SDF-1 dynamically mediates megakaryocyte niche occupancy and thrombopoiesis at steady state and following radiation injury

Platelet-producing megakaryocytes (MKs) are derived from megakaryocyte progenitors (MKPs), which are defined functionally by their capacity to form colonies in vitro.^1,2  MKPs are thought to reside near the bone surface in an “endosteal niche,” where environmental cues encourage expansion, but suppress terminal maturation.^3-6  Polyploid MKs mature cytoplasmically, extrude proplatelets in association with sinusoidal vasculature, and shed platelets into the peripheral blood.^7-9  This process results in past-maturity “exhausted” MKs comprised of a nucleus with a thin layer of cytoplasm surrounded by a cell membrane.^10,11  Megakaryopoiesis is primarily regulated by the cytokine thrombopoietin (TPO), which signals through its receptor Mpl to promote MKP proliferation and MK maturation.^12-14 

Although the physical association of MKs with sinusoidal vasculature was first appreciated several decades ago,^15-17  the functional significance of the “vascular niche” for MK maturation and thrombopoiesis has only more recently begun to be elucidated.^4,18-21  Several studies have implicated the chemokine stromal cell-derived factor-1 ([SDF-1] or CXCL12) signaling through receptor CXCR4 in the maturational localization of MKs to the vascular niche. CXCR4 is expressed throughout the MK lineage, and in vitro SDF-1 stimulation results in intracellular calcium mobilization, matrix metalloproteinase 9 expression, surface CXCR4 polarization, and ultimately migration along an SDF-1 gradient.^22-27  Several cell types within the bone marrow produce SDF-1, including osteoblasts, endothelial cells, and perivascular mesenchymal stromal cell populations.^28-32  Additionally, in vivo studies demonstrate that sustained elevation of SDF-1 in the blood increases thrombopoiesis, with images indicating enhanced MK association with vasculature.^20,33  Recently, VEGF-A treatment was shown to increase MK-vasculature interactions in vivo through upregulation of CXCR4 on MKs.³⁴  Despite this growing body of evidence indicating a role for SDF-1/CXCR4 in megakaryopoiesis, the acute and endogenous effects of SDF-1 on MK localization and platelet production remain unknown.

Radiation causes significant damage to the hematopoietic system, and thrombocytopenia can be a life-threatening consequence of radiation exposure.³⁵  Seminal studies in rodents identified an initial persistence of polyploid MKs after both sublethal and lethal doses of total body irradiation (TBI).^36-39  Accordingly, radioresistant MKs persist within an injured marrow microenvironment established by differential radiation damage to surrounding cell populations.^40-43  The sinusoidal vasculature, which supports platelet production, can dilate as surrounding cells succumb to damage and become leaky at higher levels of injury.^44,45  Intriguingly, SDF-1 levels increase in the marrow after TBI and radioresistant MKs relocate to the endosteal niche.^39,46,47  It is not known whether the spatial changes of MKs in the setting of TBI are regulated by microenvironmental changes in SDF-1 or have functional consequences for platelet production.

Here, we investigate the ability of SDF-1 to acutely control MK location in the bone marrow both at steady state and after radiation injury. Importantly, we demonstrate that endogenous SDF-1 regulates the association of MKs with vasculature, which contributes to platelet production in uninjured bone marrow. After sublethal TBI, fluctuations in marrow SDF-1 dynamically affect the distribution of MKs. Our data give insight into the spatial regulation of megakaryopoiesis and demonstrate that changes in MK location have functional consequences for thrombopoiesis.

Seven- to 9-week-old female C57BL/6J mice (Jackson Laboratory) were used for all experiments. Unanesthetized mice, confined in a Plexiglass restraint, were exposed to 4 Gy TBI at a dose rate of 1.6 Gy/min using a Shephard Irradiator with 6000 Ci ¹³⁷Cs source and collimating equipment. All animal experiments were approved by the University of Rochester Committee on Animal Resources.

Recombinant murine SDF-1a (400 ng; Peprotech) in 100 μL phosphate-buffered saline (PBS) was injected IV into unirradiated mice, or mice at day 2 or day 4 post-TBI. Diprotin A ([ILE-PRO-ILE]; 1.7mg; Enzo) in 250 μL PBS was injected intraperitoneally into unirradiated mice, or mice at day 2 or day 3 post-TBI. Recombinant murine TPO (0.3μg; Peprotech) in 250 μl PBS was injected intraperitoneally 2 hours after 4 Gy TBI.

Mice were killed by CO₂ narcosis and peripheral blood collected from the inferior vena cava. Femoral marrow was flushed into PB2⁴⁸  with 25 μg/mL heparin and titurated cells counted by hemocytometer.

Platelet counts were obtained using a HemaTrue Veterinary Hematology Analyzer (Heska). For platelet RNA analysis, EDTA-collected blood was stained in PBS with 1000 U/mL heparin, 1 μg/mL thiozole orange (Sigma-Aldrich), and 1:100 eFluor450-CD41 and phycoerythrin (PE) Cy7-Ter119 (eBioscience), fixed with 1% formaldehyde, and analyzed by Imagestream^X (Amnis).

MKPs were quantified by colony assay, as previously described.⁴⁹  MK analysis was performed by imaging flow cytometry, as recently described.⁵⁰  PE-CXCR4 (eBioscience) was used to determine median fluorescent intensity per unit area (MFI/A) for each individual MK with the median pixel intensity feature (IDEAS [version 4.0]; Amnis).

Marrow cells were stained with biotinylated Kit antibody (EBioscience), incubated with IMag streptavidin magnetic particles, and magnetically separated (iMagnet; BD Biosciences). Kit+ cells were cultured at 10⁶/mL with Iscove modified Dulbecco medium (Invitrogen), 20% BIT 9500 (Stem Cell Technologies), and 100 ng/mL TPO (Peprotech). After 4 days at 37°C/5% CO₂, 300 ng/mL SDF-1 (Peprotech) or vehicle was added for 1 hour. Cells were washed in PB2, stained, and analyzed by imaging flow cytometry.⁵⁰ 

Hindlimbs were fixed in 4% paraformaldehyde for 24 hours, decalcified in either 10% EDTA for 10 to 14 days or Richard Allen Decalcifying Solution (Thermo-Fisher) for 5 hours, and embedded in paraffin. Femoral sections (5 μm) underwent heat antigen retrieval (Dako), blocking with 5% goat serum (Vector Labs) and 5% bovine serum albumin (Gemini Bio-Products), and sequential antibody staining: 1:100 anti-panendothelial cell antigen (MECA32; Biolegend), 1:500 AlexaFluor647 goat anti-rat (Life Technologies), 1:300 anti-Gp1Bβ (Emfret), and 1:500 IRDye 800CW goat anti-rat (LI-COR). Images were acquired with an ORCA-R2 digital camera (Hamamatsu) with Nikon NIS-Elements software on an Eclipse 80i microscope (Nikon) with a ×20 objective. Total Gp1Bβ+ MKs and MKs physically associated with MECA32+ vessels were manually enumerated for 8 to 12 images per section (mean, 94 total MK/sample; range, 45-148). MECA32+ vessels were demarcated and vascular area/total marrow area was determined by NIS-Elements.

Analysis of SDF-1 transcripts was performed on femoral sections, as previously described.^51,52  Darkfield and brightfield images were acquired with an ORCA-R2 digital camera (Hamamatsu) with Nikon NIS-Elements software on an Eclipse 80i microscope (Nikon) with a ×4 objective. For each darkfield image, signal threshold was set to capture single autoradiographic grains, and the grain area determined in a 100 μm × 1250 μm region at the endosteal boundary of diaphyseal marrow, as well as an immediately adjacent region in central marrow. The endosteal:central grain area ratio was obtained for each image. The images were processed and pseudocolored using Photoshop 6.0 (Adobe).

RNA was isolated from flushed marrow and complementary DNA prepared, as previously described.⁵³  Quantitative polymerase chain reaction (qPCR) was performed with Taqman Gene Expression Assays (Life Technologies). CXCL12 (Mm00445553_m1) expression was normalized to 18S (Hs99999901_s1).

Studies utilizing a murine model of constitutive plasma elevation of SDF-1 indicate that a sustained gradient of SDF-1 toward the vasculature results in increased MKs associated with marrow vasculature.²⁰  However, the extended timeframe of SDF-1 exposure, coupled with the capacity of the vascular niche to influence MK maturation^4,18,19  makes it difficult to determine direct SDF-1 effects on MK location. Given that the chemotactic response of MKs to SDF-1 in vitro occurs rapidly,^24,54  we asked if an exogenous gradient of SDF-1 toward the vasculature acutely regulates MK migration. A single 400 ng dose of SDF-1 was administered IV and the compartments of the MK lineage were analyzed 24 hours later. There was no change in the number of MKPs, defined functionally by colony assay, or MKs, determined by imaging flow cytometry, in the femurs of the mice treated with SDF-1 compared with vehicle controls (Figure 1A). Surprisingly, however, mice receiving SDF-1 had a 17% increase in platelet count (Figure 1A). This increase in circulating platelets was likely due to enhanced production, as platelets isolated from mice treated with SDF-1 had higher thiozole orange staining, which can label both RNA and dense granules, and indicates an increase in young, reticulated platelets^55,56  (Figure 1B). Further supporting an acute increase in thrombopoiesis, mice receiving SDF-1 had more exhausted MKs at 24 hours by imaging flow cytometry^10,11,50  (Figure 1C). Although there was no change in the total number of MKs (Figure 1A), as early as 1 hour after IV SDF-1 there was a dramatic shift in their location, with a >40% increase in the proportion of Gp1Bβ+ MKs associated with MECA32+ vasculature by immunohistochemistry (IHC) (Figure 1D-E).⁵⁷ 

As changes in CXCR4 expression have been noted after SDF-1 stimulation in many cell types,^58-62  we investigated CXCR4 surface expression on MKs by imaging flow cytometry. MKs isolated 1 hour and 24 hours after IV SDF-1 demonstrated increased surface expression of the CXCR4 receptor (Figure 1F). In vitro–derived primary MKs displayed a similar increase in CXCR4 expression at 1 hour, suggesting this is a direct response of MKs to SDF-1 stimulation (Figure 1G). Taken together, these results indicate that acute elevation of vascular SDF-1 with IV administration rapidly enhances the MK-vasculature association, which leads to increased platelet production.

Next, we sought to determine if the migration of MKs during steady-state megakaryopoiesis is regulated by a native, endogenous SDF-1 gradient. In vivo, SDF-1 contributes to chemotactic gradients in close proximity to its synthesis, as it binds many glycosaminoglycan extracellular matrix components upon secretion and can undergo rapid proteolysis.^63,64  To stabilize endogenously-produced SDF-1, we used Diprotin A, a small peptide that inhibits dipeptidyl peptidase 4 (DPP4), a soluble and cell-surface protease present in the bone marrow that truncates SDF-l and abolishes its chemotactic activity.^65-68  There were no changes in the number of marrow MKPs or MKs by 24 hours after administration of Diprotin A; however, there was a 20% increase in circulating platelets (Figure 2A). Similar to the response of MKs to exogenous SDF-1 (Figure 1F), marrow MKs had increased surface CXCR4 1 hour after Diprotin A-mediated stabilization of SDF-1 (Figure 2B). Significantly, mice receiving Diprotin A also demonstrated a 30% increase in MKs associated with the vasculature by IHC (Figure 2C-D). These results mirror the MK shift toward the vascular niche and increased platelet production seen with the acute administration of IV SDF-1 (Figure 1), supporting the concept that SDF-1 regulates the maturational movement of MKs toward the vasculature during normal megakaryopoiesis.

As SDF-1 contributes to MK localization to the thrombopoietic niche at steady state, we asked whether changes in marrow SDF-1 have consequences for the MK lineage. Given the importance of the location of SDF-1 production for its chemotactic function,⁶³  and the varied injury and recovery kinetics of different marrow cell populations after sublethal TBI,^40-43  we reasoned there might be changes in marrow SDF-1 gradients in this injury setting. To address this, we investigated the location of SDF-1 transcripts in the marrow of mice after sublethal 4 Gy TBI by radioactive in situ hybridization (Figure 3A). As expected, there was an overall increase in SDF-1 message in the marrow following TBI,^39,46  which was confirmed by qPCR (Figure 3B). There were, however, no changes in plasma SDF-1 protein levels in the first 4 days after injury (supplemental Figure 1 available on the Blood Web site). Changes in marrow distribution were quantified by comparing SDF-1 transcript levels within the endosteal region in the diaphyseal marrow to the levels in an adjacent region within the central marrow (Figure 3C). Interestingly, we detected the development of an SDF-1 gradient toward the endosteum at day 2 that was no longer present at day 3 (Figure 3C).

To determine whether MKs respond to the observed shifts in SDF-1 expression following 4 Gy TBI, we first characterized the injury kinetics of the MK lineage (Figure 4A). Platelets remained at steady-state levels for 4 days, followed by the development of thrombocytopenia, with a nadir in platelet count at day 8 following TBI (Figure 4A). MKs persisted during the first 3 days postinjury (Figure 4A), despite the loss of >85% of nucleated marrow cells (data not shown). Importantly, these radioresistant MKs had increased surface expression of CXCR4, indicating they retained the ability to respond to SDF-1 (Figure 4B; supplemental Figure 1). Megakaryocytopenia and thrombocytopenia eventually develop as a consequence of the rapid, near complete loss of the upstream MKP compartment (Figure 4A).

Since MKs interact with sinusoidal endothelium, we also examined the effects of 4 Gy TBI on marrow vasculature by IHC. Most notable was a marked vascular dilation at day 1 post-TBI, with a 200% increase in the marrow area occupied by vasculature that persisted for at least 4 days following the injury (Figure 4C-D; supplemental Figure 2). Accompanying the early, rapid vascular dilation was a 30% increase in the MKs associated with sinusoids by IHC at day 1 (Figure 4E). The temporal overlap between the vascular changes and the shift in MK-vasculature association suggests that this initial change in MK niche occupancy is likely secondary to vascular dilation. However, within the subsequent timeframe in which vascular dilation remains constant, there were small, but significant changes in MKs in the vascular niche, suggesting additional factors may influence this altered localization (Figure 4E).

Given the development of an SDF-1 gradient toward the endosteum following TBI (Figure 3A,C), we also quantified MK distribution in the endosteal niche, defined as within 100 μm of the diaphyseal endosteum⁶⁹  (Figure 4F). Strikingly, there was a 45% increase in radioresistant MKs in the endosteal niche between days 1 and 2 post-TBI (Figure 4F). MKs rapidly decreased in the endosteal niche by day 3 postinjury (Figure 4F), coincident with the diminished SDF-1 gradient (Figure 3A,C). Taken together, these results spatially and temporally implicate alterations in marrow SDF-1 after TBI in MK niche occupancy changes.

We hypothesized that changes in the spatial localization of MKs between days 2 and 3 post-TBI were regulated by shifts in marrow SDF-1 production. In light of the importance of MK-vasculature association for efficient platelet production,²¹  we reasoned that alterations in MK niche occupancy would affect platelet production. To address this, we stabilized native SDF-1 with Diprotin A at day 2 post-TBI, when SDF-1 production is increased near the endosteum, and found a 20% decrease in MKs in the vascular niche (Figure 5A, left panel) and a small, but significant decrease in circulating platelets 24 hours later (Figure 5A, right panel). However, stabilization of SDF-1 at day 3 post-TBI, when there is no longer an endosteal SDF-1 gradient, has opposing effects, with significant increases both in MK-vasculature association (Figure 5B, left panel) and in platelet count (Figure 5B, right panel). These data support the concept that changes in the location of SDF-1 production at days 2 and 3 after TBI contribute to altered MK location in the marrow (Figure 5C). Additionally, these results demonstrate that the spatial distribution of MKs has functional consequences, as MKs leaving the vascular niche decreases circulating platelets while enhanced MK-vasculature association increases platelet numbers.

Having shown that a transient increase in endosteal SDF-1 production directs MKs away from the vascular niche and negatively affects platelet production, we asked whether we could manipulate MK location and influence platelet output by altering SDF-1 gradients. To this end, we elevated vascular SDF-1 with acute IV administration at day 2 postinjury, when there is a native increase in endosteal SDF-1. Whereas Diprotin A-mediated stabilization of endogenous-produced SDF-1 at day 2 decreases both MK association with the vasculature and platelet production 24 hours later (Figure 5A), IV SDF-1 had opposing effects, increasing MKs in the vascular niche (Figure 5D, left panel) as well as circulating platelets (Figure 5D, right panel). Taken together, these data highlight the direct effects of rapidly changing SDF-1 gradients on MK spatial localization and thrombopoiesis.

Our studies indicate that appropriately timed IV administration of SDF-1 can acutely increase platelet output. To test whether this strategy could be used to improve radiation-induced thrombocytopenia, we administered SDF-1 on day 4 after 4 Gy TBI, when thrombocytopenia begins to develop (Figure 4A), and our localization studies indicate a small decrease of MKs within the vascular niche (Figure 4E). With no change in MKP or MK number 24 hours following IV SDF-1 (day 5 post-TBI), there were increases in both platelet count and MKs associated with the vasculature (Figure 6A-C). However, these effects were less robust than those achieved with the same dose of SDF-1 in uninjured mice (Figure 1A). We hypothesized this was a consequence of fewer MKs present in the marrow to move to the vascular niche at this time point postinjury, and the SDF-1-mediated platelet increase may therefore depend on MK availability.

Treatment with the cytokine TPO increases the number of MKs in the marrow and has been shown to improve MK lineage recovery when administered after sublethal TBI in mice.^12,13,70  In our model, administration of TPO 2 hours following 4 Gy TBI resulted in a >200% expansion of MKPs, a 65% increase in MKs, and a small, but not statistically significant increase in platelet count (Figure 6A). Mice receiving TPO at 2 hours followed by SDF-1 at day 4 had an additive increase in circulating platelets compared with either treatment alone (Figure 6A), and a corresponding 70% increase in MKs associated with the vasculature (Figure 6B-C). These data suggest that the platelet increase accompanying SDF-1-mediated localization of MKs to the vascular niche is dependent, at least in part, on the number of MKs available to relocate. Additionally, our results indicate that a sequential combination of TPO, which increases MK number, followed by SDF-1, which enhances MK localization in the thrombopoietic niche and platelet output, may be a potent strategy to acutely improve thrombocytopenia in the setting of radiation injury.

The working paradigm of megakaryopoiesis and thrombopoiesis has been primarily cytokine-driven, based on the potent ability of TPO to increase MK numbers and accelerate MK maturation, thereby indirectly enhancing platelet production.^13,14,71  Elegant studies by Avecilla et al²⁰  identified the chemokine SDF-1 as a TPO-independent promoter of thrombopoiesis. However, the extended timeframe of vascular SDF-1 elevation in these studies complicates the evaluation of in vivo effects on MK chemotaxis vs a direct role for SDF-1 and/or endothelium in MK lineage maturation.²⁰  Furthermore, there are conflicting in vitro reports concerning a role for SDF-1 in MKP proliferation and MK maturation.^22,72,73  Here, we directly addressed the acute effects of exogenous SDF-1 using a single IV administration, which only circulates for minutes to hours.⁶⁴  With a quantitative IHC approach, we found MKs redistribute to the vasculature within 1 hour of SDF-1 administration (Figure 1E). Importantly, this rapid SDF-1-mediated association of MKs with the vasculature is also accompanied by an increase in circulating platelets apparent at 24 hours (Figure 1). In vitro treatment of MKs with SDF-1 does not yield an increase in platelets (data not shown), suggesting that SDF-1 directly contributes to MK movement to the vasculature, whereas the increase in thrombopoiesis is likely due to instructive cues from the vascular niche.^4,18,19,74 

To our knowledge, these studies are the first to demonstrate that natively produced SDF-1 directs MK migration to the vascular niche (Figure 2). For the stabilization of endogenous SDF-1, we used tripeptide Diprotin A to inhibit DPP4, a peptidase that inactivates SDF-1.^66,75  This strategy has been used to stabilize SDF-1 in several contexts, including the homing and engraftment of hematopoietic stem cells and the recruitment of marrow-derived progenitors to sites of cardiac ischemia for myocardial regeneration.^68,76-78  DPP4 was recently shown to truncate several cytokines, including IL-6, and is predicted to truncate TPO.⁷⁹  While we cannot exclude effects of stabilization of these cytokines by Diprotin A, the short timeframe of our experiments coupled with the lack of change in the number or ploidy distribution of MKs (Figure 1A; supplemental Figure 3) supports SDF-1 as the mediator of spatial changes following Diprotin A treatment. Longer term inhibition of DPP4 may uncover cytokine-driven effects on the MK lineage as Dpp4^−/− mice have increased numbers of MKPs and MKs.⁸⁰ 

CXCR4 can be internalized after SDF-1 stimulation, and either recycled to the membrane or degraded.⁸¹  Unexpectedly, we found that acute stimulation with SDF-1 resulted in a small, but significant increase in CXCR4 on the surface of MKs both in vitro and in vivo, as well as following stabilization of SDF-1 in vivo (Figure 1F-G and 2B). It has been reported that SDF-1 induces CXCR4 promoter activity⁵⁹  and activated nuclear factor κB, which is downstream of SDF-1/CXCR4 signaling in MKs,²⁶  can also induce CXCR4 transcription.⁵⁸  In addition, SDF-1/CXCR4 signaling in MKs induces vascular endothelial growth factor production,²⁶  which can increase MK surface CXCR4.³⁴  Moreover, the observed increase in surface CXCR4 could result from the translocation of intracellular stores of CXCR4 to the cell surface, which has been reported in hematopoietic progenitor cells.⁸²  Following radiation injury, the upregulation of MK surface CXCR4 could also be due to microenvironmental hypoxia through HIF-1α⁸³  (Figure 4B).

Our studies reveal spatial fluctuations in SDF-1 transcript production in the marrow following sublethal TBI that dynamically regulate MK location. Despite loss of the vast majority of nucleated marrow cells following TBI, we confirmed a global increase in marrow SDF-1 transcripts (Figure 3B). These data are in-line with the induction of SDF-1 expression after radiation injury in rodents and in humans, through HIF1α-dependent and HIF1α-independent pathways.^39,46,84-86  More relevant for chemotactic function, we identified a transient elevation of SDF-1 transcripts near the endosteum at day 2 following injury that directs MKs away from the vascular niche (Figures 3C and 5A). Evidence pointing to a cell population that may contribute to altered SDF-1 distribution comes from the work of Dominici et al,³⁹  Olson et al,⁴⁷  and Caselli et al⁸⁷  who described an expansion of osteoblasts arising from a subset of radioresistant mesenchymal cells at the endosteal surface concomitant with MK relocation following lethal TBI. Mesenchymal stromal and progenitor cells are some of the highest producers of SDF-1 in the marrow at steady state^29,31,32  and are relatively radioresistant,⁸⁸  suggesting these populations may participate in marrow microenvironmental changes following TBI. Further investigation to specifically identify the SDF-1–producing cells pertinent for MKs at steady state and postinjury may lead to new strategies to influence platelet production by controlling MK niche occupancy. Additionally, while this work focused on the thrombopoietic vascular niche because of its functionality for MKs, it remains an open question whether alterations in microenvironmental SDF-1 following sublethal radiation injury affect the location or function of other hematopoietic cells. This may be particularly relevant for hematopoietic stem and progenitor cells, which are regulated by SDF-1^29,31,32,89  and have been found in close proximity to diverse marrow microvessels,⁹⁰  including preferentially arteriolar vasculature in the endosteal region.⁹¹ 

Significantly, our studies establish that acute manipulation of SDF-1 has functional consequences for MK location and platelet production. We tested the validity of this concept in a model of radiation-induced thrombocytopenia. While SDF-1 increases MK association with the vasculature and improves thrombocytopenia, there is an additive increase in circulating platelets if the number of marrow MKs is increased with TPO administration prior to SDF-1 treatment. These data suggest a potential sequential therapeutic approach combining an agent that first increases the number of MKs in the marrow followed by an agent that shifts their location to the vascular niche to promote platelet production. Additionally, the ability to acutely elevate circulating platelets within 24 hours may have therapeutic relevance in certain scenarios as the current clinically available Mpl agonists require 5 days to begin elevating platelet counts.^92,93 

Taken together, our data indicate that SDF-1 acutely regulates MK location in the bone marrow. At steady state, SDF-1 directs MKs to the vascular niche, resulting in thrombopoiesis. Following radiation injury, dynamic changes in microenvironmental SDF-1 alter the location of radioresistant MKs in the marrow. Ultimately, we demonstrate that rational manipulation of SDF-1 can acutely improve thrombocytopenia by enhancing MK migration to the thrombopoiesis-promoting vascular niche.

The authors thank Jennifer McLaughlin for assistance with tissue sectioning, and gratefully acknowledge the technical support of Dr Tim Bushnell and Flow Cytometry Core Facility at the University of Rochester Medical Center.

This work was supported by funding from The National Institutes of Health (NIH), National Institute of Allergy and Infectious Diseases (R01AI080401 and U19AI091036) (J.P.), NIH, National Institute of Diabetes and Digestive and Kidney Diseases (F30DK100164) (L.M.N.), and the Michael Napoleone Foundation. L.M.N. is a trainee in the Medical Scientist Training Program funded by the NIH, National Institute of General Medical Sciences (T32GM07356).

Contribution: L.M.N. designed and performed experiments, analyzed data, and wrote the manuscript; K.H.F. and P.D.K. designed and performed experiments; K.E.M. designed experiments and analyzed data; and J.P. designed experiments, analyzed data, and wrote the manuscript.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: James Palis, Department of Pediatrics, University of Rochester Medical Center, Center for Pediatric Biomedical Research, 601 Elmwood Ave, Box 703, Rochester, NY 14642; e-mail: james_palis@urmc.rochester.edu.