2026 Volume 51 Issue 3 Pages 201-213
Non-clinical safety evaluations, including those of impurities, are important for vaccine development to ensure safety in humans. However, information on the mechanisms of impurity-induced adverse effects remains limited. In a repeated-dose toxicity study of our lipid nanoparticle formulated mRNA vaccine (mRNA-LNP vaccine) candidate, severe anemia was observed in rats after multiple administrations. In this study, we conducted hematological analyses and bone marrow examinations in vivo to investigate the cause and mechanism of test article-related delayed anemia. In addition, we performed in vitro mechanistic studies including antibody titer measurements and colony-forming unit assays. We found that test article-related anemia was caused by the inhibition of erythroid differentiation in the bone marrow, mediated by antibodies against erythropoietin (EPO). Furthermore, the test article was found to contain human EPO mRNA as an impurity. Lastly, the spike study showed that a minute quantity of human EPO mRNA present in mRNA-LNP vaccines as an impurity induced anemia in rats. Taken together, our data demonstrate that immune-mediated delayed anemia can be induced by impurity-oriented anti-EPO antibodies that neutralize endogenous EPO and inhibit erythroid differentiation. Our presented approach of determining the mechanism of delayed toxicity caused by impurities may be helpful in future safety evaluations.
Vaccination is one of the most effective and successful public health interventions (Stone Jr et al., 2019). Vaccines generally require lower doses and are better tolerated than small molecules. However, serious and unpredictable immune-mediated adverse events can occur during development, complicating and delaying market availability of these therapeutic agents, as well as potentially frightening the public, leading to concerns about the safety of inoculation. Improving our understanding of the mechanistic risk factors for severe adverse events associated with vaccines and conducting mechanistic causal assessments are therefore critical for sustaining this essential public health intervention.
The lipid nanoparticle formulated mRNA vaccine (mRNA-LNP vaccine) is an mRNA-based agent administered to organisms to express antigens. The first gene transfer using mRNA in vivo was reported in 1990 (Wolff et al., 1990). In 2020, the COVID-19 pandemic prompted the most rapid vaccine development in history (Chaudhary et al., 2021), including mRNA-based agents. During vaccine development, it is essential to conduct non-clinical safety evaluations in animals to ensure human safety by evaluating injection site reactions, body weight loss, changes in body temperature, immune responses, and other factors before starting clinical trials (Phase 1) (Al-Humadi, 2017; Sellers et al., 2020). A clear observation of adverse immune responses in humans and animals generally requires a long time, with these responses often identified as delayed toxicity in clinical trials, post-marketing surveillance, and non-clinical drug toxicity studies. To date, several adverse immune reactions have been reported, including autoantibody-induced Raynaud’s-like phenomenon in patients with cancer treated with immune checkpoint inhibitors and immune-mediated hemolytic anemia induced by antibiotics and non-steroidal anti-inflammatory drugs (Khan et al., 2020; Renard and Rosselet, 2017). According to the WHO guidelines on non-clinical safety evaluation of vaccines for the prevention and treatment of infectious diseases, a repeated toxicity study in animals is required to assess systemic toxicity and local reactions caused by the vaccine, including immune responses or unintended sensitization due to vaccine-induced antibodies (WHO, 2005). In addition, not only should the toxicological potential of active pharmaceutical ingredients such as antigens and adjuvants be evaluated, but also that of impurities in vaccine drug products. However, information on the mechanisms of impurity-induced adverse effects remains limited.
During the production of mRNA-LNP vaccines, physical property analysis, including the verification of mRNA length, encapsulation efficiency, and particle diameter, is performed to ensure their quality. However, there are technical limitations in detecting and excluding all impurities, especially in the early stages of research where the throughput is priority. For example, repeated use of equipment increases the risk of contamination, and limited information about the properties of a compound reduces the chance to notice abnormalities in their quality. In addition, when the length of mRNA impurity is similar to that of the intended mRNA, it is difficult to detect such impurities through aforementioned analysis. If the amounts of mRNA impurities are very low, they may not be detected even if their length differs from that of the intended mRNA. In fact, sub-microgram quantities of mRNA-LNP vaccines, approximately 1%-10% of usual dosages, can induce neutralizing antibodies in the rodents (Lin et al., 2025; Lelis et al., 2023). Therefore, even trace amounts of mRNA impurities not detectable by the above analysis could induce unexpected antibodies and cause adverse effects. In our laboratory, we evaluated the safety of several mRNA-LNP vaccines in non-clinical repeated-dose studies and unexpectedly observed that a mRNA-LNP vaccine encoding an antigen of a certain pathogen (compound A) caused anemia in rats after multiple administrations. We did not observe any abnormalities in the mRNA length, encapsulation efficiency, or particle diameter of compound A. Therefore, in this manuscript, we report our approach in investigating the cause and mechanism of mRNA-LNP vaccine-related anemia in rats. We conducted hematological analyses and bone marrow examinations in vivo, and antibody titer measurements and colony-forming unit (CFU) assays in vitro. Finally, we demonstrated that human erythropoietin-encoding mRNA (hEPO-mRNA), unintentionally contained in mRNA-LNP vaccines as an impurity, induces anti-erythropoietin (EPO) antibodies and causes anemia in rats. Our approach for examining the immune-mediated toxicity may be helpful for future toxicological evaluations of mRNA-LNP vaccines.
All studies were conducted with the approval of the Institutional Animal Care and Use Committee (approval numbers S22031D-0003, S22120D-0001, and S22165D-0000) and performed in accordance with the animal welfare bylaws of the Shionogi Pharmaceutical Research Center, Shionogi & Co., Ltd., which is accredited by AAALAC International.
In all in vivo studies, male Crl:CD (SD) rats (Jackson Laboratories Japan, Inc., Kanagawa, Japan) were used, and were 7 weeks old at dose initiation. The animals were housed in plastic cages with absorbent paper bedding. Environmental conditions were maintained at a temperature between 20 and 26°C, and relative humidity between 30 and 70%. Room air was ventilated at a rate of 10 changes/hr or more. Room lighting was controlled at a 12-hr light-dark cycle (lighting: 8:00 to 20:00). Wood bricks were used for environmental enrichment.
The animals were provided food pellets (CRF-1: ORIENTAL YEAST Co., Ltd., Tokyo, Japan) and water filtered through 30 and 3 µm filters, followed by UV irradiation, ad libitum.
Control and test articlesIn all experiments, 20 mM Tris-HCl / 350 mM sucrose was used as the control or vehicle. The mRNA-LNP vaccine was prepared using an ionizable lipid (Shionogi original lipid, manufactured in-house) and mRNA (compound A: encoding antigen A, compound B: encoding antigen B, and hEPO-mRNA-LNP: encoding the human EPO protein). The original lipids were prepared from distearoylphosphatidylcholine (Nippon Fine Chemical Co., Ltd., Osaka, Japan), cholesterol (Nippon Fine Chemical Co., Ltd., Osaka, Japan), 1,2-dimyristoyl-sn-glycerol, and methoxypolyethylene glycol 2000 (NOF Corporation, Tokyo, Japan). mRNA was synthesized and characterized, and particle size and mRNA encapsulation of the mRNA-LNPs were verified. The mean particle size (Z-average) and polydispersity index (PDI) were measured using a Zetasizer Nano ZS (Malvern, Worcestershire, U.K.), ranging 80-120 nm for the Z-average and within 0.2 for PDI. mRNA encapsulation efficiency (%EE) was determined using a Quant-iT Ribogreen RNA assay Kit (Thermo Fisher Scientific, Waltham, MA, USA). Briefly, total and free mRNA concentrations in LNPs were measured by fluorescence intensity (excitation/emission: 480/520 nm) of Ribogreen dye in the presence and absence of 2% Triton X-100, resulting %EE was over 85.0%.
Overall experimental design (Fig. 1)The control, compound A, and compound B were intermittently administered intramuscularly at a 4-week interval (twice in total) to rats at a dose of 40 µg/body as mRNA in experiment 1. In experiment 2, the control, hEPO-mRNA-LNP, and compound B spiked with two different amounts of hEPO-mRNA were intermittently administered intramuscularly at a 4-week interval (twice in total) to rats. Doses of each compound were 1, 40.16, and 41 µg/body as mRNA, respectively, for hEPO-mRNA-LNP, compound B containing 0.16 µg of hEPO-mRNA, and compound B containing 1 µg of hEPO-mRNA. The dose volume was 0.2 mL/body in experiments 1 and 2. The first day of administration was designated as Day 1. The following examinations were performed in experiments 1 and 2: clinical observation (every day), hematology (approximately 1 and 3 weeks after each dose [Days 8, 22, 35, and 50 in experiment 1; Days 22, 36, and 55 in experiment 2]), and bone marrow analysis (approximately 3 weeks after the first dose [Day 23] in experiment 1 and approximately 3 weeks after the second dose [Day 55] in experiment 2). Anti-human or anti-rat erythropoietin antibody measurements were conducted 1 week after the second dose [Day 35] in experiment 1.
Experimental schedules. The first day of administration was designated as Day 1. The control, compound A, compound B, hEPO-mRNA-LNP, or compound B spiked with hEPO-mRNA were administered intramuscularly at a 4-week interval (twice, Days 1 and 29; black arrows) to rats. Gray arrows show sampling points for hematology, bone marrow analysis, antibody measurements, and CFU assay in experiments 1 and 2. Details of the experiments and evaluation items are provided in the Materials and Methods section.
In repeated dose studies (experiments 1 and 2), blood (approximately 500 µL/rat) was collected from the tail vein with a syringe containing 5 µL of heparin sodium (Terumo Corporation, Tokyo, Japan) for antibody measurements and part of the blood was treated with EDTA-2K (Terumo Corporation, Tokyo, Japan) for hematology examination except for approximately 3 weeks after the second dose (Week 7). At Week 7, blood was drawn from the posterior vena cava using a disposable syringe under anesthesia and treated with EDTA-2K for hematological examination. Blood treated with heparin sodium was centrifuged at 3000 rpm for 15 min at 4°C to obtain plasma samples.
HematologyWhole blood samples treated with EDTA-2K were analyzed using an ADVIA 2120i Hematology System (Siemens K.K., Tokyo, Japan). The measured parameters were RBC count, HGB, HCT, and Retic count.
Anti-human erythropoietin (EPO) antibody measurement (ELISA)To measure anti-human EPO antibody levels in plasma, a 96-well plate was coated with 5 μg/mL human EPO (Proteintech Group, Inc, Rosemont, IL, USA) in PBS overnight at 4°C. The plate was washed with PBS containing 0.05% Tween 20 (PBST) and blocked with PBST containing 10 mg/mL bovine serum albumin (BSA) at 37°C for 1 hr. Plasmas were serially diluted with PBST containing 3% BSA. After washing the plates with PBST, plasma samples were added and incubated at 37°C for 1 hr. After another wash with PBST, anti-Rat IgG HRP (Thermo Fisher Scientific, Waltham, MA, USA) (1000-fold diluted with PBST containing 3% BSA) was added and incubated at 37°C for 1 hr. After another wash with PBST, ABTS Peroxidase substrate (Abcam, Cambridge, UK) was added, and the optical density was measured at 405 nm using EnSpire (PerkinElmer, Waltham, MA, USA).
Anti-rat EPO antibody measurement (ELISA)To measure anti-rat EPO antibody levels in plasma, a 96-well plate was coated with 5 μg/mL rat EPO (ProSpec, Rehovot, Israel) in PBS overnight at 4°C. The plate was washed with PBS containing 0.05% Tween 20 (PBST) and blocked with PBST containing 10 mg/mL bovine serum albumin (BSA) at 37°C for 1 hr. Plasma was serially diluted with PBST containing 3% BSA. After washing the plates with PBST, plasma samples were added and incubated at 37°C for 1 hr. After another wash with PBST, anti-Rat IgG HRP (Thermo Fisher Scientific, Waltham, MA, USA) (1000-fold diluted with PBST containing 3% BSA) was added and incubated at 37°C for 1 hr. After another wash with PBST, ABTS Peroxidase substrate (Abcam, Cambridge, UK) was added, and the optical density was measured at 405 nm using EnSpire (PerkinElmer, Waltham, MA, USA).
Bone marrow analysisIn repeated dose studies (experiments 1 and 2), bone marrow extracted from the femur was diluted approximately 2-fold with physiological saline containing 50% fetal bovine serum (Cytiva, Marlborough, MA, USA) and suspended. Bone marrow smears were prepared from the suspensions, air-dried, and fixed in methanol for 5 min. After air-drying, the specimens were stained with May–Grünwald stain (Merck, Darmstadt, Germany) diluted 2-fold with phosphate buffer for 5 min, followed by staining with Giemsa stain (Merck, Darmstadt, Germany) diluted 15-fold with phosphate buffer for 15 min. Five hundred cells per specimen were examined under a microscope and classified as dividing erythroblasts, pro-erythroblasts, basophilic erythroblasts, polychromatic erythroblasts, orthochromatic erythroblasts, dividing myelocytes, myeloblasts, pro-myelocytes, myelocytes, meta-myelocytes, stab neutrophils, seg neutrophils, eosinophils, basophils, monocytes, lymphocytes, mast cells, mega-karyocytes, macrophages, plasmacytes, and unidentifiable cells. The erythrocytic series comprises dividing erythroblasts and immature cells in the following order: pro-erythroblasts, basophilic erythroblasts, polychromatic erythroblasts, and orthochromatic erythroblasts. Burst-forming unit erythroid (BFU-E) and colony-forming unit erythroid (CFU-E) were not distinguishable under the microscope. The granulocytic series was composed of dividing myelocytes and immature cells in the following order: myeloblasts, pro-myelocytes, myelocytes, meta-myelocytes, stab neutrophils, seg neutrophils, eosinophils, and basophils. Monocytes, lymphocytes, mast cells, mega-karyocytes, macrophages, plasmacytes, and unidentifiable cells were classified as other cells and the M/E ratio was calculated using cell counts from the erythrocytic and granulocytic series.
Rat hematopoietic colony-forming unit (CFU) assaysThe CFU assay was performed using MethoCult SF M3436 (StemCell Technologies, Vancouver, Canada) containing erythropoietin to promote BFU-E differentiation according to the manufacturer’s instructions. To evaluate the differentiation of BFU-E from rats with anemia, bone marrow (BM) cells were harvested from each vehicle- or compound A-treated rat 1 week after the second dose (Day 35) in experiment 1 of the repeated-dose study. Red blood cells were removed following treatment with 1X RBC Lysis Buffer (Thermo Fisher Scientific, Waltham, MA, USA). After red blood cell lysis, BM cells were cultured in MethoCult SF M3436 in duplicate at a density of 5.0 ×104 cells/well in 6-well plates. Colonies were scored on day 13 of culture.
To evaluate whether plasma from rats with anemia inhibited BFU-E differentiation, plasma was collected from rats treated with vehicle or compound A 1 week after the second administration (Day 35) in experiment 1 of the repeated-dose study. Frozen SD rat BM cells from normal rats were purchased from iQ Bioscience (Alameda, CA, USA). BM cells were cultured in MethoCult SF M3436 in duplicate at a density of 3.75×105 cells/well in 6-well plates using plasma from each rat. The plasma from vehicle- or compound A-treated rats was 16.5-fold diluted in culture conditions. Colonies were scored on day 13 of culture.
Reverse transcriptase-polymerase chain reaction (RT-PCR)cDNA was synthesized from each mRNA (hEPO, compound A and compound B) using PrimeScript IV 1st strand cDNA Synthesis Mix (Takara Bio, Shiga, Japan) and Ribonuclease H (Takara Bio) according to the manufacturer’s protocol. cDNA from 10 ng of mRNA was amplified by PCR using hEPO-specific primers (hEPO_Fw: AAGAGCCACCATGGGCGTGCACGA, hEPO_Rv: CATCACCGGTCGCCGGTC) and PrimeSTAR Max DNA Polymerase (Takara Bio). PCR products were electrophoresed on a 2% agarose gel, followed by visualization with GelRed (Biotium, Inc., CA, USA) and LAS-3000 (Fujifilm, Tokyo, Japan).
Statistical analysisStatistical analyses were performed using GraphPad Prism 10.5.0 software (GraphPad Software, San Diego, CA, USA). Hematological analysis in the repeated-dose study with compounds containing hEPO-mRNA (Figs. 4B–E and 5A–D) was performed using one-way ANOVA with Dunnett’s multiple comparison test. All other measurements except for antibody measurements were analyzed using an unpaired t-test with Welch’s correction. The threshold for significance was set at p<0.05.
To determine the time course of hematological changes, we intermittently administered compound A to rats at a 4-week interval (Fig. 1, experiment 1). Statistically significant decreases in Retic count, HGB, and HCT in whole blood were noted from 3 weeks after the first dose (Day 22, Week 3) to 3 weeks after the second dose (Day 50, Week 7, the end of the study) in compound A-treated rats (Figs. 2A, C, and D). Notably, the Retic count decreased to 16% on average compared to vehicle-treated rats, even at Week 3, whereas both HBG and HCT only decreased to 91% on average. Subsequently, a statistically significant decrease in RBC count and anemia-related abnormal clinical signs, such as pale eyes and auricles, were noted 1 week after the second dose (Week 5) and continued until Week 7 (Fig. 2B). In addition, we observed trends for decrease in polychromatic erythroblasts and total erythroid cells in the bone marrow of compound A-treated rats 3 weeks after the first dose (Day 23, Week 3), as early as the decrease in Retic count in whole blood (Table 1). There were no clear differences in other erythrocytic series (dividing erythroblasts, pro-erythroblasts, basophilic erythroblasts, and orthochromatic erythroblasts) between vehicle- and compound A-treated rats, probably because of the small population size of these cell subsets. The following granulocytic series and other cells, except for myeloblasts, were also comparable: dividing myelocytes, pro-myelocytes, myelocytes, meta-myelocytes, stab neutrophils, seg neutrophils, eosinophils, basophils, monocytes, lymphocytes, mast cells, mega-karyocytes, macrophages, and plasmacytes. A significant decrease in myeloblasts was noted in compound A-treated rats, but no changes were noted in other granulocytic series or other cells, suggesting that inhibition of erythrocyte proliferation rather than bone marrow suppression was induced. Morphological abnormalities suggestive of cell injury were not observed in bone marrow analysis. In addition, the CFU assay with bone marrow cells from vehicle- or compound A-treated rats showed comparable colony formation (Fig. 2E). This implies that there was no difference in the number and differentiation potency of BFU-E cells, which are the most upstream of erythroid differentiation. Taken together, these results suggest that erythroid differentiation from BFU-E to erythroblasts was inhibited in compound A-treated rats that developed anemia.
Hematology and rat hematopoietic CFU assays using bone marrow cells in a repeated-dose study of compound A. (A–D) Changes in the Retic count (A), RBC count (B), HGB (C), and HCT (D) after administration of the vehicle (black column) or compound A (open column). Each column represents mean values (n=4); error bars indicate standard deviation; * p<0.05 and ** p<0.01.(E) Colony counts of CFU assay using bone marrow cells harvested from each vehicle- or compound A-treated rats 1 week after the second dose (Week 5). Each column represents mean values (n=3); error bars indicate standard deviation.Cmpd A = compound A.
;%0A%09%09%09newWindow.document.open();%0A%09%09%09newWindow.document.write('<img src=%22./Graphics/51_201-t1.jpg%22>');%0A%09%09%09newWindow.document.close();%0A%09%09)
Because delayed anemia and possible inhibition of erythroid differentiation from BFU-E to erythroblasts were noted in compound A-treated rats, we hypothesized that compound A administration induced antibodies that neutralized endogenous rat EPO. EPO is an essential growth factor for the proliferation and differentiation of erythroid progenitor cells, and its neutralization can inhibit erythroid differentiation. To test our hypothesis, we performed ELISA to detect anti-EPO antibodies in plasma. Antibodies binding to both rat EPO and human EPO were detected in the plasma of compound A-treated rats (Fig. 3A). In contrast, almost no antibodies were detected in the plasma of vehicle-treated rats. Next, to evaluate whether the plasma from each compound A-treated rat inhibited BFU-E differentiation, we performed a CFU assay using bone marrow cells from healthy rats and culture medium containing EPO with plasma from vehicle- or compound A-treated rats. We observed that plasma from compound A-treated rats almost completely inhibited colony formation, whereas that from vehicle-treated rats did not (Fig. 3B). These data suggest that the induced anti-EPO antibodies inhibited BFU-E differentiation and caused delayed anemia in compound A-treated rats in the repeated-dose study. Notably, antibodies against the antigen encoded by compound A mRNA (antigen A) were not induced in the rats (data not shown). We also observed that intradermal injections of the antigen A peptide did not induce the antibodies against antigen A, suggesting antigen A is weakly immunogenic (data not shown). Besides, no abnormality was observed in the mRNA length, encapsulation efficiency, or particle diameter of compound A. Furthermore, because antigen A is entirely different from EPO in terms of both size and amino acid sequence, we speculated that trace amounts of an impurity in compound A, which was not detected by physical property analysis, induced anti-EPO antibodies.
Anti-rat EPO and anti-human EPO antibody measurements and rat hematopoietic CFU assays using plasma in the repeated dose study of compound A. (A) Optical density (OD) of anti-rat EPO IgG ELISA (black column) and anti-human EPO IgG ELISA (open column). Plasma collected from each vehicle- or compound A-treated rats 1 week after the second dose (Week 5) was diluted 50-fold and analyzed. Each bar represents mean values (n=3); error bars indicate standard deviation. (B) Relative colony counts of a CFU assay using bone marrow cells from normal rats with plasma collected from each vehicle- or compound A-treated rats 1 week after the second dosing (Week 5). Each column represents mean values (n=3); error bars indicate standard deviation; ** p<0.01. Cmpd A = compound A.
Because anti-human and anti-rat EPO antibodies were detected in compound A-treated rats, we suspected that hEPO-mRNA used in other extraneous experiments was an impurity of compound A. To examine whether hEPO-mRNA was present in compound A, we performed RT-PCR using hEPO-mRNA-specific primers. We detected hEPO-mRNA in compound A and consequently assumed it to be the causative agent of delayed anemia (Fig. 4A). Meanwhile, we analyzed compound B, another mRNA-LNP vaccine encoding the same antigen as compound A but with a different pathogen strain. We prepared two different lots of compound B and confirmed that one lot (lot. 1) did not contain hEPO-mRNA, while the other lot (lot. 2) unintentionally contained hEPO-mRNA similar to compound A (Fig. 4A). Furthermore, the repeated-dose study with lots. 1 and 2 of compound B showed that only lot. 2 caused the significant decreases in Retic count, RBC count, HGB, and HCT 3 weeks after the second dose (Week 7), similar to compound A (Figs. 4B–E). These results strongly suggest that hEPO-mRNA, and not antigen-encoding mRNA, induces anemia in rats. On the other hand, during the mRNA length analysis of compound A mRNA using capillary electrophoresis, we did not observe a clear peak for hEPO-mRNA (data not shown). Thus, we estimated that the amount of unintentionally contained hEPO-mRNA is less than 1% of the total mRNA. Therefore, to investigate whether trace amounts of hEPO-mRNA could reproduce anemic symptoms in rats, we conducted an additional repeated-dose study using hEPO-mRNA alone and compound B artificially spiked with hEPO-mRNA (Fig. 1, experiment 2). Since compounds A and B included 40 µg of mRNA, we prepared hEPO-mRNA-LNP containing just 1 µg of hEPO-mRNA alone. Additionally, we spiked pure compound B with 0.16 or 1 µg hEPO-mRNA (0.4% or 2.5% of total mRNA) to mimic the conditions in experiment 1. In this experiment, the rats were intermittently administered these compounds at a 4-week interval as described in experiment 1. As expected, in rats administered 1 µg of hEPO-mRNA-LNP or compound B spiked with 0.16 or 1 µg of hEPO-mRNA, Retic count in whole blood was significantly decreased from 3 weeks after the first dose (Week 3) to 3 weeks after the second dose (Week 7, the end of the study), as was the case with compound A-treated rats in experiment 1 (Fig. 5A). The decrease in the Retic count was dose-dependent on the amount of spiked hEPO-mRNA in two different groups of compound B. Following the decrease in Retic count, the significant decreases in RBC count, HGB, and HCT and anemia-related abnormal clinical signs such as pale eyes and auricles were also noted from 1 week after the second dose (Week 5) to Week 7 in rats administered hEPO-mRNA-LNP or compound B spiked with 1 µg hEPO-mRNA (Fig. 5B–D). Similar significant decreases and abnormal clinical signs were noted at Week 7 in rats administered compound B spiked with 0.16 µg hEPO-mRNA. In addition, the trends for decrease in polychromatic erythroblasts and total erythroid cells in the bone marrow was observed at Week 7 in rats administered hEPO-mRNA-LNP (Table 2). There was no clear difference in other erythrocytic series, granulocytic series, and other cells between vehicle- and hEPO-mRNA-LNP-treated rats. Thus, we confirmed that both trace amounts of hEPO-mRNA alone and hEPO-mRNA-spiked compound B produce anemic symptoms in rats. In summary, our data indicate that minute quantities of hEPO-mRNA in mRNA-LNP vaccines can induce neutralizing antibodies against endogenous rat EPO, consequently leading to delayed anemia in rats.
Detection of hEPO-mRNA from compounds, and hematology in a repeated-dose study of compounds containing hEPO-mRNA. (A) Agarose gel electrophoresis of RT-PCR products amplified with hEPO-mRNA specific primers. Lane 1, hEPO-mRNA (positive control); lane 2, compound A; lane 3, compound B (lot. 1); lane 4, compound B (lot. 2). (B–E) Retic count (B), RBC count (C), HGB (D), and HCT (E) in rats at 3 weeks after the second dose of the vehicle (black column), compound A (open column), and compound B (lots. 1 and 2) (gray column and light gray column) (Week 7). Each column represents mean values (n=4); error bars indicate standard deviation; ** p<0.01. Cmpd A = compound A. Cmpd B = compound B.
Hematology in a repeated-dose study of compounds containing hEPO-mRNA. (A–D) Changes in the Retic count (A), RBC count (B), HGB (C), and HCT (D) after administration of the vehicle (black column), hEPO-mRNA-LNP (open column), compound B spiked with 0.16 (gray column) or 1 (light gray column) µg of hEPO-mRNA. Each column represents mean values (n=3); error bars indicate standard deviation; * p<0.05 and ** p<0.01. Cmpd B = compound B.
;%0A%09%09%09newWindow.document.open();%0A%09%09%09newWindow.document.write('<img src=%22./Graphics/51_201-t2.jpg%22>');%0A%09%09%09newWindow.document.close();%0A%09%09)
In the present study, we examined the cause and mechanism of mRNA-LNP vaccine-related delayed anemia in a non-clinical toxicity study in rats. We demonstrated that hEPO-mRNA, unintentionally contained in mRNA-LNP vaccines as an impurity, induces anti-EPO antibodies and causes anemia in rats. Importantly, a minute amount (0.16 µg) of hEPO-mRNA contained in the vaccine (0.4% of total mRNA) can lead to anemic symptom in rats, showing the importance of toxicological evaluation of impurities included in vaccine drug products.
In the repeated-dose study of compound A, a decrease in Retic count was noted as the first sign of anemia 3 weeks after the first administration (Day 22, Week 3) (Figs. 2A–D). However, hematopoietic changes caused by the direct effect of the test article were generally observed soon after the first administration; thus, we suspected that the observed anemia was immune-mediated. The trend of a decrease in bone marrow erythroid cells was also noted as early as the decrease in Retic count, followed by a decrease in RBC count (Table 1 and Fig. 2B). The CFU assay with bone marrow cells from vehicle- or compound A-treated rats showed that BFU-E cells were intact, even in the bone marrow of rats with anemia (Fig. 2E). In addition, morphological abnormalities suggestive of cell injury were not observed in the bone marrow of these animals. Furthermore, the cell rates of the granulocytic series and other cells were comparable between vehicle- and compound A-treated rats, suggesting that inhibition of erythrocyte proliferation, rather than bone marrow suppression, was induced. Based on these results, we suspected that delayed anemia in compound A-treated rats was a consequence of immune-mediated inhibition of BFU-E differentiation or more mature cells and not immune-mediated cytotoxicity.
Over 10 growth factors are responsible for the proliferation and survival of blood cells (Kaushansky et al., 2010). Cytokines such as thrombopoietin, interleukin-3, and stem cell factors are involved in hematopoietic stem cell differentiation into red blood cells, with EPO being an essential growth factor for the proliferation and differentiation of committed erythroid progenitor cells (Kaushansky et al., 2010; Ridley et al., 1994). The EPO-responsive compartment primarily comprises erythroid-committed progenitors and early erythroblasts. When cultured with EPO, BFU-E differentiates into CFU-E, which is highly sensitive to EPO. Moreover, EPO is essential for colony formation and terminal differentiation of CFU-E (Ridley et al., 1994; Jelkmann 1992). In patients treated with recombinant human EPO, pure red cell aplasia is caused by anti-EPO antibodies (Casadevall et al., 2002). In addition, it has been reported that the induction of homologous EPO using viral vectors in cynomolgus monkeys causes severe delayed anemia mediated by the induction of neutralizing antibodies against endogenous EPO (Chenuaud et al., 2004). Therefore, as a mechanism of delayed anemia in compound A-treated rats, we hypothesized that compound A induced antibodies to neutralize endogenous rat EPO, and consequently, erythroid differentiation was inhibited. By analyzing the plasma of rats using ELISA, we detected antibodies binding to rat and human EPO in rats that developed anemia following compound A administration, but not in vehicle-treated rats (Fig. 3A). Furthermore, the CFU assay showed that only the plasma from compound A-treated rats inhibited erythroid differentiation, although antibodies against antigen A were not induced (Fig. 3B). Given the differences in the size and amino acid sequence between EPO and antigen A, it is not plausible that antigen A induced antibodies that bound to EPO. Therefore, we explored the possibility of impurity inducing anti-EPO antibodies in compound A.
To explore the impurity as a causative agent of delayed anemia, we focused on hEPO-mRNA because anti-rat EPO and anti-human EPO antibodies were detected in compound A-treated rats. We also used hEPO-mRNA in other extraneous experiments and detected it in compound A by RT-PCR (Fig. 4A). This may have been a residual substance in the reverse-phase HPLC column used for the purification of the synthesized mRNA that was co-eluted with compound A during the purification process. Compound A and lot. 2 of compound B, in which hEPO-mRNA was detected, were purified using the same HPLC column as hEPO-mRNA, whereas another column was used for lot. 1 of compound B. In the repeated-dose study with trace amounts of hEPO-mRNA-LNP alone and compound B artificially spiked with hEPO-mRNA, the decreases in hematological parameters and anemia-related abnormal clinical signs was noted, similar to findings from compound A (Figs. 5A–D). Surprisingly, our experiment showed that the repeated intramuscular administration of just 0.16 or 1 µg hEPO-mRNA caused severe anemia in the rats. In contrast, for a toxicological evaluation, 40 µg mRNA-LNP vaccines, similar dose to a repeated-dose toxicity study of COVID-19 vaccine candidates where rats were intramuscularly administered 30 and/or 100 µg mRNA vaccine (Rohde et al., 2023), were administered. Although analytical parameters for impurities in mRNA vaccines have been established, specific regulations for impurity levels have yet to be established (USP, 2022). Currently, approved mRNA vaccines were required to present only analytical data. In the ICH guidelines for chemically synthesized new drug substances, impurities containing more than 0.15% should be evaluated for their biological safety (ICH, 2006). Although the guidelines do not cover mRNA vaccines, the amount of spiked mRNA in our experiments was close to the threshold. On the other hand, in a study by Sedic et al. (2018), the intravenous administration of hEPO-mRNA to rats at doses of 0.1 and 0.3 mg/kg induced human EPO in plasma, but its concentration subsequently decreased. This change was thought to be a consequence of the induction of neutralizing antibodies against the human EPO protein expressed by mRNA. Consistent with these observations, our data showed that the administration of hEPO-mRNA to rats induces anti-EPO antibodies. In summary, our study revealed that trace amounts of impurity mRNA can induce unexpected immune reactions and cause severe adverse effects, highlighting the importance of toxicity assessments that include impurities.
Non-specific immune reactions can be induced by the presence of microbial antigens and other contaminants that are inadvertently present during the manufacturing and purification of bio-nanopharmaceutical products. For example, nonspecific modulators may induce the production of anti-drug antibodies that affect drug efficacy (Bitounis et al., 2024). It has been reported that impurities in bio-nanopharmaceuticals derived from host cell proteins cause toxic changes, thus the toxicological evaluation of impurities in mRNA vaccines is important (Vanderlaan et al., 2018). The immune-mediated anemia observed in our study was caused by antibodies induced by unintended mRNA sequences that were impurities in the mRNA vaccine. In contrast, in a non-clinical toxicity study of an antibody drug against HM1.24, which is highly expressed on the surface of myeloma cells, the anti-HM1.24 antibody caused red blood cell agglutination in cynomolgus monkeys due to a non-specific reaction (Watanabe et al., 2003). These reports suggest that immune-mediated toxicity, as an adverse effect of vaccines, can be induced by both vaccine-expected antibodies and antibodies against impurities, such as proteins expressed from unintended mRNA sequences or host cell-derived proteins. Therefore, to assess all side reactions, vaccine formulations used in non-clinical safety evaluations should be equivalent to those used in clinical trials. In addition to the systemic toxicity and local reactions to vaccine administration, the immune response or unintended sensitization due to vaccine-induced antibodies, as well as toxicity due to impurities, should be evaluated (WHO, 2005).
In conclusion, our study demonstrates that trace amounts of hEPO-mRNA contained in mRNA-LNP vaccines can induce anti-EPO antibodies, which inhibit erythroid differentiation in the bone marrow and cause severe delayed anemia in rats. Immune-mediated anemia as an adverse effect of mRNA-LNP vaccine administration can be induced not only by antibodies against impurities but also by non-specific reactions of antibodies against the target protein. Immune-mediated toxicity is difficult to assess in non-clinical studies; thus, our toxicological evaluation approach for delayed toxicity may be useful in future safety evaluations.
The authors would like to thank Dr. Yuki Kato for the insightful feedback during manuscript writing and Ms. Masako Miyazaki for hematology and bone marrow analyses. We would like to thank Editage (www.editage.jp) for English language editing.
FundingNo funding was provided for the work.
Conflict of interestThe authors declare that there is no conflict of interest.
Data availabilityThe data in this study are included in the article/supplementary materials. Contact the corresponding author(s) directly to request the underlying data.
Author contributionsConceptualization: Yukie Murata, Shingo Kitamura, and Tomokazu Yoshinaga
Funding acquisition: Takafumi Sato, Osamu Yoshida and Tomokazu Yoshinaga
Investigation: Yukie Murata, Shingo Kitamura, Shun Kitahata, Keisaku Wakabayashi, Hidetoshi Kouno, Norikazu Kuroda, Kayo Ishida, and Shinobu Miki
Project administration: Takafumi Sato, Osamu Yoshida, Tomokazu Yoshinaga, and Tamio Fukushima
Supervision: Takafumi Sato, Osamu Yoshida, Akira Kugimiya, Tomokazu Yoshinaga, and Tamio Fukushima
Visualization: Yukie Murata, Shingo Kitamura, and Norikazu Kuroda
Writing – original draft: Yukie Murata, Shingo Kitamura, and Norikazu Kuroda
Writing – review & editing: Yukie Murata, Shingo Kitamura, Yoshiji Asaoka, Akira Kugimiya, Tomokazu Yoshinaga, and Tamio Fukushima
Ethical approval and consent to participateAll studies were conducted with the approval of the Institutional Animal Care and Use Committee (approval numbers S22031D-0003, S22120D-0001, and S22165D-0000) and performed in accordance with the animal welfare bylaws of the Shionogi Pharmaceutical Research Center, Shionogi & Co., Ltd., which is accredited by AAALAC International.
Patient consent for publicationNot applicable.
