α2β1 Integrin Is Required for Optimal NK Cell Proliferation during Viral Infection but Not for Acquisition of Effector Functions or NK Cell–Mediated Virus Control

This article is featured in In This Issue, p.1419

Natural killer cells are leukocytes of the innate immune system that belong to the group 1 of innate lymphoid cells (ILCs). After developing in the bone marrow, NK cells circulate between blood and tissues patrolling the body for infections and tumors. Following viral infections, NK cells are rapidly recruited from the blood to secondary draining lymph nodes (dLNs) and other infected tissues, where they produce IFN-γ, kill infected cells, and proliferate, playing a critical role as a first line of antiviral defense (1, 2). For example, C57BL/6 (B6) mice, which are normally resistant to mouse CMV (MCMV) and ectromelia virus (ECTV), become susceptible when NK cells are eliminated or dysfunctional (3–7). Also, in humans, NK cell–deficient individuals become sick or succumb to normally non–life-threatening infections with human CMV or with varicella zoster (8–10).

The phenotypes and functions of NK cells overlap extensively with those of ILC1s, the only other member of the group 1 ILCs. For example, both ILC1s and NK cells express the typical NK cell–activating receptors NKp46 and, in B6 mice, NK1.1. Both ILC1s and NK cells need the transcription factor T-bet for their development (11). Moreover, NK cells and ILC1s produce IFN-γ upon stimulation. Because of this, ILC1s can participate in virus control in tissues (12). Yet, ILC1s and NK cells also exhibit differences that make them unique. ILC1s are tissue resident and do not circulate in the blood, whereas NK cells circulate in the blood and are transient within tissues. NK cells but not ILC1s are cytolytic and express the transcription factor eomesodermin (Eomes) (13–16). At the cell surface, the major differences between ILC1s and NK cells is in the expression of integrins. ILC1s but not NK cells express the integrin β3 and some but not all ILC1s express integrin α1 (12). In contrast, NK cells but not ILC1s express the integrin αM (CD11b, also expressed by other unrelated leukocytes) and the integrin α2 (in this study α2; also known as CD49b).

Integrins are cell surface heterodimeric receptors formed by an α- and a β-chain (17). Integrins function in the adhesion of leukocytes to the endothelium and their extravasation into lymph nodes (LNs) and inflamed tissues. Integrins are also important for the displacement of leukocytes within tissues and can participate in their development and activation (18). α2 dimerizes with the integrin β1 (in this study β1; also known as CD29) to form α2β1 (also known as VLA-2), a receptor for collagens type I and III. Of note, β1 is thought to be the only partner of α2, whereas β1 can heterodimerize with 12 different α-chains (19, 20), including α1 (α1β1) in ILC1s. Notably, α2β1 is also expressed by human NK cells (21).

Despite α2 being a reliable NK cell marker, the role of α2β1 in the biology of NK cells is not well understood. When the Lanier group discovered α2 on murine NK cells, they also showed that two different anti-α2 mAbs, DX5 and HMα2, do not block the binding of NK cells to collagen-coated plates, despite that HMα2 blocks the α2β1-dependent binding of platelets to collagen. In addition, the Lanier group demonstrated that immobilized anti-α2 mAbs did not stimulate IFN-γ production in NK cells, nor did they block NK cell killing of YAC-1 cells (22, 23). Together, these data suggested that α2β1 may not be essential for NK cell function. Later, others showed that positive purification of NK cells using DX5 mAb negatively affected their ability to produce IFN-γ in vitro, kill target cells in vitro and in vivo, and impaired their motility in LNs (24). However, these data did not directly demonstrate a physiological role for α2β1. More recently, it was shown that NK cells interacted with collagen fibers in the dLNs of mice infected in the ear flaps with Toxoplasma gondii, and that administration of HMα2 in the ear flap decreased the presence of NK cells in T. gondii–infected foci in dLNs (25). This suggested that α2β1 interactions with collagen might be involved in NK cell migration to infected foci. Similarly, a study with human NK cells in vitro also suggested that α2β1 might participate in the interaction of NK cells with collagen, but the data were correlative (21).

To further understand the role of α2β1 in NK cell biology, we produced mice with specific genetic ablations of α2 in NKp46⁺ cells. Using these mice, we show that α2β1 affects but is not critical for NK development and maturation. Moreover, we also show that during ECTV infection α2β1 is not necessary for NK recruitment to or distributing within the dLN of ECTV-infected mice or at becoming effectors after ECTV or MCMV infection. Notably, α2β1 is required for optimal NK proliferation in response to ECTV and increases in numbers in response to MCMV but is dispensable for survival to ECTV and the control of MCMV titers in the spleen. These results indicate that α2β1 is not essential for NK cell differentiation or effector function. Moreover, we demonstrate that NK cells with defective proliferation can still protect from ECTV lethality and can control MCMV replication.

All the procedures involving mice were carried out in strict accordance with the recommendations in the eighth edition of the Guide for the Care and Use of Laboratory Animals of the National Research Council of the National Academies. All protocols were approved by the Thomas Jefferson University Institutional Animal Care and Use Committee. All mice used in experiments were 6–12-wk-old Ncr1Cre⁺-Itgb1^fl/fl or Ncr1Cre⁺-Itga2^fl/fl with appropriate littermate controls. No sex differences were observed. C57BL/6 (B6) and B6.CD45.1 mice were purchased from Charles River Laboratories directly for experiments or as breeders. Aged B6 mice (18–20 mo) were obtained from the National Institute of Aging aged colony at Charles River Laboratories. B6.Cg-Itga2^tm1.1Tkun/J (α2^fl/fl) and B6;129-Itgb1^tm1Efu/J (β1^fl/fl) mice were purchased from Jackson Laboratories. Of note, the β1^fl/fl mice originally obtained from Jackson were in mixed 129/B6 background and carried the NK complex in chromosome six from the 129 strain. Therefore, they lacked NK1.1. To solve this problem, the original β1^fl/fl mice were backcrossed to C57BL/6 (B6) mice for two generations to generate β1^fl/fl mice expressing NK1.1 homozygously. C57BL/6-Ncr1^tm1.1(iCre^)Viv/Orl mice (Ncr1-Cre) were a gift of Dr. E. Vivier (Marseille, France). Colonies were bred at Thomas Jefferson University under specific pathogen–free conditions.

ECTV-Moscow strain (American Type Culture Collection; ATCC VR-1374) and ECTV-mCherry were propagated in tissue culture as previously described (26). Mice were infected in the footpad with 3000 or 100,000 PFUs of ECTV as indicated. For the determination of survival, mice were monitored daily, and, to avoid unnecessary suffering, mice were euthanized and counted as dead when imminent death was certain as determined by lack of activity and unresponsiveness to touch. Euthanasia was carried out according to the 2013 edition of the American Veterinary Medical Association Guidelines for the Euthanasia of Animals. For virus titers, the entire spleen or portions of the liver were homogenized in 2.5% FBS RPMI 1640 medium (Corning) using a TissueLyser (QIAGEN). Virus titers were determined on BSC-1 cells as previously described (26).

For MCMV infections, indicated strains were infected i.p. with 2.5 × 10⁵ PFU of MCMV K181 (K181) or 1.0 × 10⁴ PFU of MCMV V70 (V70) for 120 h. K181 stocks were produced by infecting 4 × 10⁶ M210-B4 (American Type Culture Collection; ATCC CRL-1972TM) with a multiplicity of infection of 0.01 for 4–5 days postinfection (dpi) or until cytopathic effect in culture was observed. Cells and supernatant were collected, centrifuged at 2600 × g for 10 min, and resuspended in 10 ml of complete RPMI 1640 (Corning) supplemented with 10% FBS, 10 mM HEPES (Corning), 1 mM sodium pyruvate (Corning), 100 U penicillin (Corning), and 100 μg/ml streptomycin (Corning). Debris were then lysed with a Dounce homogenizer while on ice and spun. Supernatant was collected, transferred to ultracentrifuge tubes (Beckman Coulter), and spun at 50,000 × g for 1 h at 4°C. The supernatant was discarded, and the pellet was resuspended in 500 μl of complete RPMI 1640. V70 stocks were produced by infecting BALB/c mice (>3 wk old) with 10 PFU of tissue-cultured V70 for 21 d. Whole salivary glands were harvested and placed in 1 ml of sterile PBS (Corning), processed with a tissue grinder, and diluted to a 10% mixture of homogenate to media with a final concentration of 5% DMSO. Virus titers were determined by infecting M210-B4 monolayers for 4–5 d and counting the resulting plaques. This protocol was adapted from a published protocol (27).

NK cells were depleted by i.p. inoculation of 100 μg NK1.1 mAb (PK136; Bio X Cell) 1 d before injection and 1 dpi with ECTV. For MCMV infection, NK cells were depleted similarly 1 d before and 1 and 3 dpi with MCMV.

Preparation of nondraining LNs (ndLNs) and dLNs was performed as previously described (28). Ten-micrometer sections were stained with allophycocyanin–NK1.1 (PK136; BioLegend) and mounted in ProLong Diamond plus DAPI (Thermo Fisher Scientific). Images were collected with a Nikon A1R laser scanning microscope. For better visualization, all the photographs were assembled in a single file, and the contrast and brightness was increased in unison using Adobe Photoshop. In addition, allophycocyanin–NK1.1 is shown in the green channel to distinguish NK cells more clearly from virally infected cells.

Mice were euthanized by cervical dislocation. Single‐cell suspensions were prepared from spleen and bone marrow and lysed for RBCs using ammonium‐chloride‐potassium (ACK) lysis buffer, and cells were washed with RPMI 1640 (Corning) supplemented with 5% FCS and later used for flow cytometric analysis. To obtain single-cell suspensions, LNs were first incubated in Liberase TM (1.67 Wünsch units/ml) (Sigma-Aldrich) in PBS with 25 mM HEPES for 30 min at 37°C before adding PBS with 25 mM HEPES plus 10% FBS to halt the digestion process, followed by mechanical disruption of the tissue through a 70-μm filter.

Flow cytometry to characterize NK cells in the bone marrow and spleen was performed as previously described (29). To determine NK cell responses in the LNs, intact LNs were incubated at 37°C for 1 h in media containing 10 μg/ml brefeldin A and then made into single-cell suspensions. The cells were then stained for cell surface molecules, fixed, permeabilized, and stained for intracellular molecules using the Cytofix/Cytoperm Kit (BD Biosciences) or the eBioscience Foxp3/Transcription Factor Staining Kit (Invitrogen) according to the manufacturer’s instructions. The following Abs were used: FITC–CD3ε (clone 145-2C11; BioLegend), allophycocyanin/Fire750-CD11b/BV605-CD11b (clone M1/70; BioLegend), PerCP/Cy5.5-CD27 (clone LG.3A10; BioLegend), PE/Cy7-CD29 (clone HMβ1-1; BioLegend), PE–CD45.1 (clone A20; BioLegend), Percp-cy5.5-CD45.2 (104; BioLegend), BV421–CD49b (clone RB6-8C5; BioLegend), AF488–Eomes (clone Dan11mag; eBioscience), PE/Cy7-IFN-γ (XMG1.2; BioLegend), allophycocyanin–Ki67 (16A8; BioLegend), PE/FITC-Ly49H (3D10; BioLegend), allophycocyanin–NK1.1/BV605-NK1.1 (PK136; BioLegend), allophycocyanin–NKp46 (clone 29A1.4; BioLegend), BV786–TCRβ (clone H57-597; BD Biosciences), and either Pacific Blue–granzyme B (GzmB) (clone GB11; BioLegend) or PE-labeled anti–human GzmB (Thermo Fisher Scientific) that cross-reacts with mouse GzmB. For analysis, samples were acquired using a BD LSRFortessa flow cytometer (BD Biosciences), and data were analyzed with FlowJo software (Treestar).

Splenocytes were obtained from the indicated strains of mice, mixed in a 1:1 ratio, then labeled with 4 μM CFSE (Thermo Fisher Scientific), and a total of 2 × 10⁷ cells (1 × 10⁷ cells each) in 0.2 ml of PBS were inoculated i.v. into the recipient mice.

Total RNA from LNs was obtained with the RNeasy Mini Kit (QIAGEN), as previously described (30, 31). First-strand cDNA was synthesized with High Capacity cDNA Reverse Transcription Kit (Life Technologies). For EVM003 and Gapdh, quantitative RT-PCR (qRT-PCR) was performed using iTaq Universal SYBR Green with the following primers: Gapdh forward: 5′-TGTCCGTCGTGGATCTGAC-3′, reverse: 5′-CCTGCTTCACCACCTTCTTG-3′; EVM003 forward: 5′-TCTGTCCTTTAACAGCATAGATGTAGA-3′, reverse: 5′-TGTTAACTCGGAAGTTGATATGGTA-3′. For viral loads, RNA (ECTV) from naive mice was used as control, and no amplification was observed. Thus, for quantification purposes, their cycle threshold values were adjusted to 40.

Total DNA was extracted from the spleen using the Gentra Puregene Tissue Kit (QIAGEN), following the manufacturer’s instructions for extraction of DNA from tissues. The spleen was homogenized in 10 ml of 2.5% FBS RPMI (Corning), and 1 ml was taken for DNA extraction. RNA and protein were removed by adding RNase A Solution and Proteinase K. Extracted DNA was quantified by nanodrop, and 2 μl of DNA was used as a template in each qRT-PCR reaction using the TaqMan Fast Advanced Master Mix (Thermo Fisher Scientific) with the following primer for MCMV-E1 forward: 5′-TCGCCCATCGTTTCGAGA-3′, reverse: 5′-TCTCGTAGGTCCACTGACGGA-3′. Genome copy numbers were calculated based on a standard curve of a plasmid containing the MCMV-E1 gene.

Statistical analysis was performed with Prism Software (GraphPad Software, La Jolla, CA). For survival studies, p values were obtained with the log‐rank (Mantel–Cox) test. The p values were determined using the Mann–Whitney U test, and when multiple groups had to be compared, we used one‐way ANOVA and Fisher least significant difference test for multiple comparisons.

To investigate the role of α2β1 in NK cell maturation and function, we crossed Ncr1Cre⁺ mice (Ncr1 encodes NKp46) (32) with either Itgb1^fl/fl or Itga2^fl/fl mice to, respectively, generate mice with NK cells deficient in β1 (Ncr1Cre⁺-Itgb1^fl/fl, in this study Cre⁺-Itgb1^fl/fl) or α2 (Ncr1Cre⁺-Itga2^fl/fl, in this study Cre⁺-Itga2^fl/fl). Flow cytometry analysis of bone marrow and spleen cells showed that most but not all CD3⁻ NK1.1⁺ cells in Cre⁻-Itgb1^fl/fl littermate control mice but only a few Cre⁺-Itgb1^fl/fl mice expressed α2 and β1. This indicates that in NK1.1⁺ cells, α2 is only expressed at the cell surface if β1 is also expressed. T cells (CD3e⁺ NK1.1⁻) did not express α2 and had low levels of β1 in either Cre⁻- or Cre⁺-Itgb1^fl/fl mice (Fig. 1A). Meanwhile, α2 in CD3⁻ NK1.1⁺ was expressed at high levels in Cre⁻-Itga2^fl/fl mice but was absent in Cre⁺-Itga2^fl/fl mice, which mostly maintained β1. In contrast, α2 was not expressed by T cells in either Cre⁻- or Cre⁺-Itga2^fl/fl mice (Fig. 1B). This demonstrates that α2 is efficiently deleted in NK1.1⁺ cells of Cre⁺-Itga2^fl/fl mice. Thus, Cre⁺-Itga2^fl/fl mice have α2β1 deficiency at the surface of NK1.1⁺ cells. However, because β1was still expressed, β1 likely pairs with other integrins at the surface of NK1.1⁺ cells. Thus, subsequent experiments are described with Cre⁺- and Cre⁻-Itga2^fl/fl mice, as in addition to α2β1 loss, NK1.1⁺ cells in Cre⁺-Itgb1^fl/fl have deficiencies in additional β1 pairs. Also, Itgb1^fl/fl mice are in a B6;129 mixed background, which could complicate the analysis. Of note, the deficiency of α2 in NK1.1⁺ cells did not redirect NK cells to the ILC1 linage because the frequencies of NK cells (Eomes⁺) and ILC1s (Eomes⁻) within group 1 ILCs (NK1.1⁺ CD3⁻) was similar in the bone marrow, spleen, and peripheral LN of Cre⁻- and Cre⁺-Itga2^fl/fl (Fig. 1C), where NK cells were the predominant population.

We next investigated whether intrinsic α2β1 deficiency affects NK cell development. Expression of CD27 and CD11b defines different maturational stages of NK cells in which CD27⁺ CD11b⁻ (in this study R1) are immature, CD27⁺ CD11b⁺ (in this study R2) are transitional but with effector capacity, and CD27⁻ CD11b⁺ (in this study R3) are the most mature and have the strongest cytolytic function (33) (Fig. 2A). We found that in Cre⁺-Itga2^fl/fl mice, the frequency of NK1.1⁺ cells was decreased in the bone marrow (Fig. 2B) and in the spleen (Fig. 2C) but not in the blood (Fig. 2D). However, the frequencies of the different maturation stages (R1, R2, and R3) were not altered in any of the organs (Fig. 2B–D). Thus, the absence of α2β1 seems to slightly decrease the frequency of NK cells in the bone marrow and spleen but has no effect on NK cell maturation.

NK cells are present in LNs at low numbers at steady state and increase their numbers by migrating from the circulation during inflammation. We have previously shown that mature R3 NK cells rapidly enter the dLN after ECTV infection (7). Given the importance of integrins in lymphocyte adhesion and migration it was possible that α2β1 was required in this process. Cre⁺- and Cre⁻-Itga2^fl/fl mice were challenged with ECTV and the recruitment of NK cells into the dLNs was measured by comparing frequencies of NK cells in the ndLN (i.e., at steady state) and dLN at 48 h postinfection (hpi). We observed that Cre⁺-Itga2^fl/fl mice preferentially recruited NK cells to the dLN but slightly significantly less than Cre⁻-Itga2^fl/fl mice, which could be an effect of the decreased number of NK cells in the bone marrow and spleen of Cre⁺-Itga2^fl/fl. Nevertheless, similar to Cre⁻ controls, the accumulation of NK cells in the dLN of Cre⁺ mice resulted in a relative increase in the frequency of mature R3 NK cells and a concomitant decrease in immature R1 NK cells (Fig. 3A, 3B).

It was possible that in Cre⁺-Itga2^fl/fl mice NK cells deficient in α2β1 migrated normally to the dLN because of a lack of competition. Thus, splenocytes from Cre⁺- and Cre⁻-Itga2^fl/fl mice were mixed in a 1:1 ratio, labeled with CFSE, and transferred into B6 mice, which were subsequently infected with ECTV in the footpad. At 48 hpi, the ratios of Cre⁺/Cre⁻Itga2^fl/fl NK cells in the ndLN and the dLN were similar, confirming that deficiency in α2β1 does not affect the constitutive or virus-induced entry of NK cells into the dLN (Fig. 3C).

Next, we looked at NK cell localization within LNs. As mentioned previously, Ab blockade suggested that α2β1 directs NK cells to sites of bacterial infection within the dLN via interactions with collagen (24). Thus, we analyzed whether absence of α2β1 results in NK cell mislocalization within the dLN during a viral infection. Cre⁺- and Cre⁻-Itga2^fl/fl mice were infected with ECTV expressing mCherry, and ndLNs and dLNs were visualized by confocal microscopy after staining with anti-NK1.1 mAb. The analysis confirmed recruitment because there was an increase in NK1.1⁺ cells (green) in the dLN versus the ndLN of both Cre⁺ and Cre⁻ mice. Furthermore, similar to Cre⁻ controls, NK1.1⁺ cells in Cre⁺ mice distributed throughout the dLN, including infected and uninfected areas (Fig. 3D).

We also analyzed effector functions. At 60 hpi, the NK cells in the dLN of B6 mice are highly activated as determined by GzmB and IFN-γ expression, and these effector molecules are critical for the control of ECTV dissemination from the dLN to the liver and the spleen (7). Thus, we tested whether α2β1 deficiency affects expression of GzmB and IFN-γ in NK cells. We found that Cre⁻- and Cre⁺-Itga2^fl/fl had similarly increased frequencies of both IFN-γ⁺ and GzmB⁺ NK cells in the dLN at 60 hpi with ECTV when compared with the ndLN (Fig. 3E).

In addition to curbing virus spread from LNs, NK cells play an important role in the control of ECTV by becoming activated in the spleen and liver until the T cell response develops (3, 7). Thus, we also looked at NK cell responses in the spleen. At 120 hpi with ECTV, which is the peak of the NK response, the frequency and absolute numbers of NK cells in the spleens of Cre⁺ mice were significantly lower than in Cre⁻-Itga2^fl/fl and naive mice (Fig. 4A, 4B). A similar effect was observed in the liver and blood (Supplemental Fig. 1). However, the frequency of GzmB⁺ NK cells was not significantly different between Cre⁺ and Cre⁻ mice and was significantly higher than in naive mice (Fig. 4A, 4C), indicating that NK cells do not need α2β1to acquire effector functions in the spleen or liver during ECTV infection.

To determine whether reduced numbers of NK cells with efficient effector differentiation was a characteristic of the ECTV NK cell response or could be extended to other viral infections, we infected Cre⁺- and Cre⁻-Itga2^fl/fl mice with MCMV. As with ECTV, Cre⁺-Itga2^fl/fl mice had reduced frequencies and numbers of NK cells than Cre⁻-Itga2^fl/fl mice at 120 hpi. Compared with naive mice, MCMV-infected Cre⁻-Itga2^fl/fl had an increased total number (but not frequency) of NK cells (Fig. 4D, 4E). As with ECTV, the upregulation of GzmB in Cre⁺-Itga2^fl/fl mice remained intact (Fig. 4D, 4F).

During MCMV infection of B6 mice, NK cells expand in two phases (the early phase occurs within the first 2 d of infection), is dependent on cytokines, and most NK cells expand (34, 35). The late phase occurs ∼120 hpi and happens mostly in Ly49H⁺ NK cells that recognize the viral protein m157 at the surface of infected cells (36–40). Analysis of Ly49H⁺ and Ly49H⁻ NK cells revealed that Ly49H⁺ NK cells from Cre⁺-Itga2^fl/fl mice did not increase in number, whereas the Ly49H⁺ NK cells from Cre⁻ mice did (Fig. 4G, 4H). Yet, similar frequencies of NK1.1⁺ Ly49H⁺ cells produced GzmB in Cre⁻- and Cre⁺-Itga2^fl/fl mice (Fig. 4G, 4I). Therefore, during MCMV infection, Ly49H⁺ NK cells deficient in α2β1 failed to increase in numbers, but MCMV-specific NK cells (i.e., Ly49H⁺) still acquired an effector phenotype.

Given the reduced NK cell numbers, we next used the ECTV model to test whether α2β1 is necessary for NK cell proliferation. At 120 hpi with ECTV, Cre⁺-Itga2^fl/fl mice had a significantly lower frequency of Ki67⁺ NK1.1⁺ cells in the spleen than Cre⁻-Itga2^fl/fl mice, indicating that in the absence of α2β1, fewer NK cells entered into G1 (Fig. 4J). To determine whether reduced Ki67 expression translated to impaired cell division, we adoptively transferred CFSE-labeled splenocytes from Cre^+­- and Cre⁻-Itga2^fl/fl mice in a 1:1 ratio into a CD45.1 host, which was then infected with ECTV in the footpad. Proliferation of Cre⁺ and Cre⁻ NK cells (identified by α2 staining) was assessed by CFSE dilution at 144 hpi. The results indicated significantly decreased proliferation in Cre⁺-compared with Cre⁻-Itga2^fl/fl mice (Fig. 4K). This indicates that α2β1 is required for optimal proliferation of NK cells during ECTV infection.

B6 mice are naturally resistant to lethal mousepox but become highly susceptible when depleted of NK cells with anti-NK1.1 or Asialo GM Ab or when NK cells are dysfunctional, as demonstrated by increased virus titers in spleen and liver at 3 and 5 dpi, and by high lethality (29, 41). When challenged with ECTV, Cre⁻- and Cre⁺-Itga2^fl/fl had similar levels of viral gene expression in the dLNs at 72 hpi (Fig. 5A) and similar titers of replicating virus in the spleen and liver at 120 hpi, as determined by plaque assay (Fig. 5B). Treatment with anti-NK1.1 mAb resulted in a significant increase in lethality in both Cre⁻-and Cre⁺-Itga2^fl/fl mice, whereas undepleted Cre⁻- and Cre⁺-Itga2^fl/fl mice were resistant, indicating that α2β1 is not required for NK cell–mediated resistance to lethal mousepox (Fig. 5C). Moreover, Cre⁺-Itga2^fl/fl mice were resistant when challenged with a significantly higher dose of ECTV, but control aged B6 mice, which have immature and dysfunctional NK cells, succumbed to ECTV infection (Fig. 5D). Also, at 5 dpi with tissue-cultured (K181) and salivary gland passaged (V70) MCMV, Cre⁻ and Cre⁺ mice had similar virus loads in the spleen, which increased similarly after NK cell depletion (Fig. 5E, 5F). These results indicate that the α2β1 in NK cells and optimal NK cell proliferation are dispensable for effective NK cell–mediated control of two mouse-specific viruses in the acute phase of infection.

In this study, we show that α2β1 expression by NK cells is not essential for their maturation or commitment to the NK cell lineage, entry to and distribution within LNs, or the induction of an anti-ECTV effector response. However, α2β1 is required for optimal NK cell proliferation during ECTV infection and for increase in NK cell during MCMV infection. Despite this, α2β1-deficient NK cells were still able to protect from ECTV and control MCMV titers in the spleen, demonstrating that mature and functional NK cells can mediate protection without extensive proliferation in two viral infection models in which the antiviral role of NK cells is very well established.

Previous research has shown that the VLA family of integrin heterodimers has roles in metastasis of cancer cells, immunomodulation, and leukocyte extravasation and retention in tissue (42–47). Whereas α2β1 is used as a marker to identify NK cells across mouse strains as well as to distinguish NK cells from ILC1s, its role in NK cell biology remains unclear. Bone marrow stromal cells synthesize type 1 and 3 collagen fibers (48), and it has been shown that their reduction in bone marrow stromal cells leads to decreased collagen synthesis (49, 50). Previously, we have shown that the NK cells in the bone marrow and other organs of aged mice are immature with increased R1 and decreased R3 populations (29, 51). We also showed that NK cells in the bone marrow of aged mice have lower expression of α2β1 (51) and that NK cell extrinsic factors in the bone marrow stroma are responsible for the defective maturation of NK cells in aged mice, leading us to hypothesize that α2β1 and its interaction with collagen are necessary for NK cell maturation. Thus, we also hypothesized that expression of α2β1 on developing NK cells and their interaction with collagen could be required for their proper linage commitment and maturation in the bone marrow. However, we found that α2β1 does not influence lineage commitment of the common innate lymphoid progenitor to NK cells versus ILC1s. Also, although α2β1 deficiency slightly decreases the frequency of NK cells in the bone marrow and spleen, it does not affect NK cell maturation because the proportions of R1s and R3s were similar within the bone marrow, peripheral blood, and spleen of Cre⁺- and Cre-Itga2^fl/fl mice. Thus, the data strongly suggest that the deficient NK cell maturation in aged mice (51) is not because of deficient α2β1 expression.

It is known that CXCR3 directs homing of NK cells to the dLN during viral infection (28, 52), but the identity of the integrin(s) involved in entry into the dLN remains unknown. We show that Cre⁺-Itga2^fl/fl NK cells enter ndLNs and dLNs normally whether or not they are in competition with Cre⁻-Itga2^fl/fl NK cells. This indicates that α2β1 is not required for NK cell constitutive entry into the ndLN and ECTV-induced entry into the dLN. Also, using confocal microscopy, we found that α2β1-deficient NK cells distribute similarly within the ndLN and dLN, including areas of virus infection, indicating that α2β1 is not required for intranodal displacement. Thus, adhesion molecules other than α2β1 may be responsible for the NK cell extravasation into and distribution within ndLNs and dLNs.

It has been reported that α2 in mouse NK cells and β1 in human NK cells are needed to control T. gondii and Cryptococcus neoformans infections, respectively, via direct or indirect activation of NK cells (25, 53). In this study, we show that during viral infections α2β1 is not required for NK cell activation because Cre⁺-Itga2^fl/fl NK cells upregulated IFN-γ or GzmB normally during ECTV and MCMV infections. Furthermore, Cre⁻- and Cre⁺-Itga2^fl/fl mice controlled ECTV and MCMV similarly, whereas depletion of α2β1 NK cells resulted in high susceptibility to lethal mousepox and higher MCMV titers in spleen, indicating α2β1 is dispensable at conferring NK cell–mediated control against ECTV and MCMV.

Interactions with β1 and collagen have previously been shown to affect the expansion of pancreas β cells as well as the morphogenesis of submandibular epithelial cells (54, 55). Thus, we believe α2β1 induces proliferation of NK cells but does not affect their trafficking or localization during viral infections because of the global reduction in NK cells within the blood, liver, and spleen of infected Cre⁺-Itga2^fl/fl mice. It remains possible, however, that NK cells are increased at other sites that we did not test. Importantly, we showed that α2β1 is required for optimal NK cell proliferation. Future experiments should identify the ligands and signaling mechanisms of this α2β1-dependent proliferation and whether β1 plays a role in the proliferation of other leukocytes.

It is well established that survival to ECTV and efficient control of MCMV requires NK cells and that the NK cell responses to both viruses include acquisition of effector functions and proliferation. Our results show that the NK cell–mediated resistance to ECTV and control of MCMV virus loads can still occur with significantly decreased proliferation. It is also known that innate immune mechanisms, including NK cell effector functions in the dLN, play a critical role in protection against lethal ECTV infection (29, 56). Notably, at 48–72 hpi, the frequency of NK cells in the dLN of Cre⁺-Itga2^fl/fl is only slightly decreased. Thus, the successful migration of mature and functional α2β1-deficient NK cells to the dLN at 48–72 hpi and their activation in the spleen allows for ECTV control, even when NK numbers are altered. These findings further highlight the importance of controlling viral replication in the dLN during an acute virus infection.

During primary infection with MCMV, Ly49H⁺ NK cells interacting with the viral protein m157 expand and protect mice from MCMV (5, 39, 57, 58). These MCMV-specific NK cells then contract and give rise to long-lived memory NK cells that contribute to protection upon reinfection (59–61). Our data suggest that, in addition to Ly49H interacting with m157, α2β1 acts as an additional signal for optimal Ly49H⁺ NK cell proliferation. Future experiments could look into whether this affects the induction of memory NK cells.

In summary, α2β1 is viewed as a phenotypic hallmark of NK cells, but its role in NK cell biology is not well understood. Our work shows that α2β1 is not required for NK cell maturation, migration to LNs, or acquisition of effector functions. However, α2β1 is required for optimal virus-induced proliferation and accumulation of NK cells. Unexpectedly, despite impaired proliferation, α2β1-defficient NK cells still protected from two viral infections in their natural hosts.

We thank Lingjuan Tang for technical assistance and the Flow Cytometry and Laboratory Animal at Thomas Jefferson University for services.

The authors have no financial conflicts of interest.