Early viral replication in lymph nodes provides HIV with a means by which to escape NK-cell-mediated control

Abstract

Acute HIV infection is marked by dramatic viral replication associated with preferential replication within secondary lymphoid tissues, such as lymph nodes (LNs), that is rapidly but incompletely contained to a viral setpoint. Accumulating evidence supports a role for natural killer (NK) cells in the early control of HIV infection; however, little is known about the location of their antiviral control. Given that HIV replicates profusely in LNs during early infection, we sought to define whether changes occurred in the NK cell infiltrate within these sites during the first year of HIV infection. Surprisingly, NK cell numbers and distribution were unaltered during early HIV infection. LN NK cells expressed decreased inhibitory receptors, were more highly activated, and expressed elevated TRAIL, potentially conferring a superior capacity for NK cells to become activated and control infection. Most noticeably, KIR⁺ NK cells were rarely detected in the LN during HIV infection, associated with diminished migratory capacity in the setting of reduced expression of CX3CR1 and CXCR1. Thus, incomplete control of HIV viral replication during early disease may be due to the inefficient recruitment of KIR⁺ NK cells to this vulnerable site, providing HIV a niche where it can replicate unabated by early NK-cell-mediated innate pressure.

Introduction

Early in HIV infection, viral replication occurs preferentially in tissues, including both the gastrointestinal tract and secondary lymphoid organs such as lymph nodes (LNs) [1, 2]. In fact, viral replication in LNs exceeds the levels in plasma by several orders of magnitude [1] and is associated with a gradual destruction of the highly ordered architecture of the LN, marked by a follicular hyperplasia, massive CD4⁺ T-cell apoptosis, and the final involution of the LN [3].

Following acute infection, viral replication in the peripheral circulation is brought down within the first few weeks of infection to a viral setpoint [4], at a time when adaptive immune responses are just being induced [5, 6]. Thus, it is likely that innate immune mechanisms, including natural killer (NK) cells, may play a critical role in the early control of viral replication. NK cells are innate effector lymphocytes that play a pivotal role in a number of viral infections [7]. Ninety percent of the NK cells in peripheral blood are CD3^negCD56^dimCD16^pos (CD56^dim), exhibit highly cytotoxic functions, and express killer immunoglobulin receptors (KIRs), whereas 10% are CD3^negCD56^brightCD16^neg (CD56^bright) and are considered immunoregulatory as they instead secrete large quantities of cytokines [8, 9]. In contrast, the majority of NK cells in secondary lymphoid tissues, such as the LNs, are CD56^bright cells. NK cells expand rapidly following acute HIV-1 infection [6], and KIR-expressing NK cells have been implicated in both direct and indirect control of HIV-1 replication in vitro [10]. Given that HIV replicates profusely within LNs during early infection, early antiviral effector cells such as NK cells might play a critical role in the early control of HIV viral replication.

While many studies have examined changes in NK cell phenotype and function in the peripheral blood, little has been done in the way of characterizing NK cells in the LN, where HIV replicates preferentially and thus may be most vulnerable. We, therefore, investigated whether particular populations of NK cells accumulate in the LNs during HIV-1 infection that may be associated with early containment of virus. In this study, we show that neither NK cell levels nor subset distribution were altered in HIV⁺ individuals identified during the first year of HIV infection. However, LN NK cells in HIV⁺ individuals exhibited a higher activation status, marked by increased expression of TRAIL, and lower expression of 2B4 and CD161. Furthermore, few KIR⁺ NK cells were present in the LNs, despite active viral replication, due to low expression of the chemokine receptors CX₃CR1 and CXCR1, resulting in compromised KIR⁺ NK cell chemotaxis during HIV infection. These data suggest for the first time that NK cells exhibit a reduced capacity to migrate to LNs during early HIV infection, potentially providing the virus with an early environment enriched with target cells, but devoid of cytolytic effectors, where the virus may replicate unabated by cytolytic pressure.

Results

NK cells do not accumulate in LNs during HIV-1 infection

Early work suggested that lymphoid organs, and in particular LN, are the primary site of HIV replication and propagation, characterized by high tissue-associated viral loads, massive apoptosis of CD4⁺ T cells, accumulation of effector T cells, and production of proinflammatory cytokines [11, 12]. However, given that NK cells are rare within LNs in healthy controls, we first aimed to define whether the frequency of NK cells was altered in the LNs during early HIV infection. No difference was observed in the frequency of NK cells in the peripheral blood of HIV⁺ individuals compared with uninfected controls (Fig. 1A and B), potentially due to the fact that these individuals were beyond the acute phase of the infection. As previously reported [13], NK cells represent a remarkably small fraction of the lymphocytes in the LNs in healthy controls (Fig. 1A). However, surprisingly, despite active viral replication and associated inflammation at this site, the proportion of NK cells in the LNs of HIV⁺ subjects did not increase and the ratio remained similar among the blood and LN NK cells (Fig. 1A). Given that early HIV infection-associated lymphadenopathy leads to an enlargement of the LNs, due to increased lymphocyte recruitment and proliferation, we sought to determine whether the absolute number of NK cells changed in HIV⁺ subjects (Fig. 1C). As expected, early HIV infection was associated with an increase in the absolute frequency of both total lymphocytes and NK cells compared with healthy controls (Fig. 1C). However, the overall absolute number of NK cells was not altered. Thus, NK cells do not preferentially accumulate in LNs during the first year of HIV infection, potentially allowing the virus to replicate unabated by these critical innate antiviral cells.

Figure 1.

NK cells do not accumulate in LNs during HIV-1 infection. (A) The number of NK cells as a percentage of the total lymphocyte population in PBMC (circle) and LN (square) samples of individuals with HIV infection (open shapes) or controls (closed shapes) is shown. (B) Representative primary flow data of CD3^neg cells from PBMC and LN samples of acute HIV⁺ individuals and controls. The gate shows NK cells (CD56^pos and/or CD16^pos). (C) The absolute number of NK cells and lymphocytes recovered from LNs of HIV-infected (open symbols) and HIV^– (closed symbols) subjects is shown. (D) The percentage of the NK subsets, including the CD3^negCD56^brightCD16^neg, CD3^negCD56^dimCD16^neg/pos, and CD3^negCD56^negCD16^pos NK cells, in the total NK population in individuals in early HIV infection. Horizontal lines indicate the mean±SEM. Number of samples for each cohort: LN, 14 HIV⁺ and 4 HIV^–; PBMC, 7 HIV⁺ and 6 HIV^–. Statistical analysis was performed using an unpaired t-test. *p<0.05; **p<0.01; ***p<0.001.

No differences in overall NK subset distribution between HIV⁺ and HIV^– individuals

Although we did not observe a difference in the overall proportion of NK cells in the LNs of HIV⁺ patients compared to controls, we were interested to next determine whether HIV infection was associated with a re-distribution of NK cell subsets in the LN. NK cells in the LN predominantly belong to the immunoregulatory CD56^bright NK cell subset [13]. Given that early HIV infection is associated with inflammation in lymphoid tissues, we speculated that active viral replication may be associated with the preferential expansion of the cytolytic CD56^dim NK cells, typically involved in the antiviral response, that are rarely detectable in healthy LNs. Therefore, we compared the proportion of CD56^bright, CD56^dim, and CD3^negCD56^negCD16^pos (CD56^neg) NK cells in LNs and PBMCs of HIV⁺ individuals and negative controls. In line with previous work [9], we observed an increased frequency of CD56^dim compared with CD56^bright NK cells in the peripheral blood of healthy individuals (Fig. 1D). However, in our hands, the distribution of NK cells in the LNs of HIV⁺ subjects was on average closer to 30% CD56^bright and 40% CD56^dim, rather than the 90:10% distribution that had been previously reported [13]. Surprisingly, there was no significant difference in the distribution of NK cell subsets in the LNs in HIV⁺ individuals and controls (Fig. 1D). These data strongly suggest that early infection is not associated with a dramatic shift in the overall NK cell subset distribution in the LN, and that there is no preferential recruitment of CD56^dim cytolytic NK cells in the LN within the first year of HIV infection.

Alterations in LN NK cell receptor profiles in early HIV infection

As no redistribution in NK subsets was observed in untreated HIV⁺ LNs, we were next interested to determine whether the antiviral potential of NK cells may instead increase, through a shift in the NK cell repertoire, in the LN during early HIV infection. Thus, we examined the NK cell receptor repertoire in the LNs and PBMCs of infected and uninfected individuals. We did not observe any difference in the percentage of overall cells expressing CD94, NKG2A, NKp44, and NKp46 on NK cells in LNs between HIV⁺ and control individuals (Fig. 2). However, NK cells in the peripheral circulation did express lower levels of NKp30 [14] and NKG2D in HIV-infected individuals. Additionally, NK cells in HIV⁺ LNs also expressed significantly less CD161 (p<0.01) (Fig. 2B) when compared with control LNs or PBMCs. CD161 inhibits NK cell activation upon binding to its ligand, LLT1 [15], suggesting that LN NK cells may be less inhibited during HIV infection. As KIR⁺ NK cells have been implicated in the control of several viral infections [16, 17], including HIV infection [18, 19], we also assessed the expression of KIRs (CD158a, CD158b, and NKB1) on NK cells. We speculated that KIR⁺ NK cells may accumulate in the LN during HIV infection in an attempt to control viral replication. As previously reported, there was a slight increase in the proportion of KIR⁺ NK cells in the peripheral circulation of HIV⁺ individuals compared with uninfected controls (Fig. 2C). However, KIR⁺ NK cells were not only nearly absent from healthy LNs, in line with the previous reports of low KIR⁺ NK cell frequencies in healthy LNs [13], but were surprisingly also absent from the LNs of HIV-infected individuals (Fig. 2C). These data strongly suggest that KIR⁺ NK cells that are abundant in the blood from the same study subjects may be excluded from the LNs in both healthy and HIV-infected individuals.

Figure 2.

NK cells do not express KIR or elevated levels of typical NK cell markers in LNs during HIV-1 infection. (A–H) Representative primary flow data and plots show the expression of NK cell-surface markers on NK cells from HIV⁺ individuals and controls. Flow graphs in (A–H) show the proportion of NK cells expressing the respective cell-surface marker in a representative HIV-infected and uninfected control. Plots represent the percentage of bulk NK cells expressing the respective cell-surface marker in PBMC (circle) and LN (square) samples of individuals with HIV infection (open shapes) or controls (closed shapes). Horizontal lines indicate the mean±SEM. Number of samples for each cohort: LN, 14 HIV⁺ and 4 HIV^–; PBMC, 7 HIV⁺ and 6 HIV^–. Statistical analysis was performed using an unpaired t-test. *p<0.05; **p<0.01; ***p<0.001.

NK cells in LNs are highly activated

Previous work suggests that lymphocytes within LNs are highly activated and are phenotypically distinct to peripheral NK cells [20, 21]. Thus, we next investigated whether the activation profile of NK cells in LNs and PBMCs differed in early HIV infection. As previously reported, NK cells in LNs expressed significantly higher levels of CD69 in both HIV⁺ and HIV^– individuals, when compared with PBMC (p<0.001; Fig. 3B). In contrast, the co-activating NK cell receptor 2B4 was expressed at significantly lower levels on NK cells in LNs compared with PBMCs in both HIV⁺ and HIV^– individuals (p<0.01 and <0.05, respectively; Fig. 3A). 2B4 was initially identified as a co-activator of NKp46, NKp44, and CD16 [22]; however, more recently it has been shown to have both activating and inhibitory functions, depending on its level of expression [23]. At low 2B4 densities, 2B4 acts as an activating receptor, and thus reduced expression in HIV⁺ LNs might allow LN NK cells to be more prone to activation, allowing them to respond more readily to virally infected cells. Surprisingly, we observed an inverse correlation between the frequency of circulating 2B4⁺ NK cells and the absolute number of CD4⁺ T cells (p<0.01, Fig. 3F), suggesting that more activatable 2B4^lo NK cells may provide enhanced antiviral control, and therefore consequently enhanced CD4⁺ T-cell preservation.

Figure 3.

NK cells express altered levels of activation markers in LNs during HIV-1 infection. (A–E) Representative primary flow data and plots show the expression of cell-surface activation markers on NK cells from individuals with acute HIV infection. Flow plots show representative staining of NK cells for the expression of each cell-surface activation marker for a representative HIV-infected and uninfected control. Dot plots represent the percentage of bulk NK cells expressing the respective cell-surface marker in PBMC (circle) and LN (square) samples of individuals with HIV infection (open shapes) or controls (closed shapes). Horizontal lines indicate the mean±SEM. Number of samples for each cohort: LN, 14 HIV⁺ and 4 HIV^–; PBMC, 7 HIV⁺ and 6 HIV^–. Statistical analysis was performed using an unpaired t-test, with *p<0.05; **p<0.01; ***p<0.001. (F) Correlation between CD4⁺ T-cell counts and the expression of 2B4 on peripheral NK cells. Statistical analysis performed an using Pearson correlation coefficient r = –0.868; p<0.01.

PD-1 and HLA-DR serve as activation markers of lymphocytes, associated with persistent antigen exposure [24, 25] and have been implicated in both activation [26] and exhaustion of CD8⁺ T cells in HIV infection [27–29]. NK cells in the LNs of HIV^–individuals expressed elevated frequencies of HLA-DR (Fig. 3C); however, this difference was lost in HIV⁺ individuals, likely due to the fact that HLA-DR expression increased on NK cells in the blood during early HIV infection (Fig. 3C). However, no difference was observed in the expression of PD-1 between infected and uninfected patients, although there was a slight increase in PD-1 expression on the surface of LN NK cells in both infected and uninfected donors (Fig. 3D). These data suggest that NK cells in the LN are more activated but do not exhibit signs of exhaustion during early HIV infection.

NK cells mediate their cytolytic activity in a number of different manners, including the release of cytolytic granules, FAS/FAS-ligand binding, or through TNF-regulated apoptosis inducing ligand (TRAIL)/receptor interactions [30]. NK cells upregulate TRAIL upon activation [31]. Therefore, we quantified TRAIL expression on NK cells in the LNs and PBMCs in HIV-infected individuals compared with negative controls. TRAIL was poorly expressed on NK cells in the peripheral blood of both HIV⁺ individuals and negative controls (Fig. 3E). However, TRAIL was expressed at higher levels on NK cells in HIV⁺ LNs compared with NK cells in healthy LNs (p<0.05). Thus, it is plausible that LN resident NK cells may upregulate TRAIL during early HIV infection in an effort to contribute to the control of viral replication.

Decreased expression of CX₃CR1 and CXCR1 on HIV⁺ NK cells may affect their recruitment to LN

We were intrigued to next define whether alterations in homing receptors on the surface of NK cells could account for the lack of NK cell, and specifically KIR⁺ NK cells accumulation in the LN in early HIV infection. In mice, recruitment of NK cells to LNs depends on CD62L and CXCR3 [32]. Furthermore, CCR1 [33], CCR5 [34, 35], CCR7 [36], CXCR1 [37], CX₃CR1 [38], LPAM-1 (α4β7 integrin) [39], and CD103 (αEβ7 integrin) [40] have also been implicated in recruitment of NK cells to sites of inflammation. Surprisingly, NK cells in HIV⁺ individuals exhibited reduced CX₃CR1 (p<0.01) and CXCR1 (p<0.01) expression compared with uninfected controls, suggesting that despite HIV-associated inflammation, NK cells may have a reduced capacity to migrate to the LN due to reduced expression of chemokine receptors that are critical for homing to lymphoid tissues and sites of inflammation (Fig. 4A). Moreover, KIR⁺ NK cells expressed significantly lower levels of CD62L compared with KIR^– NK cells in both HIV⁺ individuals and controls (p<0.05, Fig. 4B). Despite the reduced expression of this LN homing marker on KIR⁺ NK cells, the overall population of NK cells in the LNs of HIV⁺ individuals expressed higher levels of CD62L compared with controls (p<0.01; Fig. 4C), suggesting that KIR^– NK cells with elevated CD62L expression may have preferential access to LNs during HIV infection. Furthermore, KIR⁺ NK cells, specifically in peripheral blood of HIV⁺ individuals, expressed lower levels of CXCR1 and CX₃CR1 when compared with controls (p<0.01 for both chemokine receptors; Fig. 4B), likely contributing to reduced NK cell chemotaxis to these vulnerable sites despite the inflammation associated with active viral replication.

Figure 4.

Distinct chemokine receptor profiles on NK cells during HIV-1 infection. (A) Expression of chemokine markers on bulk NK cells in PBMCs from HIV⁺ individuals (open circles) and negative controls (closed circles). (B) Percentage of total KIR⁺ or KIR^– NK cells expressing individual chemokine receptors in PBMC from HIV⁺ individuals (open circles) and negative controls (closed circles). Seven HIV^– and 10 HIV⁺ PBMC samples were analyzed for the expression of CD103, CCR1, and CXCR3; eight HIV^– and 16 HIV⁺ PBMC samples were analyzed for LPAM-1, CD62L, CCR5, and CCR7; seven HIV^– and 7 HIV⁺ PBMC samples were analyzed for CXCR1 and CX3CR1 expression. Horizontal lines represent the mean value. Statistical analysis was performed using an unpaired t-test, *p<0.05; **p<0.01. (C) CD62L expression on NK cells in PBMC (circle) and LN (square) samples of individuals with HIV infection (open shapes) or controls (closed shapes). A total of seven HIV⁺ and four HIV^– LN and eight HIV⁺ and five HIV^– PBMC samples were analyzed for the expression of CD62L. Horizontal lines represent the mean value. Statistical analysis was performed using an unpaired t-test. *p<0.05; **p<0.01.

To investigate whether the lower expression of CXCR1 and CX₃CR1 on peripheral KIR⁺ and KIR^– NK cells directly accounted for their reduced recruitment to the LN in HIV-infected subjects, we next assessed the chemotactic/migratory potential of these two NK cell subsets in response to IL-8 (ligand of CXCR1) and CX₃CL1 (ligand of CX₃CR1). Although there was a tendency towards a slight KIR^–, but not KIR⁺, NK cell migration in the presence of IL-8, consistent with increased KIR^– NK cells in the LNs in both HIV⁺ and HIV^– LNs, the overall level of migration was limited in the presence of this chemotactic factor (Fig. 5). More interestingly, in line with the decreased level of expression of CX₃CR1 on KIR⁺ NK cells in HIV-infected subjects (Fig. 4B), we observed a trend towards reduced KIR⁺ NK cell migration towards the compartment containing fractalkine compared with KIR⁺ NK cells from healthy controls (Fig. 5, p = 0.1). Furthermore, as expected, KIR⁺ NK cells migrated more robustly to the chemotactic signals compared with KIR^– NK cells in the healthy controls (Fig. 5, p = 0.04). However, while there was a tendency towards reduced KIR⁺ movement compared with KIR^– NK cells, in the HIV-infected patient population based on CX₃CR1 expression levels, this difference was nearly negligible, potentially reflecting an overall diminished capacity of these KIR⁺ NK cells to migrate to inflamed tissue due to additive activity of reduced expression of both CX₃CR1 and CXCR1. These data support the notion that reduced expression of chemotactic receptors on KIR⁺ NK cells during early HIV infection may result in reduced recruitment of KIR⁺ NK cells to the LN, despite aggressive viral replication and inflammation in this vulnerable tissue during early disease.

Figure 5.

KIR⁺ NK cells in HIV-infected subjects show lower levels of migration in response to CX3CR1 ligand. In vitro migration of sorted KIR⁺ and KIR^– NK cells obtained from six HIV⁺ (white bars) and six HIV^– (gray bars) individuals in response to medium only (background level of chemotaxis), CX3CL1/fractalkine (CX3CR1 ligand), and IL-8 (CXCR1 ligand). The horizontal dashed line represents the background level of NK cell migration. Data are shown as mean+SEM.

Discussion

Increasing evidence points to a central role for NK cells in the control of HIV disease progression [10, 18, 19]; however, little is known about the mechanism or location of their antiviral control. Given the fact that HIV replicates primarily in secondary lymphoid organs [1], including LNs and the gut [41], it is likely that the key(s) to defining the correlates of protection are hidden within these tissues. Thus, in this study we describe the first characterization of the distribution and phenotype of NK cell populations in the LNs of individuals identified within the first year of infection and show for the first time that despite alterations in the activation properties of NK cells, these cells, and specifically KIR⁺ NK cells, do not accumulate in LNs during early HIV infection, potentially due to a reduction in homing receptors required for recruitment to these vulnerable sites. Thus, in the absence of early innate immune pressure on the virus within this vulnerable immunological site, the virus may exploit a niche in which it can infect many target cells early on in the absence of cytolytic pressure, providing a fertile ground for early viral dissemination.

Despite the fact that NK cells do not expand in the LN in response to early viral infection, a number of HIV-associated phenotypic changes were observed in NK cells in LNs including elevated expression of CD69 and TRAIL and reduced expression of inhibitory receptors such as 2B4 and CD161, reflecting an exquisitely activated NK cell population poised to respond more effectively to infected cells. Diminished CD161 expression has been observed on CMV-, HIV- or influenza-specific CD8⁺ T cells [42], linked to increased CD8⁺ T-cell functionality [43]. Likewise, CD161^lo NK cells may be less inhibited, allowing them to respond more aggressively within the LN during early HIV infection. Most interestingly, we had initially hypothesized that cytolytic CD56^dim NK cells would expand in the LN during early infection; however, we did not observe any change in the total CD56^dim NK cell frequency within LNs. Instead, TRAIL expression increased on LN NK cells potentially allowing them to eliminate virally infected cells, in a non-cytolytic manner, through the induction of apoptosis. This alternative mechanism of antiviral control may reflect a homeostatic preference to prevent immunopathology within tissues, where cytotoxicity could potentially cause non-specific destruction that could adversely affect these critical inductive sites.

Epidemiologic data suggest that particular KIRs, when co-expressed with their HLA ligands, are associated with differential risk of infection [44, 45] and protection from disease progression [10, 18, 19, 46, 47]. However, we were surprised that KIR⁺ NK cells were not detectable in the LN during early HIV infection. This lack of KIR⁺ NK cell recruitment or expansion may reflect two different scenarios: (a) KIR⁺ NK cells may be excluded from the LN to prevent KIR-induced cytotoxicity or (b) KIR⁺ NK cells may lose KIR expression upon access to the LN. However, the latter possibility is unlikely due to the fact that KIR expression is programmed during NK cell development, and once established, KIR expression is thought to be fixed on a clonal level [48]. However, it is plausible that KIR⁺ NK cells may be excluded from tissues due to the fact that once activated these cells are highly cytolytic, and thus they may have the potential to cause immunopathology within the tissue. Therefore, it is more likely that these KIR⁺ NK cells may not accumulate in tissues, through homeostatic regulation of homing markers, to prevent access of these cytolytic effectors to vulnerable sites.

The chemokine receptor CCR7 expressed by lymphocytes, including NK cells, is the key receptor responsible for lymphocyte entry to LNs, due to its capacity to interact with CCL19 and CCL21 in secondary lymphoid organs [13, 49]. While we did not observe any difference in the frequency of CCR7⁺ NK cells between HIV⁺ and HIV^– donors, we did observe differences in the expression of CX₃CR1 and CXCR1 expression (Fig. 4A). Along these lines, elevated fractalkine (the ligand of CX₃CR1) levels have been observed in hyperplastic human LNs [50], and fractalkin is further enriched specifically in LN T-cell zones during HIV infection [51], potentially providing a strong recruiting signal for CX₃CR1 expressing lymphocytes. Similarly, acute HIV infection has also been associated with a dramatic increase in IL-8 (CXCR1 ligand) in both plasma [52, 53] and lymphoid tissues [54], also potentially driving the recruitment of CXCR1⁺ lymphocytes to the sites of infection or inflamed tissues [55]. However, several studies have demonstrated that persistent expression of IL-8 or fractalkine results in the internalization and downregulation of CXCR1 [56,57] and CX₃CR1 [58] in multiple inflammatory conditions. Thus, it is likely that elevated expression of these chemokine receptor ligands during HIV infection may result in the downregulation of their receptors on NK cells, in an effort to regulate cytolytic lymphocyte recruitment to tissues.

Given that alterations in chemokine receptor expression can have a profound impact on the capacity of lymphocytes to migrate to lymphoid tissues or sites of inflammation [59–62], we aimed to determine whether reduced expression of CXCR1 and CX₃CR1 on KIR⁺ NK cells could account for altered chemotaxis of these cells in response to their ligands. Migration experiments were performed using sorted KIR⁺ and KIR^– NK cell populations, demonstrating impaired KIR⁺ NK cell migration towards fractalkine in HIV-infected individuals, but not to IL-8, suggesting that alteration in some, but not all chemokine receptors, may impact the migratory capacity of KIR⁺ NK cells during HIV infection. Interestingly, alterations in chemokine receptor expression may alter KIR⁺ NK cell migration to all inflamed tissues, not only to LNs, thereby compromising early NK-cell-mediated control in other vulnerable tissues, including the gut. However, overall, these functional studies support the notion that downregulation of particular homing receptors on KIR⁺ NK cells may affect the recruitment of these potentially key antiviral effector cells to LNs.

Accumulating evidence suggests that earliest immunological responses that result in early containment of viral replication determine the rate of disease progression [6]. Thus, the data presented here imply that NK cells may mediate the bulk of their antiviral activity in the peripheral circulation, rather than in LNs, where robust early viral control may help prevent dissemination of the virus. Thus, viruses that rapidly evade detection by NK cells, reaching multiple secondary lymphoid sites may replicate more profusely as they have successfully evaded these early innate antiviral cells, until the first emerging HIV-specific CD8⁺ T cells begin to target the virus within these sites. Therefore, replication in tissues may offer a twofold advantage to the virus, as LNs offer access to large quantities of target CD4⁺ T cells as well as offer the potential to escape cytolytic NK-cell-mediated pressure. However, it is also plausible that we may have missed an early and transient recruitment of NK cells within the LN that may have taken place during the first few weeks of infection, due to the difficulty in collecting LN from acutely/early infected patients. However, animal models of infection may provide added insights into NK cell recruitment to secondary lymphoid tissues at earlier time points than are available from HIV-infected subjects.

In conclusion, we show for the first time that NK cells, and in particular KIR⁺ NK cells, do not accumulate in the LN early in HIV infection. This lack of recruitment of KIR⁺ NK cells from LNs may be attributable to a decreased expression of homing markers that impair the migratory capacity of these innate immune effector cells for lymphoid tissues or sites of inflammation. The LN resident NK cells, however, do respond to infection and exhibit an activated profile, including elevated expression on TRAIL that may provide an alternative, albeit modest amount of antiviral pressure on the virus within this vulnerable immunological site. These novel data suggest that the lack of KIR⁺ NK cells in the LN might provide an early sanctuary for robust HIV replication, prior to the induction of CD8⁺ T-cell responses and protected from the antiviral activity of NK cells [63].

Materials and methods

Study population

PBMCs and LNs were obtained from HAART naı¨ve HIV⁺ individuals in the first year of infection and healthy controls. Among the HIV⁺ individuals, subjects were males, 27–72 years of age, with a viral load between <50 and 3 750 000 RNA copies/mL of blood, and CD4 counts between 200 and 810 cells/μL. PBMC and LN samples obtained from HIV^– donors were used as controls. For chemokine receptor and migration assays, PBMCs were obtained from a total of 14 HIV^– donors and 27 HIV-infected subjects that were off therapy with viral loads between 6120 and 122 000 RNA copies/mL and CD4 counts between 255 and 982 cells/mL. All study subjects signed institutionally approved informed consent.

Isolation of LNs

Axillary LN biopsies were collected under local anesthesia after written consent was obtained. Mononuclear cells were collected from the LNs by mechanical pressure and frozen in fetal calf serum plus 10% DMSO.

NK cell activation and subset distribution

Cryopreserved LN samples (14 HIV⁺ and 4 HIV^–) and PBMCs (7 HIV⁺ and 6 HIV^–) were stained with three different flow cytometric panels. At least 1 × 10⁶ were stained with anti-CD3-PacificBlue, anti-CD16-allophycocyanin-Cy7, anti-CD56-PECy7 (BD Biosciences, San Jose, CA USA), anti-CD4-Qdot655, and anti-CD8-PECy5.5 (Invitrogen, Carlsbad, CA, USA). In addition, to delineate the subset distribution of NK cells in the LN, at least 1 × 10⁶ cells were stained with one of the following combinations of antibodies: (A) anti-CD161-PE-Cy5, anti-CD94-FITC, anti-NKG2D-allophycocyanin (BD Biosciences), and anti-NKG2A-PE (Beckman Coulter, Fullerton, CA, USA) or (B) anti-NKp30-PE, NKp44-Alexa647, anti-KIR-FITC (anti-CD158a-FITC clone HP-3E4+anti-CD158b-FITC clone CH-L+anti-NKB1-FITC clone DX9) (BD Biosciences), and anti-NKp46-biotin (Biolegend, San Diego, CA, USA). The activation state of NK cells was assessed using anti-CD3-PE-Cy5.5 (Invitrogen), anti-CD16-PE-Cy5, anti-CD56-PE-Cy7, anti-2B4-FITC, anti-HLA-DR-allophycocyanin-Cy7 (all from BD Biosciences), anti-TRAIL-PE, anti-PD-1-Biotin, and anti-CD69-PacificBlue (Biolegend). The expression of CD62L (anti-CD62L-PE-Cy5 antibody, BD Bioscience) was assessed in a subset of patients, seven HIV⁺ and four HIV^– in the LNs and eight HIV⁺ and five HIV^– in PBMCs. Cells were stained for 15 min, washed, and fixed by 4% paraformaldehyde, while tubes containing biotin-labeled antibodies were subsequently stained with streptavidin-cascade yellow (Invitrogen) for 15 min, washed, and fixed in 4% paraformaldehyde. A minimum of 3 × 10⁵ events were acquired on a BD LSRII flow cytometer (BD Biosciences), and the data were analyzed using FlowJo (Tree Star, Ashland, OR, USA).

Chemokine receptor and integrin expression

For chemokine receptor staining, PBMCs from healthy donors (7 samples for panels (A) and (C), and 8 samples for panel (B)) and chronic HIV⁺ individuals (10 samples for panel (A), 16 samples for panel (B) and 7 samples for panel (C)) were stained with a backbone of anti-CD3-PacificBlue, anti-CD16-allophycocyanin-Cy7, anti-CD56-Alexa 700 (all from BD Biosciences), and anti-KIR-FITC (anti-CD158a-FITC clone HP-3E4+158b-FITC clone CH-L+NKB1-FITC clone DX9) (BD Biosciences), including one of the following panels of antibodies: (A) anti-CD103-Alexa 647 (Biolegend), anti-CCR1-PE (R&D Systems, Minneapolis, MN, USA), and anti-CXCR3-PE-Cy7 (BD Biosciences) or (B) anti-LPAM-1 PE (Biolegend), anti-CD62L-PE-Cy5, anti-CCR5-allophycocyanin, and anti-CCR7-PE-Cy7 (all from BD Biosciences), or (C) anti-CXCR1-PE-Cy5 (BD Biosciences) and anti-CX₃CR1-PE (MBL International, Woburn, MA, USA). A minimum of 3 × 10⁵ events were acquired on a BD LSRII flow cytometer (BD Biosciences) and the data were analyzed using FlowJo.

KIR⁺ and KIR^– NK cell migration assay

The differential migratory potential of KIR⁺ and KIR^– NK cells in response to IL-8 and fractalkine was assessed on sorted KIR⁺ and KIR^– NK cells from PBMCs. A subset of chronically infected patients with matched CD4⁺ T-cell and viral load levels to those used in chemokine receptor analyses were selected, for whom large numbers of cryopreserved PBMCs were available for sorting. PBMCs were stained with anti-CD3-PacificBlue, anti-CD16-allophycocyanin-Cy7, anti-CD56-PECy7, and anti-KIR-FITC (anti-CD158a-FITC clone HP-3E4+anti-CD158b-FITC clone CH-L +anti-NKB1-FITC clone DX9) (BD Biosciences) and sorted into two subsets: (i) CD3^–CD56⁺CD16⁺KIR⁺ and (ii) CD3^–CD56⁺ CD16⁺KIR^– NK cells using a FACSAria sorter (BD Biosciences). Sorted cells were resuspended in 100 mL of migration medium (RPMI 1640 containing 1% bovine serum albumin and 10 mmol/L HEPES buffer, pH 6.9) and seeded in the top chamber of 24-well chemotaxis chambers (6.5 mm diameter, 5 mm pore polycarbonate transwell culture insert; Costar, Cambridge, MA, USA). Six-hundred microliters of migration medium was added in the lower chamber and supplemented with 100 ng/mL of recombinant human fractalkine (Assay Biotechnology, Sunnyvale, CA, USA), 100 ng/mL of recombinant human IL-8 (PeproTech, Rocky Hill, NJ, USA), or medium alone (negative control). The plates were incubated at 371C, 5% CO₂ for 3 h. Transwells were then removed, and the cells in the upper and lower reservoir were recovered, stained with Trypan blue, and counted using a light microscope. The percentage of migrating cells was calculated as: (number of NK cells migrated into the lower chamber)/(number of NK cells in the upper+lower chamber) × 100.

Statistical analysis

Statistical analysis of the data was performed using GraphPad Prism version 4.0c (GraphPad Software, San Diego, CA, USA, www.graphpad.com). Comparison between the means of two independent groups was made using the two-sample t-test. Comparison between the means of two dependent groups was done using the paired t-test. Correlation analysis was performed using Pearson correlation coefficient. Statistical tests were considered significant at p<0.05.

Abbreviations:

KIR

killer immunoglobulin receptor

LN

lymph node

TRAIL

TNF-regulated apoptosis inducing ligand