Differential expression of interferon-lambda receptor 1 splice variants determines the magnitude of the antiviral response induced by interferon-lambda 3 in human immune cells.

Type III interferons (IFN-lambdas(λ)) are important cytokines that inhibit viruses and modulate immune responses by acting through a unique IFN-λR1/IL-10RB heterodimeric receptor. Until now, the primary antiviral function of IFN-λs has been proposed to be at anatomical barrier sites. Here, we examine the regulation of IFN-λR1 expression and measure the downstream effects of IFN-λ3 stimulation in primary human blood immune cells, compared with lung or liver epithelial cells. IFN-λ3 directly bound and upregulated IFN-stimulated gene (ISG) expression in freshly purified human B cells and CD8+ T cells, but not monocytes, neutrophils, natural killer cells, and CD4+ T cells. Despite similar IFNLR1 transcript levels in B cells and lung epithelial cells, lung epithelial cells bound more IFN-λ3, which resulted in a 50-fold greater ISG induction when compared to B cells. The reduced response of B cells could be explained by higher expression of the soluble variant of IFN-λR1 (sIFN-λR1), which significantly reduced ISG induction when added with IFN-λ3 to peripheral blood mononuclear cells or liver epithelial cells. T-cell receptor stimulation potently, and specifically, upregulated membrane-bound IFNLR1 expression in CD4+ T cells, leading to greater antiviral gene induction, and inhibition of human immunodeficiency virus type 1 infection. Collectively, our data demonstrate IFN-λ3 directly interacts with the human adaptive immune system, unlike what has been previously shown in published mouse models, and that type III IFNs could be potentially utilized to suppress both mucosal and blood-borne viral infections.


Introduction
Previously, we demonstrated that IFN-λ3 inhibited human B cell antibody production and decreased the Th2 response to an H1N1 influenza vaccine antigen [47], but it is not clear which human immune cells directly respond to IFN-λ3. Here, we investigate and quantify expression of IFN-λR on human immune cells and correlate these findings with ISGs induced by IFN-λ3. We demonstrate that human adaptive immune cells express both IFN-λR1 variants (mIFN-λR1, sIFN-λR1), where sIFN-λR1 inhibits ISG induction by IFN-λ3. In addition, we show mIFN-λR1 expression varies between cell subsets and can be upregulated by activation of immune cell receptors including the T-cell receptor (TCR), B-cell receptor (BCR) and Tolllike receptors (TLR). In purified activated CD4 + T cells, IFN-λ3 pretreatment leads to antiviral ISG induction and a significant decrease in HIV-1 infection. Taken together, these results show that unlike in mice, IFN-λ3 directly regulates the human adaptive immune system and may be exploited in the future to promote type 1 and antiviral responses and dampen type 2 immune responses.

IFN-λR1 is differentially expressed among peripheral human immune cell subsets
Recent type III IFN studies have focused on their antiviral or protective effects in epithelial cells at barrier sites, but consensus is lacking regarding which human immune cells express the IFN-λR. We first quantified transcript levels of both subunits of IFN-λR (IFNLR1 and IL10RB) in highly pure immune cell subsets from blood of healthy individuals, primary human hepatocytes, and normal human bronchial epithelial cells (NHBE). We found, as expected, ubiquitous IL10RB expression, with the highest levels found in monocytes and neutrophils ( Fig 1A). The highest expression of IFNLR1 transcripts were found in epithelial cell types and B cells, while CD4 + and CD8 + T cells expressed less IFNLR1 mRNA, and monocytes, natural killer cells, and neutrophils had little or barely detectable expression levels of IFNLR1 transcript (Fig 1A). CD8 + T cells had significantly greater expression of IFNLR1 compared to CD4 + T cells (P = 0.0021). This is the first report of differential expression of IFNLR1 in human CD4 + versus CD8 + T cells. Our results contrast those in mice where Ifnlr1 transcript expression is highest in neutrophils and little or no expression can be detected in all other immune cell types [20,25].

IFN-λ3 binds to both epithelial cells and specific immune cell subsets
To date, studies on IFN-λR1 biology have been limited due to a lack of sensitive reagents to measure receptor protein expression. We recently developed a flow cytometry binding assay [48], which measures IFN-λ3 binding to the cell surface as a surrogate of IFN-λR1 expression. We quantified the binding of 6 His-tagged IFN-λ3 to immune cell subsets within peripheral blood, as well as primary liver and lung epithelial cells. Our flow cytometry gating strategy for PBMCs and neutrophils is shown in S1 and S2 Figs. Multiple immune cell subsets bound IFN-λ3 in a dose-dependent manner, and the maximum binding percentage was dependent on the cell type ( Fig 1B). Almost 100% of hepatocytes and NHBE cells bound IFN-λ3 at the highest dose tested (Fig 1B), and the amount of IFN-λ3 bound (median PE value) was 4.4-11.7 fold greater than what bound to immune cell subsets (S3A Fig). The binding results were very reproducible within a single donor; the percentage of monocytes or B cells that bound IFN-λ3 did not vary substantially when the same assay was repeated at least six months later (S3B Fig). An unrelated 6 His-tagged protein (OBCAM) did not significantly bind any of the cell types examined (Fig 1C and S3C Fig). IFN-λ3 also did not significantly bind Huh7 IFNLR1 knockout cells, which were previously shown to be unresponsive to IFN-λ3 stimulation [49] (S3D

Fig 1. Differential IFN-λR expression and IFN-λ3 binding among primary human immune and epithelial cells. A) Normalized expression values
for each subunit of the IFN-λR as determined by RT-qPCR for normal human bronchial epithelial cells (NHBE), primary hepatocytes (hep) or immune cells purified from healthy human donor blood. Graph shows mean + SEM from 3-7 different donors for each population. Results were normalized to the geomean of HPRT1 and RPL13A reference genes. B-H) IFN-λ3 binding was quantified via flow cytometry as described in the Materials and Methods. B) Dose curves of adding 1, 2 or 5 μg/ml IFN-λ3 to epithelial cells or total human PBMCs. 0 μg/ml IFN-λ3 refers to adding the secondary antibody alone. Graph shows mean +/-SD for 3-7 different donors. C) Binding percentages as detected by flow cytometry for IFN-λ3 or a similarly His-tagged control protein (OBCAM) where means +/-SD are shown. D-H) Quantified IFN-λ3 (5 μg/ml) binding percentages to each cell type where each dot represents a different healthy individual and background binding of the secondary antibody alone was subtracted. In D),

PLOS PATHOGENS
Interferon-lambda antiviral activity in human immune cells Fig). Consistent with previous reports that pDCs respond to IFN-λ stimulation [41,43,44], IFN-λ3 bound to pDCs at high levels (Fig 1B and 1D). B cells, monocytes, pDCs, and mDCs all bound IFN-λ3, whereas little IFN-λ3 bound to total T cells, NK cells, and neutrophils (Fig 1B  and 1D). IFN-λ3 bound a significantly higher percentage of hepatocytes and NHBE cells compared with all immune cell subsets tested (Fig 1D, S1 Table). While there was no difference in IFN-λ3 binding between memory and naïve B cells, we observed significantly higher binding of IFN-λ3 to IgD + versus IgDmemory B cells (Fig 1E and 1F). CD8 + T cells also bound significantly more IFN-λ3 than CD4 + T cells (Fig 1G), consistent with the greater IFNLR1 transcripts measured in CD8 + T cells ( Fig 1A). Next, we examined whether the percentage of IFN-λ3 binding correlated between immune cell subsets within each donor. We found there was significant correlation between cell types, especially within the same immune cell lineage. For example, the levels of IFN-λ3 binding to CD4 + T cells positively correlated with the percentage bound to CD8 + T cells, and similarly, the binding of IFN-λ3 to pDCs positively correlated with the binding to mDCs (Fig 1H). Correlation results between immune cell subsets are shown in S2  (Fig 1A), significantly higher amounts of IFN-λ3 bound to NHBE cells than B cells (P<0.0001) (Fig 1B  and 1D and S3A Fig). Collectively, these results show that primary epithelial cells and specific human immune cell subsets bind IFN-λ3, but epithelial cells bind IFN-λ3 at a much higher level.

IFN-λ3 binding leads to ISG induction in both epithelial cells and specific immune cell subsets
Previous studies that examined IFN-λR expression and IFN-λ stimulation of human immune cells had conflicting results [35,39,44,45,50]. Since we observed dramatic differences in IFN-λ3 binding by immune cell subsets, we next determined whether the amount of IFN-λ3 bound by each subset mirrored the relative induction of ISGs upon IFN-λ3 stimulation. We compared IFN-λ3 responses in highly pure B cells, monocytes, CD4 + T cells, CD8 + T cells, or neutrophils freshly isolated from peripheral blood of healthy donors. Representative results of the purity of our isolated cell subsets are shown in S2 and S5 Figs. For all cell types except neutrophils, we added recombinant IFN-λ3 overnight to induce ISG expression. Neutrophils were treated with IFN-λ3 for 5 hours since the majority of neutrophils die during overnight culture in vitro [51][52][53]. Among immune cells, the greatest ISG response to IFN-λ3 was seen in B cells, consistent with their expression of IFNLR1 and IFN-λ3 binding potential (Fig 2A). Low levels of ISGs were induced by IFN-λ3 in CD4 + T cells with the highest induction of OAS1 (mean 2.9 fold upregulation in IFN-λ3 treated versus unstimulated control), whereas all ISGs tested were induced by IFN-λ3 in CD8 + T cells (mean 4.8 to 8.2 fold upregulation in IFN-λ3 treated versus unstimulated control). Baseline expression of all 3 ISGs was not statistically different between B cells, CD4 + T cells and CD8 + T cells, therefore the higher response in B cells was not due to lower ISG expression prior to IFN-λ3 stimulation (S6A Fig). In contrast, neutrophils had significantly greater baseline IFIT1 and ISG15 expression compared to B and T cells, in statistical significance results were identical when comparing either NHBE or hep to all other immune cell subsets. All comparisons are shown in S1

PLOS PATHOGENS
Interferon-lambda antiviral activity in human immune cells agreement with published microarray and single cell RNA sequencing data [54,55] (S6A Fig). OAS1 expression was not significantly different between B cells and neutrophils or monocytes. Overall, our results indicate both B and T cells can directly respond to IFN-λ3, but with different magnitudes. Both monocytes and neutrophils failed to upregulate ISGs in response to IFN-λ3 (Fig 2A), demonstrating that the IFN-λ3 we detected bound to monocytes in our binding assay does not induce ISG induction, at least under the conditions tested. All immune cell types tested responded to our positive control IFN-α2 (S6B Fig).
Since we observed greater IFN-λ3 binding to NHBE cells compared to B cells despite higher relative IFNLR1 transcript expression in B cells, we next quantified ISG induction by IFN-λ3 in NHBE cells. When NHBE cells were treated with IFN-λ3 overnight, high levels of ISGs were induced (Fig 2B). At this time point, ISG induction by IFN-λ3 was significantly higher than the levels we observed with IFN-α2 treatment for most genes tested (Fig 2B). There was up to 50-fold higher ISG induction by IFN-λ3 in NHBE cells compared to B cells (Fig 2A and 2B).

PLOS PATHOGENS
Interferon-lambda antiviral activity in human immune cells cells bound dramatically more IFN-λ3 at the cell surface compared to DG75 B cells with 7-12 fold higher median fluorescent intensities detected ( Fig 3B). This increased IFN-λ3 cell surface binding translated to 10-11 fold greater ISG induction in BEAS-2B cells compared to DG75 B cells in response to overnight IFN-λ3 treatment ( Fig 3C). Our results demonstrated that our IFN-λ3 binding assay is a useful tool to predict IFN-λ3 responsiveness. While previous studies have solely quantified IFNLR1 mRNA and because a sensitive antibody is not commercially available, one must be cautious that total IFNLR1 transcript levels do not necessarily correlate with IFN-λ3 induced ISGs.

Soluble IFN-λR1 directly binds cells to increase IFN-λ3 binding to the surface but inhibits ISG induction
To further examine the discrepancy between high and similar levels of IFNLR1 transcript seen in both B cells and lung epithelial cells, but low B cell ISG expression in response to IFN-λ3, we determined if sIFN-λR1 plays a role in controlling ISG responses. This variant lacks a transmembrane domain and was detectable in cell line supernatants when overexpressed [35]. Little is known about the regulatory role of sIFN-λR1 in IFN-λ biology except that pre-incubating 100-1000 fold excess sIFN-λR1 with IFN-λ1 inhibited MHC class I upregulation on the

PLOS PATHOGENS
Interferon-lambda antiviral activity in human immune cells HepG2 hepatocyte cell line [35]. We used our previously designed RT-qPCR assay to specifically quantify either the full length membrane form (mIFNLR1) or sIFNLR1 [48]. Using this assay, we found all immune cells tested had higher levels of sIFNLR1 relative to mIFNLR1, when compared to epithelial cells. The membrane/soluble receptor transcript ratio was 0.6-2 in immune cells, whereas in lung or liver epithelial cells it was 9.4-17.6 ( Fig 4A). B cells had similar mIFNLR1 mRNA expression, but higher sIFNLR1 levels compared to epithelial cells. CD8 + T cells expressed the next highest transcript level of mIFNLR1 at 3-fold higher levels than CD4 + T cells. Monocytes and NK cells expressed very little or no IFNLR1 mRNA of either variant. Neutrophils were the only cell type tested where there was higher sIFNLR1 than mIFNLR1 expression (Fig 4A). In agreement with our data with primary T and B cells, we had previously found Jurkat T cell and Raji B cell lines also had low membrane/soluble IFNLR1 ratios of~2-3 [48]. We extended our cell line results to demonstrate that the lung BEAS-2B and A549 epithelial cell lines had a much higher membrane/soluble IFNLR1 transcript ratio of 14.9 and 6.7 respectively, while DG75 B cells had a lower ratio of 4.0 ( Fig 4A).
Our results indicate that the varied expression of sIFNLR1 may relate to extent of the ISG response induced by IFN-λ3.
To examine if sIFN-λR1 regulates IFN-λ3-mediated ISG responses, we added recombinant sIFN-λR1 protein alongside IFN-λ3 to PBMCs and then quantified ISG induction. The addition of sIFN-λR1 dramatically inhibited ISG induction by an average of 54-78% ( Fig 4B). This inhibition of IFN-λ3 activity was not unique to immune cells because sIFN-λR1 similarly inhibited IFN-λ3-mediated ISG induction in the Huh7.5 hepatocyte cell line ( Fig 4C). Since soluble cytokine receptors can act to prevent cytokine binding to the cell surface, or can act directly at the cell surface interacting with co-receptors [56], we first determined if sIFN-λR1 could bind the cell surface in the absence or presence of IFN-λ3 cytokine. Surprisingly, sIFN-λR1 bound both primary immune cells within PBMCs (Fig 4D), and epithelial cell lines ( Fig  4E) in the absence of IFN-λ3. This binding was not due to the Fc tag on recombinant IFN-λR1, since excess human IgG was added first to block Fc receptors, and recombinant IL-10RB with the same Fc tag did not bind any cell type tested (Fig 4D and 4E). Among peripheral immune cells, more primary monocytes bound sIFN-λR1 than any other cell type, but A549 and Huh7.5 epithelial cell lines required~100-fold less sIFN-λR1 protein to bind similar levels of sIFN-λR1 as monocytes (0.01 μg/ml epithelial cells versus 1 μg/ml for monocytes) (Fig 4D  and 4E). Due to the secondary anti-Fc antibody binding to surface IgG on primary B cells, we did not include B cell results in our analysis. We visualized sIFN-λR1 binding to the cell surface of four cell lines using imaging flow cytometry. We observed a distinct cell surface staining pattern with higher levels of sIFN-λR1 on the surface of A549 and Huh7.5 epithelial cell lines compared to THP-1 monocytes and Jurkat T cells (median 8-10 fold greater despite 2-fold less sIFN-λR1 added) ( Fig 4F). Addition of IFN-λ3 protein to cells simultaneously with sIFN-λR1 did not increase sIFN-λR1 binding to Huh7.5 cells (S7A Fig). Taken together, we have shown for the first time that sIFN-λR1 can bind multiple cell types without requiring previous interaction with IFN-λ cytokine.
After observing striking differences in binding of sIFN-λR1 to various cell types, we next utilized our IFN-λ3 binding assay to determine if sIFN-λR1 binding to the cell surface affects IFN-λ3 binding. The addition of recombinant sIFN-λR1 with IFN-λ3 to PBMCs led to 5-15 fold greater binding of IFN-λ3 to the surface of B cells, monocytes, and T cells compared to IFN-λ3 alone at the highest dose tested (Fig 4G and 4H). Addition of sIFN-λR1 also increased binding of IFN-λ3 to Huh7.5 cells in a dose-dependent manner (S7B Fig). Allowing sIFN-λR1 to bind first to cells before adding IFN-λ3, or adding them simultaneously, resulted in the equivalent increased binding of IFN-λ3 compared to adding cytokine alone (S7C Fig). Recombinant IL-10RB or denatured sIFN-λR1 at the same concentrations did not increase IFN-λ3

PLOS PATHOGENS
Interferon-lambda antiviral activity in human immune cells binding to any cell type tested (Fig 4G). These results could help explain why B cells are less responsive to IFN-λ3 compared to lung epithelial cells despite similar IFNLR1 transcript expression. Taken together, our data shows sIFN-λR1 increased IFN-λ3 binding to the cell surface, potentially through interactions with IL-10RB, the second subunit of the IFN-λR. This binding would prevent ISG induction though, because the cytoplasmic tail, not present in sIFN-λR1, is required for downstream signaling [29]. Lower levels of sIFN-λR1 at epithelial barriers could explain why those cell types are especially responsive to type III IFNs.

Soluble IFNLR1 evolved late in evolution with detectable transcripts in specific primates
Our data describing which human immune cell express total IFNLR1 transcripts contrast published mouse cell Ifnlr1 expression, but previous studies had only measured sIFNLR1 transcripts in human cells [3,4,35]. We next investigated whether other primates and lower mammals such as mice also encoded for a soluble form of IFNLR1. We found that the sIFNLR1 transcript variant missing the transmembrane domain (exon 6) is only annotated in primates, specifically most old world monkeys and all apes, but not in new world monkeys, lemurs, tarsiers and lower mammals with the exception of the guinea pig ( Fig 4I and S3 Table). To examine if annotations were missing for sIFNLR1 and to find experimental evidence of sIFNLR1 expression, we BLAST searched the NCBI Sequence Read Archive (SRA) using a 120 nucleotide sequence spanning the end of exon 5 and beginning of exon 7 for each species of interest. We found detectable sIFNLR1 transcripts in experiments with cells from apes and certain old world monkeys in multiple tissues including epithelial cells and immune cells, but sIFNLR1 transcripts could not be detected in any experimental data deposited for lower mammals, including the potential variant annotated in guinea pigs. A full summary of species studied and example data accession numbers are listed in S3 Table. These results are consistent with the late evolution of the soluble variant of IFN-λR1 in mammals, with only the closest relatives of humans having experimental evidence for the expression of sIFNLR1.

Stimulation of B cells and neutrophils upregulates IFN-λR1 expression
The differential expression of IFN-λR1 on the surface of different human immune cell subsets implied that IFN-λR1 expression can be regulated. We therefore examined whether IFN-λR1 expression could be altered by adding various stimuli to B cells or neutrophils. B cells were chosen because they represent cells that express high levels of IFNLR1, while neutrophils have little IFNLR1 transcript expression. Previous work had shown that IFN-α significantly upregulated IFNLR1 mRNA in human hepatocytes [9], so we tested whether this regulation also occurred in immune cells. Stimulation of PBMCs for 3 days with anti-BCR (IgM/IgG/IgA) and anti-CD40 or the TLR7/8 ligand R848, but not the cytokines IFN-γ or IFN-α2, increased IFN-λ3 binding to total B cells ( Fig 5A). Surprisingly, IFN-α2 treatment significantly decreased the percent of total B cells binding IFN-λ3 by 62% on average (Fig 5A). We performed a cell subsets within total PBMCs (D), or to A549 or Huh7.5 cell lines (E). Binding was detected by anti-Fc PE antibody. Histograms are representative of results from 4 different PBMC donors (bar graph right panel D), or from 2-3 independent experiments (E). F) Imaging flow cytometry visualization of binding of sIFN-λR1 (500 ng/ml (A549, Huh7.5), 1 μg/ml (THP-1, Jurkat)) to 4 cell lines. The median PE intensity is shown below each representative image from 6,000-12,000 total cells acquired from 2-3 independent experiments. G-H) IFN-λ3 binding to cell subsets within PBMCs where IFN-λ3 (2 μg/ml) was added with or without sIFN-λR1 (0.5-5 μg/ml +/-boiling) or IL-10RB (5 μg/ml). Percent IFN-λ3 bound and fold increase in median PE fluorescence are representative from 2-4 donors (G) or plotted as means + SD from 2-4 different donors (H). I) The presence of a soluble IFNLR1 variant within primates and lower mammals. Gray indicates no annotation and no transcript detectable in deposited RNA sequencing data, and purple indicates both annotation and detection of RNA transcript in multiple cell types from deposited RNA sequencing data. � , P<0.05, paired t-tests (B-C). https://doi.org/10.1371/journal.ppat.1008515.g004

PLOS PATHOGENS
Interferon-lambda antiviral activity in human immune cells similar analysis to compare naïve and memory B cells within PBMCs. R848 treatment significantly upregulated IFN-λ3 binding fluorescence to naïve B cells, but not memory B cells, whereas IFN-γ addition specifically decreased IFN-λ3 binding to memory B cells (Fig 5B).

PLOS PATHOGENS
Interferon-lambda antiviral activity in human immune cells data demonstrate that receptor stimulation of B cells and neutrophils can upregulate surface expression of IFN-λR1. Therefore, during an infection or inflammatory state, multiple immune cell types could have greater responsiveness to type III IFNs.

Stimulation of CD4 + T cells via the TCR upregulates IFN-λR1 expression and IFN-λ3 induced ISG expression
Given that BCR and TLR activation upregulated IFN-λR1 expression on B cells and neutrophils, we hypothesized that the low IFNLR1 expression observed in CD4 + T cells could also be modulated with appropriate T cell specific stimulation. In addition, we and others have shown IFN-λs modulate Th2 cytokine production [6,21,57], therefore understanding if IFN-λR1 expression changes upon stimulation has direct implications in elucidating mechanistically how Th2 responses are regulated. Only stimulating the TCR with anti-CD3/anti-CD28 antibodies significantly upregulated IFN-λ3 binding to CD4 + T cells (Fig 6A). R848 has limited activation potential with human T cells in the absence of other signals [58], therefore R848 activating monocytes and B cells during the 3 day culture did not indirectly upregulate IFN-λ3 binding to CD4 + T cells. We further investigated changes in IFN-λR expression at the transcript level in sorted CD4 + T cells. We found total IFNLR1, but not IL10RB, mRNA was significantly increased by TCR stimulation (Fig 6B). Interestingly, only the signaling capable mIFNLR1, but not sIFNLR1, was upregulated by TCR stimulation (Fig 6C). The mIFNLR1/sIFNLR1 ratio for CD4 + T cells had increased from an average of 2.5 at day 0 to 8.0 at day 3, bringing it closer to ratios we previously observed in epithelial cells. This is the first report of specific regulation of mIFNLR1 expression by TCR stimulation. We next examined whether TCR upregulation of IFNLR1 enhanced IFN-λ3 responsiveness. Stimulation of PBMC cultures with anti-CD3/anti-CD28 for 3 days followed by IFN-λ3 addition to sorted CD4 + T cells led to enhanced ISG induction for 2 of 3 ISGs tested (Fig 6D). Altogether, our data demonstrate TCR stimulation significantly increases mIFN-λR1 expression on CD4 + T cells leading to greater ISG induction in response to IFN-λ3.

IFN-λ3 treatment inhibits HIV-1 infection of purified CD4 + T cells
To examine the consequence of increased IFN-λR1 expression on CD4 + T cells, we determined if IFN-λ3 could directly inhibit a viral infection of CD4 + T cells. We used HIV-1 infection of purified CD4 + T cell cultures to monitor whether pre-incubation of IFN-λ3 before HIV-1 inoculation leads to decreased viral infection in the absence of any other immune cells. Optimal HIV-1 infection requires T cell activation [59,60], therefore we purified total CD4 + T cells and activated them with PHA for 3 days. PHA activation upregulated mIFNLR1 mRNA 2.4 fold on average, but did not upregulate sIFNLR1 expression ( Fig 7A). As expected, subsequent IFN-λ3 treatment of PHA activated CD4 + T cells stimulated significant ISG expression compared to no cytokine treated controls (Fig 7B). Next, we tested whether IFN-λ3 induced an antiviral state in PHA-stimulated CD4 + T cells. We added IFN-λ3, or pegylated IFN-α2 as our positive control, for 24 hours prior to HIV-1 infection. CD4 + T cells treated with either IFN-λ3 or IFN-α2 had significantly decreased HIV-1 p24 positive cells, indicating IFN-λR or IFN-αR signaling inhibited HIV-1 infection (Fig 7C and 7D). These data indicate that type III IFNs can directly impact antiviral responses of peripheral blood CD4 + T cells to inhibit HIV-1 infection.

Discussion
The lack of a sensitive IFN-λR1 antibody has limited our understanding of IFN-λR1 biology in the context of human immune cells. Here, we have clearly identified the main cell types that

PLOS PATHOGENS
Interferon-lambda antiviral activity in human immune cells bind and respond to IFN-λ3 in human peripheral blood, and demonstrate that a unique soluble IFN-λR1 variant present in higher primates inhibits IFN-λ3-mediated ISG induction by binding the cell surface and sequestering IFN-λ3 away from a functional membrane spanning IFN-λR heterodimer complex. Activation of immune cells promotes specific mIFNLR1 expression and the direct interaction of IFN-λ3 with CD4 + T cells induces an antiviral state to decrease HIV-1 infection. Thus, unlike in mice, type III IFNs have the potential to directly impact resting and activated human adaptive immune cells.
Quantification of cell surface IFN-λ3 binding to peripheral blood immune cells correlated with IFNLR1 transcript expression in most cases. Human B and T cells can interact directly with IFN-λ3, and IFN-λ3 binding levels correlated with the magnitude of ISG induction in stimulation assays. Consistent with previous literature, human monocytes and NK cells did not respond directly to type III IFNs [35,39,61]. However, there were exceptions where IFN-

PLOS PATHOGENS
Interferon-lambda antiviral activity in human immune cells λ3 binding did not match IFNLR1 transcript expression and/or ISG induction. We noted that liver and lung epithelial cells expressed similar or lower total IFNLR1 mRNA than B cells, but bound~5-fold more IFN-λ3 and upregulated ISGs up to 50-fold greater than B cells. Upon further analysis of the variants of IFNLR1, we found every immune cell isolated expressed

PLOS PATHOGENS
Interferon-lambda antiviral activity in human immune cells greater sIFNLR1 transcripts relative to mIFNLR1, compared to epithelial cells. sIFN-λR1 bound well to every cell type tested, and surprisingly promoted increased binding of IFN-λ3 to the cell surface. Significant inhibition of ISG induction in the presence of sIFN-λR1 likely occurs because the IFN-λ3-sIFN-λR1 complex could not initiate signaling due to the lack of cytoplasmic tail in sIFN-λR1. The lower baseline expression of sIFNLR1 in epithelial cells, may also be the reason that sIFN-λR1 bound 100% of epithelial cells at a 10-100 fold lower dose than immune cells. Our data add to the understanding of IFN-λ-mediated pathogen control mechanisms by the expression of sIFN-λR1 only in upper primates, and mIFN-λR1 upregulation upon activation of B cells, CD4 + T cells, and neutrophils. The presence of the human sIFN-λR1 variant was detected at the time of receptor discovery in 2003 [3,4], and a previous report indicated pre-incubation of sIFN-λR1 with IFN-λ1 decreased stimulation of a hepatocyte cell line [35], however no mechanisms that changed the relative expression of the 2 main splice forms of IFN-λR1 have been investigated. Soluble cytokine receptors can have either inhibitory or enhancing properties. For example, IL-22BP, also an IL-10 cytokine family member, acts as a decoy blocking IL-22 binding at mucosal surfaces where IL-22R is expressed [62,63]. In contrast, sIFN-αR2a binding to IFN-α enhances IFN-α responses in vivo [64]. It is unclear why primitive primates and lower mammals do not have evidence of sIFN-λR1 expression, while both mice and humans express IL-22BP and sIFN-αR2a. Compared to type I IFNs, type III IFNs induce a slower, lower magnitude response [23, 24] with less inflammatory effects in human clinical trials or in mouse lung infections [25, [65][66][67]. The expression of sIFN-λR1 may have evolved as a mechanism to contribute to immune cells remaining in an "off" condition until they receive optimal activation signals to allow for increased type III IFN responses, unlike all immune cells responding to type I IFNs at baseline. Since our work identified stimuli that regulate mIFN-λR1, but not sIFN-λR1 expression, determining what stimuli modulate sIFN-λR1 expression will be of future interest.
We observed IFN-λ3 bound well to monocytes despite very little mIFNLR1 transcript expression, but ISGs were not induced in our assays. The binding was not an artifact of the 6 His-tag on IFN-λ3, since an unrelated, similarly tagged protein did not bind monocytes. IFN-λ3 is not likely binding IL-10RB alone, or high binding would have also been observed to neutrophils, which also express very high levels of IL-10RB. In addition, the affinity of type III IFNs to IL-10RB alone is very low [68]. Since we showed that recombinant sIFN-λR1 directly binds monocytes, we speculate that IFN-λ3 binds monocytes in our assay at least in part because sIFN-λR1 is already bound to the cell surface. Altogether, our data indicate there are multiple layers of regulation of the responses to IFN-λ3.
While many studies have demonstrated potent antiviral activity of type III IFNs at epithelial barrier sites, much less is known regarding type III IFN antiviral activity in peripheral blood. Previous studies proposed that human and mouse T cells do not directly respond to IFN-λs, and that IFN-λ dependent regulatory effects occur via a specific subset of DCs or neutrophils [20,25,35,44,[69][70][71]. Here, we show for the first time that human CD8 + T cells express significantly greater IFNLR1 transcripts, and bind IFN-λ3 to a higher degree than CD4 + T cells, leading to the upregulation of ISGs in response to IFN-λ3 in vitro. Only upon TCR or PHA activation of CD4 + T cells, did the level of IFNLR1 mRNA expression match or surpass that seen in CD8 + T cells, in turn leading to greater ISG induction and ultimately inhibition of HIV-1 infection by IFN-λ3 treatment. While two previous studies found IFN-λ1 or IFN-λ2 could inhibit or enhance HIV-1 infection of CD4 + T cells [72,73], our findings support the IFN-λ3-mediated induction of an antiviral program in CD4 + T cells, with no indication of an enhancement of infection. Our data demonstrate that CD4 + and CD8 + T cells can be directly regulated by type III IFNs to combat viral infections. While we focused on IFN-λ3 activity in this study, we predict other IFN-λ family members would be able to stimulate ISGs in the same cell types since they all bind the same receptor, but receptor affinity may affect the magnitude of the response [74].
Our results that human neutrophils bind low levels of IFN-λ3 and express low levels of IFN-λR1 are in contrast to mouse models of influenza virus or Aspergillus fumigatus infection and arthritis or colitis where mouse neutrophils directly respond to IFN-λ to modulate responses [19,20,25,46]. Limited data has been published measuring ISG induction by IFN-λs in human neutrophils. Human neutrophils responded~20-fold less than mouse neutrophils to IFN-λ2 [20], and IFIT1 was not induced in human neutrophils after IFN-λ1 treatment [75]. IFN-λ3 may induce greater ISG levels in neutrophils in vivo because neutrophils survive longer than the 5 hour time point used in the current study. Additionally, whether human neutrophils are also regulated by the unique STAT-1 independent, non-translational signaling pathway seen after Ifn-λ2 treatment of mouse neutrophils is unclear [20]. Our quantification showing that human neutrophils express low levels of IFNLR1 transcripts directly ex vivo are in agreement with a published human microarray and single cell RNA sequencing results [54,55]. Subsequent stimulation of neutrophils with TLR agonists and bacteria induced IFNLR1 mRNA expression, which complements data showing that Aspergillus fumigatus addition to neutrophils upregulated IFN-λR1 expression [46]. Therefore, human neutrophils may optimally respond to IFN-λs in vivo through upregulation of IFN-λR1 during inflammatory conditions. Future work directly measuring IFN-λR1 protein on the cell surface after stimulations will only be possible when a specific, sensitive antibody becomes available.
There is a growing list of differences between the mouse and human type III IFN system. Mice only have functional Ifnl2 and Ifnl3 genes, and here we show the sIFN-λR1 splice variant is not present in mice and other non-primates. Although initial studies found human B cells do not respond to IFN-λs [35, 39], we and another group have now clearly demonstrated human B cells directly respond to IFN-λ3 (current study) and IFN-λ1 [45]. Our data showing IFN-λ3 binds to human pDCs matches previous studies showing human pDCs respond to type III IFNs [41,43,44], but studies have found contradicting results for whether type III IFNs stimulate ISGs in mouse pDCs [7,20]. We showed that activation through TCR, BCR, or TLRs upregulated IFNLR1 mRNA expression and IFN-λ3 binding to multiple human immune cell types, but a recent study found TLR3 stimulation or addition of Sendai virus did not upregulate IFNLR1 transcript expression in a hepatocyte cell line [26]. Therefore, IFNLR1 expression may be differentially regulated in epithelial cells compared to immune cells. Interestingly, we found IFN-α2 decreased IFN-λ3 binding to B cells, whereas a previous study demonstrated IFN-α treatment upregulates IFNLR1 mRNA expression in hepatocytes [9]. IFN-γ treatment alone was also unable to upregulate IFN-λR1 expression in B cells or T cells. Our findings demonstrate IFN-λR1 expression is uniquely regulated in immune cells compared to epithelial cells, and future work should confirm if IFN-λR1-dependent regulatory mechanisms discovered in mouse models are reproducible in human studies.
In summary, our study has provided clear evidence of which human immune cells express functional IFN-λR with major differences in IFN-λR1 expression between mice and humans, especially within the adaptive immune system. The remarkable absence of IFN-λR1 expression in most mouse immune cells begs the question for how phenotypes and immunoregulatory mechanisms could differ during a viral infection or autoimmune mouse model if IFN-λR1 cell expression mimicked that seen in humans. Going forward, studying the regulation of human IFN-λR1 expression in epithelial cells and immune cells will provide critical information to guide mechanistic studies related to type III IFN regulation of immune responses, with the goal of developing therapies to fight viruses, dampen chronic inflammation at mucosal sites, and treat cancer with less side effects compared to type I IFNs.

Ethics statement
This study was approved by the University of Alberta Health Research Ethics Board (Pro00046564). All blood donors gave written informed consent in accordance with the Declaration of Helsinki. Normal human lungs that were not used for transplantation were obtained via a tissue retrieval service (International Institute for the Advancement of Medicine, Edison, NJ). The identity of donors was not provided, although basic demographic data was included. Ethical approval to obtain normal human bronchial epithelial cells (NHBE) was obtained from the Conjoint Health Research Ethics Board of the University of Calgary and from the Internal Ethics Board of the International Institute for the Advancement of Medicine.

Cell lines and primary cell isolation
Huh7.5 cells (from Dr. Charles Rice, The Rockefeller University) and Huh7 IFNLR1 knockout cells [49] (from Dr. Ram Savan, University of Washington), were cultured in DMEM containing: 10% fetal bovine serum (FBS), 100 U/ml penicillin, 100 μg/ml streptomycin, 1X MEM non-essential amino acids and 4 mM L-glutamine (media from GE Healthcare, supplements from Thermofisher). BEAS-2B and DG75 cells were obtained from ATCC and cultured using the BEGM bronchial epithelial growth medium Bulletkit (Lonza) according to the manufacturer's instructions, except hydrocortisone was not added to medium during any cytokine stimulations. PBMCs were isolated from healthy human individuals by Ficoll-Paque PLUS (GE Healthcare) gradient centrifugation. All PBMC experiments used freshly isolated cells. Immune cells were sorted using a BD FACSAria III cell sorter within a biosafety cabinet with dead cells excluded using LIVE/DEAD Near IR (Thermofisher) after gating out doublets. Cells were sorted as follows: B cells: CD20 + CD3 -CD56 -CD14 -, CD4 + T cells: CD3 + CD4 + CD8 -, CD8 + T cells: CD3 + CD8 + CD4 -, Monocytes: CD14 + CD3 -CD20 -, NK cells: CD56 + CD3 -. Purities were routinely � 97-99% and pDCs were not detectable by flow cytometry (CD123 + HLA-DR + ). Representative purity results are shown in S5 Fig. In some experiments, B cells were isolated with a negative B cell isolation kit (StemCell Technologies) or CD14 + monocytes were isolated by CD14 positive selection (Miltenyi Biotec), but results were comparable whether cells were isolated via FACS or magnetic selection. PBMCs or isolated immune cells were cultured in RPMI 1640 containing: 10% FBS, 100 U/ml penicillin, 100 μg/ml streptomycin, 10 mM HEPES and 2 mM glutamax (media from GE Healthcare, supplements from Thermofisher). Primary human hepatocytes were purchased from Biore-clamationIVT. Normal, nontransplanted human lungs were obtained via a tissue retrieval service (International Institute for the Advancement of Medicine, Edison, NJ). Bronchial epithelial cells were isolated as previously described [76], and cultured in BEGM with supplements (Lonza) as previously described [77]. In brief, cells were plated in 6 or 12 well plates and utilized at~70% confluence (typically after 10-11 days with media change every 2 days) for binding or stimulation assays. Neutrophils were isolated from healthy donor whole blood with Polymorphprep (Axis Shield) according to the manufacturer's protocol to achieve highly pure, unprimed neutrophils suitable for gene expression studies [78]. Purities were � 95% (S2 Fig) and pDCs were not detectable by flow cytometry. IFN-λ3 binding to neutrophils was also quantified after RBC magnetic depletion (StemCell Technologies). All cells were maintained in an incubator with a humidified atmosphere with 5% CO 2 at 37˚C.

PLOS PATHOGENS
Interferon-lambda antiviral activity in human immune cells

Flow cytometry
The IFN-λ3 binding assay was performed as published [48]. In brief, cells were cultured with or without His-tagged IFN-λ3 (R&D Systems) diluted in PBS containing 1% BSA on ice for 60 min at indicated doses. In some experiments, recombinant sIFN-λR1 or IL-10RB protein (R&D Systems) was added simultaneously with IFN-λ3 while FcγRs were blocked with 150 μg/ ml human IgG (Jackson Immunoresearch) or human Fc Block (BD Biosciences). Cells were then washed (PBS + 1% BSA + 0.05% sodium azide) and stained with anti-His PE (Miltenyi Biotec) in combination with multiple surface marker antibodies on ice for 40 min in the dark. Anti-His PE secondary antibody was added alone in a parallel sample to obtain background fluorescence that was subtracted from each sample ('background subtracted'), and background was always less than 1%. Cells were washed again and re-suspended in 2% paraformaldehyde (Electron Microscopy Sciences) at room temperature for 15 min in the dark. Paraformaldehyde was washed away prior to analysis. sIFN-λR1-Fc or IL-10RB-Fc (R&D Systems) binding to the cell surface was similarly quantified. sIFN-λR1 or IL-10RB was added to cells for 45 min on ice and after washing, anti-Fc PE (Biolegend) was added to detect receptor bound to the cell surface. anti-Fc PE antibody was always added alone in a parallel sample so that any background would be subtracted. The following antibodies were used to identify our subpopulations: CD3 (clone UCHT1 Biolegend), CD4 (clone SK3 BD Bioscience, RPA-T4 eBioscience), CD8 (clone SK1 Biolegend), CD14 (clone HCD14 Biolgend, 61D3 eBioscience), CD56 (clone HCD56 Biolegend), CD20 (clone 2H7 Biolegend), CD66b (clone G10F5 Biolegend), HLA-DR (clone L243 eBioscience), CD123 (clone 6G6 eBioscience), CD11c (clone Bu15 Biolegend), CD16 (clone 3G8 Biolegend), IgD (clone IA6-2 Biolegend), CD27 (clone O323 Biolegend). For each antibody panel, a dump channel, where antibodies to multiple immune cells not of interest for that assay, was used to exclude contaminating cell types. Examples of our gating strategy are shown in S1 and S2 Figs. For stimulation experiments, LIVE/DEAD Near IR (Thermofisher) was utilized within the dump gate to determine live cells for IFN-λ3 binding quantification. NHBE and hepatocytes were incubated with IFN-λ3 in the same binding assay without any surface staining antibodies since they were pure populations. Samples were analyzed using a BD LSR Fortessa X-20 or LSR Fortessa-SORP Flow Cytometer (5 laser: 375 nm, 405 nm, 488 nm, 561 nm, and 633 nm) and FlowJo software (BD Biosciences) was used for data analysis and graph generation. In certain experiments, sIFN-λR1 binding was also visualized via imaging cytometry using the Amnis ImageStream 1 X mark II Flow Cytometer (Millipore-Sigma) and Inspire 1 (Amnis) software. 6,000-12,000 total events were collected for each experiment. Samples were imaged at 60x magnification and IDEAS 1 (Amnis) software was used for single cell analysis.

PLOS PATHOGENS
Interferon-lambda antiviral activity in human immune cells guidelines, with on column DNase I digestion. Reverse transcription was performed with a Superscript VILO IV mastermix (Thermofisher). RT-qPCR was performed with a Bio-Rad CFX 96 using Thermofisher POWER SYBR mastermix (Thermofisher) according to the manufacturer's protocol: 95˚C for 10 min, and then repeating 40 times, 95˚C for 15 sec and 60˚C for 60 sec, followed by a melting curve analysis. SYBR green primer sequences are listed in S4 Table, with the exception of the B2M QuantiTect Primers (Qiagen). For Taqman analysis of IFNLR1 (Hs00417120_m1), IL10RB (Hs00175123_m1), HPRT1 (Hs02800695_m1), RPL13A (Hs04194366_g1) and B2M (Hs99999907_m1) (Thermofisher), Taqman Fast Advanced mastermix (Thermofisher) was used with the following cycling parameters: 50˚C for 2 min, 95˚C for 2 min, and then repeating 40 times, 95˚C for 1 sec and 60˚C for 20 sec. No template controls were run for every set of primers on each plate. Samples were normalized to the geometric mean of two reference genes, HPRT1 and RPL13A, as described [80], unless indicated when B2M was the reference gene that was stably expressed between samples. Fold changes in mRNA expression were calculated using the ΔΔCt method with comparisons of stimulated to unstimulated cells. Relative expression values were calculated by 2 -(ΔCT) after normalization to reference genes indicated.

CD4 + T cell HIV-1 infection
CD4 + T cells were isolated using a CD4 + T cell enrichment kit (StemCell Technologies) according to manufacturer's instructions. The enriched CD4 + T cells (2×10 6 cells/ml) were maintained in RPMI 1640 media supplemented with 10% FBS, 100 U/ml penicillin, 100 μg/ml streptomycin, 10 μg/ml phytohemagglutinin-M (PHA-M; Sigma), and 50 U/ml recombinant IL-2 at 37˚C and 5% CO 2 incubator for 3 days. Excess PHA was removed by washing with fresh media and cell density was adjusted to 7×10 5 cells/ml in 48 well plates. Activated T cells were cultured with media alone, IFN-α2 (100 IU/ml) or IFN-λ3 (100 ng/ml) for 24 hours before CXCR4-tropic HIV-1 LAI virus (AIDS Research and Reagent Program, NIH) infection via magnetofection (OZ Biosciences), as previously described [81]. Non-bound, excess HIV-1 virus was washed away after 24 hrs of infection and fresh media with or without IFNs were added back for an additional 2 days in culture. Uninfected and infected CD4 + T cells with or without IFN-λ3 treatment were stained intracellularly with HIV-1 KC57-p24 core (Beckman Coulter) antibody on day 3 post HIV-1 infection.

Statistics and data analysis
Graphs were formulated and data were analyzed in Graphpad Prism 7 or FlowJo software (BD Biosciences). A P-value less than 0.05 was considered significant. The number of healthy donors or replicates and statistical tests are specified in each figure legend.

PLOS PATHOGENS
Interferon-lambda antiviral activity in human immune cells S1