We have previously demonstrated that a thymus-dependent multiorgan autoimmune disease occurs in HIS mice generated by intravenous injection of human fetal liver (FL) CD34⁺ cells into NSG mice and that this disease develops more rapidly in mice containing a native murine thymus (termed ‘Mu/Hu mice’ [murine thymus/human CD34⁺ cells]) than in thymectomized (Khosravi-Maharlooei et al., 2020) NSG mice receiving a human thymus graft, termed ‘Hu/Hu mice’ (human thymus/human CD34⁺ cells) (Khosravi-Maharlooei et al., 2021). While Mu/Hu mice rely on the native mouse thymus for human T cell development, T cells in Hu/Hu mice develop in the human fetal thymus graft. At 20 weeks after transplantation, animals in both groups were sacrificed and their splenocytes were CFSE-labeled and tested for reactivity to autologous FL-derived human dendritic cells (DCs) and for responses to NSG mouse bone marrow-derived DCs (anti-host response) as well as responses to autologous human DCs pulsed with apoptotic NSG mouse DCs (indirect anti-host response). As shown in Figure 1A and B, T cells from Hu/Hu and Mu/Hu mice showed proliferation to autologous human DCs that was not significantly augmented by pulsing of human DCs with murine antigens. Since the responder splenocyte preparations were not depleted of murine cells, we cannot distinguish whether the baseline proliferation represents responses to human antigens or to murine antigens presented by these DCs. However, proliferation of both CD4 and CD8 T cells to NSG mouse DCs was greater in Mu/Hu mice than in three of four Hu/Hu mice, which showed little, if any, direct proliferation to murine antigens. Proliferation of Mu/Hu CD4⁺ T cells to murine DCs was significantly greater than proliferation in the absence of DCs, in contrast to Hu/Hu CD4 cells, suggesting a lack of tolerance to NSG antigens only in Mu/Hu CD4⁺ T cells (Figure 1B). This observation is consistent with the lack of cortico-medullary structure or normal negative selection in the native murine thymi (Khosravi-Maharlooei et al., 2021).

Since increases in Tfh and Tph have been associated with various human autoimmune diseases, we compared Tfh-like and Tph-like cells in groups of Mu/Hu (n=8) and Hu/Hu mice (n=11) generated from the same two FL donors (Figure 2A). CD3⁺ T cells were detected in peripheral blood around 12 weeks after transplantation (data not shown), as we previously reported (Kalscheuer et al., 2012; Khosravi-Maharlooei et al., 2021; Khosravi-Maharlooei et al., 2020).

We monitored PD-1⁺CXCR5^- Tph-like and PD-1⁺CXCR5⁺ Tfh-like CD4⁺CD45RA^- T cell reconstitution from 12 to 32 weeks post-transplantation (Figure 2B–D). Circulating Tph-like cells were detected at 13 weeks post-transplantation in both groups. The frequency and absolute numbers of Tph-like cells in blood of Mu/Hu mice tended to be greater than those in Hu/Hu mice and the difference in percentages achieved statistical significance across multiple time points (Figure 2C). Circulating Tfh-like cells appeared at 13 weeks post-transplantation in Mu/Hu mice (Figure 2D), increased progressively over time and were significantly more abundant in this group than in Hu/Hu mice (Figure 2D).

In additional experiments, we compared splenic Tfh-like and Tph-like cells in Mu/Hu (n=19) and Hu/Hu mice (n=29) (n=5 different FL donors) sacrificed before 20 weeks, between 20 and 30 weeks, and more than 30 weeks post-CD34⁺ FL transplantation (Figure 2E). While no significant differences in Tph-like cell frequency and absolute numbers were detected between the two groups (Figure 2F), percentages of Tfh-like cells were significantly greater in spleens of Mu/Hu compared to Hu/Hu mice (Figure 2G). The numbers of both Tph-like and Tfh-like cells increased over time in both HIS mouse groups and staining for Ki67 revealed high proliferative rates for both cell types (Figure 2—figure supplement 1).

These results suggest that Tph and Tfh differentiate and expand over time and do so more rapidly in Mu/Hu than Hu/Hu mice. These data correlate with the more rapid onset of autoimmune disease in Mu/Hu than Hu/Hu mice, consistent with a role for these T cell types in disease development.

To further characterize Tph-like and Tfh-like cells in HIS mice, we measured ICOS expression and IL-21 production (Figure 2H and I). These cells expressed higher levels of ICOS (Figure 2H) and, upon phorbol myristic acid (PMA)/ionomycin stimulation, secreted more IL-21 (Figure 2I) compared to control, CD45RA⁺ T cells, consistent with helper function and with their role in humans. The majority of Tfh-like and Tph-like cells in Mu/Hu and Hu/Hu mice expressed ICOS, though the percentage of ICOS⁺ cells was significantly higher in Mu/Hu Tfh-like and Tph-like cells compared to their Hu/Hu counterparts (Figure 2J).

We also compared the levels of FOXP3 within splenic Tfh-like and Tph-like populations in Hu/Hu and Mu/Hu mice. Approximately 10% of Tfh-like and Tph-like cells in both Hu/Hu and Mu/Hu groups were FOXP3⁺, with the highest proportion observed in Hu/Hu Tph-like cells (Figure 2K).

To determine whether or not the increased proportion of Tfh-like cells in Mu/Hu HIS mice impacted B cell differentiation and survival, as reported in humans (Weinstein et al., 2012), we monitored total B cell reconstitution and serum IgM and IgG levels over time. While overall B cell reconstitution (Figure 3A) and serum IgM concentration was similar between Mu/Hu and Hu/Hu mice, serum IgG levels were significantly greater in Mu/Hu mice (Figure 3B), suggesting increased B cell differentiation and IgG class switching.

Next, to compare the functional capacity of Tfh-like and Tph-like cells from HIS mice with mouse or human thymus, we measured the ability of sorted CD25^-CXCR5⁺ CD45RA^- (including Tfh and excluding regulatory follicular cells) and CD25^- CXCR5^-CD45RA^- (including Tph cells and excluding Tregs) CD4⁺ cells to induce differentiation of CD19⁺CD38^-IgD⁺CD27^- naïve B cells to CD20^-CD38⁺ plasmablasts in vitro (Figure 3C). After isolation, these T cell fractions were incubated with autologous naïve B cells. Helper cells from both Mu/Hu and Hu/Hu mice induced plasmablast formation and this activity was greater at later (>20 weeks) than earlier (<20 weeks) times post-transplant. However, Tfh-like cells from Hu/Hu mice demonstrated greater B cell helper function than those from Mu/Hu mice, particularly in the later time period. A similar trend was seen for Tph-like cells (Figure 3D), indicating that T-B cell interactions are more effective for T cells generated in a human thymus that is isogenic to the B cells than for those generated in a xenogeneic murine thymus, and consistent with previous literature (Lang et al., 2013; Danner et al., 2011; Suzuki et al., 2012).

Because B cells undergo class switching in germinal centers (GC), whose activity can be estimated by plasma CXCL13 levels, we analyzed serum concentrations of human CXCL13 in HIS mice. While human CXCL13 was detected in both groups and did not correlate with the number of splenic Tfh or Tph cells (data not shown), the levels tended to be higher in Mu/Hu compared to Hu/Hu mice (Figure 3E).

We performed histological and immunostaining analyses of spleens to quantify B cell follicles in tissue sections from Mu/Hu (n=8) and Hu/Hu (n=11) mice in three time periods (<20 weeks, 20–30 weeks, and >30 weeks post-transplantation). Tissue sections were stained for CD3 (T cells), CD20 (B cells), and peanut agglutinin (PNA, marking GC-B cells) (Figure 4A). B cell follicles were manually quantified, and their size (in pixels per area) was determined using ImageJ software.

Hematoxylin and eosin (H&E) staining revealed a semiorganized splenic structure, with distinct B and T cell localization and well-defined follicles up to 30 weeks post-transplant in both groups. However, after 30 weeks, spleens showed fewer B cell follicles and appeared disorganized (Figure 4B and C). Three-color immunofluorescence staining on spleens from four Hu/Hu and three Mu/Hu mice at <20 weeks post-transplant revealed distinct B cell areas (CD20⁺), with PNA⁺ cells concentrated within B cell zones but also distributed in other splenic regions. Analysis of spleens from four Hu/Hu and three Mu/Hu mice at 20–30 weeks post-transplant revealed smaller B cell zones compared to earlier time points, and PNA⁺ cells were dispersed throughout the spleen rather than being enriched in B cell areas. Only two Mu/Hu mice showed clear B cell zones with localized PNA⁺ areas. Beyond 30 weeks post-transplant, spleens from two Hu/Hu and two Mu/Hu mice revealed no distinct B cell zones, and PNA⁺ cells were diffusely distributed.

Despite the increased number of Tfh-like cells in the spleens of Mu/Hu mice, there were no significant differences in the number or total area of splenic follicles between Mu/Hu and Hu/Hu mice (Figure 4B).

To investigate the possible role of autoantibodies in driving autoimmune disease in HIS mice, we tested serum IgG and IgM for autoreactivity (Figure 5). Figure 5B confirms, in a separate experiment from that in Figure 3, a progressive increase in total serum IgM levels in both groups over time, while Figure 5D shows high levels of IgG antibodies in sera of Mu/Hu mice already by 20 weeks, when Hu/Hu mice still showed very low IgG levels. As shown in Figure 5A and Table 1, serum from Mu/Hu and Hu/Hu mice contained IgM antibodies that were reactive to multiple murine tissues. Several Mu/Hu mice also contained IgG antibodies with broad reactivity to multiple murine tissues. IgM in sera from HIS mice was reactive to LPS, insulin, and dsDNA, and these levels increased over time irrespective of the thymus type (Figure 5C). IgG against these antigens also increased over time in both groups (Figure 5E), even though total IgG levels tended to be higher in Mu/Hu than Hu/Hu mice (Figure 5D). To assess the possibility that IgM autoantibodies might be polyreactive, we compared the sum of IgM concentrations reacting to dsDNA, insulin, and LPS to total serum IgM concentrations and observed that the sum of these reactivities was >100% for three Mu/Hu mice (Figure 5F), consistent with polyreactivity of individual B cells in these mice.

To link the B cell types in HIS mice with autoantibody production, we performed phenotypic analysis of B cells. These studies revealed increased proportions of activated or age-associated B cells (IgD⁻CD27⁺CD11c⁺) and atypical B cells (IgD⁻CD27⁻CD11c⁺) in Mu/Hu mice compared to those from Hu/Hu mice (Figure 5G).

To assess the possible role of B cells in the development of autoimmunity in HIS mice, we treated Hu/Hu mice with rituximab (anti-CD20) once every 3 weeks from 20 to 38 weeks post-transplantation (Figure 6A). As shown in Figure 6B, this treatment successfully depleted B cells from the circulation and eliminated serum IgM. Immunofluorescent analysis of serum IgM from rituximab-treated mice showed an absence of autoreactive IgM (Figure 6C). However, B cell depletion with this method did not prevent or even delay disease development (Figure 6D), suggesting a minor, if any, role for antibodies in disease progression. However, since it remained possible that serum antibody was critical in presenting antigen to T cells and/or initiating inflammation that caused later disease, we also evaluated the impact of rituximab treatment early post-transplant. As shown in Figure 6E, starting rituximab treatment 1 week post-transplant and continuing through the post-transplant course also failed to delay or reduce the development of disease and, surprisingly, significantly accelerated its development. From these results, we concluded that antibodies were not required for disease development.

To further clarify the role of Tfh- and Tph-like cells in the development of autoimmunity, we tested their ability to expand, differentiate, promote B cell differentiation, and induce disease in a T cell adoptive transfer model. We first generated a cohort of Mu/Hu and Hu/Hu mice from the same human FL CD34 cell donor. Twenty-two weeks later, after T cells had reconstituted the spleen, we euthanized both groups of HIS mice and FACS-sorted splenic CD4⁺CD45RA^- CD45RO⁺PD-1⁺CXCR5⁺ Tfh-like and CD4⁺CD45RA^-CD45RO⁺PD-1⁺CXCR5^- Tph-like cells. These T cells were adoptively transferred intravenously into thymectomized NSG mouse recipients that had received FL CD34⁺ cells 12 weeks earlier from the same human donor. Since these latter recipients lacked an endogenous (murine) thymus and had not received a thymus graft, they reconstituted only with B cells and myeloid APCs derived from the fetal HSC donor and did not generate T cells, as we have previously reported (Khosravi-Maharlooei et al., 2021; Khosravi-Maharlooei et al., 2020). Thus, there were five different recipient groups: (a) no T cell transfer (APC only, n=3), (b) mice that received Tph-enriched (n=5) or (c) Tfh-enriched cells (n=3) from Hu/Hu mice, and mice that received (d) Tph-enriched (n=5) and (e) Tfh-enriched cells (n=3) from Mu/Hu mice (Figure 7A). Figure 7B illustrates the gating strategy used to isolate Tfh and Tph cells from donor spleens.

As shown in Figure 7C and D, both Tfh-like and Tph-like cells originating in Hu/Hu mice expanded faster and to a greater extent than those originating in Mu/Hu mice, achieving higher circulating levels. Recipients of Tfh-like cells from Hu/Hu mice contained high numbers of circulating Tph-like and much lower numbers of Tfh-like cells (Figure 7E and F), suggesting that Tfh-like cells might have lost CXCR5 expression or that Tph-like contaminants expanded preferentially. Recipients of Tph-like cells from Hu/Hu mice also contained low numbers of Tfh-like cells and much higher fractions of Tph-like cells, whose numbers and percentages were similar to those seen in recipients of Hu/Hu Tfh-like cells (Figure 7F). Recipients of Mu/Hu Tfh-like or Tph-like cells contained much lower numbers of human T cells of both types (Figure 7D–F), but T cells in the group receiving Mu/Hu Tfh-like cells also demonstrated a predominant Tph-like phenotype (Figure 7F). Similar trends were seen in spleens at the time of sacrifice, though differences did not achieve statistical significance (Figure 7—figure supplement 1, top row).

These differences in Tfh-like and Tph-like cell expansion from Hu/Hu vs Mu/Hu donor mice led us to compare B cell differentiation in the adoptive recipients. Proportions of B cells among human CD45⁺ cells (Figure 8A) reflected the degree of dilution due to expansion of transferred T cells noted above (Figure 7D) and overall circulating B cell counts were similar between groups (Figure 8A). No significant difference in proportions of naïve (IgD), memory (CD27), and IgG class-switched B cells was observed between the four groups and the APC-only HIS mouse control group, although there was a trend toward increased IgG⁺ B cells in adoptive recipients of T cells at later time points (Figure 8B).

By 5 weeks following adoptive transfer, the blood of recipients of Tfh and Tph cells from Hu/Hu mice exhibited increased levels of CD11c⁺ B cells, an effector B cell population associated with aging, infection, and human autoimmune diseases, compared to recipients of T cells from Mu/Hu mice (Figure 8C). This difference largely reflected an increase in the percentage of atypical B cells (IgD⁻CD27⁻CD11c⁺, DN2) in the recipients of Hu/Hu T cells. Analysis of splenic B cell populations showed similar trends (Figure 7—figure supplement 1). Both activated or age-associated B cells (IgD⁻CD27⁺CD11c⁺) and atypical B cells (IgD⁻CD27⁻CD11c⁺) were significantly increased in the spleens of recipients of Hu/Hu Tph and Tfh cells compared to recipients of Mu/Hu Tph and Tfh cells 9 weeks following adoptive transfer (Figure 8C). Splenic total and naïve B cell percentages were not statistically different between groups (Figure 7—figure supplement 1). However, spleens of recipients of Tfh-like and Tph-like cells from Hu/Hu mice contained higher percentages of memory B cells and of CD11c⁺ B cells than those of Mu/Hu T cell recipients (Figure 7—figure supplement 1, bottom row).

We also analyzed the levels of IgM and IgG from these four groups of recipient HIS mice and APC-only controls. Recipients of Tfh-like or Tph-like cells from Hu/Hu mice had higher levels of IgM and IgG antibodies compared to recipients of Tfh-like or Tph-like cells from Mu/Hu mice and the APC-only controls contained IgM at low levels and undetectable levels of IgG (Figure 8D). Collectively, these results demonstrate a role for Tfh-like and Tph-like cells in inducing both IgM and IgG antibodies in HIS mice and show that T cells from Hu/Hu mice provide help for autologous B cell antibody responses more effectively than those from Mu/Hu mice.

Adoptive recipients of Tfh-like and Tph-like cells from Hu/Hu mice lost weight beginning at 7 and 9 weeks, respectively, whereas the recipients of Mu/Hu T cells did not show obvious weight loss (Figure 8E). Clinical disease was also more apparent in recipients of Hu/Hu T cells, with the most rapid onset in recipients of Hu/Hu Tfh-like cells. In contrast, most recipients of Mu/Hu T cells did not develop clinically apparent disease during the follow-up period (Figure 8F). Together, our data suggest that T cells developing in an autologous human thymus graft interact more effectively with peripheral B cells and other APCs, presumably reflecting thymic positive selection on the same HLA molecules as those in the peripheral APCs, resulting in T cell expansion, increased CD11c expression on B cells, and eventually disease development.

NSG (NOD.Cg-Prkdcscid Il2rgtm1Wjl) mice (Strain #:005557, RRID:IMSR_JAX:005557) were purchased from Jackson Laboratory (Bar Harbor, ME, USA) and were housed and bred in specific pathogen-free helicobacter and Pasteurella pneumotropica-free conditions in the Animal Facility at Columbia University Medical Center. Human fetal thymus and FL tissues (gestational age 17–20 weeks) were obtained from Advanced Biosciences Resources. Fetal thymus fragments were cryopreserved in 10% dimethyl sulfoxide (Sigma-Aldrich) and 90% human AB serum (Gemini Bio Products). FL fragments were treated for 20 min at 37°C with 100 μg/ml of Liberase (Roche) to obtain a cell suspension. Human CD34⁺ cells were isolated from FL by density gradient centrifugation (Histopaque-1077, Sigma) followed by positive immunomagnetic selection using anti-human CD34 microbeads (Miltenyi Biotec) according to the manufacturer’s instructions. Cells were then cryopreserved in liquid nitrogen. Studies were approved by the Animal Care and Use Committee at Columbia University under the IACUC protocol AC-AABM5551. All human samples were collected with approval of the Institutional Review Board of Columbia University Medical Center in accordance with the Declaration of Helsinki.

The definition of autoimmune-like disease after humanization with HSCs or following adoptive T cell transfer to T cell-deficient HIS mice is based on a GVHD scoring system described in our earlier publication. This system evaluates five criteria: percentage of weight loss, posture, activity level, fur condition, and skin condition, with each parameter assigned a grade from 0 to 2 (Table 2). For the disease-free survival curve (Figures 6 and 8F), a total score exceeding 3 or >20% weight loss was classified as indicative of disease.

To generate HIS mice with human thymus (Hu/Hu mice), 6- to 8-week-old NSG mice were thymectomized and allowed to recover for 2 weeks, then sublethally irradiated (1.0–1.2 Gy) using an RS-2000 X ray irradiator (Rad Source Technologies, Inc, Suwanee, GA, USA). Human fetal thymus fragments of about 1 mm³ were implanted beneath the kidney capsule and 2×10⁵ human FL CD34⁺ cells were injected IV. Recipient mice received intraperitoneal injections of 400 μg of anti-human CD2 monoclonal antibody BTI-322 or Lo-CD2 26 (Bio X Cell, Inc) on days 0, 7, and 14 post-transplantation. HIS mice with intact mouse thymus (Mu/Hu mice) were generated simultaneously and received similar treatment without thymectomy or thymus transplantation.

Human reconstitution, Tfh cells, Tph cells, and B cells were monitored in PBMC from 8 weeks after transplantation and in spleens of HIS mice until endpoint. Single-cell splenocyte suspensions and blood samples were treated with ACK lysis buffer (Life Technologies) to remove erythrocytes. After ACK erythrocyte lysis, remaining spleen cells were passed through a 70 μm filter prior to staining for FCM analysis. All cells were stained for detection of surface ICOS (ISA-3), CD45RA (HI100), CCR7 (G043H7), PD-1 (EH12.2H7), CXCR5 (RF8B2), CD8 (RPA-T8), CD25 (2A3), CD3 (SK7), CD4 (RPA-T4), CXCR3 (1C6), CD127 (A019D5), CCR2 (1D9), CCR5 (2D7), CD20 (L27), CD19 (HIB19), IgM (G20-127), IgG (G18-145), IgA (IS11-8E10), IgD (IA6-2), CD27 (O323), CD38 (HIT2), CD138 (MI15), CD11c (B-ly6), CD14 (M5E2), CD21 (Bu32), hCD45 (HI30), and mCD45 (30-F11). To detect IL-21, cells were stimulated for 3 hr with 50 ng/ml phorbol myristic acid (PMA; Sigma-Aldrich, St Louis, MO, USA), 1 μg/ml ionomycin (Sigma-Aldrich) and 3 μg/ml brefeldin A in RPMI supplemented with 10% FBS, 1% HEPES (Sigma-Aldrich), 1% penicillin-streptomycin (Life Technologies), and 0.05% gentamicin (Life Technologies) in 37°C, 5% CO₂ incubator conditions. For intracellular staining, single-cell suspensions were fixed-permeabilized with the FOXP3 Fixation Kit (Invitrogen) for 45 min at RT and then stained with mAbs for IL-21 (3A3-N2.1) for 30 min at RT in permeabilization buffer. After washing, cells were acquired on an Aurora Cytek (Fremont, CA, USA) or on a FACS FORTESSA (BD Biosciences), and analyzed with FlowJo v10 (Tree Star, USA) software.

Mu/Hu and Hu/Hu mice were sacrificed 20 weeks after transplantation and their splenocytes (not purified, so multiple murine and human cell populations were present) were CFSE-labeled and tested for reactivity to various antigen-presenting cells. To test direct reactivity to autologous human DCs, FL CD34⁺ cells used to generate both Hu/Hu and Mu/Hu mice were differentiated into DCs. Briefly, HSCs were cultured in Aim-V medium containing 10% human serum and stem cell factor (SCF, 50 ng/ml) and granulocyte-macrophage colony-stimulating factor (GM-CSF, 50 ng/ml) for 3 days. After 3 days, additional GM-CSF (100 ng/ml) and SCF (50 ng/ml) and low dose TNF-α (1 ng/ml) were added to the culture. At day 7, cells were washed and cultured in medium containing GM-CSF (100 ng/ml), IL-4 (20 ng/ml), and TNF-α (2 ng/ml). One week later, differentiated DCs were harvested and either activated with prostaglandin E2 (PG-E2, 3000 ng/ml) and TNF-α (20 ng/ml) or incubated with apoptotic (irradiated at 30 Gy) NSG mouse DCs for 12 hr for indirect presentation of mouse antigens, followed by activation with PG-E2 and TNF-α. To generate NSG mouse DCs, mouse bone marrow cells were cultured in AIM-V media containing 10% FBS and murine IL-4 (5 µg/ml) and GM-CSF (3 µg/ml). After 1 week, mouse DCs were harvested and either activated with LPS (1 μg/ml) for 8 hr for direct presentation or irradiated (30 Gy) for co-culture with human DCs (for indirect presentation of mouse antigens on human DCs).

Splenocytes from Hu/Hu and Mu/Hu mice were stained with CFSE and cultured with various DCs (Auto human DCs, NSG DCs, Auto DCs loaded with mouse antigens, or no DCs) at a 10:1 ratio of splenocytes:DCs (100K CFSE-stained splenocytes and 10K DCs) in U-bottom 96-well plates. After 6 days, cells were stained with antibodies against human CD3, CD4, and CD8 and analyzed with a flow cytometer.

Single-cell splenocyte suspensions were stained with a panel to detect CD4 (RPA-T4), CXCR5 (RF8B2), CD45RA (HI100), CD19 (HIB19), IgD (IA6-2), CD27 (O323), and CD38 (HIT2) and prepared for sorting on a BD FACSAria Fusion II. Cells were first gated based on single cells and lymphocytes, and then Tfh cells (CD4⁺CD45RA^-CXCR5⁺CD25^-), Tph cells (CD4⁺CD45RA^-CXCR5^-CD25^-), and naive B cells (CD19⁺CD38^-CD27^-IgD⁺) were FACS sorted. Cells were collected in RPMI 1640 (Life Technologies) supplemented with 10% FBS (Life Technologies). After collection, 20×10³ Tfh cells or Tph cells and naïve B cells (CD19⁺CD38^-CD27^-IgD⁺) were incubated in the presence of staphylococcal enterotoxin B (0.1 μg/ml) in a 1:1 ratio (T cells:B cells). Cultures were performed in V-shaped 96-well plates in RPMI 1640 (Life Technologies) supplemented with 10% FBS, 1% HEPES (Sigma-Aldrich), 1% penicillin-streptomycin (Life Technologies), and 0.05% gentamicin (Life Technologies) in 37°C, 5% CO₂ incubator conditions. After 7 days, cell pellets were harvested and stained to analyze plasmablast differentiation (CD19⁺CD20^-CD38⁺).

Whole peripheral blood (in heparin) was centrifuged at 1000 rpm for 15 min to collect plasma. Plasma was further centrifuged at 13,000 rpm for 10 min to remove debris. CXCL13 was evaluated in EDTA plasma by ELISA assay (Human CXCL13/BLC/BCA-1 Quantikine ELISA Kit, R&D Systems) following the manufacturer’s instructions.

To quantify human antibodies in sera of HIS mice, diluted samples were added to plates (Corning Inc, Corning, NY, USA) coated with goat anti-human IgG Fcγ fragment (Jackson) or goat anti-human IgM (Southern Biotech) overnight at 4°C. Plates were then washed in PBS with 0.05% Tween 20, and blocked with 2% Bovine Serum Albumin (BSA, Fisher Scientific). Bound human Ig was detected using biotin-conjugated mouse anti-human IgG (BD Pharmingen) or biotin-conjugated mouse anti-human IgM (BD Pharmingen) secondary antibodies, followed by streptavidin-horseradish peroxidase (HRP) (Thermo Scientific). Colorimetric change by 3,3′,5,5′-tetramethylbenzidine substrate solution (Thermo Scientific) was stopped by 2 M sulfuric acid (Sigma), and optical densities determined by spectrophotometer absorbance at 450 nm. Human serum with known IgM and IgG concentrations (Bethyl) was used to establish standard curves.

To determine the reactivity of serum antibodies to insulin, dsDNA, and LPS, polystyrene plates (Corning) were coated overnight at 4°C with 10 μg/ml LPS, 10 μg/ml dsDNA, or 10 μg/ml recombinant human insulin (Sigma-Aldrich, St. Louis, MO, USA). After five washes in PBS with 0.05% Tween 20, serum was added and incubated for 2 hr at room temperature. Plates were then washed in PBS with 0.05% Tween 20, incubated for 1 hr with either HRP-conjugated goat anti-human IgM or IgG (Invitrogen, Camarillo, CA, USA), washed again, and developed using 3,3′,5,5′-tetramethylbenzidine (eBioscience, San Diego, CA, USA). Optical density was read at 450 nm absorbance. Supernatants from polyreactive IgM and IgG B cell cultures generated as described (Porcheray et al., 2012) and provided by Dr. Emmanuel Zorn (Columbia University) were used to establish standard curves.

For postmortem histopathology, spleen from HIS mice were excised, fixed in zinc-formalin (Sigma), and embedded in paraffin. Serial 5 µm sections were stained with H&E and images were acquired using Aperio AT2 (Leica Biosystems, Wetzlar, Germany).

Spleens from HIS mice were fixed with formalin and embedded in paraffin. FFPE tissues were cut using a Leica microtome at 5 μm. Cuts were done sequentially, placed on slides and left to dry for 24 hr prior to antibody staining. Tissues were deparaffinized and rehydrated by warming at 60°C for 1 hr followed by a serial bath of xylene and ethanol dilutions (100%, 95%, 80%, 70%, water). Antigen retrieval was performed using Borg RTU de-cloaking solution and a pressure cooker. The pressure cooker was set for 15 min at 11°C. Antigen retrieval was followed by permeabilization with confocal buffer for 1 hr. Titrated primary antibodies were added and left at 40°C overnight. Tissues were washed three times in 1× PBS and blocked with mouse and/or goat serum for 1 hr. Then, anti-human CD3 Alexa Fluor 488 (UCHT1, BioLegend, 1:50 dilution), anti-human CD20 Alexa 647 (Abcam, EPR1622Y), and peanut agglutinin (PNA/GC) antibodies were incubated for 1 hr at room temperature followed by three washes, then incubation with secondary antibodies for 2 hr. Tissues were washed again in 1× PBS followed by staining with nuclear marker DAPI and Fluoromount G mounting media.

For detection of mouse-reactive human IgM and IgG antibodies, sera from HIS mice were incubated on pancreas, liver, spleen, kidney, thymus, small intestine, large intestine, skin, bone, and lung sections taken from a naïve NSG mouse. After incubation for 2 hr at room temperature, serum was washed away and secondary anti-human IgM and IgG antibodies were added for detection for 2 hr at room temperature. Negative controls were incubated with either no serum or naïve NSG mouse serum followed by secondary antibodies.

A group of Mu/Hu mice were generated (Figure 6A) with irradiation (1 Gy) followed by injection with FL CD34⁺ HSCs. Starting at week 20 post-transplantation, mice were injected with rituximab (1 mg diluted in 1 ml of PBS, injected intraperitoneally) every 3 weeks until week 38. Blood was drawn at various time points before and after rituximab injection to measure human immune cell reconstitution in the blood and IgM levels in the serum. Mice were followed and evaluated for development of autoimmunity.

In another experiment (Figure 6E), Hu/Hu mice generated as described above received intraperitoneal injections of rituximab (1 mg/mouse/injection) starting from week 1 and every 3 weeks until week 31. Serum IgM was checked with ELISA and mice were followed and evaluated for the development of autoimmunity.

Donor Mu/Hu and Hu/Hu mice were sacrificed at 22 weeks and CD4⁺CD45RA^-CD45RO⁺PD-1⁺CXCR5⁺ Tfh and CD4⁺CD45RA^-CD45RO⁺PD-1⁺CXCR5^- Tph cells were FACS sorted from their pooled (for each donor type) splenocyte suspensions. 2.5×10⁵ Tfh cells or Tph cells were adoptively transferred intravenously into recipient mice. The recipients were thymectomized NSG mice reconstituted 12 weeks earlier with the same HSCs as the Tfh/Tph donors, without a thymus transplant, which therefore contained autologous human APCs but no T cells. Adoptive recipient mice were monitored for disease and peripheral blood was collected from the tail vein once a week. After 9 weeks, adoptive recipient mice were euthanized. Spleens and blood were collected and stained for Tfh, Tph, and B cell analysis. Plasma was also collected for IgM and IgG evaluation.

Statistical analyses and comparisons were performed with GraphPad Prism 8.0 (GraphPad Software). All data are expressed as average ± standard error of mean. The nonparametric Mann-Whitney U test was used to compare two groups. One-way analysis of variance (ANOVA) with post hoc Tukey test was performed to compare three groups. Two-way ANOVA was used to resolve overall effects between transplant groups over time. When two-way ANOVA was significant, Bonferroni’s multiple comparison test was run. Spearman’s correlation coefficient was computed to study the strength of correlation between quantitative variables. Wilcoxon matched pairs signed rank test was used when analyzing paired groups. Mantel-Cox tests were used to measure significance of differences in Kaplan-Meier survival curves. p<0.05 was considered to be statistically significant.