Dual fluorescence reporter mice for Ccl3 transcription, translation, and intercellular communication

Chemokines facilitate immune cell encounters, for example, during the defense against pathogens and tumors (Proudfoot, 2002; Bachmann et al., 2006; Zlotnik and Yoshie, 2012; Bird, 2017; Nagarsheth et al., 2017). While the production of secreted chemokines can be determined in a straightforward manner, for example, by intracellular flow cytometry, so far no methods exist to directly determine the cells that have sensed them.

Cytomegalovirus (CMV) infection is a common viral infection that causes mild symptoms in healthy individuals but may lead to severe illness in immunologically immature or immunosuppressed patients, resulting in pneumonia, hepatitis, or encephalitis (Griffiths and Reeves, 2021). The mouse model of infection with murine CMV (mCMV) reproduces many human CMV symptoms, especially hepatitis and pneumonia (Brizić et al., 2018; Reddehase and Lemmermann, 2018). The early immune defense against CMV is mediated by natural killer (NK) cells, which play a critical role in defense against many viral infections (Salazar-Mather et al., 1998; Mitrović et al., 2012; Mujal et al., 2021). They can directly recognize infected cells by detecting stress-induced ligands on their surface (Pyzik et al., 2011a, 2011b). NK cells release cytotoxic granules and secrete cytokines to eliminate the infected cells and enhance the immune response (Yokoyama and Kim, 2006; Quatrini et al., 2021; Wolf et al., 2023).

Upon recognition of CMV, immune cells produce type I interferons (IFNs), interleukins (ILs), and chemokines to activate and recruit immune effector cells to the site of infection (Brinkmann et al., 2015; Müller et al., 2017; Biolatti et al., 2018; Dell’Oste et al., 2020). The early response critically hinges on the chemokine Ccl3 (also known as macrophage inflammatory protein-1/MIP-1α), which has been shown to be produced mainly by resident macrophages (Menten et al., 2002; Maurer and von Stebut, 2004) and serves to recruit and activate immune cells, particularly NK cells, T cells, and monocytes, to sites of mCMV infection, like liver and lung (Salazar-Mather et al., 2000; Salazar-Mather and Hokeness, 2006). Blocking Ccl3 in mCMV-infected mice prevented NK cell recruitment and exacerbated infection (Salazar-Mather et al., 1998). Ccl3 binds to the chemokine receptors Ccr1, Ccr4, and Ccr5 expressed by multiple immune cell types, including NK cells, T cells, and monocytes (Hughes and Nibbs, 2018). Chemokine receptors are G-protein coupled receptors that upon binding their cognate chemokines undergo conformational changes, resulting in internalization of the chemokine/receptor complex and activation of intracellular signaling pathways that initiate, for example, chemotaxis or cell adhesion (Griffith et al., 2014; Hughes and Nibbs, 2018; Shao et al., 2022).

A challenge to understanding the coordination of immune responses against CMV and other pathogens is the difficulty in identifying chemokine-producing and -responding cells. Here, we developed a novel class of genetically modified mice that allows for identifying chemokine-producing and -responsive cells. Focusing on Ccl3 and mCMV infection, we provide new insight into the role of this chemokine by showing that it is first produced by NK cells rather than by macrophages and is also sensed by NK cells in an auto/paracrine manner. In addition, we show that IFN-I promotes Ccl3 production on the posttranscriptional level and suggest that it may also regulate sensing of this chemokine.

To investigate Ccl3-dependent immune responses, we generated a dual reporter knock-in mouse strain named CCL3-EASER (ErAse, SEnd, Receive) that allows the identification of both Ccl3-producing and -receiving cells, as well as conditional depletion of Ccl3-producing cells. To this end, the endogenous Ccl3 gene locus was modified to drive expression of (1) a fusion protein of native Ccl3 and Venus fluorescent protein, (2) tdTomato fluorescent protein, and (3) the human diphtheria toxin receptor (DTR), all expressed as individual proteins by virtue of P2A sequences (Fig. 1 A).

We validated CCL3-EASER mice using the well-studied mCMV infection model, given that the early immune response against this virus is mediated by and depends on Ccl3 (Cook et al., 1995; Salazar-Mather et al., 1998). Homozygous CCL3-EASER and C57BL6/J (BL6) wild type (WT) mice showed similar susceptibility to mCMV infection (Fig. S1 A), presenting a 100% survival rate (Fig. S1 B), no significant body weight alteration (Fig. S1 C), and controlling infection by day 14 (Fig. S1, D and E), exactly as BL6 mice did, indicating that the transgene did not compromise Ccl3 functionality.

Next, we demonstrated the functionality of the DTR transgene by administering diphtheria toxin (DT) on two consecutive days (Fig. 1 B), which effectively depleted Ccl3-producing cells, as revealed by a 98–99% reduction of tdTomato-expressing cells (Fig. 1, C and D). In the context of mCMV, depleted mice showed about 1,000-fold higher mCMV load in the liver compared with DT-treated BL6 WT or untreated CCL3-EASER mice on days 4 and 7 after infection (Fig. 1 E). Their livers contained more mCMV infection foci (Fig. 1 F) as well as increased immune infiltration (Fig. 1 G) compared with mice harboring normal Ccl3⁺ cell numbers. Depleted CCL3-EASER mice were unable to control the infection and succumbed about a week after mCMV infection (Fig. S1 B). These results were consistent with the known higher susceptibility of Ccl3^−/− mice to the virus (Cook et al., 1995; Salazar-Mather et al., 1998). Thus, CCL3-EASER mice produce functional Ccl3 and allow investigating Ccl3-dependent immune responses during viral infections such as mCMV.

Having demonstrated the importance of Ccl3-producing cells during early mCMV infection, we next examined the tdTomato reporter of CCL3-EASER mice to identify cells that actively transcribe this chemokine. When we analyzed ex vivo isolated cells from non-infected CCL3-EASER mice, the majority of the cells with active Ccl3 transcription (>90%) were NK cells in the liver (Fig. 2, A and B), spleen, and lung (Fig. 2 B and Fig. S2, A and B). In addition, smaller Ccl3⁺ subsets, including NKT cells, CD8 T cells, and monocytes, were detected in all analyzed organs (Fig. 2 B and Fig. S2, A and B).

When CCL3-EASER mice were infected with mCMV, more than 90% of the cells transcribing Ccl3 at the early time points after infection were still NK cells (Fig. 2, C and D; and Fig. S2, C and D). Only at 48 h after infection, cells other than NK cells slightly increased their contribution in the spleen, resulting in a reduction in the proportion of NK cells contributing to global Ccl3 expression to about 65% (Fig. 2 D and Fig. S2 C). These findings implied that early Ccl3-dependent NK cell recruitment may be strongly influenced by NK cells themselves rather than by macrophages, as commonly believed (Salazar-Mather et al., 1998, 2000, 2002).

The finding that NK cells are the main cell population that actively transcribe Ccl3 suggests that DT application in CCL3-EASER mice should primarily deplete these cells. We therefore enumerated immune cell subsets after DT injection (Fig. 3 A). Indeed, 90–95% of the NK cells were lost from liver, spleen, and lung compared with PBS-treated littermates (Fig. 3 B). However, various other cell populations were depleted at higher proportions (Fig. 3 B) than suggested by their Ccl3-tdTomato expression (Fig. 2, A and B; and Fig. S2, A and B). This may be explained by the observation made in previous studies showing that a single DT molecule is sufficient to kill a DTR-expressing cell (Yamaizumi et al., 1978; Saito et al., 2001). Low-level background transcription of Ccl3 may drive expression of a few DTR molecules, sufficient to render cells killable, but not enough to produce sufficient tdTomato molecules for detection by flow cytometry. In support of this, direct comparison of tdTomato expression in different immune cells showed that the tdTomato signal of NK cells was more than one magnitude higher than, for example, that of B cells (Fig. 2 A). Thus, the depletion of cells that lack detectable tdTomato expression does not argue against the faithfulness of the tdTomato reporter because DTR expression reaches functionally relevant levels when Ccl3 production is still negligible. Nonetheless, to test for the necessity of other Ccl3⁺ cells in mCMV infection, we adoptively transferred WT NK cells into DT-treated CCL3-EASER mice (Fig. 3 C), thereby partially reconstituting them (Fig. 3 D). This significantly improved protection against mCMV infection (Fig. 3 E), verifying the critical role of Ccl3-producing NK cells in the defense against mCMV infection.

We next examined the Ccl3-Venus reporter signal as a readout for Ccl3 translation. We showed above that NK cells from naive EASER mice actively transcribe Ccl3 (Fig. 2, A and B). However, they expressed no Venus reporter (data not shown), suggesting a lack of translation. This was confirmed by quantitative RT-PCR (RT-qPCR) and ELISA analyses of NK cells from non-transgenic BL6 mice (Fig. S2, E and F). Therefore, we applied a standard intracellular cytokine staining protocol to measure Ccl3-Venus, where isolated cells are stimulated in the presence of monensin/brefeldin A, to prevent the rapid secretion of Ccl3-Venus (Fig. 4 A). During mCMV infection, Ccl3-Venus indeed became clearly detectable and increased with time (Fig. 4 B and Fig. S3, A and B). The frequency of tdTomato⁺ NK cells remained stable during the course of mCMV infection (Fig. 4 C), but the intensity of Ccl3 transcription (tdTomato mean fluorescence intensity [MFI]) increased at later time points in all organs examined (Fig. 4 D). Venus⁺ cells became more abundant at later time points in liver and lung, but not in spleen (Fig. 4 E). The amount of Ccl3 protein (Venus MFI) started to increase 8 h after infection in liver and lung, but not in the spleen (Fig. 4 F). These findings showed organ-specific regulation of Ccl3, suggesting that Ccl3 was posttranscriptionally regulated during mCMV infection.

We next analyzed such posttranscriptional regulation by correlating Venus and tdTomato signals at the single-cell level at different time points after infection. At all time points, a statistically significant linear correlation between tdTomato and Ccl3-Venus expression in NK cells was found in liver (Fig. 4 G), spleen, and lung (Fig. S3, C and D). The slope of the regression lines indicated how much Ccl3 protein had accumulated within the NK cells per Ccl3 transcription reporter and thus represented an approximate parameter for the translational activity. This slope increased with time in the liver and lung but stayed unaltered in the spleen (Fig. 4, G and H), indicating that Ccl3 was more actively translated in these organs at later time points of infection (8 and 16 h). Also, comparison of the linear regression at equal transcription revealed different levels of Ccl3 translation in liver and lung, but not spleen (Fig. 4 H). These results indicated that other factors beyond promoter activity regulated Ccl3-Venus protein expression, suggesting that Ccl3 production is posttranscriptionally fine-tuned at sites of infection. Similar results were found for monocytes (Fig. S3, E–I), demonstrating that the uncoupling of Ccl3 transcription–translation was not specific for NK cells. Viral transcripts in the different organs were similar during these time points, arguing against a direct effect of the viral infection severity on this mechanism (Fig. S3, J and K). Taken together, CCL3-EASER mice revealed a posttranscriptional uncoupling between Ccl3 transcription and translation.

We next investigated the mechanism underlying this infection-induced uncoupling of Ccl3 transcription and translation. To clarify whether direct recognition of mCMV-infected cells by NK cells was required, we infected CCL3-EASER mice with either mCMV-WT or with mCMV-∆m157 lacking the viral glycoprotein m157, which activates NK cells via Ly49H binding (Fig. 5 A and Fig. S4 A) (Smith et al., 2002). Neither the percentage nor the intensity of tdTomato- and Venus-expressing NK cells were altered in the livers of mCMV-∆m157–infected compared with those from mCMV-WT–infected mice (Fig. 5, B and C, respectively, and Fig. S4, B and C). Thus, direct NK cell contact with infected cells was not required to upregulate Ccl3 promotor activity or Ccl3 protein expression.

We next investigated whether soluble factors induce Ccl3 translation and focused on IFN-I, which are central immune activators against many viruses including mCMV (Salazar-Mather et al., 2002; Holzki et al., 2015; Farrell et al., 2016). Antibody-mediated IFNAR-1 blockade (Fig. 5 D) significantly decreased Ccl3 promotor activity (Fig. 5 E) and protein expression (Fig. 5 F) after mCMV infection. Although IFNAR-1 blockade downregulated both processes, linear regression analysis in individual mice between tdTomato and Ccl3-Venus expression showed that protein expression was further downregulated in cells with similar promotor activity levels (Fig. 5, G and H). Consistent with these findings, exogenous addition of IFNβ to naive CCL3-EASER NK cells in vitro preferentially increased Ccl3 protein expression (Fig. S4 D), supporting the finding that IFN-I increased Ccl3 translation in vivo (Salazar-Mather et al., 2002).

We next asked whether infection-induced Ccl3 expression was confined to distinct subsets of NK cells in the liver before or 16 h after mCMV infection. Most cells expressing Ccl3 protein remained mature (CD11b⁺CD27⁻) or intermediate (CD11b⁺CD27⁺), with the latter becoming more predominant during infection (Fig. 6 A). No significant differences were seen in NK cells classified based on expression of Ly49D, Ly49H, or KLRG1 activating receptors (Fig. 6, B and C). Among the effector molecules, while 60% of naive NK cells produced IFNγ, only a few were granzyme B⁺. Yet, during mCMV infection, the proportion of cytotoxic (granzyme B⁺) NK cells increased to ∼60% (Fig. 6 D). Thus, Ccl3 expression and regulation did not show major differences between NK cell subsets.

Besides identifying the cellular source of Ccl3, identifying Ccl3-responding cells is necessary to understand the role of this chemokine during infection. CCL3-EASER mice were designed to also answer this question since the tdTomato reporter is retained intracellularly and only Ccl3-Venus is secreted. Therefore, Ccl3-producing cells are tdTomato⁺Venus⁺, whereas tdTomato^–Venus⁺ cells are those that do not produce Ccl3 but must have internalized it. To validate that CCL3-EASER mice allow identifying Ccl3-recipient cells, we first cocultured CCL3-EASER NK cells as Ccl3-Venus producers together with different non-reporter cell types as potential Ccl3-receivers and stimulated Ccl3 production with activating anti-NK1.1 antibodies (Fig. 7 A). We performed cocultures with non-reporter NK cells, NKT cells, activated T cells, and thioglycolate-elicited macrophages as Ccl3 receivers, because these cells can express Ccr1, Ccr4, or Ccr5 (Hughes and Nibbs, 2018). Indeed, we detected uptake of Ccl3-Venus protein by NK and NKT cells (Fig. 7 B and Fig. S5 A), but not by activated OVA-specific transgenic CD4⁺ OT-II cells (Fig. 7 C and Fig. S5 B) nor CD8⁺ OT-I T cells (Fig. 7 D and Fig. S5 C). Notably, NK and NKT cells only took up Ccl3 when they were isolated from infected mice (Fig. S5 E), suggesting that responsiveness to Ccl3 needs to be activated, for example, by IFN-I or other infection-induced cytokines. By contrast, macrophages showed both Venus and tdTomato signals (Fig. 7 E and Fig. S5 D), suggesting engulfment of CCL3-EASER NK cells rather than Ccl3-Venus protein uptake. We confirmed this interpretation by coculturing macrophages in a non-cell permeable transwell to prevent them from interacting with fluorescent cells and observed that they remained tdTomato⁻Venus⁻ (Fig. S5 F).

Next, we wished to clarify whether the acquisition of Venus in CD45.1 NK cells was a functional readout of Ccl3 signaling and not simply an endocytosis event. Therefore, we cocultured stimulated EASER-NK cells with non-reporter CD45.1 NK cells separated by a cell-permeable transwell insert. After 6 h of incubation, we observed that cells in the bottom of the transwell were enriched with CD45.1⁺Venus⁺ NK cells (Fig. 7 F). This indicated that Venus uptake correlated with NK cell migration, a typical chemokine response. In summary, CCL3-EASER mice are a suitable tool for identifying cells that sense Ccl3.

We finally used CCL3-EASER mice to identify cells that sense Ccl3 during mCMV infection in vivo. NK cells are known to be recruited by this chemokine (Bernardini et al., 2008; Bekiaris and Lane, 2010), but since these cells are also those that produce Ccl3, we had to create a situation where endogenously produced Ccl3-Venus cannot eclipse endocytosed Ccl3-Venus produced by other cells. To this end, we generated mixed bone marrow (BM) chimeras in which CD45.2 mice were reconstituted with a 60:40 mixture of BM from CD45.1/2 CCL3-EASER and CD45.1/1 non-reporter mice ([EASER:WT→WT] mBMx). This allowed for discriminating CD45.1/2 Ccl3-producing cells from CD45.1/1 Ccl3-recipient cells (Fig. 8 A panel 1).

Intravital imaging of infected livers in (EASER:WT→WT) mBMx mice identified tdTomato single positive-producing cells actively migrating along the sinusoids, whereas double positive Ccl3 protein-producing cells appeared arrested in the tissue (Fig. 8 B and Videos 1 and 2). Also, Ccl3-Venus⁺tdTomato⁻ recipient cells were arrested, suggesting homing or activation of Ccl3-responding cells.

To further identify cells receiving Ccl3-Venus cues, we analyzed the Venus signal in CD45.1/1 recipient cells by flow cytometry in peripheral blood, spleen, and liver (Fig. 8 C). To ensure that we detect a genuine Venus signal in non-reporter cells, we utilized as autofluorescence controls mCMV-infected BM chimeras, reconstituted only with BM from CD45.1/1 non-reporter mice ([WT→WT] BMx, Fig. 8 A panel 2). Using this approach, we noted that 3.19 and 3.48% of the NK cells in liver and spleen, respectively, but none in the blood, were Venus⁺ tdTomato⁻, whereas neglectable proportions of all other investigated cell types showed this phenotype (Fig. 8 C), indicating that NK cells are the main Ccl3 receivers during mCMV infection in vivo. Consistent with the retention of recipient cells in the infected livers observed by intravital microscopy (Fig. 8 A), the absence of Ccl3-Venus in peripheral blood leukocytes (Fig. 8 C) suggested that cells taking up Ccl3 may become arrested at the site of secretion rather than migrating between organs.

In summary, CCL3-EASER mice are a suitable tool to identify both Ccl3-producing and -receiving cells. Using this model, we demonstrate that NK cells are the main producers and direct responders to this chemokine during mCMV infection.

Chemokines facilitate immune cell encounters, which are critical for the defense against pathogens and tumors. While chemokine production and secretion can be directly measured, e.g., by flow cytometry, ELISA, or western blot, no methods exist to identify cells that sense chemokines. Single-cell transcriptomics and proteomics allow interactome analyses on the basis of the expression of receptor-ligand pairs, but this does not prove that an individual cell has, in fact, sensed a chemokine. Various mathematical algorithms consider downstream transcriptional events, but these are often not specific for a distinct chemokine, nor do they take chemokine doses or cell positioning in vivo into account (Brohée and van Helden, 2006; Plewczyński and Ginalski, 2009; Armingol et al., 2021; Malec et al., 2022). Furthermore, such -omics approaches often require the destruction of the cell for analysis, and reliable tools for validation are lacking. Here, we present the EASER mouse model as a novel approach to determine production and sensing of a given chemokine in vivo and in vitro, allowing for further downstream applications. Furthermore, this approach permits simultaneous evaluation of transcriptional and translational regulation of chemokine production, which can help in understanding mechanisms of chemokine induction and secretion.

We applied this approach to Ccl3, a chemokine critically important for the NK cell–dependent defense against viral infections, and obtained novel insights into the role of this chemokine during mCMV infection, a widely studied model of human CMV infection. It is presently assumed that an important source of Ccl3 is tissue-resident macrophages (Wang et al., 2000; Menten et al., 2002). However, most previous studies had analyzed Ccl3 later than 48 h after infection and utilized methods like ELISA from total tissue homogenates, so that the producing cells were not directly identified. One previous study noted that also NK cells secrete Ccl3 in mCMV infection (Salazar-Mather et al., 1998). CCL3-EASER mice revealed that these cells are in fact the first and primary source of Ccl3 during the very early response after infection. Thus, NK cells perform a sentinel role, implying that they have been previously recruited to sites of infection like the liver, perhaps during previous viral infections.

We noted that NK cells produced Ccl3 transcripts also in uninfected mice, whereas translation occurred mainly after viral infection. Thus, NK cells are prepared to readily produce and secrete Ccl3 during infections, consistent with a sentinel role. We calculated a linear regression line between the channel intensities of the reporters for Ccl3 transcription and translation and found that its slope was reduced when IFN-I signaling was blocked. IFN-I is known to cause Ccl3 production (Bug et al., 1998; Zang et al., 2001; Salazar-Mather et al., 2002), and our study reveals that it does so primarily on the posttranscriptional level. Furthermore, we noted that the slope of the regression curve increased with time in organs of viral replication. This may indicate that local IFN-I stimulates Ccl3 production to attract more NK cells. However, it has to be kept in mind that Ccl3 transcripts and their tdTomato reporters are physically separated and may possess different half-lives within NK cells. Thus, it is possible that after resolving a viral response, the reporter might persist longer than the transcripts, thereby lowering the slope of the regression curve. In this case, our system would overestimate transcription and underestimate translation of Ccl3.

The other main novelty of EASER mice is their ability to report for cells that receive chemokine cues by showing uptake of fused Venus-Ccl3 protein. In the case of Ccl3, measuring such uptake is not straightforward because the cells producing and responding to this particular chemokine are the same, i.e., NK cells, which was not known when we generated these mice. EASER mice principally do not allow discriminating whether a cell that produces a chemokine also senses it. We overcame this challenge by generating mixed BM chimeras in which 40% of the NK cells do not produce Ccl3, so their chemokine uptake can be detected. This confirmed that NK cells are the main Ccl3 addressees during early mCMV infection. The dual role of producer and recipient implies that Ccl3 facilitates auto/paracrine NK cell communication. This promoted not only NK recruitment to the site of infection but may also facilitate a positive feedback loop early after mCMV infection by which NK cells can quickly amplify their response. At later time points, other cells, including monocytes, contributed to Ccl3 production, consistent with previous findings (Hokeness et al., 2005). Interestingly, we noted that NK cells mainly internalized Ccl3 when they were derived from virus-infected but not from uninfected mice. Thus, Ccl3-mediated NK cell recruitment in viral infections is regulated at least at two levels: by facilitating translation of preexisting Ccl3 transcripts and also by rendering NK cells responsive to secreted Ccl3.

The EASER approach is principally applicable also to other chemokines. For example, Ccl19 is produced by stroma cells and sensed via Ccr7 by mature dendritic cells and naive T cells. Thus, the generation of CCL19-EASER mice might improve understanding of T cell priming. Notably, many chemokines resemble Ccl3 in being produced and sensed by the same cell type, suggesting that the autocrine feedback identified here for Ccl3 loop may be a general principle in chemokine networks. For instance, the chemokine Ccl2 attracts inflammatory monocytes and is also produced by these cells. If CCL2-EASER mice were generated, mixed BM chimeras might be needed to identify Ccl2-responsive cells. In principle, also cytokines other than chemokines or even hormones can be investigated by the EASER approach, as long as they are internalized by sensing cells.

Mouse strains utilized: C57BL6/J (RRID:IMSR_JAX:000664, referred here as to BL6), B6.SJL-Ptprc^a Pepc^b/BoyJ (RRID:IMSR_JAX:002014, referred here as to CD45.1/1), B6.SJL-Ptprc^b/a Pepc^b/BoyJ (CD45.1/1 crossed with C57BL6/J, referred here as to CD45.1/2), C57BL/6-Tg(TcraTcrb)1100Mjb/J (RRID:IMSR_JAX:003831, crossed with CD45.1/1, referred here as to OT-I.CD45.1), and B6.Cg-Tg(TcraTcrb)425Cbn/J (RRID:IMSR_JAX:004194, crossed with CD45.1/1, referred here as to OT-II.CD45.1). The CCL3-EASER transgenic mice were generated in cooperation with Ozgene Pty Ltd. A fusion construct composed of the DNA sequences for the linker (G-G-S-G)x2, the Venus fluorescent protein (Versus) from Aequorea victoria, P2A from porcine teschovirus, the tdTomato fluorescent protein from Discosoma striata, the human DTR, and the neomycin resistance cassette from Escherichia coli K12 Bos taurus were generated and inserted in the pMB1 vector. The final construct was electroporated into BL6/J embryonic stem (ES) cells. Neomycin-resistant ES cells were injected into goGermline blastocysts (Koentgen et al., 2016). GoGermline chimeras were implanted into BL6/J surrogate mothers. The generated mice were then crossed with Oz-Flp mice for deletion of the neomycin resistance cassette. The resulting transgenic mouse line was designated CCL3-EASER (BL6-Tg[Ccl3-Venus-2A-tdTomato-2A-HBEGF]/J). CCL3-EASER mice were CD45.2 homozygous. BL6.Cg-Ptprc^a Ccl3^{Venus-2A-tdTomato-2A-HBEGF}/J and BL6.Cg-Ptprc^a/b Ccl3^{Venus-2A-tdTomato-2A-HBEGF}/J (referred here as to CCL3-EASER-CD45.1/1 and CD45.1/2, respectively) were generated by crossing JAX stock #002014 with CCL3-EASER mice. Genotyping was carried out by PCR from genomic DNA of ear biopsies using the following primers: 5′-AGA​TGG​GGG​TTG​AGG​AAC​GTG​T-3′ and 5′-ATC​TGT​CTG​TCT​GCT​GGT​CAT​CG-3′. For targeted-ablation of CCL3-producing cells, 25 ng/g/body weight (BW) of DT from Corynebacterium (#D0564; Sigma-Aldrich) was applied intraperitoneally (i.p.) on two consecutive days.

All mice were housed under specific pathogen–free conditions at the animal facilities of the University Hospital Bonn. Mice were used at 10–12 wk of age and randomized in groups for each experiment. Sex- and age-matched BL6 were bred in the same facility and used as control mice. All animal experiments were approved by the corresponding government authority (Landesamt für Natur, Umwelt und Verbraucherschutz from North-Rhein Westphalia).

Mixed BM chimera mice were generated as previously described (Hochweller et al., 2009) by transferring 2 × 10⁶ donor BM cells into 9 Gy–irradiated BL6 recipient mice. Donor BM consisted of a 60:40 mixture of CCL3-EASER-CD45.1/2 and CD45.1 mice. CD45.1 and CD45.2 markers enabled for discrimination of cells from CCL3-EASER-CD45.1/2, CD45.1, and BL6 (CD45.2) recipient mice. Experiments were performed 10 wk after BM transfer when reconstitution of the NK cell compartment reached 90–95%.

mCMV.WT (strain Smith, VR1399; ATTC) or mCMV.∆m157 (Bubić et al., 2004) were cell culture propagated and purified as previously described (Podlech et al., 2002). Mice were infected by a single intravenous (i.v.) inoculation of 5 × 10⁵ PFU of either strain in 100 µl of endotoxin-free PBS.

IFNAR-1 was blocked by injecting mice with 200 μg of anti-IFNAR-1 (clone MAR1-5A3, #BE0241, RRID:AB_2687723; Bioxcell) monoclonal antibody i.p. 1 h before mCMV infection. Control mice were treated equally but with irrelevant isotype control (clone MOPC-21, #BE0083, RRID:AB_1107784; Bioxcell).

Single-cell suspensions were prepared from liver, spleen, lung, and kidney as previously described (Tittel et al., 2012). In general, organs were digested with collagenase IV from Clostridium histolyticum (#SCR103; Sigma-Aldrich) and DNase I from bovine pancreas (#DN-25; Sigma-Aldrich). Digested organs were pressed and subsequently filtered through 70–100-µm cell strainers. Liver and kidney interstitial cells were discarded by differential centrifugation. Liver-associated leukocytes were further enriched with a Percoll density gradient centrifugation (#P1644; Sigma-Aldrich). Samples were depleted of erythrocytes via incubation with ACK lysis buffer (#A1049201; Thermo Fisher Scientific), washed, and kept on ice until usage.

Four days prior to CD4⁺ and CD8⁺ T cell purification, OT-II.CD45.1 and OT-I.CD45.1 mice were immunized i.v. with 200 µl of endotoxin-free PBS containing 240 µg/mouse OVA protein (#A5503; Sigma-Aldrich) and 50 µg/mouse of CpG-ODN 1668 (#IAX-200-C100; Biomol).

EASER-NK cells were purified by magnetic cell sorting (MACS) following the manufacturer’s instructions (#130-115-818; Miltenyi). After magnetic separation, enriched NK cells were washed with complete medium (RPMI GlutaMax supplemented with 10% hi-FBS, 1% non-essential amino acids, 1% pyruvate, 1% Pen/Strep, and 0.05 mM β-mercaptoethanol–all from Thermo Fisher Scientific).

CD4⁺ and CD8⁺ T cells were isolated from total splenocytes via positive selection by MACS following the manufacturer’s instructions (#130-117-043, #130-177-044, respectively; Miltenyi). Thioglycolate-elicited peritoneal macrophages were isolated as previously described (Pavlou et al., 2017). Non-reporter NK and NKT cells were enriched from total liver lymphocytes after labeling with biotinylated-NK1.1 antibody (clone PK136 #13-5941-82, RRID:AB_466804; ThemoFisher Scientific) and anti-biotin MicroBeads (#130-090-485; Miltenyi), followed by magnetic separation. Population purity was determined by flow cytometry using NK1.1 (clone PK136 #551114, RRID:AB_394052) from BD Pharmigen, CD3 (clone 17A2 #100229, RRID:AB_11204249), CD11b (clone M1/70 #101256, RRID:AB_2563648), and F4/80 (clone BM8 #123128, RRID:AB_893484) antibodies from Biolegend, and CD4 (clone GK5.1 #612761, RRID: AB_2870092) and CD8a (clone 53-6.7 #612898, RRID:AB_2870186) from BD Horizon.

NK cells were expanded for 1 wk in 96-well round bottom plates in complete medium supplemented with 1.700 U/ml of mouse recombinant IL-2 (#212-12; PeproTech). Medium was refreshed every 3 days. If necessary, on day 3 cells were split 1:2.

Cytokine-expanded NK cells were coincubated 1:1 with either non-reporter NK/NKT cells, activated CD4⁺, activated CD8⁺ T cells, or thioglycolate-elicited macrophages cells in 96-well round bottom plates precoated overnight with 10 µg/ml of purified anti-NK1.1 antibody (clone PK136 #MA1-70100, RRID:AB_2296673; Thermo Fisher Scientific). Cocultures were incubated for 4 h at 37°C 5% CO₂. Negative controls were performed following the same experimental procedure but in a medium containing GolgiPlug/GolgiStop (#555029 and #554724; BD Biosciences, respectively). Ccl3-Venus secretion control corresponded to single culture of expanded EASER-NK cells stimulated with 10 ng / 0.5 µg/ml of PMA/ionomycin (#P1585 and #I3909; Sigma-Aldrich, respectively) or plate-bound NK1.1, both in the presence of GolgiPlug/GolgiStop.

CCL3-EASER NK cells were coincubated with thioglycolate-elicited macrophages separated by a non-cell-permeable transwell insert (#862640; Greiner bio-one) in 24-well flat-bottom plates precoated overnight with 10 µg/ml of purified anti-NK1.1 antibody. Cocultures were incubated for 4 h at 37°C 5% CO₂. Macrophage autofluorescence control corresponded to a single culture of these cells under the same experimental conditions. Ccl3-Venus secretion control corresponded to single culture of EASER-NK cells stimulated plate-bound NK1.1 in the presence of GolgiPlug/GolgiStop.

Transmigration assay was performed likewise, coculturing EASER-NK cells with non-reporter NK cells separated by a cell-permeable transwell insert (#3421; Costar). Co-cultures were incubated for 6 h at 37°C 5% CO₂ in 24-well flat-bottom plates precoated overnight with 10 µg/ml of purified anti-NK1.1 antibody.

DT-depleted CCL3-EASER and BL6 mice were transferred with 1 × 10⁶ of CD45.1 NK cells in 200 µl of endotoxin-free PBS 1 h before mCMV infection. Recipient mice received sex-matched donor cells. In all cases, NK cells were isolated as previously described.

To determine mCMV organ titers, spleen, lungs, liver, and salivary glands were isolated at 4 or 7 days after infection, homogenized, and assessed by standard virus plaque assay on murine embryonic fibroblasts (Brizić et al., 2018). Viral titers were expressed as log₁₀ (PFU per organ or gram organ). Viral genome copies were quantitated by M55 (encoding gB)-specific qPCR normalized to cell number by pthrp-specific qPCR, as previously described (Lemmermann et al., 2010).

Liver and spleens were fixed in 10% neutral-buffered formalin and paraffin-embedded. Tissue sections (8 μm) were stained with hematoxylin and eosin using standard methods. Image analysis was performed using a digital Zeiss Axio Observer Z1 inverted microscope. Images were acquired as tile scans with 20× magnification and automatic stitching with 5% overlap using ZEN Blue software.

For each biopsy, 8-µm-thick serial sections were cut from frozen tissue blocks and mounted on acid-cleaned glass slides. Endogenous peroxidase activity was inhibited by incubation with 3% H₂O₂ in methanol. To reduce nonspecific background staining, slides were incubated with 5% goat serum. Finally, slides were incubated with anti-MCMV to IE1 primary antibody (Croma 101, dilution 1:500, kindly provided by S. Jonjic, University of Rijeka, Rijeka, Croatia) in a moist chamber overnight at 4 °C. After washing, sections were incubated with a donkey anti-goat-HRP secondary antibody (#A15999, RRID:AB_2534673 dilution 1:500; Thermo Fisher Scientific). Peroxidase activity was detected with Pierce Peroxidase detection kit (#36000; Thermo Fisher Scientific). Finally, sections were weakly counterstained with Mayer’s hematoxylin and mounted. Image analysis was performed using a digital Zeiss Axio Observer Z1 inverted microscope.

1–2 × 10⁶ cells were centrifuged at 400 g for 5 min at 4°C. Cells were resuspended in 100 μl PBS containing 3% FCS, 0.1% NaN₃, 5% Octagam (human poly Ig), and fluorochrome-conjugated antibodies directed against cell surface antigens for 30 min on ice in the dark. Cells were washed and acquired in an LSR Fortessa flow cytometer (BD Biosciences). Sample acquisition was stopped when an equal number of CaliBRITE beads (#340486, RRID:AB_2868734; BD Biosciences) had been acquired. Dead cells and doublets were excluded with ZombieUV (#423108; BioLegend) or DAPI (#D9542; Sigma-Aldrich) and standard procedures, respectively. Viable cells were analyzed with FlowJo software V. 10.9.1. Before analysis, technical outliers were excluded using the FlowAI. Specific cell populations were identified by conventional gating. In case of intracellular staining, cells were incubated 4 h in hi-FBS supplemented RPMI GlutaMAX with 10 ng/ml of PMA, 0.5 µg/ml of ionomycin, and GolgiPlug/GolgiStop (1:500/1:750, respectively diluted). After surface marker staining, cells were fixed with cytoFix/cytoPerm (#554714, RRID:AB_2869008; BD Biosciences), washed with 1xPermBuffer (#554714; BD Biosciences), and stained for intracellular antigens. The following antibodies were utilized for flow cytometry analysis: CD11c (clone N418 #117318, RRID:AB_493568), CD19 (clone 6D5 #115530 or #115541, RRID:AB_830707 and RRID:AB_11204087), CD27 (clone LG.3A10 #124241, RRID:AB_2800595), CXCR6 (clone SA051D1 #151111, RRID:AB_2721558), I-A/I-E (clone M5/114.15.2 #107622, RRID:AB_493727), IFNγ (clone XMG1.2 #505840, RRID:AB_2734493), KLRG1 (clone 2F1/KLRG1 #138421, RRID:AB_2563800), Granzyme B (clone QA18A28 #396413, RRID:AB_2810603), Ly49D (clone 4E5 #138306, RRID:AB_10574955), Ly49H (clone 3D10 #144714, RRID:AB_2783113), Ly6C (clone HK1.4 #128035, RRID:AB_2562353), Ly6G (clone 1A8 #127643, RRID:AB_2565971), NK1.1 (clone PK136 #108720, RRID:AB_2132713) from Biolegend; CD11b (clone M1/70 #563553, RRID:AB_2738276), CD45 (clone 30-F11 #564279, RRID:AB_2651134), and TCRγδ (clone GL3 #563532, RRID:AB_2661844) from BD Horizon; CD49a (clone H131/8 #741776, RRID:AB_2871130) from Invitrogen; and SiglecF (clone E50-2440 #740158, RRID:AB_2739911) from BD OptiBuild.

Intravital microscopy of the mouse liver was performed as previously described (Dal-Secco et al., 2015). In brief, mice were anesthetized by i.p. injection of 200 mg/kg ketamine and 10 mg/kg xylazine. After anesthesia, the tail vein was cannulated for maintenance of anesthetic and to permit the delivery of fluorescently labeled antibodies. A midline incision followed by a lateral incision along the costal margin to the midaxillary line was performed to expose the liver. Exposed abdominal tissues were covered with saline-soaked gauze to prevent dehydration. The mouse was placed in a right lateral position and the ligaments attaching the liver to the diaphragm and the stomach were cut allowing the liver to be externalized onto a glass coverslip on the inverted microscope stage with a heated chamber to maintain the mouse bodies’ temperature at 37°C. Images were acquired with a Zeiss LSM710 Observer and the corresponding ZEN 2.3 (black) every 30 s for 30 min. Ccl3-producing cells expressed tdTomato, and Ccl3 recipient cells appeared with Venus signal. Chimeric cells were identified with an i.v. administered APC-conjugated anti-mouse CD45.1 (clone A20, #110714, RRID:AB_313503, 2 μg; BioLegend).

The percentage of a given cell subset among Ccl3-producing cells was determined by dividing the number of tdTomato-positive cells of that subset by the number of all tdTomato-positive cells.

Unless otherwise stated, statistical analyses were performed using unpaired Mann–Whitney U-test; *P < 0.05; **P < 0.01; ***P < 0.001; and ****P < 0.0001. Data are shown as mean ± SD. GraphPad Prism 9.5.1 (GraphPad Software) was used for the calculations.

Provided online are five supplementary figures and two supplementary videos. Fig. S1 shows that DT-treated CCL3-EASER mice are more susceptible to mCMV infection (related to Fig. 1). Fig. S2 shows further details on Ccl3-tdTomato transcription. Fig. S3 shows further details on posttranscriptional fine-tuning of Ccl3 during mCMV infection. Fig. S4 shows data on the direct influence of IFN-I on Ccl3-protein expression in NK cells. Fig. S5 shows details on CCL3-NK cell coculture controls (related to Fig. 7). Videos 1 and 2 show intravital microscopy video from two independent CCL3-EASER infected livers (related to Fig. 8).

The data in the figures are available in the published article and in the online supplemental material.

We thank Daniela Klaus, Moritz Blankart, Alice Jacob, and Marigona Sutaj for their excellent technical assistance. We thank Ozgene Pty Ltd. for excellent cooperation in generating CCL3-EASER mice. We also thank the Flow Cytometry Core Facility of the Medical Faculty at the University of Bonn for providing support and instrumentation funded by the Deutsche Forschungsgemeinschaft (German Research Foundation)—project numbers: 216372401, 387335189, 387333827, and 216372545.

Work was supported by the Deutsche Forschungsgemeinschaft (German Research Foundation) grants SFBTRR237 project number 369799452; Excellence strategy EXC2151 project number 390873048; IRTG2168 project number 272482170; SFB1454 project number 432325352; SFBTRR259 project number 397484323; FOR5427 project number 466687329; and SFB1292/2 TP11 project number 318346496.

Author contributions: M.B. Rodrigo: Experimental Design, Investigation, Data Analysis, Visualization, and Writing. A. De Min: CD45.1 NK cell transfer experiment. S.K. Jorch: Intravital microscopy analysis. C. Martin-Higueras: Preliminary studies of Ccl3 deletion. A.-K. Baumgart: Provided critical reagents and techniques. B. Owczarek: Revision experiments. S. Becker: Revision experiments. N. Garbi and N.A. Lemmermann: Experimental Design, Funding, Supervision, Data Analysis, and Writing. C. Kurts: Conceptualization, Funding, Supervision, Data Analysis, and Writing.