The Majority of SARS-CoV-2 Plasma Cells are Excluded from the Bone Marrow Long-Lived Compartment 33 Months after mRNA Vaccination

The goal of any vaccine is to induce long-lived plasma cells (LLPC) to provide life-long protection. Natural infection by influenza, measles, or mumps viruses generates bone marrow (BM) LLPC similar to tetanus vaccination which affords safeguards for decades. Although the SARS-CoV-2 mRNA vaccines protect from severe disease, the serologic half-life is short-lived even though SARS-CoV-2-specific plasma cells can be found in the BM. To better understand this paradox, we enrolled 19 healthy adults at 1.5–33 months after SARS-CoV-2 mRNA vaccine and measured influenza-, tetanus-, or SARS-CoV-2-specific antibody secreting cells (ASC) in LLPC (CD19−) and non-LLPC (CD19+) subsets within the BM. All individuals had IgG ASC specific for influenza, tetanus, and SARS-CoV-2 in at least one BM ASC compartment. However, only influenza- and tetanus-specific ASC were readily detected in the LLPC whereas SARS-CoV-2 specificities were mostly excluded. The ratios of non-LLPC:LLPC for influenza, tetanus, and SARS-CoV-2 were 0.61, 0.44, and 29.07, respectively. Even in five patients with known PCR-proven history of infection and vaccination, SARS-CoV-2-specific ASC were mostly excluded from the LLPC. These specificities were further validated by using multiplex bead binding assays of secreted antibodies in the supernatants of cultured ASC. Similarly, the IgG ratios of non-LLPC:LLPC for influenza, tetanus, and SARS-CoV-2 were 0.66, 0.44, and 23.26, respectively. In all, our studies demonstrate that rapid waning of serum antibodies is accounted for by the inability of mRNA vaccines to induce BM LLPC.


Introduction
As of December 2023, SARS-CoV-2 (SARS2) has infected over 772 million people worldwide and killed seven million, including 1.2 million in the United States alone 1 .While the original wildtype SARS2 primary vaccine series and boosters have been effective against severe disease, hospitalization, and death, protection with sterilizing immunity against infection or transmission has not been entirely evident.SARS2 vaccines appear to provide lasting T cell responses.However, waning neutralizing antibody levels within 3-6 months results in notable breakthrough infection (BTI) or reinfections with the same strain [2][3][4] .Therefore, we asked whether subjects after SARS2 vaccination develop SARS2 spike-speci city in the long-lived plasma cell (LLPC) subset (CD19 − CD38 hi CD138 + ) of the human bone marrow (BM) 5 .For clarity, the term ASC refers to all antibody-secreting cells (ASC), which include early-minted ones (oftentimes referred to as plasmablasts) and more mature ASC known as plasma cells that can contain LLPC.
Early in the pandemic, reports of SARS2 spike-speci c IgG ASC were readily identi ed in the BM after SARS2 infection or vaccination 6,7 , or in the non-human primates after SARS2 spike protein vaccination 8 , suggesting long-lived humoral protection without serologic con rmation.Interestingly, BM ASC compartments are quite heterogeneous comprising of early-minted ASC (new arrivals) of which some progressively mature into LLPC [9][10][11][12][13][14] .How LLPC are generated is not entirely clear but after vaccination, the majority of ASC released from secondary lymph nodes are destined to undergo apoptosis unless they nally arrive in the specialized BM survival niches lled with mesenchymal stromal cells and myeloid cells, which provide important factors for survival and maturation such as IL-6 and APRIL 15,16 .These new arrivals can further differentiate into a mature long-lived phenotype, LLPC (CD19 − CD138 + ), which secrete neutralizing antibodies for decades 9,10 .Although the human BM is a reservoir of LLPC, new arrivals, including CD19 + CD138 − and intermediate phenotypes of CD19 + CD138 + ASC, make it quite heterogenous 17 , such that a mere presence to this locale may not assign durability.
Tetanus vaccination generates antigen (Ag)-speci c BM LLPC and affords safeguards for decades with a serologic half-life of 10 years 9,18 .For in uenza, humoral immune protection provided by in uenza vaccines typically wanes within 4-6 months 19 ; however, natural infection provides long-lasting immunity as shown when elderly adults maintained neutralizing antibodies to the 1918 Spanish in uenza virus nearly 90 years after the primary infection 20 .Infants may have preexisting maternally derived antiin uenza antibodies although they wane over the rst 6 months of life 21 .Unvaccinated individuals are estimated to have their rst in uenza infection within 5 years of birth 22 and to be infected with a new in uenza virus strain every 3-7 years 23 .Furthermore, newly induced immune responses are enhanced owing to cross-reactive antibodies from infections and reinfections with antigenically similar in uenza virus strains [24][25][26][27][28] .Although in uenza vaccine induces short-lived protection, natural infection to in uenza viruses generates long-lasting humoral immunity to the infecting strain 19,20 .
Here, we measure SARS2 spike-speci c ASC in multiple BM compartments up to 33 months after vaccination and compare them to well-known long-lived responses such as tetanus-and in uenzaspeci c ASC by ELISpots.We also validate these results from the cultured BM ASC with secreted antibodies in the supernatants with new bead-based methods.Both assays show that SARS2 IgG responses are excluded from the BM LLPC compartment as early as two months and up to nearly three years after immunization, although they are readily detectable in the non-LLPC subsets of the human BM (CD19 + CD138 − and CD19 + CD138 + ).In contrast, from the same individuals, tetanus-and in uenzaspeci c ASC are easily detected in both the BM CD19 + (non-LLPC) and CD19 − (LLPC) compartments.In all, unlike long-lived speci cities, SARS2-speci c ASC are mainly excluded from the BM LLPC in healthy subjects, despite being present in mature CD19 + CD138 + ASC.This nding provides a mechanistic explanation for the short duration of antibody responses to SARS2 mRNA vaccines.Together with our previous demonstration that LLPC mature from earlier BM ASC precursors, this study provides an experimental model to identify the requirements for full differentiation of BM LLPC.

Demographic and clinical characteristics of the 19 BM subjects
From May 2021 until October 2023, we enrolled 19 healthy adults aged of 20-65 years old (Fig. 1a).The subjects were recruited for BM aspirates 1.5-33 months after receiving the rst dose of SARS2 mRNA vaccines.All received a total of 2 to 5 vaccine doses, and BM aspirates were obtained 0.5-21 months after receiving the last booster (the third, the fourth, or the fth dose) (Table 1).One subject provided 3 longitudinal BM samples over a period of 21 months, resulting in a total of 21 BM aspirates.Five subjects reported infection with SARS2 2-17 months prior to the BM collection, of which two subjects had infection once and three had two subsequent PCR proven SARS2 infections.These infections occurred 1-15.5 months after receiving the most recent vaccine dose (either the rst, the second, or the third booster dose).Two of these 5 subjects had infections before vaccination (hybrid immunity) and post-vaccination infection was ruled out by symptomatic cases proven by rapid antigen or PCR testing in all other subjects.All 19 individuals received the quadrivalent in uenza vaccine within 1-12 months (relative to the time of each BM aspirate) and one was delayed 1 year due to the pandemic.All received the childhood series of the tetanus toxoid vaccine with recent boosters ranging from one month to 24 years from the time of BM aspirates.
BM non-LLPC (PopA and PopB) and LLPC (PopD) subsets and antigen validation BM ASC subsets were FACS-sorted according to surface expression of CD19, CD38, and CD138, as previously described 9 (Fig. 1b).To overcome the problem with the rapid death of ASC ex vivo 15,16 , we rested the ASC overnight in a new human in vitro plasma cell survival system which is capable of maintaining ASC viability for months 15 .Since we had previously localized the BM LLPC compartment into PopD (CD19 − CD38 hi CD138 + ) 9,16 , this population was sorted out of total BM ASC together with PopA (CD19 + CD38 hi CD138 − ) and PopB (CD19 + CD38 hi CD138 + ), and tested for total IgG secretion as well as SARS2-, in uenza-(Flu-), and tetanus toxoid-(Tet-) speci c IgG secretion by bulk ELISpots.To optimize antigen detection for the BM assays, we collected early-minted blood ASC (CD27 hi CD38 hi ; Suppl.Fig. S1) 6-7 days after SARS2, Tet, or Flu vaccine, which is the peak time for enrichment of vaccine-speci c ASC in the blood after secondary immunization 29,30 , and performed ELISpots (Suppl.Fig. S2a).Of the SARS2 antigens (S1, S2, RBD, S2P, NTD, and NP proteins), S2P, a prefusion-stabilized SARS2 spike trimer 31 , generated the highest frequency, followed by S1 (with no signi cant difference; p = 0.21) (Fig. 2a,b), and so S2P was selected for BM ELISpot assays.We also validated the quadrivalent Flu vaccine (seasons of 2019-20 to 2023-24) and Tet antigen (Suppl.Fig. S2b,c), which are enriched in blood ASC at the peak of the respective vaccines 29,30 .

Exclusion of SARS2-speci c BM IgG PopD as measured by ELISpots
Since BM aspirates can yield variable cell numbers, we included BM aspirates with > 3,000 sorted cells in each of the three ASC populations and performed bulk ELISpots (Fig. 2c).Among the 21 BM samples, su cient cells to con dently measure vaccine-speci cities within PopA, PopB, and PopD were obtained from 8, 15, and 17 samples.As previously shown, all BM ASC subsets had detectable total IgG ASC.

Kinetic responses for IgG ASC in a longitudinal BM aspirate collection
We next assessed the IgG ASC kinetic response in a single subject who provided 3 sequential BM aspirates over a period of 23 months after the rst vaccine.BM aspirates were taken 10, 62, and 94 weeks after the rst SARS2 vaccine dose (or 6, 58, and 90 weeks after the second vaccine dose) (Table 1).Each BM aspirate provided > 3,000 FACS sorted ASC in each subset.Again, total IgG ASC were detected in all BM PopA, PopB, and PopD.We observed an increase in the frequencies of S2P-speci c IgG ASC in PopA and PopB at 62 weeks (1.07%and 9.02% respectively) and 94 weeks (3.98% and 6.24% respectively) compared to the rst time point (0.90% and 0.38%, respectively) (Fig. 2g,h).However, in PopD, there were no S2P-speci c IgG spots detected at the rst two time points and only 0.31% at 94 weeks post-vaccination.Notably, at the earliest time-point (10 weeks), the highest S2P-speci c IgG ASC frequency was observed in PopA which then increased in PopB at 62 and 94 weeks.In all, regardless of time points, the S2P-speci c ASC frequencies were always higher in PopA and PopB compared to PopD, which was quite rare even at 94 weeks.As expected, we observed the highest Flu-and Tet-speci c frequencies in PopD, followed by PopB, and lowest in PopA (Fig. 2g,h).Interestingly, the Flu-and Tetspeci c frequencies were quite consistent over the course of two years in these BM subsets.
Exclusion of SARS2-speci c IgG secreted in PopD as measured in culture supernatant Next, to validate the antigen-speci c ELISpot responses, we measured secreted IgG from BM ASC subsets using a novel plasma cell culture system (Fig. 3a; see also Methods).Brie y, from the 8 individuals that yielded su cient sorted cells, we cultured BM ASC subsets (PopA, PopB, and PopD) in a specialized in vitro BM mimetic system overnight 15 and measured the cultured supernatants for IgG speci c for Flu, Tet, and S2P by multiplex bead-binding assays (Luminex) 33 .The results were similar to the ELISpot numbers (Fig. 3b).3b).In comparison, the fold change within PopB for Flu:SARS2 was 1.7 and for Tet:SARS2 was 0.9, demonstrating similar quantities of IgG to Flu, Tet, and SARS2 in PopB.Ultimately, using this method of measuring secreted antibodies from the cultured BM ASC, the ratios of non-LLPC:LLPC for Flu-, Tet-, and SARS2 from BM ASC culture supernatant were 0.66, 0.44, and 23.26, which was similar to the ELISpot results (Fig. 3c).In all, we validated the antigen speci cities observed by the ELISpots using our novel in vitro plasma cell culture method which also showed exclusion of SARS2-speci c ASC in PopD.

Discussion
In this study, we show that although there is an abundance of SARS2-speci c ASC in the BM short-lived plasma cell compartment, these cells are largely excluded from the LLPC compartment.This phenomenon is in stark contrast to in uenza-and tetanus-speci cities which are inherent to the BM LLPC.Hence, the lack of SARS2-speci c ASC in the LLPC provides a mechanistic explanation of the short-lived serology of mRNA vaccines and how these mRNA platforms are unable to induce the LLPC formation.Early-minted ASC typically journey to the BM microniches and require time to fully mature into LLPC.Fundamental changes of the new ASC arrivals are required for maturation into LLPC.These BM maturation programs where an early-minted ASC undergoes dramatic morphological, transcriptional, and epigenetic modi cations together with metabolic alterations provide the nal maturation steps to become a LLPC 10 .Increased Ig transcripts 11 , increased unfolded protein response (UPR) 34 , and anti-apoptotic 10 and autophagy 35 programs are just a few of the pathways involved in ASC maturation 36 .Because this progression is arduous, not all the new arrivals can ultimately complete this entire LLPC process.Moreover, the exact kinetics of this maturation also remain unknown.In all, our study offers major insights into how the mRNA vaccines fail to induce the necessary ASC precursors with programs that are required to fully mature into LLPC.
The exact mechanisms of how LLPC are formed are still under investigation.At one time, it was thought that all human ASC had the potential to become LLPC by simply migrating to environments rich in survival factors; but recent evidence shows how imprinting of an early-minted ASC at the time of priming in addition to terminal maturation in survival niches endow particular properties for durability.LLPC are thought to come from memory B cells (mBC) 37 .Thus, a longer interval between prime-boost vaccine strategies for mBC formation may play a role in the generation of LLPC.Also, the phenotype of mBC such as FcRL5 + and the cytokine milieu i.e.IFNγ may also play key roles in the generation of LLPC 38 .It is likely that imprinting at the time of B cell induction also determines the LLPC fate but with SARS2 mRNA vaccines, they fail to imprint these LLPC programs.Finally, even after 33 months after the vaccine, we show that SARS2-spec c ASC are still excluded from the BM LLPC compartment.Thus, a longer tincture of time is unlikely to ll the LLPC subset.
In the patient with sequential BM aspirates nearly two years after the initial vaccine, there are two explanations for the abundant S2P speci city in PopA at 94 weeks and PopB at 62 and 94 weeks.Conventionally, PopA and PopB are considered the result of more recent immune responses.Thus, the very high frequency in PopA and PopB suggests another exposure to SARS2 antigens such as breakthrough asymptomatic infections temporally close to the corresponding BM sampling.Asymptomatic infections have been well described with the emergence of the highly transmissible Omicron variants 23 .Thus, assuming an asymptomatic infection, even 94 weeks after vaccination and infection(s), S2P-speci c ASC cannot ll the LLPC compartment.Alternatively, it could be argued that lymph node S2P-speci c IgG ASC take time to migrate to the BM microenvironment and re ect the product of ongoing germinal center (GC) reactions, which have been documented to last for up to 6 months after vaccination which is well after the peak of early-minted ASC responses in the blood (day 7) 7 .This argument still emphasizes the fact that even two years after the vaccines, PopB cannot differentiate into LLPC even with ongoing persistent GC.
Our results are consistent with recent BM studies by Tehrani et al. demonstrating that most spike-speci c ASC are detectable in the CD19 + compartments post-SARS2 infection alone 39 .However, in this study, BM sampling was collected only 5-8 months post-illness and not up to 3 years as in our study 39 .Most importantly, this study used frozen BM ASC which are less reliable due to the fragility of BM ASC populations upon thawing.Also, the authors did not include longitudinal samples, IgA isotypes, Flu speci city, or PopA.Nonetheless, 5-8 months after infection alone, SARS2-speci c ASC were also excluded in LLPC, similar to the nding in our study after vaccination.
In another ow cytometry-based BM study, Schulz et al. also found S1-speci c responses predominately in the BM CD19 + compartment after vaccination 40 .In this study, BM samples were collected during hip joint replacement surgery from patients of up to 17 months post-SARS2-vaccination.While no Flu-or Tetspeci cities were evaluated, the authors noted some SARS2 speci city in the CD19 neg ASC compartment and concluded they were in fact long-lived, but these SARS2 speci c ASC were notably in the CD45 + (of CD19 neg ) ASC subset 40 .The majority of LLPC as originally de ned in Halliley et al. demonstrate downregulated CD45 in our LLPC 9 (Suppl.Fig. S5), which is also consistent with previous studies 41,42 .
Interestingly, in concordance with our study, in Schulz et al, the CD19 neg CD45 neg subset which includes the majority of our previously de ned LLPC also excluded the S1-speci c responses 40 .Hence, the bona de LLPC which may be a subset of the CD19 neg BM ASC population likely harbors Flu-and Tetspeci cities as well as measles-and mumps-speci cities but appears to exclude SARS2-speci c responses.
We cannot rule out a subset of BM PopB cells which may be an intermediary population on the road to maturing into the LLPC.Our previous single-cell transcriptional data showed that the most mature BM clusters with aggregated LLPC also contained some PopB 11 .Thus, simple surface markers CD19 and CD138 may be too blunt an instrument to dissect the heterogeneity of BM PopB.PopB likely contains some new arrivals as well as some early mature BM ASC subsets since Tet-speci c responses reside in PopB and LLPC years after vaccination albeit at much lower frequencies in PopB compared to LLPC.Ultimately, additional dissection into the transcriptional and epigenetic differences in Tet-versus S2Pspeci c PopB (CD19 + CD138 + ) cells may reveal important mechanistic differences in forming long-lived ASC.
Although the emergence of new viral variants confounded serum protection, in this study, we focused on responses against the original virus and the wildtype vaccines, knowing that they rapidly wane within 3-6 months regardless of the vaccine platform (mRNA and viral vector-based vaccines like adenovirus (Ad) vectors) 3,4 .The unstable nature of mRNA and the resultant transient expression of the spike protein during induction might explain the lack of sustained antibody responses.Interestingly, the Ad vectors persist for weeks, yet speci c humoral immunity is also short-lasting 3,4 .Furthermore, recent evidence showed that the mRNA vaccine platforms do not induce strong type I IFN responses 43 .In vitro and animal studies suggest that type I IFN may play a role in LLPC 44,45 , but whether it is important for the generation of in vivo human LLPC will need further studies.Given both the mRNA and Ad vector vaccine platforms induce strong germinal center reactions and interactions with Tfh cells, the mechanisms underlying their failure to generate LLPC are even more puzzling 2 suggesting dysfunction in the maturation process in BM microniche.In all, we merely suggest that before moving the current vaccines to the mRNA platforms, more studies on the durability of humoral immunity are needed.
Could the limited durability of neutralizing antibody responses be due to the structural nature of the spike protein itself and thus limited only to coronavirus vaccines?Coronaviruses lack highly repetitive organized structures or pathogen-associated structural patterns 46 .Most RNA viruses that induce longlasting antibody immunity have on their surface rigid repetitive structures spaced 5-10nm 47 .In coronaviruses, the long spike proteins are embedded in a uid membrane which are often loosely oating and widely spaced at 25nm apart 46 .Therefore, the inherent nature of the spike protein on coronaviruses itself may be an issue in B cell activation 47 since neutralizing antibody responses to seasonal human coronaviruses, as well as SARS-CoV-1 and MERS-CoV, are also short-lived 2 .Perhaps the long widely spaced spike protein structure of the coronaviruses may play an important role in antibody durability.
There are limitations in our study.First, our sample size is relatively small especially of those after vaccine and infection.Second, the infections were self-reported symptoms that warranted testing, so any asymptomatic infections were not con rmed.Third, primary BM ASC are rare cell types and BM aspirates are di cult to obtain and interrogate; thus, not all samples provided su cient cells in each BM subset.Lastly, we had limited longitudinal and sequential samples with the longest at 33 months since the rst SARS2 vaccine dose.Of course, it would be important to assess the BM compartment decades after the primary vaccine and as new variant SARS2 viruses continue to circulate.
In conclusion, the holy grail of vaccinology is the generation of LLPC.Our ndings demonstrate the exclusion of SARS2 speci city in the BM LLPC compartment and provide novel insights of how the mRNA vaccines fail to induce necessary precursor programs to fully mature into BM LLPC.These ndings have implications for the need to improve COVID-19 vaccination.Whether optimizing vaccine regimens or immunization schedules, engineering different spike proteins, or formulating vaccine adjuvants and delivery systems will need better understanding.Finally, interrogating the LLPC require invasive BM aspirates and a greater tincture of time (likely years); thus, identifying early biomarkers of early-minted blood ASC or LLPC precursors are needed to help predict the durability of new platform vaccines.

Methods
Healthy human subjects.A total of 21 BM aspirate samples were obtained from 19 healthy adult donors who received Tdap, in uenza, COVID-19 primary and booster vaccines.Select demographic and clinical characteristics of the subjects can be seen in Fig. 1a.Detail information on BM subjects and BM samples can be found in Table 1.For antigen validation, peripheral blood samples were obtained from 64 healthy subjects who received vaccines for Tdap, in uenza, or COVID-19 (the 3 rd , 4 th , or 5 th dose) at 5-7d prior to sample collection.For CD45 ow cytometric staining, an additional of ve healthy BM subjects were obtained, acquired, and analyzed.All studies were approved by the Emory University Institutional Review Board Committee.
Blood and BM ASC bulk cultures and ELISpot assays.HumanASC cultures were conducted in MSC secretome (ASC survival medium) and in hypoxic conditions (2.5% O 2 ) at 37°C, as previously described 15 .
This culture system is called plasma cell survival system (PCSS) 48 .IgG and IgA secretion of cultured ASC was assessed by ELISpot assays, which quantitated IgG-and IgA-secreting cells.These assays used goat anti-human IgG or IgA for total IgG or IgA capturing, respectively, and alkaline phosphatase-conjugated goat anti-human IgG or IgA, respectively, for detection, and were performed as previously described 15 .
Multiplex bead binding assays (MBBA).Multiplex bead binding assays (MBBA) were performed on the supernatants collected from culture of BM ASC puri ed from 8 individuals who provided su cient postsort cells for all 3 subsets.For total IgG, biotinylated goat anti-human IgG (Southern Biotech) was conjugated to avidin-coupled MagPlex-avidin microspheres of spectrally distinct regions as previously described 49 .For antigen-speci c MBBA, antigens were conjugated to MagPlex microspheres of spectrally distinct regions (Luminex) via standard carbodiimide coupling procedures as previously described 33 .Antigen speci c and total IgG MBBA were performed using a Luminex FLEXMAP 3D instrument (Luminex) as previously described 33 .All viral protein coupled microspheres were tested together as a combined multiplex antigen-speci c immunoassay; all anti-human immunoglobulin coupled microspheres were tested together as a combined multiplex total Ig immunoassay.Median uorescent intensity (MFI) using combined or individual detection antibodies was measured using the Luminex xPONENT 4.3 software at enhanced PMT setting.The net MFI was obtained by subtracting the background value.The culture supernatant MFI values were normalized to the relative IgG concentrations (pg/mL) based on the total human IgG standard curves, followed by normalization of these resultant IgG concentrations (pg/mL) to the ASC input numbers and duration of culture (day).The MFI normalization was performed based on the equation as shown in Suppl.Fig. S4.Data were expressed as the percents or ratios of the titers of Ag-speci c IgG to those of total IgG.All the supernatants were collected after 1 day in culture of off-sorter BM ASC subsets and were tested undiluted (neat) or 1:2 diluted -except for the total IgG titrations which were also assayed at further dilutions (e.g.1:4, 1:8, 1:16, 1:32, 1:64, 1:128, and 1:256).Statistics.All statistics were assessed using Student's t-test (two-tailed unpaired t-test) or one-way ANOVA performed with GraphPad Prism (v8.4.2;GraphPad Software).Data are presented as mean±SD and differences were considered signi cant at p values less than 0.05.
Author Contributions Conceptualization: DCN and FEL.Methodology: DCN, ITH, AMP, DS, NSH, CC.Resources and sample acquisition: PAL, JA, DR, SL, IS, and FEL.Funding acquisition: FEL.Supervision: FEL.Writing -original draft: DCN and FEL.All authors have reviewed, edited, and approved the nal manuscript.

Figures
Figures