STAT6 D419 mutation neither increases the proliferative rate of GCB-type DLBCL cells nor STAT6 phosphorylation in response to IL-4

The enrichment of STAT6D419 mutations at relapse suggests they contribute to cell survival in DLBCL. Indeed, knock-down of STAT6 in DLBCL, PMBCL, cHL, and FL reduces cell viability in the absence of IL-4, which suggests that STAT6 has tonic survival functions [13, 27, 28]. To test this, we generated OCI-Ly8 [29], SU-DHL-4 [30], and DB [31] cell lines mimicking the heterozygous expression of mutations found in patient biopsies by overexpressing STAT6WT or STAT6D419 mutants (Fig. 1A). Each of these cell lines are known to be derived from GCB-type DLBCL, and each of their mutational profiles further suggests that they are of the EZB subtype, as defined by Wright et al. [4] and characterized by mutations in genes such as EZH2, CREBBP, and KMT2D. STAT6 mutations are typically associated with the EZB subtype, and our patient data further asserts that STAT6 mutant rrDLBCL is of the EZB subtype (Fig. S1). Overexpression of STAT6WT or STAT6D419A/G/H/N had no effect on DLBCL growth, as compared to empty vector controls (Figs. 1B, S2A). In addition, expression of STAT6D419 mutation alone had no impact on cellular response to any of the individual components of R-CHOP therapy (Fig. S2C). Together, these results indicate that STAT6D419 mutant cells do not have enhanced proliferation or increased therapeutic resistance in the absence of cytokine stimulation.

Fig. 1

STAT6D419 cells have similar response to IL-4 as STAT6WT cells. A Schematic demonstrating the generation of STAT6 mutant DLBCL cell lines. B Growth curve of OCI-Ly8-pLX304 cell lines, expressing empty vector, STAT6WT, or STAT6D419A/G/H/N. Cells were counted once every 24 h for 96 h using a haemocytometer and trypan blue exclusion. Data shown are 4 biological replicates, consisting of 3 technical replicates each. C Representative western blot of 3 biological replicates demonstrating OCI-Ly8 sensitivity to increasing concentrations of IL-4 for 30 min. (S.E: Short exposure; L.E: Long exposure). D Representative western blot of 3 biological replicates demonstrating OCI-Ly8 time course of 200 pg/mL IL-4 stimulation. E Representative western blot of 3 biological replicates demonstrating OCI-Ly8 200 pg/mL IL-4 wash off time course. F Schematic illustrating the OCI-Ly8 and 293-EV/IL-4 co-culture system. G Western blot demonstrating that co-culture with 293-IL-4 induces sustained STAT6 phosphorylation. H Growth curve of OCI-Ly8 cell lines co-cultured with 293-EV or 293-IL-4 for 72 h. Data shown are 4 biological replicates, consisting of 2 technical replicates each. In all cases, bar graphs show densitometry of phospho-STAT6 expression, normalized to GAPDH or β-actin loading control expression

We observed enriched phospho-STAT6 staining in DLBCL patient biopsies with a STAT6D419 mutation [9], suggesting that these mutations lead to a constitutively active STAT6, altered kinetics of response to cytokine, or increased cytokine signaling from the microenvironment. To test the first two of these possibilities, phospho-STAT6 was assessed with increasing concentrations of IL-4. STAT6 was not phosphorylated in the absence of IL-4 (Fig. 1C), and each cell line had similar STAT6 phosphorylation in response to different concentrations of IL-4 (Fig. 1C). We next assessed whether STAT6D419N mutation changed the kinetics of STAT6 phosphorylation. IL-4-induced phospho-STAT6 expression peaked between 30 min and 1 h, and gradually declined to background levels by 24 h in all cell lines (Figs. 1D, S2B). Furthermore, when cells were treated with IL-4 for 3 h and media was removed, the loss of phospho-STAT6 was similar between cell lines (Fig. 1E). These findings indicate that STAT6D419N is not constitutively active nor is STAT6D419N phosphorylation altered in response to IL-4. Of note, we found that exogenous expression of STAT6WT and STAT6D419N had no impact on total STAT6 expression levels in OCI-Ly8 and SU-DHL-4 cells (Figs. 1C, S2), suggesting that there is a buffering mechanism that keeps total STAT6 protein levels static.

Our data demonstrate that cytokine stimulation is required for mutant STAT6 phosphorylation. To test whether D419 mutation alters cell growth in the presence of sustained cytokine signaling, we developed a co-culture model wherein DLBCL cells are cultured in a transwell above a stromal cell line that constitutively produces IL-4 (293-IL-4; Fig. 1F), leading to sustained STAT6 phosphorylation (Fig. 1G). Using this co-culture system, OCI-Ly8, DB, and SU-DHL-4 expressing STAT6WT or STAT6D419A/G/H/N cells have a similar proliferative rate with and without IL-4 induction (Figs. 1H, S2D-F). Thus, IL-4 does not enhance the proliferative rate of STAT6D419 cells in vitro.

STATD419N can heterodimerize with STAT6WT.

In the STAT6 canonical signaling cascade, STAT6 dimerizes upon phosphorylation, and translocates to the nucleus [15]. This dimerization step is critical within the STAT6 activation cascade; thus, we questioned whether D419N mutation impacts the ability of STAT6 to form dimers and whether STAT6D419N could dimerize with STAT6WT. To test the ability of STAT6WT and STAT6D419N to homo- and heterodimerize, we used co-immunoprecipitation (co-IP) in HEK293ft cells transiently expressing tagged STAT6WT or STAT6D419N capable of increasing phospho-STAT6 in response to IL-4 (Fig. S3). We transfected Flag-tagged and V5-tagged STAT6WT or STAT6D419N and stimulated cells for 1 h with IL-4. By immunoprecipitating with anti-Flag and probing for V5, and the reciprocal, we found that STAT6WT and STAT6D419N can both homodimerize and heterodimerize (Fig. 2A). Interestingly, STAT6WT-STAT6D419N heterodimers and STAT6D419N homodimers were immunoprecipitated from nuclear extracts more readily upon IL-4 stimulation, as compared to STAT6WT homodimers (Fig. 2B). The latter suggests an increased nuclear localization of STAT6 dimers when one or both monomers contain the D419N mutation.

Fig. 2

Increased presence of STAT6D419N hetero- and homodimers in the nucleus upon IL-4 stimulation. A Representative western blot of 3 biological replicates from co-IP experiments. Lysates used for co-IP were from nuclear fractions. Labelling indicates transfection conditions. “Input” immunoblots are from whole cell extracts, to confirm equal transfection conditions. B Densitometry of Flag IP. Band intensity was normalized to input control, and then represented as fold change in intensity from the STAT6WT homodimerization -IL-4 condition. C Representative western blot of 3 biological replicates from OCI-Ly8-EV, OCI-Ly8-STAT6WT, and OCI-Ly8-STAT6D419N cellular fractions stimulated with 200 pg/mL IL-4 over 6 h. The nuclear fraction shows phospho-STAT6 nuclear accumulation with D419N mutation. D Densitometry of nuclear phospho-STAT6 expression, normalized to H1 loading control expression and expressed as fold change from the EV without IL-4. E, F Representative western blots of 3 biological replicates from cytoplasmic and nuclear cell fractions of transfected 293-EV cells stimulated with IL-4 for 3 h (acute stimulation; E) and transfected 293-IL-4 cells (chronic stimulation; F). Labelling indicates transfection conditions. Bar graphs show densitometry of p-STAT6 expression, normalized to GAPDH or H3 loading control. Bars in the graph are presented corresponding to western blot loading. Statistics on all bar graphs are 2-way ANOVA with multiple comparison (*p < 0.05, ** = p < 0.01, *** = p < 0.005, **** = p < 0.001)

STAT6D419N displays increased nuclear translocation.

Next, we assessed STAT6D419N phosphorylation kinetics in cellular fractions following IL-4 stimulation in OCI-Ly8 cell lines. Cells were stimulated with IL-4 for 1, 3, and 6 h, and phospho-STAT6 expression was assessed in whole cell, cytoplasmic, and nuclear protein extracts. In whole cell extracts, phospho-STAT6 induction by IL-4 was similar between STAT6WT and STAT6D419N, consistent with our initial findings (Figs. 1D, 2C). Likewise, within the cytoplasmic fraction, phospho-STAT6 was similar between STAT6WT and STAT6D419 mutant cells. However, within the nuclear fraction, STAT6D419N displayed increased phosphorylation in response to IL-4 (Fig. 2C–D). Similar results were obtained with SU-DHL-4 cells (Fig. S4A) and when STAT6D419N mutant OCI-Ly8 cells were stimulated with IL-13 (Fig. S4B). We also confirmed HEK293ft cells transiently transfected with STAT6D419N have increased nuclear expression both upon acute (3 h; Fig. 2E) and chronic (72 h; Fig. 2F) exposure to IL-4. Thus, STAT6D419 mutants have increased nuclear expression following cytokine activation, concordant with patient data showing strong nuclear phospho-STAT6 staining [9].

STAT6D419N upregulates a more stringent set of genes with increased magnitude

We next hypothesized that the increased nuclear expression of STAT6D419N might lead to altered transcription factor activity and thereby, altered gene expression. To assess differential gene expression signatures, we extracted RNA from OCI-Ly8 cells expressing either STAT6WT or STAT6D419N grown in co-culture with 293-EV or 293-IL-4 cells for 3 h (acute stimulation) or 72 h (chronic stimulation) and performed RNA sequencing. In the absence of IL-4, gene expression between STAT6WT and STAT6D419N was similar and clustered by exposure time rather than genotype by hierarchical clustering (Fig. 3A). Only 4 genes were differentially expressed in the absence of IL-4 between STAT6WT and STAT6D419N cells, including PRKCB, IL1RAPL1, ZNF492, and RN7SL396P. We compared the IL-4-dependent differentially expressed genes (DEGs) between STAT6WT and STAT6D419N cells and identified only 84 DEGs after acute IL-4 treatment and 2,185 DEGs after 72 h (Fig. 3B). Of note, the DEGs which distinguished STAT6WT and STAT6D419N cells following 72 h IL-4 treatment were often upregulated in STAT6D419N as compared to STAT6WT (Fig. 3C). However, we also found that IL-4 induced DEGs are often shared by both cell lines (35% in acute and 42% in chronic exposure; Fig. 3B), and include the STAT6 transcriptional targets FCER2, IL4R, AMICA1, SOCS1, and DPP4. Using qPCR, we validated the expression of these genes, as well as some novel genes found to be upregulated by IL-4 induction, such as NR4A3, LTBP1, and MOB3C. Each of these genes were upregulated in STAT6WT and STAT6D419N cells upon chronic IL-4 stimulation, and gene expression of SOCS1, DPP4, and NR4A3 was induced more in STAT6D419N cells than in STAT6WT cells (Fig. 3D). Furthermore, expression of each of these genes depends on sustained IL-4 signaling, as transcript levels decline when IL-4 is removed from cell culture media, leading to pre-stimulation levels by 6–12 h post-IL-4 wash off (Fig. S5).

Fig. 3

STAT6D419N upregulates a restricted set of genes with increased magnitude. A Dendrogram demonstrating that OCI-Ly8-STAT6WT and OCI-Ly8-STAT6D419N groups cluster by IL-4 treatment time. B Venn diagram showing the number of DEGs upon 3 h and 72 h IL-4 treatment. Upon IL-4 treatment, STAT6WT has more unique DEGs than STAT6D419N. C Heatmap showing gene expression differences between STAT6WT and STAT6D419N cells, upon 72 h IL-4 treatment. Blue or red intensity represents Z-score. D qPCR validations of RNAseq data demonstrate that STAT6D419N upregulates transcription of gene targets with increased magnitude. Data consist of 4 biological replicates, with 3 technical replicates each (2way ANOVA; *p < 0.05, ** = p < 0.01, *** = p < 0.005, **** = p < 0.001)

Interestingly, there were more DEGs exclusive to STAT6WT than STAT6D419N cells at both time points. After chronic IL-4 exposure, there were 5 × more DEGs selective for STATWT (1062 in WT vs 205 in D419N; Fig. 3B). These results were surprising, given that STAT6D419N demonstrated enhanced nuclear localization. Together, our RNA sequencing and qPCR data demonstrate that STAT6D419N upregulates the expression of a more selective gene set, but with increased magnitude, upon IL-4 stimulation.

STAT6D419 mutation does not uniformly impact DNA-binding dynamics.

The D419 residue locates along the major groove of the protein-DNA-binding interface, and it has therefore been previously predicted that mutation of the D419 residue may influence DNA binding by STAT6 [10, 13]. Thus, we next questioned how D419 mutation of STAT6 might alter DNA interactions. When MD simulations were used to model the molecular structure of endogenous DNA-bound STAT6WT (Figs. 4A, S6A) and STAT6D419N (Fig. 4B), no remarkable differences in the backbone geometry were observed. Likewise, MD were also used to describe STAT6 DNA binding by monitoring the evolution of three distances: first, the distance between both SH2 domains; second, the distance between D419 residue and DNA centres of mass (chain A); and third, the distance with DNA chain B. No major differences were found for STAT6WT and STAT6D419N systems (Figs. 4C, S6B, C). Once bound, STAT6WT and STAT6D419N showed similar ability to unravel the DNA structure, as illustrated by the predicted number of hydrogen bonds between base pairs and the base paired ratio (Figs. 4D, S6D). Furthermore, umbrella sampling simulations were used to model STAT6 release from the DNA. Starting from the stable DNA-bound conformation, the DNA structure was pulled out of the STAT6 structure, by increasing the distance from SH2-SH2 and DNA from 30 to 80 Å, until the DNA was released from the protein. These simulations found that STAT6 released from DNA around 50 Å, and D419N mutation did not induce a change in the free energy of binding (Figs. 4E, S6E). While STAT6D419N showed no remarkable differences in DNA-binding dynamics from STAT6WT, STAT6D419A and STAT6D419G did have significantly changed DNA binding affinity (Fig. S6), with STAT6D419A and STAT6D419G showing greater ability to unravel the DNA secondary structure, and STAT6D419G additionally showing an increase in the free energy required for DNA unbinding. These findings indicate that D419 mutation does not uniformly change STAT6 DNA interactions, and that increased DNA-binding affinity does not necessarily underlie the increased nuclear presence of STAT6 mutants or the increased transcription of STAT6 target genes.

Fig. 4

STAT6D419N recognizes an altered DNA-consensus motif. A, B Ribbon diagrams showing the DNA-bound conformation of STAT6WT and STAT6D419N. C Molecular dynamic simulations showing STAT6 DNA-binding dynamics. SH2-SH2 shows the distance between SH2 domains measured between their centre of mass, A-DNA shows the distance between one of the D419 residues and the DNA centre of mass (chain A, resID 291), and B-DNA shows the distance between the other D419 residue and the DNA centre of mass (chain B, resID 814), during the process of STAT6WT or STAT6D419N binding to DNA. D Evolution of the number of DNA intermolecular hydrogen bonds and the paired bases ratio along the MD simulation. When purple and orange lines jump from one limit to the other, it denotes a higher ratio of frames without H-bonds and a lower BP-ratio. E Graphical representation of the umbrella sampling study, and Potential of Mean Force (PMF) obtained after umbrella sampling simulations. The distance between SH2-SH2 and DNA was considered as the collective variable under study. Fifty independent simulations were conducted for each STAT6 DNA complex to explore all the molecular states when changing the collective variable from 30 to 80 Å. The overlap in density plots along the coordinate assure the thermodynamic reliability of the study. PMF approximates the free energy landscape when following a coordinate of interest (a.k.a. collective variable). On the right, the PMF involving the DNA unbinding process is shown. F Venn diagram showing the number of identified STAT6-binding sequences identified within proximity of genes upregulated by STAT6WT or STAT6D419N upon 3 h IL-4 treatment. G STAT6WT and STAT6D419N novel consensus motifs, as identified with MEME using STAT6 DNA-binding sequences obtained from GTRD

STAT6D419N recognizes a restricted DNA-consensus sequence.

STAT6D419N regulates the expression of a more stringent set of genes than STAT6WT, so we next questioned whether there was a difference in DNA-binding motifs recognized by STAT6WT and STAT6D419N. Genes upregulated upon 3 h of IL-4 stimulation are likely to be primary targets of STAT6 transcription factor activity, thus, this dataset was used for this analysis. ChIP-seq data compiled by GTRD [20] were used to identify STAT6 DNA-binding sequences upstream and downstream of genes upregulated in STAT6WT only (72 sequences), STAT6D419N only (23 sequences), and in both STAT6WT and STAT6D419N (58 sequences); thus 130 STAT6-binding sequences were associated with STAT6WT, and 81 STAT6-binding sequences were associated with STAT6D419N (Fig. 4F). Within these STAT6-binding sequences, the canonical palindromic STAT6 5′-TTC(N)3…4GAA-3′ consensus motif [32] was identified in 10% of STAT6WT-associated sequences (13/130; Evalue = 9.29e−1), and in 8.6% of STAT6WT-associated sequences (7/81; Evalue = 6.14e−1), serving as an internal validation of this methodology, but also suggesting the canonical STAT6-binding consensus motif is not significantly enriched within our dataset. However, a novel DNA-consensus sequence motif was identified within our dataset (Fig. 4G) at 70 sites with an E value of 1.5e−014 in STAT6WT-associated binding sequences and at 59 sites with an E value of 5.9e−005 in STAT6D419N-associated binding sequences. Interestingly, this binding motif is restricted to a guanine residue at positions 3 and 6 in sequences regulated by STAT6D419N (G-NN-G motif), while the residues at positions 3 and 6 are more variable in sequences regulated by STAT6WT. These results suggest that the binding of STAT6D419N to DNA is restricted to a more specific sequence than STAT6WT.

STAT6D419 mutants upregulates CCL17 expression to recruit CD4 + T-Cells

To further understand the DEGs selective to STAT6WT or STAT6D419N, we performed pathway analyses of the genes significantly altered in STAT6WT or STAT6D419N OCI-Ly8 cells upon chronic IL-4 induction. We found that STAT6D419N cells have increased activation of pathways that have genes related to cell viability and survival, and less activation of pathways that have genes related to organismal death and cell morbidity/mortality as compared to STAT6WT, suggesting that STAT6D419N cells have improved survival in response to IL-4. Notably, we found that expression of STAT6D419A/G/H/N does not confer a proliferative advantage in GCB-DBCL cell lines, although these cell lines are culture-adapted and highly proliferative, potentially masking a mutation-induced phenotype. STAT6D419N cells also have increased activation in pathways related to chemotaxis and cell migration, including cell movement and migration of mononuclear cells and lymphocytes (Fig. 5A; Table 1). Together, these data indicate that STAT6D419N cells have a survival advantage in response to IL-4 stimulation and may have increased capability to attract or migrate to specific immune cells that could be IL-4 secreting.

Fig. 5

STAT6D419 mutation results in increased CCL17 secretion and increased invasion of CD4 T-Cells. A IPA analysis of RNAseq data showing STAT6D419N cells have enrichment in pathways related to Cell Viability and Cell Migration and Chemotaxis. B qPCR showing CCL17 transcription is increased in STAT6D419N cells upon IL-4 stimulation. Data consist of 4 biological replicates in technical triplicate (2-way ANOVA; ****p < 0.001). C ELISA showing CCL17 secretion is increased in STAT6D419N cells compared to STAT6WT cells upon IL-4 stimulation. Data consists of 3 biological replicates in technical triplicate (2-way ANOVA; **p < 0.01). D CCL17-reporter luciferase assay showing CCL17 transcription is increased 30-fold when STAT6D419N is expressed in 293-IL-4 cells. (RLU: relative luminescence units). E CCL17 expression obtained from gene microarray in GCB rrDLBCL patients who had negative or positive phospho-STAT6 staining (two-tailed unpaired t test; *p < 0.05). Stars indicate the presence of a STAT6D419 mutation. F CCL17 expression obtained from a compendium of RNA sequencing of 598 de novo DLBCL patients and 1 transformed lymphoma patient, comparing CCL17 expression in GCB STAT6WT patients with CCL17 expression in STAT6 non-D419 mutant patients (STAT6Mut) and STAT6D419 mutant patients (1-way ANOVA; **p < 0.01). Of the STAT6Mut patients, 4 were GCB and 2 were ABC. Of the STAT6D419 mutant patients, 9 were GCB and 1 was unclassified G–I CD3, CD4, and CD8 staining in GCB rrDLBCL patient biopsies. CD3, CD4, or CD8 staining intensity score was compared between phospho-STAT6 negative and phospho-STAT6 positive patients (scored by a blinded pathologist; 1 = negative staining, 2 = weak staining, 3 = moderate staining, 4 = strong staining). Scale bars below each image indicate 100 µm. Stars indicate the presence of a STAT6D419 mutation. J CCL17 expression obtained from gene microarray in GCB rrDLBCL patients who had low CD4 staining (score 1–2), or high CD4 staining (score 3–4; unpaired t test; ***p < 0.005). Stars indicate the presence of a STAT6D419 mutation

Table 1 IPA analysis of OCI-Ly8-STAT6WT and OCI-Ly8-STAT6D419N cells under chronic IL-4 stimulation

STAT6D419 mutant rrDLBCL patient biopsies have increased phospho-STAT6 staining and an increased STAT6 expression signature [9], suggesting that STAT6D419-mutants are able to recruit IL-4 secreting cells to maintain an activated state. This aligns with the pathway analyses showing enrichment in chemotactic signals. One interesting candidate from our dataset and a known STAT6 transcriptional target is CCL17 (aka TARC), a chemokine that can attract various CD4+ T-cells, such as Th2 cells and Tregs. CCL17 is upregulated in human FL samples that express mutated STAT6 [10, 13], but this has not been shown in STAT6 mutant DLBCL cell lines. To this end, qPCR determined that IL-4-induced CCL17 mRNA expression is significantly upregulated by STAT6D419A/G/H/N as compared to control, in OCI-Ly8, SU-DHL-4, and DB cells (Figs. 5B, S7A-C). In addition, CCL17 was more highly secreted by STAT6D419A/G/H/N than STAT6WT cells following chronic IL-4 induction (Figs. 5C, S7D-F). Using a CCL17-reporter luciferase assay in 293-EV and 293-IL-4 cells, we confirmed the upregulation of CCL17 by STAT6D419N was indeed transcriptional (Fig. 5D).

We assessed whether this observation could be translated to clinical samples, using our previously published gene microarray data [9] from CD19 + cells selected from rrDLBCL biopsies that were also annotated for phospho-STAT6 expression, indicating STAT6 activation by IL-4. Patients were stratified into “phospho-STAT6 positive” and “phospho-STAT6 negative” groups, with the “phospho-STAT6 positive” group enriched in STAT6D419 mutations, but also including mutations in other members of the STAT6 signaling axis (i.e., IL4R, SOCS1, and JAK2). This allowed us to expand our analyses to all rrDLBCL samples with the same phenotype. We found that phospho-STAT6 positive rrDLBCL tumors had increased CCL17 mRNA expression (Fig. 5E). In addition, using an independent cohort from BC Cancer, where phospho-STAT6 histology data was not available, CCL17 expression was compared in STAT6WT and STAT6D419 mutant patient samples. Although these samples were obtained from de novo and not relapsed/refractory patients, CCL17 was still found to be significantly increased when a D419 mutation was present (Fig. 5F). Furthermore, in this dataset, CCL17 expression was not found to be increased in patients with a non-D419 STAT6 mutation, indicating a unique role for the mutated D419 residue.

Given the role of secreted CCL17 as a chemoattractant, we hypothesized that the surrounding microenvironment of phospho-STAT6 positive tumors might be enriched in T-cells. Using serial sections from rrDLBCL biopsies, we assessed the presence of CD3, CD4, and CD8 antigens with IHC, and found that phospho-STAT6 positive rrDLBCL tumors have a significantly higher proportion of CD3+ T-cells (Fig. 5G). Half of phospho-STAT6 positive tumors had a high percentage (25–50%) of CD4+ T-cells, and of the 3 phospho-STAT6+/CD4High samples, 2 had a STAT6D419 mutation (Fig. 5H). Moreover, even the phospho-STAT6 negative samples with a STAT6D419 mutation were found to have high infiltration of CD4+ T-cells. Interestingly, there were no significant differences in the presence of CD8+ T-cells when samples were stratified by phospho-STAT6 status (Fig. 5I). However, when we stratified the data by STAT6 mutation, CD8+ T-cells were significantly increased in STAT6 mutant biopsies, in addition to the CD4+ T-cells (Fig. S8). Finally, to establish a link between CCL17 in rrDLBCL tumor cells and CD4+ T-cell infiltration, CCL17 mRNA expression was compared between CD4High and CD4Low samples, and there was a striking increase in tumor cell CCL17 mRNA expression in samples with CD4High staining (Fig. 5J). Together, these results indicate that STAT6 mutant rrDLBCL cells remodel their tumor microenvironment (TME) via the secretion of the chemokine CCL17 to attract CD4+ T-cells, although the subtype of CD4+ T-cells which were recruited were not determined in this study.