T follicular helper (TFH) cells are a specialized subset of CD4 T cells that deliver critical help signals to B cells for the production of high-affinity Abs. Understanding the genetic program regulating TFH differentiation is critical if one wants to manipulate TFH cells during vaccination. A large number of transcription factor (TFs) involved in the regulation of TFH differentiation have been characterized. However, there are likely additional unknown TFs required for this process. To identify new TFs, we screened a large short hairpin RNA library targeting 353 TFs in mice using an in vivo RNA interference screen. Yin Yang 1 (YY-1) was identified as a novel positive regulator of TFH differentiation. Ablation of YY-1 severely impaired TFH differentiation following acute viral infection and protein immunization. We found that the zinc fingers of YY-1 are critical to support TFH differentiation. Thus, we discovered a novel TF involved in the regulation of TFH cells.
Germinal centers (GCs) are essential for the generation of high-affinity Abs following infection or vaccination. Activated B cells that receive sufficient help from T cells differentiate into GC B (BGC) cells and undergo somatic hypermutation and isotype class switching. These processes are uniquely dependent on signals delivered by a specialized subset of CD4 T cells called T follicular helper (TFH) cells. Differentiation of TFH cells is a multistep process beginning with interaction of CD4 T cells with Ag-presenting dendritic cells. Upon reception of critical signals such as IL-6 (1), activated CD4 T cells begin differentiating into TFH cells. Differentiation of TFH cells depends on expression of the transcriptional repressor Bcl6 (2–5). Bcl6 plays a central role in the specification of the TFH genetic program by suppressing alternative fates (5, 6). Direct repression of Blimp-1 and Id2 by Bcl6 allows expression of TFH-associated genes such as CXCR5 (6, 7). There is substantial evidence that multiple transcription factors (TFs) act in a coordinated manner to specify cell fate and function (8). This is illustrated by work showing that retinoic acid–related orphan receptor (ROR)γt is at the center of a TF network that includes STAT3, Maf, IFN regulatory factor (IRF)4, and Batf to establish the genetic program supporting the differentiation of TH17 cells (9).
In recent years, multiple TFs regulating TFH cell differentiation have been identified. LEF-1 and TCF-1 are both essential for the early induction of Bcl6 and suppression of Blimp1 expression (10–12). Batf (13) and Maf (14–16) promote TFH differentiation and function by driving expression of IL-21, a cytokine essential for B cell help (6). Expression of IRF4 needs to be tightly regulated to ensure TFH differentiation, as both absence and overexpression of IRF4 prevent formation of TFH cells (17). TFH-promoting TFs are balanced by suppressors of TFH formation. Blimp-1 and Foxo1 both block Bcl6 expression (2, 5, 18). Klf2 inhibits TFH differentiation by promoting expression of T cell zone homing receptors and suppressing TFH-associated genes (5, 19, 20). Id2 impairs TFH differentiation by blocking the E protein–mediated expression of CXCR5 (7, 21). All of these TFs form a large network to establish the TFH genetic program.
In this study, we report the identification of Yin Yang 1 (YY-1) as an essential TF for TFH differentiation. YY-1 is involved in the development and differentiation of multiple immune cells. Deletion of YY-1 in B cells results in a developmental block at the pro–B cell stage caused by defective rearrangement of the IgH locus (22). Similarly, ablation of YY-1 in thymocytes results in a severe block in T cell development at the double-negative stage (23). YY-1 is also required for the differentiation of BGC cells following protein immunization (24–26). Knockdown of YY-1 completely abrogated early differentiation of TFH cells. Using a series of YY-1 mutants, we demonstrate that the zinc finger of YY-1 is essential to promote TFH cells. Furthermore, we show that the defective differentiation of YY-1–deficient TFH cells can be partially rescued by Bcl6 overexpression.
Materials and Methods
Mice and lymphocytic choriomeningitis virus infection
SMARTA mice (27), Bcl6fl/fl CreCD4 SMARTA mice (28), OT-II mice, and CD45.1+ mice were on a full C57BL/6 background and bred in-house. C57BL/6 mice were purchased from The Jackson Laboratory. Both male and female mice were used throughout the study, with sex- and age-matched T cell donors and recipients. All mice were maintained in specific pathogen-free facilities and used according to protocols approved by the Animal Care and Use Committees of the La Jolla Institute for Immunology. Recipient mice were infected by i.p. injection of 2 × 105 (day 6), 5 × 105 (day 3), or 1 × 106 (days 2.5 or 1) PFU of LCMV (lymphocytic choriomeningitis virus) Armstrong (LCMV-Arm). Ten micrograms of OVA conjugated to 4-hydroxy-3-nitrophenylacetyl (NP) hapten (NP-OVA) mixed with Alhydrogel (InvivoGen) in 20 μl of PBS was injected in each foothock.
Transfection, transduction and cell transfer for short hairpin RNA screen
MicroRNA-adapted short hairpin RNA (shRNAmir) vectors (LMPd-Ametrine) were described previously (29). The shRNAmir retroviral vector (shRNAmir-RV) library targets 353 TFs expressed in mice. Each gene is targeted by three or four shRNAmirs for a total of 1414 shRNAmir-RVs in the library. The library was arrayed at a concentration of 50 ng/ml in 21 individual 96-well plates (60 shRNAmir-RV plasmids per plate). We included a set of 42 shRNAmir-RVs targeting 14 genes with known outcomes on TFH differentiation (Supplemental Table I). The shRNA screen was performed and analyzed as previously described (S. Bélanger, S. Haupt, and C.E. Faliti, manuscript in preparation).
Retroviral vectors, transductions, and cell transfers
Yy1, Yy1 resistant to shYy1 no. 1, and the Yy1 deletion mutants were cloned into pMIG. Transductions were performed as previously described (29). Transfer of sorted cells into recipient mice was performed by i.v. injection via the retro-orbital sinus. Transferred cells were allowed to rest in host mice for 3–4 d before infection or immunization. Then, 2 × 104 (day 6), 4 × 105 (day 3), or 1 × 106 (days 2.5 and 1) transduced CD4+ T cells were transferred into each mouse. In certain experiments, cells were labeled with 5 μM CellTrace Violet (Life Technologies) prior to adoptive transfer.
Single-cell suspensions of spleen or draining inguinal lymph nodes were prepared by standard gentle mechanical disruption. Surface staining for flow cytometry was done with mAbs against CD4, CD8, CD45.1, and B220 (eBioscience), CD69 (BD Biosciences), and SLAM and CD25 (BioLegend). CXCR5 staining was done using biotinylated anti-CXCR5 (BioLegend), followed by PE-Cy7–, BV421- or allophycocyanin-labeled streptavidin (1:1000; BioLegend). Intracellular staining was performed with an mAb to Bcl6 (BD Biosciences), TCF1 (Cell Signaling Technology), LEF-1 (Cell Signaling Technology), YY-1 (Santa Cruz Biotechnology), or Tbet (eBioscience), as well as the Foxp3 intracellular cytokine staining kit buffers and protocol (Invitrogen). Stained cells were analyzed using FACSCelesta (BD Biosciences) and FlowJo software (Tree Star).
Statistical tests were performed using Prism 9.0 (GraphPad). Significance was determined by an unpaired Student t test with a 95% confidence interval or by two-way ANOVA with a Sidak’s multiple comparison test (*p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001, and ****p ≤ 0.0001).
YY-1 is a positive regulator of TFH differentiation
We hypothesized that a number of TFs regulating TFH differentiation are yet to be discovered. We designed a large shRNAmir library (1414 shRNAmirs) targeting 353 TFs. This library was screened using our previously detailed in vivo shRNA screening protocol (S. Bélanger et al., manuscript in preparation) (Supplemental Fig. 1A). Briefly, the shRNAmir library was expressed in SMARTA CD4 T cells by retroviral transduction. shRNAmir+ SMARTA CD4 T cells were adoptively transferred into C57BL/6 mice that were subsequently infected with LCMV-Arm. At 3 d postinfection, SMARTA CD4 T cells were sorted into CXCR5+SLAMlo TFH and SLAM+CXCR5− TH1 populations (Supplemental Fig. 1B), and the distribution of the shRNAmir library in both populations was interrogated by next-generation sequencing. We calculated a Z score value of the TFH/TH1 ratio for each gene targeted by the shRNAmir library and observed good reproducibility of both screens (Fig. 1A). We first examined the Z scores of control genes included in the library. Bcl6 and Itch had negative Z score values (Fig. 1A) denoting depletion of shRNAmir targeting Bcl6 and Itch from the TFH population as expected because of their roles as positive regulators of TFH differentiation (2–4, 30). Prdm1, Tbx21, and Ccnt1 are all inhibitors of TFH cells or positive regulators of TH1 formation (2, 29, 31) and had positive Z score values indicative of enrichment of their shRNAmir constructs in TFH cells. We next filtered genes with negative Z score values in both independent screens. Yy1 had Z scores of −2.46 and −3.67 in the first and second screen, respectively, suggesting that YY-1 positively regulates differentiation of TFH cells (Fig. 1A). Yy1 codes for the polycomb group (PcG) transcription factor YY-1, which acts as a transcriptional activator or repressor (32–34). YY-1 is important for the commitment of CD8 T cells to the effector fate (35) and for NKT cell differentiation (36). In CD4 T cells, YY-1 cooperates with Foxp3 (37, 38) and GATA3 (39) to specify regulatory T cell and TH2 cell fate, respectively, but its role in TFH cells is unknown.
YY-1 deficiency severely impairs early TFH formation
We first sought to confirm the results of our screen. SMARTA CD4 T cells transduced with shRNAmir-RV targeting Yy1 (Fig. 1D) or a control gene (Cd8) were transferred into C57BL/6 mice. At this time postinfection, early TFH and TH1 cells can be identified by differential expression of CXCR5 and CD25. CXCR5+CD25− cells are early TFH cells whereas CD25+CXCR5− cells are early TH1 cells (40). Loss of Yy1 expression severely impaired differentiation of CXCR5+CD25− TFH cells 3 d after LCMV-Arm infection (Fig. 1B). The reduction in CXCR5+CD25− TFH differentiation was accompanied by an increase in the proportion of CD25+CXCR5− TH1 cells in shYy1+ SMARTA CD4 T cells (Fig. 1B). Expression of the essential TFH-promoting transcription factor Bcl6 was impaired in the absence of YY-1. Yy1-deficient CD4 T cells could not differentiate efficiently into CXCR5+Bcl6+ TFH cells (Fig. 1C). We also observed a substantial reduction in the accumulation of shYy1+ SMARTA CD4 T cells (Fig. 1E, Supplemental Fig. 2A, 2B). Yy1-deficient cells did not show a proliferation defect after in vitro culture (Supplemental Fig. 2C, 2D). Importantly, expression of Tbet in CD25+CXCR5− TH1 cells was not altered by loss of YY-1 (Fig. 1F), suggesting TH1 differentiation can proceed in Yy1-deficient CD4 T cells. These data confirmed the shRNA screening results and strongly suggest that YY-1 is essential for TFH formation.
We next tested whether the requirement for YY-1 to support early TFH differentiation is limited to an acute viral infection model. shYy1+ or shCd8+ OT-II CD4 T cells were transferred into C57BL/6 mice. Draining lymph nodes were analyzed 3 d after immunization with NP-OVA in alum. Few CXCR5+Bcl6+ TFH cells were observed in Yy1-deficient OT-II CD4 T cells after immunization (Fig. 1G, 1H). Accumulation of shYy1+ OT-II was also dramatically reduced (Fig. 1I, Supplemental Fig. 2E). We conclude that YY-1 is a novel critical regulator of TFH differentiation in multiple in vivo models.
Impaired activation of YY-1–deficient CD4 T cells
To test whether early activation of CD4 T cells requires YY-1 expression, we transferred shCd8+ or shYy1+ SMARTA CD4 T cells in C57BL/6 mice and analyzed the transferred cells at 1 d postinfection with LCMV-Arm. The frequency of shYy1+ SMARTA CD4 T cells was comparable to that of control cells (Fig. 2A). Control shCd8+ SMARTA CD4 T cells had initiated upregulation of CXCR5 but shYy1+ cells did not, confirming the requirement for YY-1 to support early TFH formation (Fig. 2B). Expression of activated CD4 T cells markers CD69 and CD25 was reduced on Yy1-deficient CD4 T cells (Fig. 2C, 2D). Thus, disruption of YY-1 expression substantially impairs early activation of CD4 T cells.
We next examined whether the impaired TFH differentiation in the absence of YY-1 is a consequence of reduced proliferation. We assessed proliferation of transferred shCd8+ and shYy1+ SMARTA CD4 T cells at 2.5 d postinfection with LCMV-Arm. Most control CD4 T cells had divided three or four times (Fig. 2E). In contrast, Yy1-deficient CD4 T cells had divided only once or twice, and substantial fractions of cells were still undivided (Fig. 2E). We next analyzed TFH differentiation in cells that were in the same cell division to control for the differences in proliferation between shCd8+ and shYy1+ SMARTA CD4 T cells. Whereas ∼15% of shCd8+ SMARTA T cells in their third division were CXCR5+CD25− TFH cells, only ∼5% of shYy1+ SMARTA T cells in the same cell division had differentiated into TFH cells (Fig. 2F, Supplemental Fig. 2F). Loss of CXCR5+Bcl6+ TFH cells in shYy1+ SMARTA CD4 T cells in their third division was also observed (Fig. 2G, Supplemental Fig. 2G). These data suggest a distinct YY-1 requirement for TFH differentiation.
Ectopic expression of YY-1 reverts Yy1 deficiency
We then asked whether we could rescue the defective TFH differentiation of shYy1+ CD4 T cells with ectopic expression of YY-1. Wild-type Yy1 cDNA could not be expressed in shYy1+ CD4 T cells because all shYy1 constructs were specific for sequences in the coding sequence of Yy1. To circumvent this issue, we generated a shRNA-resistant Yy1 RV (rYy1-RV) by introducing silent mutations that abrogate silencing by the shYy1 no. 1 construct. Ectopic expression of rYy1-RV in shYy1+ CD4 T cells rescued YY-1 expression to levels similar to those in control cells (Supplemental Fig. 2H). Expression of rYy1 in shYy1+ CD4 T cells reverted the severe TFH differentiation defect observed in shYy1+ CD4 T cells 3 d postinfection with LCMV-Arm (Fig. 2H). Importantly, re-expression of YY-1 also corrected the reduced accumulation of Yy1-deficient CD4 T cells (Fig. 2I, Supplemental Fig. 2I, 2J). These data demonstrate that the shYy1+ TFH deficiency phenotype was due to on-target effects of the shYy1 construct.
We next asked whether constitutive expression of Yy1 potentiates TFH formation. We hypothesized that overexpression of YY-1 would enhance differentiation of TFH cells similar to what is seen in cells overexpressing LEF-1 (10) or TCF-1 (12). We transduced SMARTA CD4 T cells with Yy1-RV or a control GFP-RV, transferred the RV+ cells into C57BL/6 mice, and analyzed the transferred cells 3 d after LCMV infection. Constitutive expression of Yy1 did not augment differentiation of CXCR5+CD25− TFH cells (Fig. 3A, Supplemental Fig. 3A) or affect accumulation of SMARTA CD4 T cells (Supplemental Fig. 3B, 3C). We next tested whether ectopic expression of Yy1 increased TFH differentiation 6 d after LCMV-Arm infection. At 6 d after LCMV infection, TFH cells can be distinguished from TH1 cells by differential expression of CXCR5 and SLAM, with TFH cells being CXCR5+SLAMlo and TH1 cells being SLAM+CXCR5−. TFH cells can further differentiate into GC TFH cells, characterized by higher expression of Bcl6 (2). Yy1-RV+ SMARTA CD4 T cells had a minor increase in the proportion of CXCR5+SLAMlo TFH cells in comparison with GFP-RV+ SMARTA CD4 T cells (Fig. 3B, Supplemental Fig. 3D–F), but the frequency of CXCR5+Bcl6+ GC TFH cells was similar between Yy1-RV+ and control SMARTA CD4 T cells (Fig. 3C, Supplemental Fig. 3G). These results demonstrate that forced expression of YY-1 is not sufficient to enhance TFH differentiation.
Zinc fingers of YY-1 are required to support TFH differentiation
We next generated a series of Yy1 deletion mutants using rYy1 cDNA as the starting sequence to ask which domains of YY-1 are required to support TFH differentiation (Fig. 4A). Each Yy1 deletion mutant was expressed in shYy1+ SMARTA CD4 T cells and transferred into C57BL/6 mice. TFH differentiation of SMARTA was analyzed 3 d after LCMV infection. In shYy1+ SMARTA CD4 T cells, expression of YY-1 lacking the REPO domain fully restored proliferation of the cells, but it could only partially support TFH differentiation compared with complete rYy1 (Fig. 4B, 4C, Supplemental Fig. 4B). The partial rescue of TFH differentiation was not a consequence of impaired expression of the YY-1–ΔREPO protein (Supplemental Fig. 4A).
A Yy1 construct lacking the transactivation domain (TD) could only partially rescue the TFH differentiation of Yy1-deficient CD4 T cells (Fig. 4B, Supplemental Fig. 4B). We could not detect YY-1 expression when the Yy1-ΔTD construct was expressed (Supplemental Fig. 4A). Presumably, the anti–YY-1 Ab used for flow cytometry binds to an epitope in the TD. The rescue of proliferation of YY-1–ΔTD+shYy1+ SMARTA CD4 T cells (Fig. 4C, Supplemental Fig. 4B) suggested that YY-1–ΔTD was efficiently expressed. Deletion of the zinc fingers completely abrogated the ability of YY-1 to restore TFH differentiation and accumulation of Yy1-deficient CD4 T cells (Fig. 4B, 4C, Supplemental Fig. 4B). YY-1–ΔZn was expressed at levels similar to those of rYY-1 (Supplemental Fig. 4A). These data demonstrate that the zinc fingers of YY-1 are absolutely required for TFH differentiation whereas the REPO domain and TD are dispensable for CD4 T cell proliferation.
Bcl6 is required for YY-1–mediated TFH differentiation
To test the relationship between YY-1 and Bcl6, we transduced Bcl6fl/fl CD4Cre SMARTA CD4 T cells with GFP-RV or Yy1-RV and transferred the cells into C57BL/6 mice. Three days after LCMV infection, we did not observe CXCR5+CD25− TFH cells in both groups of Bcl6fl/fl CD4Cre SMARTA CD4 T cells (Fig. 5A, 5B, Supplemental Fig. 4C). Therefore, YY-1 cannot compensate for loss of Bcl6, suggesting that YY-1 does not act in a dominant manner downstream of Bcl6 to support TFH differentiation.
To test whether Bcl6 acts downstream of YY-1, shCd8+ or shYy1+ SMARTA CD4 T cells were transduced with GFP-RV or Bcl6-RV, transferred into C57BL/6 mice, and analyzed 3 d following LCMV-Arm infection. Forced expression of Bcl6 could partially rescue the ability of shYy1+ SMARTA CD4 T cells to differentiate into CXCR5+CD25− TFH cells (Fig. 5C, Supplemental Fig. 4D). TFH differentiation of Bcl6-RV+shYy1+ SMARTA CD4 T cells was rescued to the level of GFP-RV+shCd8+ SMARTA CD4 T cells, suggesting that Bcl6 could not completely compensate for loss of YY-1. Forced expression of Bcl6 had no influence on the accumulation of Yy1-deficient SMARTA CD4 T cells nor on YY-1 expression (Fig. 5D, 5E, Supplemental Fig. 4D). Importantly, the partial rescue in TFH differentiation was not caused by inadequate Bcl6 overexpression in Bcl6-RV+shYy1+ SMARTA CD4 T cells (Fig. 5F). This suggests that Bcl6 cannot fully support differentiation of TFH cells in the absence of YY-1.
In this study, to our knowledge we provide the first evidence that YY-1 is required for proper differentiation of TFH cells. The involvement of YY-1 in both BGC and TFH cells adds one more common TF shared between these two cell types. We observed impaired accumulation of Yy1-deficient CD4 T cells after infection or protein immunization. The lack of TFH differentiation of Yy1-deficient CD4 T cells could be a consequence of the severe proliferation defect in these cells. Our data show that TFH differentiation in Yy1-deficient cells can be partially rescued by ectopic expression of Bcl6, without correcting the proliferation defect. It is possible that restoration of the proliferative capacity of Yy1-deficient cells could correct the TFH differentiation defect. Similarly, loss of YY-1 increases apoptosis in BGC cells (26) and in thymocytes (23). YY-1 directs p53 ubiquitination (41) and deletion of p53 reverts the developmental defect of Yy1-deficient thymocytes (23). TFH cell analysis in the present study revealed that the reduced formation of TFH cells observed in Yy1-deficient CD4 T cells was maintained in cells in the same cell division, suggesting that the TFH cell–promoting activities of YY-1 are independent of its role in supporting CD4 T cell activation and survival.
We studied three different domains of YY-1 for their roles in supporting TFH cell differentiation. The REPO domain of YY-1 recruits PcG proteins (42). PcG proteins are transcriptional repressors owing to their ability to methylate lysine 27 of histone 3 (43). We observed near normal TFH differentiation when a YY-1 mutant lacking the REPO domain was re-expressed in Yy1-deficient CD4 T cells, indicating that the REPO domain is mostly dispensable for this process. This contrasts with the essential requirement of the REPO domain of YY-1 for proper B cell development (44). EZH2 is the catalytic subunit of the PRC2 PcG protein group (43) and EZH2 is required for TFH differentiation (45, 46). In myoblasts, YY-1 recruits EZH2 to the chromatin, permitting H3K27 methylation of muscle genes (47). It is therefore possible that the interaction between YY-1 and PRC2-EZH2 is required for TFH differentiation. The TD mediates the transcriptional activity of YY-1 (32, 48). A YY-1 mutant in which the TD is deleted can partially rescue the defective TFH differentiation of shYy1+ CD4 T cells, suggesting that the transcription-promoting activities of YY-1 contribute to TFH formation. The requirement for the REPO domain and TD of YY-1 implies that transcriptional repression and activation by YY-1 are both involved in regulating TFH differentiation. The normal accumulation of SMARTA CD4 T cells expressing these constructs further supports an essential role of YY-1 in supporting TFH cells independently of CD4 T cell activation and proliferation.
We found that the zinc finger domain of YY-1 is necessary for TFH differentiation, suggesting that YY-1 DNA binding is an absolute requirement to support TFH cells. YY-1 has four zinc fingers. Further work is required to dissect which zinc fingers of YY-1 are essential for TFH cells. Zinc fingers of YY-1 are required to repress Foxp3 expression and transcriptional activity (37) and to enhance Il4 promoter activity (49). The zinc fingers of YY-1 also mediate interaction with ATF (50). YY-1 interacts with Foxp3 to suppress regulatory T cell differentiation (37), with GATA3 to regulate expression of TH2 cytokine genes (39), and with PLZF to promote NKT cell development (36). It is therefore tempting to speculate that YY-1 may interact with Bcl6 to support TFH differentiation. Forced expression of Bcl6 cannot fully compensate for the loss of YY-1. Conversely, YY-1 does not drive TFH differentiation in the absence of Bcl6. These data suggest that YY-1 is required for the full TFH-promoting activities of Bcl6. Further experiments are required to understand the interplay between Bcl6 and YY-1 and how these two TFs work together to promote TFH cells.
In sum, we identified YY-1 as a novel critical regulator of TFH differentiation and show that the zinc fingers of YY-1 are essential for this function. Our work extends the network of TFs involved in the regulation of the TFH cell fate.
We thank members of the Crotty laboratory for critical reading of the manuscript and D. Hinz, M. Haynes, S. Sehic, S. Ellis, and C. Dillingham of the LJI Flow Cytometry core for cell sorting.
The work of S.H. was partially supported by the Achievement Rewards for College Scientists Foundation of San Diego. This work was supported by the National Institutes of Health, including National Institute of Allergy and Infectious Diseases Grants U19 AI109976 and R01 AI72543 and National Institutes of Health Grant S10 RR027366 (to La Jolla Institute for Immunology), as well as by internal La Jolla Institute institutional funds to S.C.
The online version of this article contains supplemental material.
S.B. designed and performed the experiments, analyzed data, and wrote the manuscript; S.H. and B.L.F. performed the experiments and analyzed data; A.J.G. and H.D. performed and analyzed the next-generation sequencing of the shRNA screen; M.E.P. supervised sequencing analysis; and S.C. designed the study, supervised the work, and wrote the manuscript.
Abbreviations used in this article:
IFN regulatory factor
lymphocytic choriomeningitis virus
OVA conjugated to NP hapten
shRNA-resistant Yy1 RV
microRNA-adapted short hairpin RNA
shRNAmir retroviral vector
T follicular helper
S.B. is a current employee of VIR Biotechnology and may possess shares of VIR Biotechnology. The other authors have no financial conflicts of interest.