Innate Control of Adaptive Immunity Via Remodeling of Lymph Node Feed Arteriole

Home , Adaptive immune system, Lymph node

Innate control of adaptive immunity via remodeling of lymph node feed arteriole

Kelly A. Soderberg*, Geoffrey W. Payne†‡, Ayuko Sato*, Ruslan Medzhitov*§, Steven S. Segal†, and Akiko Iwasaki*¶

*Section of Immunobiology, †John B. Pierce Laboratory and Department of Cellular and Molecular Physiology, and §Howard Hughes Medical Institute, Yale University School of Medicine, New Haven, CT 06520

Edited by Douglas T. Fearon, University of Cambridge, Cambridge, United Kingdom, and approved September 22, 2005 (received for review July 21, 2005) The adaptive immune system relies on rare cognate lymphocytes to propriate microbial peptides must occur rapidly to provide the host detect pathogen-derived antigens. Naı¨ve lymphocytes recirculate with survival advantage. Thus, we hypothesize that such a process through secondary lymphoid organs in search of cognate antigen. must be inducible upon infection and regulated by the innate Here, we show that the naı¨ve-lymphocyterecirculation pattern is immune system. controlled at the level of innate immune recognition, independent In the present study, we tested these hypotheses using a combi- of antigen-specific stimulation. We demonstrate that inflamma- nation of approaches involving minimal TLR agonists and a tion-induced lymphocyte recruitment to the lymph node is medi- physiologically relevant model of mucosal viral infection with ated by the remodeling of the primary feed arteriole, and that its herpes simplex virus 2 (HSV-2). We report that during the initiation physiological role is to increase the efficiency of screening for rare of an adaptive immune response, a large number of naı¨ve lympho- antigen-specific lymphocytes. Our data reveal a mechanism of cytes are recruited specifically to the LNs draining the site of innate control of adaptive immunity: by increasing the pool of immunization or infection, a process we call ‘‘inflammation- naı¨velymphocytes for detection of foreign antigens via regulation induced recirculation.’’ We demonstrate that inflammation- of vascular input to the local lymph node. induced naı¨ve lymphocyte recruitment to the dLN occurs via remodeling of the primary arteriole feeding the LN to a larger o mount protective immunity, the adaptive immune system diameter, and that its physiological role is to increase the efficiency Trelies on the detection of foreign antigens by rare lymphocytes of screening for rare antigen-specific lymphocytes. possessing appropriate antigenic specificity. After local infection, antigens from the pathogen are taken up by dendritic cells (DCs) Methods (1). Concomitant recognition of pathogen-associated molecular Cells, Viruses, and Reagents. Thymidine kinase mutant HSV-2 patterns (PAMPs) by the Toll-like receptors (TLRs) induces DC strains 186TK⌬Kpn (13) were used to infect mice. Depo-Provera- activation and migration into the draining lymph nodes (dLNs), treated female mice were infected intravaginally (ivag) with 106 where they present antigenic peptides on MHC molecules to naı¨ve PFU of HSV-2 or with uninfected Vero cell lysate (mock infection) T lymphocytes (1–3). TLR-mediated recognition of PAMPs is as described in ref. 14. All antibodies were obtained from eBio- required for the initiation of Th1 immunity (4–7), and incorpora- science. All other reagents were from Sigma, unless otherwise tion of TLR agonists in vaccine adjuvants is often required for the indicated. generation of robust adaptive immunity. TLR signals contribute to the generation of adaptive immunity by several distinct mechanisms Mice. Female C57BL6 ϫ 129 F2, C57BL6, and BALB͞c mice, 6–8 including the activation of antigen presenting cells (APC) (2) and wk old, were purchased from The Jackson Laboratory. TLR9Ϫ/Ϫ stromal cells (7). TLR pathways are also required for the activation (15) or TLR2Ϫ/Ϫ (16) mice were bred and maintained in our of adaptive responses by blocking the suppressive effect of regula- facilities under standard conditions. All procedures used in this tory T cells (8). study complied with federal guidelines and institutional policies set A major unresolved mystery of the adaptive immune system by the Yale animal care and use committee. IMMUNOLOGY relates to how the rare cognate lymphocytes are screened for specificity toward pathogen-derived antigens in a timely manner. Immunization. TLR agonists CpG (50 ␮g) or LPS (5 ␮g) were Although several in vivo studies have elegantly demonstrated the injected into the hind footpad with or without OVA323–339 dynamic interactions of naı¨ve cognate T cells and antigen-loaded peptide in saline (Fig. 1A) or in incomplete Freund’s adjuvant (IFA) DCs in the lymph node (LN) (9–12), to visualize such events, these (Fig. 1B). Four days later, popliteal LNs were collected and experimental approaches often require introduction of a large analyzed. number of lymphocytes bearing a specific T cell receptor. In a naı¨ve mouse, each LN contains Ϸ106 cells, of which only 1 in 106 to 105 Measurement of Cell Proliferation. Mice were continuously fed has the specificity for a given antigen. This screening process BrdUrd at a concentration of 0.8 mg͞ml in their drinking water for represents a daunting task for the antigen-presenting DC to seek the duration of infection. All dLNs (inguinal and iliac) were out the rare cognate lymphocytes within the dLN. We hypothesized harvested, and cell suspensions were stained for BrdUrd and that a mechanism must exist that facilitates the search for cognate lymphocytes within the secondary lymphoid organs. In this regard, the encounter of cognate T cells with the APC can be enhanced in Conflict of interest statement: No conflicts declared. the dLN by several mutually nonexclusive processes. These include This paper was submitted directly (Track II) to the PNAS office. increases in (i) the motility of the APC and͞or T cells within the Abbreviations: ACh, acetylcholine; APC, antigen-presenting cells; CFSE, carboxyfluorescein dLN, (ii) the retention of naı¨ve lymphocytes in the dLN, and (iii) diacetate succinimdyl ester; DC, dendritic cell; dLN, draining lymph node; HEV, high endothelial venule; IFA, incomplete Freund’s adjuvant; ivag, intravaginal(ly); LN, lymph the recruitment of additional naı¨ve lymphocytes to the dLN. The node; ndLN, nondraining LN; NOS, NO synthase; PAMP, pathogen-associated molecular latter two processes would result in cellular accumulation and pattern; p.i., postinfection; SNP, sodium nitroprusside; TLR, Toll-like receptor. enlargement of the dLN. Thus, we propose that an important ‡Present address: University of Northern British Columbia, Northern Medical Program, 3333 immunological purpose of the LN hypertrophy is to facilitate University Way, Prince George, BC, Canada V2N 4Z9. screening for cognate lymphocytes during the initiation of adaptive ¶To whom correspondence should be addressed at: Yale University School of Medicine, immune responses. Moreover, in the face of invasion by a replicat- 300 Cedar Street, TAC S655B, New Haven, CT 06520. E-mail: [email protected]. ing pathogen, the search for lymphocytes that recognize the ap- © 2005 by The National Academy of Sciences of the USA

www.pnas.org͞cgi͞doi͞10.1073͞pnas.0506190102 PNAS ͉ November 8, 2005 ͉ vol. 102 ͉ no. 45 ͉ 16315–16320 Downloaded by guest on September 26, 2021 Fig. 1. TLR agonists alone are sufficient to induce LN hypertrophy. (A) The number of cells in popliteal LN was measured in WT or TLR9Ϫ/Ϫ mice (n ϭ 4 per condition) 4 days after footpad injection of TLR ligands in saline. (B) The number of cells in popliteal LN was measured in WT, TLR9Ϫ/Ϫ, or TLR2Ϫ/Ϫ mice (n ϭ 3 per condition) 4 days after footpad injection of TLR ligands with or without OVA323–339 peptide in IFA. (C) The total cell number of dLNs (inguinal and iliac) after ivag HSV-2 (106 PFU) infection was measured at the indicated time points. (D) Photographic depiction of dLNs of HSV-2-infected mice. Similar results were obtained in five separate experiments. *, P Ͻ 0.05, significant difference between the indicated groups.

surface lymphocyte markers according to the manufacturer’s in- Results structions (BD Pharmingen). For analysis of DNA content, dLNs Innate Signals Are Sufficient to Induce Lymph Node Hypertrophy. To were harvested at indicated time points after HSV-2 infection, and test our hypothesis that selective accumulation of naı¨ve lympho- single cells were labeled with propidium iodide according to a cytes to the dLN would facilitate screening for cognate lymphocytes standard protocol. The percentage of cells incorporating propidium during the initiation of an immune response, and that such process ͞ iodide at 2N (G1)or4N(G2 M) was determined by flow cytometry must be induced rapidly by the innate immune system, we first according to the manufacturer’s instructions (Becton Dickinson). examined the nature of the stimulus required to promote cellular accumulation in the dLN. Specifically, we asked whether cellular Adoptive Transfer. Single-cell suspensions were prepared from the accumulation depends on antigen, TLR signals, or both. Mice were Ϫ Ϫ LNs of WT or CD62L / mice. The cells were labeled with 0.5 ␮M injected s.c. with LPS or CpG, agonists for TLR4 and TLR9, carboxyfluorescein diacetate succinimdyl ester (CFSE) (Molecular respectively. This procedure resulted in LN hypertrophy and an Probes), and 107 cells were transferred into the lateral tail vein of increase in cellularity (Fig. 1A). Importantly, inoculation of TLR the recipient mice. Mice were infected2hafteradoptive transfer agonist alone induced cellular accumulation in the dLNs, revealing and killed at 4 days postinfection (p.i.). In some experiments, the the sufficiency of innate signals in this process. These responses donor lymphocytes were pretreated with 200 ng͞ml pertussis toxin required the presence of a specific TLR, because mice deficient in for 45 min at 37°C, extensively washed, and labeled with CFSE TLR9 failed to accumulate cells in the dLNs after CpG injection. before transfer at Ϫ1 day or ϩ2 days of infection. The rates of To test the contribution of the antigen-specific T cells in LN lymphocyte entry were assessed by injecting 107 CFSE-labeled hypertrophy, an antigenic peptide OVA323–339 was co-injected syngeneic naı¨ve donor lymphocytes into mice that have been with TLR agonists (Fig. 1B). Because productive activation of naı¨ve infected with HSV-2 for 2 days or 5 days and collecting dLNs T cells requires prolonged antigen exposure (19), these agents were (inguinal and iliac) and nondraining lymph nodes (ndLNs) (axillary emulsified in IFA. Although inoculation of IFA alone resulted in a andcervical)2hlater.Therate of entry was calculated as the measurable LN hypertrophy, LPS and CpG induced significant number of CFSEϩ lymphocytes in the LN per hour. For OT-II TLR-dependent increase in cellularity in the dLN (Fig. 1B). Im- transfer experiments, recipient C57BL6 mice were injected with portantly, inclusion of a highly immunogenic peptide did not result 104,105,106,or107 OT-II TCR-transgenic CD4ϩ T cells i.v., in additional increase in cellularity in the 4 days after injection. followed by immunization in the hind footpad with 100 ␮gof Thus, TLR-mediated innate recognition of PAMPs alone is suffi- ovalbumin and 4 ␮g of LPS emulsified in IFA. CD4ϩ T cells were cient to trigger LN hypertrophy, and antigen-specific T cell re- isolated from the dLNs (popliteal) at the indicated time points, and sponses are not required for this process. How can this antigen- 105 dLN CD4ϩ T cells were restimulated for 72 h in vitro with 2 ϫ independent PAMP-induced accumulation of lymphocytes 105 irradiated syngeneic splenocytes in the presence or absence of contribute to the generation of adaptive immunity? 1 ␮M OVA323–339 peptide. Cytokines were measured by ELISA. To address this question in the context of a physiological infection, we used a mouse model of genital herpes. Ivag infection of Intravital Microscopy. These experiments were modeled after the mice with HSV-2 represents an ideal model to dissect immune method described by von Andrian (17) and evaluated with proce- inductive mechanisms because the location and the kinetics of dures as described in ref. 18 (see Supporting Text, which is published cell-mediated immune responses have been well defined (14, 20, as supporting information on the PNAS web site). 21). We confirmed that hypertrophy occurs in the inguinal and iliac LNs draining the infected genital mucosa, both in terms of the Statistical Analyses. All statistical analyses were performed with increase in cellularity (Fig. 1C) and gross organ enlargement (Fig. SIGMASTAT 2.03 (SPSS, Chicago). Paired t tests and two-way repeat- 1D). Thus, we decided to use this model of physiological mucosal ed-measures analysis of variance were used to evaluate the effects HSV-2 infection to address key questions regarding how adaptive of treatment, with statistical significance defined as P Ͻ 0.05. immune responses are initiated within the defined sets of dLNs.

16316 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0506190102 Soderberg et al. Downloaded by guest on September 26, 2021 Fig. 2. dLN enlargement is caused by the accumulation of nonproliferating, resting naı¨velymphocytes. (A) Cumulative BrdUrd incorporation by lymphocytes in the dLNs (inguinal and iliac) during 4 days of HSV-2 infection was assessed by flow cytometric analysis of BrdUrdϩ vs. BrdUrdϪ lymphocyte subsets. (B) The number of dividing cells was determined by propidium iodide labeling of the dLN cells at the indicated time points after HSV-2 infection. Each bar represents an average of five mice in each group. (C) The resting vs. blast phenotype of the dLN cells after HSV-2 infection was assessed by forward scatter vs. side scatter analysis of the lymphocyte subsets. (D) The number of resting vs. blast cells in popliteal LN was measured in WT, TLR9Ϫ/Ϫ, or TLR2Ϫ/Ϫ mice (n ϭ 3 per condition) 4 days after footpad injection of TLR agonists in IFA with or without OVA323–339 peptide (100 ␮g) as in Fig. 1B.(E) dLN cells from HSV-2-infected mice were collected and analyzed for surface expression of CD69 and CD62L. The bar depicts the number of CD69ϩ and CD62Lhi cells at various time points after infection. These data are representative of three similar experiments. *, P Ͻ 0.05, significant difference between control and infected groups.

Inflammation Induces Recruitment of Nondividing Naı¨ve Lymphocytes numbers of naı¨ve lymphocytes are recruited from the peripheral to the dLN. Because TLR agonists alone are sufficient to induce LN blood, we examined the rate of lymphocyte entry into the dLNs. hypertrophy, this process likely reflects the accumulation of non- CFSE-labeled naı¨ve lymphocytes were injected i.v. into mice that specific naı¨ve lymphocytes rather than antigen-induced prolifera- had been infected with HSV-2 for 2 or 5 days, and the rate of entry tion. To determine whether the increase in cellularity is caused by into iliac and inguinal nodes was measured (Fig. 3). The rate of the proliferation of lymphocytes within the dLNs or accumulation entry of the transferred naı¨ve lymphocytes into the dLNs increased of nondividing lymphocytes, we gave mice a continuous dose of steadily, reaching a 5-fold increase by day 5 p.i. compared with that BrdUrd and measured the number of proliferating lymphocytes in of noninfected mice or to the ndLNs of the infected mice (Fig. 3A). the dLNs after ivag HSV-2 infection. There was a marked increase The rate of entry was proportional to the total cellularity in the dLN in all lymphocyte subsets. We found that Ͼ95% of the cells found or the ndLN (Fig. 3B). Thus, these data indicated that during the in the dLNs at 4 days p.i. were nondividing lymphocytes (Fig. 2A). first few days of priming, there is a dramatic increase in the rate of IMMUNOLOGY This finding was also confirmed by measuring DNA content in the naı¨ve lymphocyte recruitment from the peripheral blood pool, dLN cells by propidium iodide staining at various time points after consistent with previous observations (22–25). Thus, the recircu- HSV-2 infection (Fig. 2B). Thus, these data indicated that LN lation pattern of the naı¨ve lymphocytes is significantly altered so as hypertrophy occurs as a result of the accumulation of nondividing to recruit more naı¨ve lymphocytes selectively to the LNs draining lymphocytes and is not due to the proliferation of activated the site of infection. Moreover, because naı¨ve resting lymphocytes lymphocytes. constitute the majority of the dLN cells after TLR agonist inocu- The nature of the nondividing lymphocytes that accumulate in lation (Fig. 2D), innate recognition of PAMPs is sufficient to induce the dLNs early postinfection was further examined. The majority of this altered recirculation. these cells had a small resting naı¨ve lymphocyte phenotype (Fig. 2C), even at the peak of HSV-2-specific Th1 responses (14). Inflammation-Induced Recirculation Occurs via the High Endothelial Similarly, small resting lymphocytes constituted the majority of the Venule (HEV). To determine the molecular requirement for the dLN cells after TLR agonist inoculation (Fig. 2D). To confirm the inflammation-induced recirculation, we examined the entry of activation status of these small resting lymphocytes, they were naı¨ve lymphocytes deficient in specific molecules known to be analyzed for several cell surface markers for naı¨ve and activated involved in homeostatic recirculation. During homeostatic recircu- lymphocytes. The majority of the lymphocytes in the dLNs were lation, naı¨ve lymphocytes enter the LNs by transendothelial migra- CD62Lhi and CD69Ϫ until day 5 after HSV-2 infection (Fig. 2E). tion across the HEVs (26, 27). This process requires (i) the Similar accumulation of naı¨ve resting lymphocytes was observed in interaction of CD62L with its ligand, molecules collectively known the dLNs of mice infected ivag with HSV-1, as well as postimmu- as the peripheral node addressins (PNAd), and (ii) the stimulation nization with antigen emulsified in complete Freund’s adjuvant (see of cells through the chemokine receptor CCR7 by chemokines Fig. 6, which is published as supporting information on the PNAS (CCL19 and CCL21), which induces conformational changes in the web site). These data suggested that LN hypertrophy in general is integrins such as LFA-1 (26). Whereas WT lymphocytes migrated due to the accumulation of naı¨ve lymphocytes. into the LNs draining HSV-2 infection site, those lacking CD62L Naı¨ve lymphocyte accumulation can be mediated by increase in failed to accumulate in the dLNs (Fig. 3C). Furthermore, pretreat- their retention or recruitment to the dLN. To test whether greater ment of naı¨ve lymphocytes by pertussis toxin, an inhibitor of Gi␣

Soderberg et al. PNAS ͉ November 8, 2005 ͉ vol. 102 ͉ no. 45 ͉ 16317 Downloaded by guest on September 26, 2021 Fig. 3. Inﬂammation-induced recirculation utilizes CD62L and Gi␣- dependent pathway. The rate of cell entry into the dLN was measured by adoptive transfer of CFSE- labeled naı¨ve lymphocytes into mice at days 2 and 5 after HSV-2 infection. Two hours later, cells were collected from either the dLNs or ndLNs, and the number of CFSEϩ cells (A) and the total number of cells (B) were measured. Each bar represents an individual mouse. Data are representative of two separate experiments. (C) Lympho- cytes from WT or CD62LϪ/Ϫ mice were labeled with CFSE and adoptively transferred into WT or CD62LϪ/Ϫ recipients. Mice were either infected or not infected with HSV-2 ivag, and 4 days later, CFSEϩ cells were enumerated in the dLNs. Data represents the average number of cells per dLN. (D) Donor lymphocytes were treated with pertussis toxin or with PBS, labeled with CFSE, and adoptively transferred to mice at Ϫ1 day or ϩ2 days p.i. with HSV-2. The number of CFSEϩ donor cells was analyzed in the dLN at 4 days p.i. and plotted as the average number of donor cells per dLN. These data are representative of three similar experiments. *, P Ͻ 0.05.

signaling, rendered cells incapable of accumulating into the dLNs lymphocytes that enter and survey the incoming antigens within (Fig. 3D). These data showed that inflammation-induced recircu- the dLNs. lation requires both CD62L and chemokine responsiveness through G protein-coupled receptors. Mechanism of Inflammation-Induced Recirculation. Enhanced recruitment of naı¨ve lymphocytes into dLNs can be mediated by a The Cognate Lymphocyte Number Controls the Onset of Th1 Immunity. change in the ability of the HEV to allow immigration of more naı¨ve Our data indicated that local LNs are designed to collect a large lymphocytes per given surface area and͞or by an increase in number of naı¨ve lymphocytes during the initiation of an immune vascularity (28) leading to increase in the number of HEVs. To this response. A major potential advantage of such a system is that by end, we examined the number and size of the HEVs in the dLNs. allowing more naı¨ve lymphocytes to specifically survey the LNs Quantitative analysis of serial tissue sections revealed that although draining the site of infection, it provides a larger repertoire from the density of HEVs in the dLNs stayed constant, the total number which to screen for lymphocytes with specificity to the microbial of HEVs in the dLN increased proportional to the LN size (see Fig. antigens. Because only 1 in 106 to 105 lymphocytes has the specificity for a given antigen, selective increase in the total number of naive lymphocytes in the dLNs would facilitate the screening process for cognate lymphocytes. This would provide a significant immunological advantage because unchecked pathogen replication can proceed exponentially in the infected host, and the survival of the host critically depends on the timely generation of effector responses. To test whether the total number of naive cognate CD4ϩ lymphocytes in the dLNs affects the onset of Th1 immunity, we examined the time course of Th1 generation in mice that had been injected with defined numbers of antigen-specific CD4ϩ T lymphocytes. Mice were adoptively transferred with 104,105,106,or107 OT-II TCR-transgenic CD4ϩ T cells and subsequently immunized in the footpad with ovalbumin plus LPS in IFA. CD4ϩ T cells from the dLNs were isolated at various time points and restimulated in vitro with irradiated syngeneic splenocytes in the presence or absence of the cognate OVA323–339 peptide. We found that the onset of the effector cytokine secretion by CD4ϩ T cells depended on the number of OT-II cells transferred (Fig. 4). Mice that received 107 OT-II cells generated an IL-2 response in the dLNs as early as 1 day postimmunization. The onset of IFN-␥ secretion occurred as early as day 2 postimmunization in mice receiving 106 or 107 OT-II cells, whereas those that received 105 OT-II cells had a delay in the onset of effector responses until day 6. Mice that did not receive OT-II cells failed to induce Th1 response until day 8 postimmunization. These data indicated that the time required to generate Th1 Fig. 4. The total number of cognate T cells determines the onset of Th1 responses. C57BL6 mice injected i.v. with the indicated number of OT-II CD4ϩ response is inversely proportional to the number of cognate lym- T cells were immunized in the hind footpad with ovalbumin (100 ␮g) and LPS phocytes present in the LN. Thus, an immunological purpose of the (4 ␮g) in IFA. dLN CD4ϩ T cells were restimulated and assayed for IL-2 (A) and inflammation-induced recirculation is to facilitate the induction of IFN-␥ (B) secretion in the presence (ﬁlled bar) or absence (open bar) of adaptive immune responses by increasing the total number of naı¨ve OVA323–339 peptide. Data are representative of four similar experiments.

16318 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0506190102 Soderberg et al. Downloaded by guest on September 26, 2021 Fig. 5. Inﬂammation-induced recirculation occurs through remodeling of the arteriole feeding the dLN. The internal vessel diameter of the primary arteriole feeding the dLN was measured in naı¨ve or HSV-2- infected mice at 3 days p.i. by using intravital microscopy. (A) Resting and maximal (ϩSNP) diameters of the arteriole proximal to dLN in noninfected control (open bar; n ϭ 7) vs. infected (ﬁlled bar; n ϭ 5) mice. (B) Vasoconstriction of the arteriole after application of varying doses of norepinephrine (NE). (C) Smooth muscle function of the arteriole feeding the dLN was measured in uninfected control (n ϭ 6) and HSV-2-infected (n ϭ 5) mice. Maximal vasoconstriction was measured after depolarization of vascular smooth muscle with potassium chloride. Change is calculated as the difference between resting and KCl treatment. (D) Vasodi- latation of the arteriole was measured after administration of varying doses of ACh. (E) The effect of NOS ␻ inhibition by N -nitro-L-arginine (L-NNA) on vascular diameter was assessed over time. Data represent mean diameter (A, C, and E) or change in diameter (B and D) Ϯ SEM. *, P Ͻ 0.05.

7, which is published as supporting information on the PNAS web sive vasoconstriction observed in control mice, arteriole of the ␻ site). These results indicated that the increased input of naı¨ve infected mice failed to constrict in response to N -nitro-L-arginine lymphocytes to the dLN is at least in part mediated by a rapid (Fig. 5E), indicating a defect in NO synthesis by the dLN arteriolar increase in the number of HEVs. We reasoned that the increase in endothelium. Collectively, these results revealed that a rapid re- lymphocyte recruitment through additional HEVs must be modeling and expansion of the primary feed vessel to the dLN takes achieved by an increase in blood supply into the LNs, as suggested place during the initiation of the adaptive immune responses, and by previous studies in larger animals (24, 29, 30). If so, the arteriole that this process is accompanied by the modification of the vascular feeding the dLN must expand to accommodate an increase in blood endothelial but not smooth muscle function. flow. To test this hypothesis, we measured the diameter of the arteriole Discussion feeding the inguinal LN by intravital microscopy. This analysis Owing to their strategic location and architectural design (31), the revealed that by day 3 p.i., the arteriole feeding the inguinal LN was LNs serve as optimal site for the induction of adaptive immune enlarged by nearly 50% (Fig. 5A, Resting). The increase in vessel responses. Naı¨ve lymphocytes recirculate through the LNs in search diameter can be caused by either a loss of basal tone or by structural of cognate antigens. Upon infection, DCs are activated by recog- remodeling of the arteriole. To distinguish these possibilities, we nition of PAMPs and are induced to migrate from the site of examined the effect of local administration of a nitric oxide (NO) infection to the LN via the afferent lymph (1). After interaction donor, SNP, which induces maximal vessel dilation. If the feed with activated DCs, antigen-specific T cells are preferentially re- arteriole of the dLN was simply dilated because of loss of tone, the tained within the LN by down-regulating their sphingosine-1 phos- SNP-induced maximal diameters will be the same between the phate receptor (32). This process prevents cognate T cells from infected and the control groups. However, after SNP administra- exiting via the efferent lymph, thus allowing them to maintain IMMUNOLOGY tion, the arteriole feeding the dLN in the infected mice exhibited contact with the APC. Such prolonged interactions with the APC greater maximum diameter (161 Ϯ 3 ␮m) compared with the induce T cells to undergo proliferation and differentiation into control (128 Ϯ 5 ␮m), indicating that arteriolar enlargement effector cells. However, although these events become detectable occurred as a result of vascular remodeling (Fig. 5A, ϩSNP). within the dLN after 5 days p.i., in animals with a natural TCR Vascular remodeling can result in altered function of the arte- repertoire, the majority of the cells that constitute the hypertrophic riolar smooth muscle cells and͞or the endothelial cells. The nature LN at earlier time points were naı¨ve resting lymphocytes (Fig. 2). of adaptation was investigated by topical administration of agonists Here, we showed that through the recruitment of a large number that are selective for respective cell types. Vasoconstriction to of naı¨ve lymphocytes, the LN allows productive surveillance and norepinephrine (a smooth muscle agonist) revealed normal respon- rapid induction of effector responses. This altered recirculation of siveness of the arteriolar smooth muscle cells (Fig. 5B). Moreover, the naı¨ve lymphocytes is induced by innate signals alone (Fig. 1) and depolarization of smooth muscle cells with KCl demonstrated leads to an increase in the rate of naı¨ve lymphocyte entry specifi- similar, dramatic arteriolar constriction in both infected and control cally to the dLN (Fig. 3). Furthermore, enhanced recruitment of mice (Fig. 5C). Collectively, these data indicated that the smooth naı¨ve lymphocytes is mediated by remodeling of the arteriole muscle function in the enlarged dLN arteriole is not affected by feeding the LN to a larger diameter (Fig. 5), which will increase the inflammation. To examine whether remodeling was associated with blood flow and delivery of circulating lymphocytes. altered endothelial cell function, responses to ACh (endothelium- Our study demonstrates a process by which the innate immune dependent vasodilator via NO production) were evaluated. Re- recognition of pathogens controls the generation of adaptive im- markably, the arteriole feeding the dLN underwent much reduced mune responses via remodeling of the arteriolar vessel. According dilation in response to ACh, contrasting with the pronounced to the Poiseuille–Hagen equation, the volumetric flow rate through vasodilation observed in control mice (Fig. 5D). This reduced a cylindrical blood vessel is proportional to the forth power of the responsiveness to ACh by the dLN feed arteriole suggested that the vessel radius. Thus, with constant arterial perfusion pressure, the endothelial cells are defective in NO production. To test for the increase in arteriolar diameter from 101 Ϯ 9 ␮m (naı¨ve) to 147 Ϯ ␻ constitutive activity of endothelial NOS, N -nitro-L-arginine (an 3 ␮m (day 3 p.i.) leads to a 4.5-fold increase in the rate of naı¨ve inhibitor of NOS) was administered topically. Unlike the progres- lymphocyte flow rate through the dLNs. This conclusion is consis-

Soderberg et al. PNAS ͉ November 8, 2005 ͉ vol. 102 ͉ no. 45 ͉ 16319 Downloaded by guest on September 26, 2021 tent with an earlier study showing that there is a 4-fold increase in lation. In the case of HSV-2 infection, MyD88Ϫ/Ϫ mice underwent the cardiac output into the LN draining the site of inflammation at similar LN enlargement compared with WT mice (data not shown), 3 days after oxazolone painting (22). Thus, the net capacity for naı¨ve suggesting that multiple pathways (TLR-dependent and -indepen- cognate lymphocyte screening increases from 3 ϫ 106 per LN per dent) contribute to mediating this process after live viral infection. day (17) to 14 ϫ 106 per LN per day after 3 days of infection (see This finding is consistent with our previous observation that calculation in Supporting Text). Whereas transient reduction in the MyD88-dependent pathways critically control Th1 differentiation egress of lymphocytes, or ‘‘lymphocyte shutdown,’’ occurs within but not T cell migration into the dLNs (7). 6–18 h after challenge, cell output in the efferent lymph actually After natural infection or immunization, our data suggest that rises rapidly, reaching a 3-fold increase over the steady-state levels the 6–7 days required for the generation of observable Th1 by day 3 (33). Thus, the gross effect of the contribution of the immunity in the local LNs reflect the time required for the rare enlarged arteriole in cell input to the dLN is likely underestimated lymphocytes to be screened for its specificity to microbial antigens. because of the mass egress via the efferent lymph. Combined with The present findings show that this screening process is augmented the 10-fold increase in the total number of lymphocytes in the local by the increase in the recruitment of naı¨ve lymphocytes specifically dLNs at day 3 p.i. (Fig. 1), the increase in the total number of to the dLN via remodeling of its arterioles, and not simply due to lymphocytes that transit the relevant LNs greatly enhances the loss of vascular tone. These results demonstrate that remodeling of chance for encounter of naı¨ve lymphocytes with cognate antigens. the feed arteriole occurs with LN hypertrophy, revealing an im- Consistent with this interpretation, by artificially introducing dif- portant link between the microvascular and immune systems. fering numbers of cognate lymphocytes, we show that indeed the Moreover, enlargement of the feed arteriole is associated with an increase in the total number of cognate CD4ϩ T lymphocytes impaired NO production by endothelium while smooth muscle dramatically accelerates the onset of Th1 response induction. function is preserved. This loss of NOS function is particularly Because s.c. injected antigens reach the dLNs within 15–30 min (34, intriguing, given the known role of the endothelial NO in main- 35), with ample cognate T cells in the LN, a Th1 response developed taining homeostatic vasodilation. Thus, this change in endothelial within 48 h of immunization with ovalbumin and LPS (Fig. 4). function likely reflects a unique stage in vascular remodeling. These results indicated that the rate-limiting step in the generation Furthermore, increase in LN cellularity was accompanied by the of Th1 effector responses is not the time required for naı¨ve T cells increase in the HEV within the dLNs (Fig. 7). Although the details to undergo activation and differentiation but, rather, the time of the sequence of events leading to this rapid remodeling of both required for the rare lymphocytes to find their cognate antigen the arterioles and HEV remain to be elucidated, our study presented by the appropriate DCs in the LN. demonstrates a key control of microcirculation by the innate The enlargement of the regional LN is a hallmark of infection immune system. Thus, our data provide evidence that the in- with a variety of pathogens, and as with any other biologically nate immune system not only enhances the generation of adaptive important processes, it is likely mediated by multiple mechanisms immune responses through the activation of DCs and lymphocytes, invoked by innate stimuli. After bacterial challenge in the skin, but also through increasing the efficiency of screening for cognate TNF-␣ produced by mast cells has been shown to mediate LN lymphocytes within the dLN by enlarging the primary arteriole hypertrophy (36). We observed normal LN hypertrophy in TNF- feeding the dLN. ␣-deficient mice after HSV-2 infection (data not shown), suggesting that distinct and͞or multiple pathways for LN hypertrophy are We thank Drs. J. Pober and W. Sessa for insightful discussions. K.A.S. was triggered by viral infection in the vaginal mucosa. We showed that supported by the National Institutes of Health Minority Supplement the accumulation of naı¨ve lymphocytes to the local LN is mediated CA16885S1, and A.S. was supported by a James Hudson Brown–Alexander Brown Coxe postdoctoral fellowship. A.I. is a recipient of the Wyeth– by innate signals alone in the absence of antigenic stimulation. TLR Lederle Vaccine Young Investigator Award and a grant from the Ethel F. agonists either can directly bind to the vascular endothelial cells and Donaghue Women’s Health Investigator Program at Yale University, and induce remodeling or may activate DCs, which in turn will activate is a Burroughs Wellcome Investigator in Pathogenesis of Infectious Dis- the feed arteriole. TLR stimulation by PAMPs represents a key, but eases. This work was supported by National Institutes of Health Grants R01 not exclusive, innate signal that induces naı¨ve lymphocyte accumu- AI054359 (to A.I.), AI062428 (to A.I.), and R21 AG19347 (to S.S.S.).

1. Banchereau, J. & Steinman, R. M. (1998) Nature 392, 245–252. 19. Kaech, S. M., Wherry, E. J. & Ahmed, R. (2002) Nat. Rev. Immunol. 2, 251–262. 2. Iwasaki, A. & Medzhitov, R. (2004) Nat. Immunol. 5, 987–995. 20. Milligan, G. N. & Bernstein, D. I. (1995) Virology 212, 481–489. 3. Itano, A. A. & Jenkins, M. K. (2003) Nat. Immunol. 4, 733–739. 21. Parr, M. B., Kepple, L., McDermott, M. R., Drew, M. D., Bozzola, J. J. & Parr, 4. Jankovic, D., Kullberg, M. C., Hieny, S., Caspar, P., Collazo, C. M. & Sher, A. E. L. (1994) Lab. Invest. 70, 369–380. (2002) Immunity 16, 429–439. 22. Ottaway, C. A. & Parrott, D. M. (1979) J. Exp. Med. 150, 218–230. 5. Muraille, E., De Trez, C., Brait, M., De Baetselier, P., Leo, O. & Carlier, Y. 23. Cose, S. C., Jones, C. M., Wallace, M. E., Heath, W. R. & Carbone, F. R. (1997) (2003) J. Immunol. 170, 4237–4241. Eur. J. Immunol. 27, 2310–2316. 6. Schnare, M., Barton, G. M., Holt, A. C., Takeda, K., Akira, S. & Medzhitov, 24. Mackay, C. R., Marston, W. & Dudler, L. (1992) Eur. J. Immunol. 22, 2205–2210. R. (2001) Nat. Immunol. 2, 947–950. 25. Palframan, R. T., Jung, S., Cheng, G., Weninger, W., Luo, Y., Dorf, M., Littman, 7. Sato, A. & Iwasaki, A. (2004) Proc. Natl. Acad. Sci. USA 101, 16274–16279. D. R., Rollins, B. J., Zweerink, H., Rot, A. & von Andrian, U. H. (2001) J. Exp. 8. Pasare, C. & Medzhitov, R. (2003) Science 299, 1033–1036. Med. 194, 1361–1373. 9. Bousso, P. & Robey, E. (2003) Nat. Immunol. 4, 579–585. 26. Campbell, D. J., Kim, C. H. & Butcher, E. C. (2003) Immunol. Rev. 195, 58–71. 10. Miller, M. J., Wei, S. H., Parker, I. & Cahalan, M. D. (2002) Science 296, 27. von Andrian, U. H. & Mempel, T. R. (2003) Nat. Rev. Immunol. 3, 867–878. 1869–1873. 28. Herman, P. G., Yamamoto, I. & Mellins, H. Z. (1972) J. Exp. Med. 136, 11. Stoll, S., Delon, J., Brotz, T. M. & Germain, R. N. (2002) Science 296, 1873–1876. 697–714. 12. Mempel, T. R., Henrickson, S. E. & Von Andrian, U. H. (2004) Nature 427, 29. Csanaky, G., Kalasz, V. & Pap, T. (1991) Acta Morphol. Hung. 39, 251–257. 154–159. 30. Hay, J. B. & Hobbs, B. B. (1977) J. Exp. Med. 145, 31–44. 13. Jones, C. A., Taylor, T. J. & Knipe, D. M. (2000) Virology 278, 137–150. 31. Gretz, J. E., Anderson, A. O. & Shaw, S. (1997) Immunol. Rev. 156, 11–24. 14. Zhao, X., Deak, E., Soderberg, K., Linehan, M., Spezzano, D., Zhu, J., Knipe, 32. Matloubian, M., Lo, C. G., Cinamon, G., Lesneski, M. J., Xu, Y., Brinkmann, D. M. & Iwasaki, A. (2003) J. Exp. Med. 197, 153–162. V., Allende, M. L., Proia, R. L. & Cyster, J. G. (2004) Nature 427, 355–360. 15. Hemmi, H., Takeuchi, O., Kawai, T., Kaisho, T., Sato, S., Sanjo, H., Matsumoto, 33. Cahill, R. N., Frost, H. & Trnka, Z. (1976) J. Exp. Med. 143, 870–888. M., Hoshino, K., Wagner, H., Takeda, K. & Akira, S. (2000) Nature 408, 740–745. 34. Gretz, J. E., Norbury, C. C., Anderson, A. O., Proudfoot, A. E. & Shaw, S. 16. Takeuchi, O., Hoshino, K., Kawai, T., Sanjo, H., Takada, H., Ogawa, T., (2000) J. Exp. Med. 192, 1425–1440. Takeda, K. & Akira, S. (1999) Immunity 11, 443–451. 35. Itano, A. A., McSorley, S. J., Reinhardt, R. L., Ehst, B. D., Ingulli, E., 17. von Andrian, U. H. (1996) Microcirculation 3, 287–300. Rudensky, A. Y. & Jenkins, M. K. (2003) Immunity 19, 47–57. 18. Payne, G. W., Madri, J. A., Sessa, W. C. & Segal, S. S. (2004) FASEB J. 18, 36. McLachlan, J. B., Hart, J. P., Pizzo, S. V., Shelburne, C. P., Staats, H. F., Gunn, 280–286. M. D. & Abraham, S. N. (2003) Nat. Immunol. 4, 1199–1205.

16320 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0506190102 Soderberg et al. Downloaded by guest on September 26, 2021