as a pool of Th1 cells and contributed to the protective effects of neonatal BCG vaccination on allergen-induced airway inflammation. Our data have several implications for understanding the mechanisms of BCG vaccination for the prevention of allergic airway inflammation. First, we have confirmed that enhanced Th1 immune responses induced by neonatal BCG vaccination play a central role. Importantly, we have demonstrated, for the first time, that it is the distant draining lymph nodes, ILN, but not the spleen or MLN, that serve as the source of Th1 cells induced by BCG vaccination. Upon OVA challenge, the increased Th1 cells migrate from the ILN into the inflamed lung. Meanwhile, we have examined the possible role of CXCR3 and its chemoattractant, CXCL9, in the regulation of Th1-cell migration.

Previous studies have suggested that the IFN-γ+ T cells in the inflamed lung come from the migration of circulating T cells and/or the rapid expansion of residual memory CD4+ and CD8+ T cells in the lung after BCG vaccination or M. tb chronic infection [18, 26, 27]. However, adoptive transfer of effector/memory T cells from the lung shows no evidence of substantial and rapid expansion and beneficial effect in Rag-/- mice [18, 26]. It is possible that these cells are not long-lived memory T cells and cannot generate a rapid recall response to secondary pathogen challenge [28]. Indeed, live BCG, following s.c. vaccination, cannot reach distant lung tissue [26], making it impossible to establish effector T cells in the lung. Furthermore, treatment with antibiotics usually does not effectively remove all mycobacterial antigens [26, 29, 30], and the continual presence of mycobacterial antigens makes difficult to generate mycobacteria-specific memory T-cell immunity [17, 18]. We found a higher frequency of IFN-γ+CD4+ T cells in the ILN of BCG-vaccinated mice under the steady-state condition, which was significantly reduced following OVA challenge in mice, accompanied by increased number of IFN-γ+CD4+ T cells in the inflamed lung. Furthermore, these IFN-γ+CD4+ T cells displayed Tef/em phenotype and have potentials to activate, proliferate, and secrete IFN-γ following stimulation in vitro. These data suggest that the stored Th1 cells already migrated from the ILN to the lung. Indeed, surgical removal of the ILN prior to OVA challenge almost completed abrogated an increase in the number of IFN-γ+CD4+ T cells in the inflamed lung of mice. These data indicate that the ILN serves as a pool of BCG-induced Th1 cells.

Adoptive transfer experiments show that CD4+ T cells from the ILN of BCG-vaccinated mice inhibit the increase in allergic airway inflammation and mucus production. The protective effects of adoptive transfer CD4+ T cells from the ILN of BCG-vaccinated mice are associated with an increase in the levels of IFN-γ in BALF (Fig. 5C), which suggests the possibility that more IFN-γ-producing T cells move into the inflamed lung. To address this issue, we labeled CD4+ T cells with CFSE and found that more CD4+ T cells from the BCG-vaccinated ILN move into the inflamed lung tissue. Similar to our results, Fonseca et al. [31] have also found that BCG antigen–induced Th1 cells migrate into the OVA-challenged lung and secrete IFN-γ to counteract the OVA-induced Th2 response and subsequent allergic airway inflammation. These studies suggest that the accumulation of IFN-γ-producing CD4+ T cells in the lung by s.c. BCG vaccination plays a critical protective role in the subsequent development of asthma. To further confirm the idea that the importance of increased IFN-γ-expressing T cells in the ILN against allergic airway inflammation, we performed a surgical excision of the ILN before OVA challenge to completely deplete the effect of IFN-γ-expressing T cells in the ILN. As expected, the protective effects of BCG vaccination on allergic airway inflammation are lost following the ILN excision, and the loss of this protection is associated with a decrease in the frequency of IFN-γ-expressing CD4+, but not CD8+, T cells in the inflamed lung.

Chemokines and their corresponding receptors are crucial for the migration of CD4+ T cells into inflamed tissue. Thus, we measured Th1-associated chemokines in the OVA-challenged lung. After OVA challenge, the levels of CXCL9, a ligand for CXCR3, expression in the lung were significantly up-regulated in the current study regardless of BCG vaccination; however, increased frequency of Th1 cells only occurred in the lung of BCG/OVA mice. This is partially explained by the fact that BCG vaccination promoted the expression of CXCR3 on CD4+ T cells in the ILN. This result supports the idea that CXCR3 expression level is a principal factor determining both the ability to respond and the intensity of response to CXCR3 ligands in environmental antigen responses [32]. Transwell migration assays further demonstrate that CD4+ T cells from the ILN of BCG-vaccinated mice can potentially migrate in response to CXCL9. Therefore, these results suggest that high levels of CXCR3 expression on CD4+ T cells in the ILN of BCG-vaccinated mice and high concentrations of CXCL9, the Th1-associated chemokine, in the lungs may be the basis for the migration of Th1 cells from the ILN of BCG-vaccinated mice into the inflamed lung upon OVA challenge. However, further in vivo experiments are needed to determine the exact role of CXCR3 and CXCL9 in the migration of Th1 cells.

In summary, our data indicate the distant draining lymph node (ILN) as a ‘weapon’ pool of IFN-γ-expressing CD4+ T cells induced by neonatal BCG vaccination. These Th1 cells are resting in steady state and are ready for migrating into the inflamed lung through the CXCR3–CXCL9 axis, subsequently inhibiting allergen-induced airway inflammation.

Acknowledgments

We would like to thank Drs. Ming Yan (College of Biomedical Engineering & Instrument Science, Zhejiang University), Zhijian Cai (Institute of Immunology, Zhejiang University), Yungui Wang (Institute of Hematology, Zhejiang University), and Youfa Zhu (Department of Pathology, Zhejiang University) for their technical assistance. This study was supported in part by the grants from National Natural Science Foundation of China for Distinguished Young Scholars (No. 30825019, Shen HH) and National Natural Science Foundation of China (No. 81070013, Shen HH), and Major and Key Program of Zhejiang Province (2008C03002-2, Li W).

Author contributions

Gen-sheng Zhang and Ping-li Wang contributed to the design of the experiments, acquisition and analysis of the data and also contributed to the initial draft writing of this manuscript. Zhang-wei Qiu, Xue-jun Qin, Xiao-ping Lin, Na Li, Hua-qiong Huang, Hui Liu, and Wen Hua contributed to the collection and analysis and interpretation of data. Wen Li and Zhi-hua Chen contributed to the design and assessment of data and revised the manuscript. Hang Zhao provided a critical review and revision of the manuscript. Hua-hao Shen contributed to conception and design of the experiments, involved in the interpretation of experimental results, revised this manuscript and contributed to final approval of the version to be published.

Conflict of interest

The authors have declared no conflict of interest in the publication of this manuscript.

Ancillary