IL-21 signaling promotes the establishment of KSHV infection in human tonsil lymphocytes by increasing early targeting of plasma cells

Factors influencing Kaposi’s sarcoma-associated herpesvirus (KSHV) transmission and the early stages of KSHV infection in the human immune system remain poorly characterized. KSHV is known to extensively manipulate the host immune system and the cytokine milieu, and cytokines are known to influence the progression of KSHV-associated diseases. Here, using our unique model of KSHV infection in tonsil lymphocytes, we investigate the influence of host cytokines on the establishment of KSHV infection in human B cells. Our data demonstrate that KSHV manipulates the host cytokine microenvironment during early infection and susceptibility is generally associated with downregulation of multiple cytokines. However, we show that IL-21 signaling promotes KSHV infection by promoting both plasma cell numbers and increasing KSHV infection in plasma cells as early as 3 days post-infection. Our data reveal that this phenotype is dependent upon a specific milieu of T cells, that includes IL-21 producing Th17, Tc17 and CD8+ central memory T cells. These results suggest that IL-21 plays a significant role in the early stages of KSHV infection in the human immune system and that specific immunological states favor the initial establishment of KSHV infection by increasing infection in plasma cells.

syndrome (KICS) [5]. KSHV is linked to 1% of all human tumors, and the World Health 23 Organization (WHO) has classified it as class I carcinogen [6,7]. KSHV infection is asymptomatic 24 in most healthy individuals, and KSHV-associated malignancies arise primarily in 25 immunocompromised patients. Indeed, KS remains one of the most common cancers in people 26 living with HIV/AIDS [8]. 27 The geographical distribution of KSHV is not ubiquitous. KSHV infection is endemic in sub-28 Saharan Africa and in the Mediterranean basin. KSHV prevalence is also high in subpopulations 29 in other parts of the world such as men who have sex with men (MSM). Saliva is the only secretion 30 where KSHV DNA is commonly detected [9], and, based on this, person-to-person transmission 31 of KSHV is thought to occur via saliva. The oral lymphoid tissues are rich in KSHV target cell 32 types including lymphatic endothelial cells and B cells, and are therefore a likely site for the initial 33 establishment of KSHV infection in a new human host. However, the exact mechanisms for KSHV 34 transmission and how environmental, behavioral and host factors influence transmission and early 35 infection events remain to be definitively established. This gap in our understanding dramatically 36 affects our ability to find efficient strategies to decrease the transmission or influence host-level 37 susceptibility to KSHV infection. We previously analyzed susceptibility to KSHV infection in a 38 cohort of human tonsil samples with diverse race, sex and age distributions and found that these 39 samples displayed high variability in susceptibility that could not be linked to demographic factors 40 [10]. Our ongoing research seeks to identify, and mechanistically characterize, host-level 41 susceptibility factors that influence this variable susceptibility. It is important to note that, in this 42 context, it is the highly susceptible and highly refractory "outlier" specimens that may ultimately 43 prove the most informative in identifying these critical susceptibility factors. 44 diamonds and right y-axis scale). However, many of the most susceptible samples included in this 113 dataset, display lower levels of multiple cytokines in KSHV-infected

IL-21 supplementation increases KSHV infection in tonsil B lymphocytes
Because IL-21 production was positively correlated with plasma cell infection in our initial dataset 135 ( Fig 1E), we wanted to examine the impact of manipulating IL-21 levels on the establishment of 136 KSHV infection. To do this, we performed Mock infection or KSHV infection in 12 unique tonsil 137 samples and supplemented the resulting cultures with varying concentrations of recombinant IL-138 21. At 3 dpi, we analyzed these cultures for GFP+ B lymphocytes by flow cytometry to assess the 139 magnitude of KSHV infection (Fig 2A & B). Although the specimens included in this data set had 140 high variability in their baseline susceptibility, we can see increased infection in response to IL-141 21 treatment, and the effect seems to be particularly strong in the more susceptible samples (Fig 142 2A). Normalization of the data to each specimen's untreated control reveals that at 10/12 samples 143 show increased infection upon treatment with 100pg/ml of IL-21. Importantly, most of these 144 concentrations were higher than what was observed by 3 dpi in our initial dataset quantitating 145 native cytokine secretion in our culture system (Fig 1A), which may explain why we didn't observe 146 an association of IL-21 secretion with overall infection at that timepoint ( Fig 1D). We then 147 repeated these supplementation experiments with only the 100pg/ml dose of recombinant IL-21 in 148 an additional 12 tonsil specimens and examined both overall infection and subset-specific 149 responses in these cultures at 3 dpi using our B cell immunophenotyping panel (Table 1 and  150 Supplemental Fig 1A). Similar to the initial dataset, this analysis shows increased infection in 151 response to recombinant IL-21 in the majority of tonsils, and the difference in GFP+ B 152 lymphocytes was statistically significant in IL-21 treatment compared to control (p=0.02, F=6.4) 153 ( Fig 2C). This increase in infection was not associated with alterations in the frequency of viable 154 B cells in the cultures (Supplemental Table 1B). 155

IL-21 increases plasma cell frequency and susceptibility in primary human tonsil B 156 lymphocytes
To examine whether IL-21 treatment altered the B cell subset-specific distribution of KSHV 158 infection in these experiments, we quantitated the percent of each B cell subset that was GFP+ to 159 determine the within-subsets distribution of KSHV infection in control or IL-21 treated cultures. 160 One-way repeated measures ANOVA analysis indicates that IL-21 supplementation did not 161 significantly affect KSHV infection for most subsets (Supplemental Table 1A). However, we 162 observed a significant increase in plasma cell targeting with IL-21 treatment (p=0.02, F=6.6) and 163 notable, but non-significant, increases in the infection of germinal center and plasmablast subsets 164 ( Fig 2D). We next examined whether IL-21 supplementation was associated with alterations in B 165 cell frequencies in either Mock or KSHV-infected cultures. Two-way repeated measures ANOVA 166 analysis (Supplemental Table 1B) revealed a highly significant increase in total plasma cell 167 frequency associated with both IL-21 treatment and KSHV infection with a significant interaction 168 of the two variables ( Fig 2E). Neither infection nor treatment had a significant effect on germinal 169 center cell frequencies, but there was a significant main effect of KSHV infection on frequencies 170 of plasmablasts in these cultures (p=0.03, F=6). Post hoc paired T tests revealed significant 171 differences with IL-21 treatment on total plasma cells (p=0.0002) for the KSHV-infected 172 conditions only, and in IL-21 treated conditions there was a significant difference between Mock 173 and KSHV cultures for total plasma cells (p=0.0003). In order to determine whether the increase 174 in total plasma cell frequency and/or increased infection of plasma cells was directly correlated 175 with the effect of IL-21 on total KSHV infection, we performed linear model regressions and 176 analyzed the results using Pearson's method ( Fig 2F). These results reveal a significant linear 177 correlation between total GFP and plasma cell frequency (r=0.7, p=0.007) and a weaker, but still 178 significant, correlation between total GFP and the frequency of GFP+ cells within the plasma cell 179 subset (r=0.56, p=0.04). Taken together this data shows that IL-21 treatment promotes the establishment of KSHV infection in human tonsil lymphocytes, and that this increased infection 181 is correlated with both increased plasma cell frequencies and increased plasma cell infection at 3 182 dpi. 183

Neutralization of IL-21 inhibits KSHV infection in primary tonsil B lymphocytes 184
We next wanted to determine whether neutralization of the natively-secreted IL-21 in our tonsil 185 lymphocyte cultures would affect the establishment of KSHV infection. To do this, we performed 186 infections with Mock or KSHV-infection in 11 unique tonsil specimens, included varying 187 concentrations of an IL-21 neutralizing antibody in the resulting cultures, and assessed the 188 magnitude and distribution of KSHV infection at 3 dpi by flow cytometry. These results reveal 189 decreased KSHV infection in the presence of IL-21 neutralizing antibodies ( Fig 3A).  repeated measures ANOVA revealed a significant effect of IL-21 neutralization on GFP+ cells in 191 KSHV infected cultures (p=0.00001, F=9.4) and post-hoc Dunnett test revealed significance at the 192 100µg/ml dose (p=0.03). When each sample was normalized to its untreated control, we observed 193 that 9/11 samples had decreased infection in the presence of 100µg/ml IL-21 neutralizing antibody 194 and this increased to 10/11 samples at higher antibody doses ( Fig 3B). 195 Interestingly, when we examined whether this decrease in KSHV infection was associated with 196 alterations in infection of any particular B cell subsets by one-way repeated measures ANOVA 197 (Supplemental Table 2A), we only observed a significant decrease in GFP+ transitional B cells 198 (p=0.01,F=4.0). However, there are non-significant trends showing lower frequencies of infection 199 within plasmablast and CD20+ plasma cell subsets in most samples. (Fig 3C). Moreover,there 200 were no KSHV-specific effects of IL-21 neutralization on the total frequency of any B cell subsets 201 in these experiments via two-way repeated measures ANOVA (Supplemental Table 2B). However, total and CD20+ plasma cell frequencies were significantly reduced at the 200µg dose only in 203 mock cultures (Fig 3D). The observation that the effect of IL-21 neutralization on plasma cell 204 frequencies is restricted to mock-infected cultures is interesting in the context of our IL-21 205 supplementation data where we observed significant main effects of both IL-21 and KSHV 206 infection on plasma cell frequencies as well as a significant interaction between the two factors 207 We believe it is the sample-specific variability in responses to IL-21 neutralization that resulted in 216 non-significant effects when the data was parsed to infection within subsets. This variability is not 217 surprising given that magnitude of any effect of neutralization is dependent upon the quantity of 218 IL-21 signaling in the particular culture, which is variable depending upon the sample (Fig. 1A) 219 We hypothesized that if IL-21 signaling is affecting overall infection by contributing to 220 differentiation of KSHV-infected cells, subsets whose differentiation is important to the 221 establishment of infection would accumulate within the GFP+ population with IL-21 222 neutralization, and this accumulation would correlate with decreased overall levels of infection in 223 response to neutralization. Conversely, subsets whose targeting promotes overall infection in an To examine this, we performed RT-PCR for LANA (latent) and K8.1 (lytic) on total RNA from 244 untreated, IL-21 supplemented or IL-21 neutralizing antibody treated, KSHV-infected cultures 245 from 8 unique tonsil specimens. GAPDH was used as a housekeeping gene and normalizing factor 246 for the viral gene expression data. This data is consistent with our previous data showing a mix of 247 lytic and latent transcripts in infected lymphocytes [10]. These data reveal no significant influence of either supplementation or neutralization on lytic gene expression ( Fig 3F). In the majority of 249 samples K8.1 expression remained unchanged or changes were also reflected in LANA transcripts, 250 indicating higher overall infection rather than increased lytic activity (Fig 3G).  Table 3B). However, there were no significant correlations between the MFI of IL-280 21R at baseline and total GFP at 3 dpi for any subset (Supplemental Table 4D). These results 281 suggest that IL-21+ plasmablasts are important for susceptibility to KSHV in the absence of high 282 levels of IL-21 at early timepoints in untreated cultures while naïve B cells contribute to the effect 283 of IL-21 supplementation. 284 Interestingly, there was no positive correlation seen between baseline IL-21R expression and the 285 response of plasma cell frequencies to IL-21 (Supplemental Table 3C), suggesting that the plasma 286 cell response to IL-21 at 3 dpi in KSHV infected cultures may be a product of IL-21 receptor up-287 regulation in response to infection instead of intrinsic baseline levels of IL-21 on plasma cells or 288 plasma cell precursors in our tonsil lymphocyte cultures. Indeed, modulation of IL-21 receptor 289 expression by KSHV infection is one possible mechanism for the synergistic promotion of plasma 290 cell numbers we observe with both IL-21 treatment and infection ( Fig 2D).

IL-21R+ plasmablasts increase in response to KSHV infection and IL-21R+ Plasma cells 292 increase in response to IL-21 only in KSHV+ cultures. 293
In order to examine this hypothesis, we analyzed IL-21R expression on B cell subsets at 3 dpi in 294 our culture system with or without IL-21 supplementation to determine whether KSHV and/or IL-295 21 can modulate the response to IL-21 during infection. There were no statistically significant 296 differences in either the frequency ( Table 4C). However, 310 there was no significant effect of KSHV infection or IL-21 treatment on the fluorescence intensity 311 of IL-21R for any subset (Supplemental Table 4D). Interestingly, there was a significant 312 correlation between the frequency of plasmablasts within IL-21+ and overall infection at 3dpi (Fig  313   5F, left), which was driven by samples where large increases in GFP in response to IL-21 treatment corresponded to large increases in IL-21R+ plasmablasts (Fig 5F, right). This result is particularly 315 interesting taken together with the correlation between baseline plasmablast frequencies and 316 overall infection at 3 dpi in untreated conditions. 317 These results may indicate that infection and IL-21 treatment is affecting IL21R expression on 318 existing plasmablasts and plasma cells, or that KSHV and IL-21 synergistically drive 319 differentiation of IL-21R+ cells to plasmablast and plasma cell phenotypes. Our observation that 320 IL-21R+ naïve B cells at day 0 are correlated with the response of KSHV infection to IL-21 321 treatment ( Fig 4F) is one indication that differentiation may be playing a role in the IL-21 response. 322 However, our data do not exclude the possibility that a combination of both receptor modulation 323 and differentiation are contributing to the observed 3 dpi phenotypes in the presence of both IL-324 21 and KSHV infection. 325

Characterization of T cell subsets producing IL-21 in primary human tonsil B lymphocytes 326
We next wanted to determine the source of native IL-21 secretion in our culture system, and 327 determine whether the production of IL-21 is affected by KSHV infection. To accomplish this, we 328 utilized an additional immunophenotyping panel for T cell subsets (Table 2 and Supplemental Fig  329 1B), and performed intracellular cytokine staining (ICCS) on unstimulated T cells at 3 dpi to 330 identify T cell subsets that are producing IL-21 (Supplemental Fig 1C) in Mock and KHSV-331 infected cultures from 14 unique tonsil samples. This data shows that IL-21 secretion in T cells is 332 highly variable between tonsil lymphocyte cultures (range=1.4-24.7%, mean=10.6, median=10.55, 333 standard deviation=6.9), but is not significantly affected by KSHV infection (Fig 6A). More of the 334 IL-21+ cells were CD4+ T cells vs. CD8+ T cells and this distribution was also not affected by 335 the contribution of T cell subsets to IL-21 secretion. This value is calculated by multiplying the 337 mean fluorescence intensity of IL-21 in each T cell subset by the subset's frequency within total 338 T cells. Thus, iMFI integrates the amount of IL-21 being secreted by a subset with the frequency 339 of that subset; more correctly quantitating the contribution of low frequency subsets that are high 340 IL-21 producers. Unlike the frequency of IL-21+ T cells, the iMFI of IL-21 within CD4+ and 341 CD8+ T cells displayed some overlap (Fig 6C), indicating that, although they are low frequency, 342 CD8+ T cells can be high producers of IL-21. Indeed, there were two samples in our analysis 343 where, based on iMFI, CD4+ and CD8+ T cells were contributing equally to total IL-21 secretion 344 Importantly, KSHV infection did not significantly change the overall levels and subset distribution 352 of T cells within these cultures, indicating these are biological changes within T cell subsets and 353 not due to changes in the T cell population during infection (Supplemental Fig 2A). When we 354 performed the iMFI calculation on the subset level we observed that, as expected based on their 355 established function, CD4+ Tfh had an increased contribution to IL-21 secretion relative to their 356 frequency. CD45RO+ T cells contributed more to IL-21 in CD4+ T cells whereas CD45RA+ T 357 cells contributed more to IL-21 secretion in the CD8+ population. Interestingly, among the CD8+ 358 T cell subsets, RoRT+ and central memory subsets showed increased contribution to IL-21 secretion relative to their frequency indicating that these cells are high IL-21 producers (Fig 6F). 360 Finally, KSHV infection significantly decreased the iMFI of CD4+ CD45RA+ and CD4+ RoRT+ 361 T cell subsets, which is likely related to their significantly decreased frequency (Fig 6E) rather 362 than an effect on the level of IL-21 secretion from the subset. 363   Table 5A) we found that only the frequency of IL-21+ CD8+ central memory cells were 371 significantly correlated (r=0.57, p=0.03) ( Fig 7A) and there were no significant correlations 372 between the iMFI of IL-21 in T cell subsets and overall GFP+ B cells at 3dpi (Supplemental Table  373 5B). When we examined correlations between baseline T cell subsets and infection at 3dpi, the 374 only significant correlation was a negative impact of CD4+ CD45RO+ T cells (r=-0.57, p=0.03) 375 (Supplemental Table 5C). Given that these experiments rely on native IL-21 secretion over time 376 in the culture system, as opposed to high levels of recombinant IL-21 added at day 0 in our previous 377 experiments (Fig 2), we hypothesized that impacts on total GFP may be absent at this timepoint 378 because KSHV targeting and manipulation of plasma cell frequencies in response to IL-21 379 precedes the effect on total infection. Thus, we examined correlations between plasma cell 380 infection in the context of subset-specific IL-21 secretion by T cells. In this analysis, we found that 381 (1) the baseline frequency of CD8+ central memory cells (r=0.59, p=0.02), (2) their frequency 382 within IL-21+ at 3dpi (r=0.82, p=0.0003) and (3)  lymphocytes. In addition, the frequency of both CD4+ and CD8+ T cells that express RoRT+ (the 388 Th17/Tc17-defining transcription factor) within IL-21+ T cells were significantly correlated with 389 plasma cell targeting by KSHV (Fig 7C and Supplemental Figure 2B). These correlations were 390 stronger for CD4+ RoRT+ cells and were coupled with a significant negative correlation with 391 Figure 2B). In this data, we noticed that the same 392 two tonsil specimens were driving the positive correlations between plasma cell targeting by 393 KSHV and IL-21 secretion by CD8+ central memory, CD4+ RoRT+, and CD8+ RoRT+ T cells 394 ( Fig 7B&C). We next correlated the iMFI of these three T cell subsets (CD8+ central memory, 395 CD4+ RoRT+, and CD8+ RoRT+; hereafter referred to as "subsets of interest") with both GFP+ 396 plasma cells and total plasma cells, which were significantly elevated in KSHV-infected conditions 397 in response to IL-21 supplementation (Fig 2E). Consistent with our previous dataset (Fig 2F), the 398 targeting of plasma cells by KSHV is directly correlated with the total frequency of plasma cells 399 ( Fig 7D, top left panels). The iMFI of our T cell subsets of interest was significantly correlated 400 with GFP+ plasma cells but was not correlated to total plasma cell numbers. However, the same 401 two tonsil samples driving the previous correlations did show elevated plasma cell frequencies 402 together with elevated T cell iMFI in this data (Fig 7D, red boxes) while the lack of correlation in 403 this data was driven by samples with high plasma cell frequencies where high KSHV targeting of 404 plasma cells was absent and the iMFI for at least one of the T cell subsets was low. This may 405 indicate that other factors can influence total plasma cell numbers, but both total plasma cells and 406 GFP+ plasma cells are only simultaneously elevated when the IL-21 producing T cells are also 407 present. Moreover, this data indicates that all three IL-21 producing T cells are necessary for the 408 combined plasma cell phenotype. Indeed, we observed direct correlations between the iMFI values 409 of all three T cell subsets of interest, further supporting the conclusion that it is actually the 410 combination of factors rather than independent contributions of each T cell subset that is driving 411 the combined increases in plasma cell frequencies and plasma cell targeting in KSHV-infected 412

KSHV-infection of naïve T cells (Supplemental
conditions. 413 The sample numbers in our dataset were not sufficient for true multivariate analysis with the large 414 number of T cell subsets analyzed. However, multiple pairwise correlations can provide some 415 further insight into this phenomenon. Indeed, we observe highly significant correlations between 416 all three T cell subset based on their frequency at baseline and their frequency at 3dpi (Fig 8A). 417 Interestingly, there were distinctively separate populations of samples in our analysis where the 418 three T cell subsets were either all low or all high, and this distinction was particularly obvious 419 within the 3 dpi frequencies (Fig 8A, right). Similarly, the frequency of these subsets within IL-420 21+ and the iMFI of IL-21 in these subsets are also significantly correlated, but only in the KSHV-421 infected samples (Fig 8B). However, the IL-21 correlations were relatively weak compared to the 422 frequency correlations indicating that additional functions of this T cell milieu that are not directly 423 related to IL-21 secretion may also influence the plasma cell targeting phenotype. Finally, we 424 wanted to determine whether any other T cell subsets correlated with all three subsets, and may be 425 additionally contributing to the T cell milieu which promotes plasma cell targeting by KSHV. 426 Since 3 dpi frequency yielded the strongest correlations within the T cell data (Fig 8A, right), we 427 performed pairwise correlations between our three subsets of interest and the remaining T cell 428 subsets in the analysis (Fig 8C). The data reveals that the frequencies of all three subsets of interest are also significantly correlated with the frequency of CD4+ stem cell memory and samples where 430 BCL6+ cells predominate within the CD4+ Tfh subset. Taken together, these results indicate that 431 there is a defined T cell milieu in some tonsil samples that includes elevated frequencies of IL-21-432 producing CD8+ central memory, CD4+ RoRT+, and CD8+ RoRT+ T cells, as well as other T 433 cell subsets, and this particular milieu correlates with the ability of KSHV to increase plasma cell 434 numbers and target plasma cells for infection at early timepoints. 435

Discussion 436
Our results presented in this study indicate that KSHV can influence cytokine production in tonsil-437 derived lymphocytes and that the host inflammatory state contributes to the dramatic variation in 438 susceptibility we observe among our tonsil lymphocyte specimens However, these results are novel and interesting in that they demonstrate direct correlations 458 between plasma cell frequencies, plasma cell infection, and overall susceptibility to KSHV 459 infection (Fig 2F and 7D). Specifically, the synergistic effect of KSHV infection and IL-21 460 signaling increasing plasma cell frequencies ( Fig 2E)  Both IL-21 and IL-6, which is highly induced in our KSHV-infected cultures (Fig 1) (Fig 7C and  493 Supplemental Fig 2C), suggesting that the Th17/Tc17 environment in the tonsil may be a critical 494 factor influencing donor-specific susceptibility to KSHV infection. Indeed, as an important site for 495 mucosal immunity in the oral cavity, the Th17/Tc17 environment in tonsil is highly dynamic and 496 physiologically important. In fact, Th17 cells play a major role in host defenses against several pathogens and immunopathogenesis [33,34] center cells in these observations (Fig 3E). However, although our ex vivo model of KSHV 503 infection in primary lymphocytes is a powerful tool, it certainly does not recapitulate the complex 504 interactions that are needed for a functional germinal center reaction, so further examination of 505 these particular mechanisms will require the utilization of an alternative model system, such as a 506 humanized mouse. However, the participation of CD8+ central memory, Th17 and Tc17 cells in 507 the T cell milieu associated with increased plasma cell targeting indicates that an extrafollicular 508 pathway may also be active in this process, which is consistent with literature implicating 509 extrafollicular maturation of KSHV-infected B cells in the pathogenesis of MCD [38]. 510 Although CD8+ T cells are minor contributors to IL-21 secretion compared to CD4+ T cells, our 511 data strongly indicates that they participate in the inflammatory milieu that promotes KSHV 512 dissemination in our model (Fig 7 and Fig 8). Previous studies have shown that CD8+ T cells can 513 be observed in B lymphocyte areas of tonsil and provide co-stimulatory signals and cytokines to 514 support B cell survival [39]. Interestingly, recent studies have shown that IL-6 regulates IL-21 515 production in CD8+ T cells in a STAT3-dependent manner, and that CD8+ T cells induced in this 516 way can effectively provide help to B cells [40]. Thus, the induction of human IL-6 during KSHV 517 infection may modulate the function of CD8+ T cells in a way that favors the establishment and 518 dissemination of KSHV infection within the lymphocyte compartment independent of traditional 519 CD4+ helper T cells, which would be an interesting dynamic in the context of CD4+ T cell 520 immunosuppression associated with HIV infection where KSHV-mediated malignancies are 521 common. 522

Material and Methods 523
Ethics Statement. Human specimens used in this research were de-identified prior to receipt, and 524 thus were not subject to IRB review as human subjects research. 525 Reagents and Cell Lines. CDw32 L cells (CRL-10680) were obtained from ATCC and were 526 cultured in DMEM supplemented with 20% FBS (Sigma Aldrich) and Penicililin/Streptomycin/L-527 glutamine (PSG/Corning). For preparation of feeder cells CDw32 L cells were trypsinized and 528 resuspended in 15 ml of media in a petri dish and irradiated with 45 Gy of X-ray radiation using a 529 Rad-Source (RS200) irradiator. Irradiated cells were then counted and cyropreserved until needed 530 for experiments. Cell-free KSHV.219 virus derived from iSLK cells [39] was a gift from Javier G. 531 Ogembo (City of Hope). Human tonsil specimens were obtained from the National Disease 532 Research Interchange (NDRI; ndriresource.org). Human fibroblasts for viral titering were derived 533 from primary human tonsil tissue and immortalized using HPV E6/E7 lentivirus derived from 534 PA317 LXSN 16E6E7 cells (ATCC CRL-2203). Antibodies for flow cytometry were from BD 535 Biosciences and Biolegend and are detailed below. Recombinant human IL-21 was from 536 Preprotech (200-21) and IL-21 neutralizing antibody was from R&D Systems (991-R2). 537

Isolation of primary lymphocytes from human tonsils. De-identified human tonsil specimens 538
were obtained after routine tonsillectomy by NDRI and shipped overnight on wet ice in 539 DMEM+PSG. All specimens were received in the laboratory less than 24 hours post-surgery and 540 were kept at 4˚C throughout the collection and transportation process. Lymphocytes were 541 extracted by dissection and maceration of the tissue in RPMI media. Lymphocyte-containing media was passed through a 40µm filter and pelleted at 1500rpm for 5 minutes. RBC were lysed 543 for 5 minutes in sterile RBC lysing solution (0.15M ammonium chloride, 10mM potassium 544 bicarbonate, 0.1M EDTA). After dilution to 50ml with PBS, lymphocytes were counted, and 545 pelleted. Aliquots of 5(10)7 to 1(10)8 cells were resuspended in 1ml of freezing media containing 546 90% FBS and 10% DMSO and cryopreserved until needed for experiments. 547 Infection of primary lymphocytes with KSHV. Lymphocytes were thawed rapidly at 37˚C, diluted 548 dropwise to 5ml with RPMI and pelleted. Pellets were resuspended in 1ml 549 RPMI+20%FBS+100µg/ml DNaseI+ Primocin 100µg/ml and allowed to recover in a low-binding 550 24 well plate for 2 hours at 37˚C, 5% CO2. After recovery, total lymphocytes were counted and 551 naïve B cells were isolated using Mojosort Naïve B cell isolation beads (Biolegend 480068) or 552 Naïve B cell Isolation Kit II (Miltenyi 130-091-150) according to manufacturer instructions. 553 Bound cells (non-naïve B and other lymphocytes) were retained and kept at 37˚C in RPMI+20% 554 FBS+ Primocin 100µg/ml during the initial infection process. 1(10) 6 Isolated naïve B cells were 555 infected with iSLK-derived KSHV.219 (dose equivalent to the ID20 at 3dpi on human fibroblasts) 556 or Mock infected in 400ul of total of virus + serum free RPMI in 12x75mm round bottom tubes 557 via spinoculation at 1000rpm for 30 minutes at 4˚C followed by incubation at 37˚C for an 558 additional 30 minutes. Following infection, cells were plated on irradiated CDW32 feeder cells in 559 a 48 well plate, reserved bound cell fractions were added back to the infected cell cultures, and 560 FBS and Primocin (Invivogen) were added to final concentrations of 20% and 100µg/ml, 561 respectively and recombinant cytokines or neutralizing antibodies were also added at this stage, 562 depending upon the specific experiment. Cultures were incubated at 37˚C, 5% CO2 for the 563 duration of the experiment. At 3 days post-infection, cells were harvested for analysis by flow cytometry and supernatants were harvested, clarified by centrifugation for 15 minutes at 15,000 565 rpm to remove cellular debris, and stored at -80˚C for analysis. 566 Bead-based immunoassay for supernatant cytokines. Clarified supernatants were thawed on ice 567 and 25µl of each was added to a 13-plex LEGENDplex (Biolegend) bead-based immunoassay 568 containing capture beads for the following analytes: IL-5, IL-13, IL-2, IL-9, IL-10, IL17A, IL-569 17F, IL-6, IL-21, IL-22, IL-4, TNF-, and IFN-. These assays were performed according to the 570 manufacturer's instructions, data was acquired for 5000 beads per sample (based on approximately 571 300 beads per analyte recommended by the manufacturer) using a BD FACS VERSE flow 572 cytometry analyzer and cytokine concentrations in the experimental supernatants was calculated 573 from standard curves using the LEGENDPlex software. 574   unique tonsil specimens and can be compared between these panels and red diamonds indicate the 695 mean value for all tonsil specimens. RT-PCR analysis of KSHV transcripts at 3 dpi in 8 unique 696 tonsil specimens with either IL-21 supplementation at 100 pg/ml (Fig 2) or IL-21 neutralizing quantities and means (red diamonds) and bottom panels show trends of increase/decrease 721 comparing GFP+ to GFP-within the same culture. (E) Distribution of IL-21 receptor on B cell 722 subsets at day 0 (baseline) or 3dpi within Mock, Mock+100pg/ml IL-21, KSHV or KSHV+ 723 100pg/ml IL-21 conditions. Red diamonds indicate the mean values for each condition and 724 significant differences were assessed via two-way repeated measures ANOVA (Supplemental 725 Table 4A) and post-hoc paired T-tests for both culture/infection conditions (Supplemental Table  726 4B) and IL-21 treatment (Supplemental Table 4C). (F) Pearson correlation analysis of the 727 frequency of plasmablasts within IL-21+ and overall infection at 3dpi (left). GFP response (y=axis) 728 and plasmablast response x-axis to IL-21 treatment for each sample (right). 729 For (A) and (B) ***p<0.001, **p<0.01, *p<0.05 and grey shading indicates 95% confidence intervals (C) correlogram of pairwise correlations between 3 dpi frequencies of the T cell subsets