key: cord-0430526-85zzw8nu
authors: Waickman, Adam T.; Victor, Kaitlin; Newell, Krista; Li, Tao; Friberg, Heather; Foulds, Kathy; Roederer, Mario; Bolton, Diane L.; Currier, Jeffrey R.; Seder, Robert
title: mRNA-1273 vaccination protects against SARS-CoV-2 elicited lung inflammation in non-human primates
date: 2021-12-27
journal: bioRxiv
DOI: 10.1101/2021.12.24.474132
sha: e78ec02c04ff187bdaffb3fa480f3389b2265d9d
doc_id: 430526
cord_uid: 85zzw8nu

Vaccine-elicited SARS-CoV-2 antibody responses are an established correlate of protection against viral infection in humans and non-human primates. However, it is less clear that vaccine-induced immunity is able to limit infection-elicited inflammation in the lower respiratory tract. To assess this, we collected bronchoalveolar lavage fluid samples post-SARS-CoV-2 strain USA-WA1/2020 challenge from rhesus macaques vaccinated with mRNA-1273 in a dose-reduction study. Single-cell transcriptomic profiling revealed a broad cellular landscape 48 hours post-challenge with distinct inflammatory signatures that correlated with viral RNA burden in the lower respiratory tract. These inflammatory signatures included phagocyte-restricted expression of chemokines such as CXCL10 (IP10) and CCL3 (MIP-1A) and the broad expression of interferon-induced genes such as MX1, ISG15, and IFIT1. Induction of these inflammatory profiles was suppressed by prior mRNA-1273 vaccination in a dose-dependent manner, and negatively correlated with pre-challenge serum and lung antibody titers against SARS-CoV-2 spike. These observations were replicated and validated in a second independent macaque challenge study using the B.1.351/beta-variant of SARS-CoV-2. These data support a model wherein vaccine-elicited antibody responses restrict viral replication following SARS-CoV-2 exposure, including limiting viral dissemination to the lower respiratory tract and infection-mediated inflammation and pathogenesis. One Sentence Summary Single cell RNA sequencing analysis demonstrates that mRNA-1273 vaccination limits the development of lower respiratory tract inflammation in SARS-CoV-2 challenged rhesus macaques

mRNA-based vaccine platforms -such as Moderna's mRNA-1273 and Pfizer/BioNTech's BNT162b2 -141 which encode a stabilized version of the SARS-CoV-2 spike glycoprotein (23) show >90% efficacy 142 against symptomatic COVID-19 in initial Phase 3 analyses and in large-scale prospective studies 143 performed after their global rollout (24, 25) . However, the efficacy of these vaccines against severe 144 lower airway disease wanes over time after the initial prime and boost (26) (27) (28) . Pre-clinical and clinical 145 studies have strongly suggested that vaccine-elicited serum levels of SARS-CoV-2 neutralizing antibody 146 titers are a mechanistic immune correlate of vaccine efficacy (29, 30) . Despite the abundance of clinical 147 and pre-clinical efficacy data for these mRNA-based vaccine platforms, there is little prospective 148 information currently available on how these vaccines impact SARS-CoV-2-elicited inflammation in the 149 lower respiratory tract with any degree of spatial or temporal resolution. 150 151

In this study we sought to understand the impact of mRNA-1273 vaccination on the cellular 152 inflammatory response to SARS-CoV-2 infection in the lower respiratory tract of nonhuman primates 153 (NHPs,) and whether vaccination is capable of breaking the inflammatory feedback loop that 154 characterizes severe COVID-19. We used scRNAseq to analyze BALF cells from rhesus macaques 155 challenged with SARS-CoV-2 strain USA-WA1/2020 after vaccination with two doses of 30µg or 1µg 156 of mRNA-1273 or PBS. mRNA-1273 vaccination limited SARS-CoV-2 elicited inflammation in the 157 lower respiratory tract as defined by the expression of pro-inflammatory chemokines and cytokines in 158 multiple cell types, as well as the broad reduction in expression of interferon gene products such as 159 MX1, ISG15, and IFIT1. Additionally, SARS-CoV-2 elicited inflammation was directly associated with 160 post-challenge viral titers and inversely associated with pre-challenge antibody levels in unvaccinated 161

and mRNA-1273 vaccinated animals. The ability of mRNA-1273 to limit SARS-CoV-2 elicited 162 inflammation in the lower reparatory tract was independently verified using the antigenically disparate 163 B.1.351/beta variant. Collectively, these results demonstrate that vaccination with mRNA-1273 not only 164 limits SARS-CoV-2 viral replication, but restricts inflammation in NHPs. Additionally, these data 165 support a model wherein neutralizing antibody at the site of virus inoculation reduces the viral burden, 166 constraining upper respiratory tract viral replication and secondary viral dissemination to the lower 167 respiratory tract and infection-associated inflammation. 168 169 RESULTS 170 171

Frequency of BALF resident cells following SARS-CoV-2 challenge. It has previously been 172 demonstrated that vaccination of macaques with mRNA-1273 results in robust serum antibody responses 173 and high-level protection from subsequent SARS-CoV-2 challenge in a dose-dependent fashion (19, 29) . 174 To extend these observations and to assess the impact of vaccination on SARS-CoV-2 elicited 175 inflammation in the lower respiratory tract, we performed scRNAseq analysis of fresh BALF obtained 176 on days 2 and 7 post SARS-CoV-2 challenge in animals which previously received either 30 µg (n=4) or 177 1 µg (n=6) of mRNA-1273 in a prime-boost series administered four weeks apart. Control animals 178 received PBS (n=6). All animals were challenged intranasally/intratracheally (IN/IT) with 8x10 5 PFU of 179 SARS-CoV-2 (strain USA-WA1/2020) four weeks after the last vaccine dose. In addition, BALF cells 180

were analyzed from naïve uninfected animals to serve as controls. 181 182

scRNAseq was used to classify and quantify the cell composition and dynamics within the BALF post-183 challenge. A total of 65,226 viable and high quality BALF cells from all animals were recovered after 184 filtering and quality control steps (Fig. 1A, 1B) . Of note, epithelial cells (Fig. 1C) , lymphocytes ( Fig.  185 1D), dendritic cells (Fig. 1E) and macrophages (Fig. 1F) were identified in all time points from all 186 animals. Alveolar macrophages were further separated into either MARCOor MARCO + populations, 187

corresponding to interstitial and tissue-resident alveolar macrophages, respectively (14, 31). Following 188 SARS-CoV-2 challenge, CD4 + and CD8 + T cells increased in frequency between days 2 and 7 post-189 challenge in unvaccinated animals. Several DC populations also trended higher among unvaccinated 190 infected animals at one or more time points relative to uninfected controls. No significant changes were 191 observed in the frequency of epithelial cell or macrophage populations. context of acute SARS-CoV-2 infection in humans (4). In addition, elevated expression of cytotoxic 205 factors GZMA and PRF1 was observed in CD8 + T cells following SARS-CoV-2 challenge on day 2 and 206 maintained 7 days post challenge. Notably, the expression of these pro-inflammatory chemokines and 207 chemokines was dramatically suppressed in vaccinated animals in a dose-dependent manner across all 208 time points. 209 210

To reduce the complexity of the data and provide more direct insight into the dynamics of SARS-CoV-2 211 elicited inflammation, we defined a transcriptional "inflammation index" which could be used to 212 quantify the level of enrichment for inflammatory gene products in a given sample and cell type. This 213 index was developed by selecting 8 genes (MX1, MX2, IFIT1, IFIT2, IFIT3, IFI6, ISG15 , and ISG20) 214 that were 1) previously known to be regulated at a transcriptional level by viral infection and/or 215 interferon stimulation, 2) highly induced in our dataset following SARS-CoV-2 challenge, and 3) 216

consistently observed in all cell types captured in our analysis. Using this reductionist approach, we 217 observed a dose-dependent suppression of SARS-CoV-2 elicited inflammation in epithelial cells 218

(pneumocytes, club cells), myeloid cells (MARCO + macrophages, MARCOmacrophages, mast cells), 219 dendritic cells (cDC.1, cDC.2, pDC, Mig DC), and lymphocytes (B cells, CD8 + T cells, CD4 + T cells) 220

with increasing mRNA vaccination dose ( Fig. 2C-F) . Furthermore, inflammation in animals that 221 received the full dose of vaccine was nearly equivalent to that of the unchallenged control animals in all 222 cell types assessed. Inflammation returned to baseline in all groups by day 7 post vaccination. These 223 results establish a single metric for quantifying the transient inflammatory transcriptional response 224 elicited following SARS-CoV-2 infection across multiple cell populations using scRNAseq, and by 225 extension provide a measurement of the site-specific host-response to the virus. 226 227

Cell-associated viral RNA burden following SARS-CoV-2 infection. Having defined the impact of 228 mRNA-1273 vaccination on SARS-CoV-2 associated inflammation in the lower respiratory tract of 229 macaques, we next attempted to quantify the cell-associated SARS-CoV-2 viral RNA burden by 230 aligning scRNAseq reads that failed to align to the macaque genome against the SARS-CoV-2 USA-231 WA1/2020 reference genome. BALF contained widespread SARS-CoV-2 RNA + cells on day 2 in 232 unvaccinated animals (Fig. 3A, 3B) . SARS-CoV-2 RNA + positive cells were seen in all annotated cell 233 types with the exception of mast cells in unvaccinated animals, although the greatest number of viral 234 RNA + cells were found in the MARCOmacrophage cluster. Similar to the inflammation index, the 235 frequency of viral RNA + cells was suppressed by vaccination in a dose-dependent fashion and mostly 236 resolved by day 7 post infection. The frequency of viral RNA + cells in the BALF on day 2 correlated 237 well with contemporaneous viral subgenomic RNA (sgRNA) load in the BALF as quantified by PCR of 238 the E and N gene (Fig. 3C, fig. S1 ). Notably, the correlation between the frequency of viral RNA + cells 239

in the BALF was weaker with the upper respiratory tract (nasopharyngeal swab) sgRNA loads (Fig. 3D,  240 fig. S1). These results show that SARS-CoV-2 viral burden in lung cells is abrogated by mRNA 241 vaccination and is consistent with reduced soluble viral RNA measures in BALF.

Relationship between SARS-CoV-2 RNA load and cell type-specific inflammation. To examine the 244 relationship between the observed dose-dependent reduction in SARS-CoV-2 viral burden and 245 inflammation in the BALF of mRNA-1273 vaccinated animals after SARS-CoV-2 challenge, we 246 compared viral RNA measures to cellular inflammatory responses. Strikingly, cell-free SARS-CoV-2 247

RNA load positively correlated with the previously defined inflammation index score of both BALF 248 dendritic and myeloid cell compartment on day 2 post infection across all study groups (Fig. 4A, Fig.  249 4D, fig. S2 ). However, nasal swab viral RNA load poorly correlated with dendritic cell inflammation, 250

and correlated only weakly with myeloid inflammation (Fig. 4B, Fig. 4E, fig. S2 ). Cell-associated viral 251 RNA loads in the BALF also correlated with the inflammation score for both dendritic and myeloid 252 compartments (Fig. 4C, Fig. 4F ). These results demonstrate that viral burden in the BALF, but not nasal 253 environment, correlates with the amount of lower respiratory tract inflammation following SARS-CoV-2 254 challenge. 255 256

Pre-challenge immune profiles predict lung inflammation following SARS-CoV-2 challenge. 257 SARS-CoV-2-specific antibody titers have been implicated in mRNA vaccination-mediated protection 258

from SARs-CoV-2 infection in both humans and NHPs (29, 30, 32) , but the relationship between 259 specific antibody titers and lower respiratory tract inflammation is not clear. To this end, we 260 incorporated previously published data (29) on serum levels of full-length spike protein and receptor-261 binding domain (RBD)-specific titers IgG present immediately before SARS-CoV-2 challenge in these 262 animals into our analysis. Pre-challenge (8 week post initial vaccine dose) serum titers of both spike-263

and RBD-specific IgG were negatively correlated with dendritic cell inflammation scores in the lower 264 respiratory tract 2 days post SARS-CoV-2 challenge across all study groups ( Fig. 5A to B) . This 265 relationship was also observed with pre-challenge spike-specific IgG titers in the BALF (Fig. 5C ).

Furthermore, serum neutralizing antibody responses assessed by both pseudovirus and live-virus 267 neutralization assays were also associated with reduced DC inflammatory responses ( Fig. 5D to E).

These data suggest that SARS-CoV-2 specific antibody titers in mRNA-1273 vaccinated macaques 269 function as a powerful predictor of SARS-CoV-2 elicited inflammation in the lower respiratory tract. following WA-1 challenge were observed following B.1.351/beta challenge (Fig. 6A, fig. S3 ). However, 279

unlike USA-WA1/2020 challenge, infection with B.1.351/beta resulted in a significant perturbation in 280 the abundance of multiple cell types including pDCs and migratory DCs ( fig. S4 ). These changes in 281 cellularity were not observed in mRNA-1273 vaccinated animals, and the production of chemokines, 282 cytokines, and cytolytic factors were again suppressed in vaccinated animals relative to their 283 unvaccinated counterparts (Fig. 6B) . Vaccination with mRNA-1273 also suppressed SARS-CoV-2 284 associated inflammation observed in epithelial cells (Fig. 6C) , dendritic cells ( Fig. 6D) , myeloid cells 285 (Fig. 6E) , and lymphocytes (Fig. 6F) . The frequency of SARS-CoV-2 RNA positive cells in BALF was 286 also reduced by mRNA-1273 vaccination, with the greatest number of viral RNA + cells again found in 287 the MARCOmacrophage cluster ( fig. S5 ). In their totality, these results indicate that vaccination with 288 mRNA-1273 is capable of limiting lower airway inflammation in macaques following challenge with 289 multiple antigenically and evolutionally divergent strains of SARS-CoV-2. 290 291 DISCUSSION 292 293

In this study we sought to provide functional and mechanistic insight into the properties of mRNA-1273 294 elicited protection from SARS-CoV-2 challenge in a widely used nonhuman primate model of mild to 295 moderate COVID-19 disease. While immune correlates of protection from symptomatic SARS-CoV-2 296

infection are currently being assessed and defined in both clinical and pre-clinical studies, there is a 297 more limited information on the impact of vaccine-elicited immunity on SARS-CoV-2 induced 298 inflammation in the lungs at the single cell level. Here, we utilized scRNAseq technology to analyze 299 BALF cells from mRNA-1273 vaccinated animals that were subsequently challenged with SARS-CoV-300 2 to define the transcriptional signatures of infection and to ascertain how this inflammatory response is 301 modulated by vaccine-elicited adaptive immune responses in a dose-dependent fashion. SARS-CoV-2 302 infection induced a robust inflammatory response in all unvaccinated animals that was suppressed in a 303 dose-dependent fashion by mRNA-1273 vaccination. Notably, migratory DCs and MARCO -304 macrophages appeared to be the most responsive cell types in the lower respiratory tract to SARS-CoV-305 2 infection, as indicated by chemokine and cytokine production. Cell-associated SARS-CoV-2 viral 306 RNA was readily detected in the BALF of unvaccinated animals, restricted by mRNA-1273 vaccination, 307

and correlated with dendritic and myeloid cell inflammation. 308 309

It was previously established that S-specific antibody responses elicited by mRNA-1273 vaccination 310

correlates with upper and lower airway control of SARS-CoV-2 replication in macaques after challenge 311 (29). Here we further demonstrate, in these same animals, that the pre-challenge antibody profile, 312

including titers of binding and neutralizing antibody, predicted and inversely correlated with the 313 inflammatory profile within the lung across multiple cell types. The high degree of correlation between 314 pre-challenge antibody titers, post-challenge viral loads, and post-challenge inflammation suggests a 315 model of mRNA-1273-mediated protection from SARS-CoV-2 challenge in nonhuman primates. 316

Namely, neutralizing antibody levels determine the burden of viral replication, and the amount of virus 317 persisting in the upper respiratory tract drives secondary viral dissemination to the lower respiratory 318 tract and infection-attendant inflammation. The absence of inflammation and viral RNA in the lower 319 respiratory tract of SARS-CoV-2 challenged animals just 2 days post infection supports the high level of 320 efficacy of mRNA vaccination against lower respiratory infection and pathology. 321 322

Examination of human BALF from mild and severe COVID-19 patients has been used to distinguish an 323 inflammatory signature associated with severity. This signature includes expression of genes for 324 chemokine production, proinflammatory cytokines, and activated phenotypic markers within resident 325 and infiltrating cells (5-10). Several studies have described the potential role of neutrophils and NETosis 326

in local lung pathology, as well as shifts in monocyte and macrophage populations toward an 327 inflammatory phenotype (5-7, 9, 10) . Reinforcing the critical role of type I interferon in the antiviral 328 response, many studies have also identified strong type I IFN signatures in single immune cells from 329 COVID-19 patients, although the relationship of this cytokine profile and disease severity is still 330 uncertain (12, 33). Accordingly, our observation of stronger correlation between BALF inflammatory 331 immune cell gene signatures and BALF viral burden than that of nasopharyngeal swabs suggests that 332

inflammation-driven lung pathology is directly influenced by local viral replication. However, given the 333 migratory nature of these cell populations and the relatively low abundance of viral RNA in the lower 334 respiratory tract, the possibility that these cells were stimulated by viral ligands at other anatomical sites 335 cannot be discounted. Furthermore, the cells that were found to be positive for SARS-CoV-2 RNA 336

represented a range of cell types. Although the range of cell types expressing ACE2 and therefore 337

permissive to viral entry is wide (34), these populations may not represent bona fide productively-338 infected cells. Rather, cells may acquire viral RNA through phagocytic mechanisms, for example. 339

Despite these potential caveats, our findings were validated by challenge of animals with the SARS-340

CoV Bioqual Inc. Post-vaccination antibody titers generated as previously described (29, 38, (40) (41) (42) (43) , and 375

previously reported by Corbett et al. (29) . 376 377

USA-WA1/2020 challenge: At week 8 post initial vaccination (4 weeks after boost), all animals were 378 challenged with a total dose of 8 × 10 5 PFUs of SARS-CoV-2 as previously described (29). The stock of 379

1.99 × 10 6 TCID 50 or 3×10 6 PFU/mL SARS-CoV-2 USA-WA1/2020 strain (BEI: NR-70038893) was 380 diluted and administered in 3-mL doses by the intratracheal route and in 1-mL doses by the intranasal 381 route (0.5 mL per nostril). Post-challenge SARS-CoV-2 sgRNA burden in nasal swabs and BAL were 382 determined as previously described (19, For the USA-WA1/2020 dataset, the first 40 resultant PCs were initially used to perform a UMAP 430 dimensional reduction of the dataset (RunUMAP()) and to construct a shared nearest neighbor graph 431 (SNN; FindNeighbors()). This SNN was used to cluster the dataset (FindClusters()) with default 432 parameters and resolution set to 0.7. From this initial clustering a population of low-viability cells was 433 identified and removed from the anaylsis, after which the dataset PCA was re-run and the first 35 434 resultant PCs were used to perform a UMAP dimensional reduction of the dataset (RunUMAP()) and to 435 construct a shared nearest neighbor graph (SNN; FindNeighbors()). This SNN was used to cluster the 436 dataset (FindClusters()) with default parameters and resolution set to 1.5. The resultant clusters were 437 assigned to following cell types based on the expression of the indicated gene products: For the B.1.351/beta dataset, the first 31 resultant PCs were initially used to perform a UMAP 444 dimensional reduction of the dataset (RunUMAP()) and to construct a shared nearest neighbor graph 445 (SNN; FindNeighbors()). This SNN was used to cluster the dataset (FindClusters()) with default 446 parameters and resolution set to 1.7. From this initial clustering a population of low-viability cells was 447 identified and removed from the anaylsis, after which the dataset PCA was re-run and the first 31 448 resultant PCs were used to perform a UMAP dimensional reduction of the dataset (RunUMAP()) and to 449 construct a shared nearest neighbor graph (SNN; FindNeighbors()). This SNN was used to cluster the Following dataset integration and dimensional reduction/clustering, gene expression data was log 458 transformed and scaled by a factor of 10,000 using the NormalizeData() function. This normalized gene 459 expression data was used to determine cellular cluster identity by utilizing the Seurat application of a 460

Wilcoxon rank-sum test (FindAllMarkers()), and comparing the resulting differential expression data to 461 known cell-linage specific gene sets. Differential gene expression analysis between study time points 462 was performed using normalized gene expression data and the Wilcoxon rank-sum test with 463 implementation in the FindMarkers() function, with a log 2 fold change threshold of 0. Jolla, CA). A P-value < 0.05 was considered significant. ,  C  i  r  c  u  i  t  s  b  e  t  w  e  e  n  i  n  f  e  c  t  e  d  m  a  c  r  o  p  h  a  g  e  s  a  n  d  T  c  e  l  l  s  i  n  S  A  R  S  -C  o  V  -2  p  n  e  u  m  o  n  i  a  .  506   N  a  t  u  r  e   5  9  0  ,  6  3  5  -6  4  1  (  2  0  2  1  ) . 507

. J .  S  c  h  u  l  t  e  -S  c  h  r  e  p  p  i  n  g   e  t  a  l  .   ,  S  e  v  e  r  e  C  O  V  I  D  -1  9  I  s  M  a  r  k  e  d  b  y  a  D  y  s  r  e  g  u  l  a  t  e  d  M  y  e  l  o  i  d  C  e  l  l  508   C  o  m  p  a  r  t  m  e  n  t  .   C  e  l  l   1  8  2  ,  1  4  1  9  -1  4  4  0  e  1  4  2  3  (  2  0  2 

Jackson Foundation for the Advancement of Military Medicine, Inc., and the U.S. Department 600 of Defense (DOD). The views expressed are those of the authors and should

designed the study. K.V. and T.L. generated data

All data supporting the findings of this study are available within the 608 manuscript or from the corresponding author upon request. Data tables for expression counts and 609 unprocessed raw data from the scRNAseq analysis are deposited in NCBI's Gene Expression Omnibus 610 and are accessible through GEO accession GSE190913 (USA-WA1/2020 challenge)

Fig 1. Identification and quantification of BALF cells by scRNAseq. A) UMAP projection of BALF 616 cells captured by scRNAseq analysis. B) Expression of key linage specific genes in all annotated cell 617 types. C) Frequency of epithelial cell populations. D) Frequency of lymphocyte cell populations. E) 618 Frequency of dendritic cell populations. F) Frequency of macrophage populations 619 620 Fig 2. Transcriptional signatures of SARS-CoV-2 WA-1 elicited inflammation. A) Expression of 621 inflammatory markers and cytokines

Expression of inflammatory markers and cytokines/chemokines in all annotated cells day 7 post 623 infection. C) Inflammatory index scores in epithelial cells. D) Inflammatory index scores in dendritic 624 cells. E) Inflammatory index scores in myeloid cells. F) Inflammatory index scores in lymphocytes 625

Fig 3. Identification and quantification of SARS-CoV-2 RNA positive cells. A) Location of SARS-627

CoV-2 RNA positive cells. B) Frequency of SARS-CoV-2 RNA positive cell. C) Relationship between 628 BALF RNA load and frequency of SARS-CoV-2 RNA + cells. D) Relationship between NS RNA load 629 and frequency of SARS-CoV-2 RNA + cells

A) DC Inflammation index score vs BALF 632 sgRNA (E gene). B) DC Inflammation index score vs Nasal swab sgRNA (E gene). C) DC 633 Inflammation index score vs SARS-CoV-2 RNA + cell fraction. D) Macrophage Inflammation index 634 score vs BALF sgRNA (E gene). E) Macrophage Inflammation index score vs Nasal swab sgRNA (E 635 gene). F) Macrophage Inflammation index score vs SARS-CoV-2 RNA + cell fraction

Fig 5. Relationship between antibody titers and SARS-CoV-2 elicited inflammation. A) 639

DC 640 inflammation on day 2 post challenge. B) Relationship between pre-challenge serum RBD-specific IgG 641 titers (wk 8 post vaccination) and DC inflammation on day 2 post challenge. C) Relationship between 642 pre-challenge BALF S-specific IgG titers (wk 6 post vaccination) and DC inflammation on day 2 post 643 challenge. D) Relationship between pre-challenge serum pseudovirus neut titers titers (wk 8 post 644 vaccination) and DC inflammation on day 2 post challenge. E) Relationship between pre-challenge 645 serum live virus FRNT (wk 8 post vaccination) and DC

Transcriptional signatures of SARS-CoV-2 beta VBM elicited inflammation. A) UMAP 649 projection of BALF cells from beta VBM challenge. B) Expression of inflammatory markers and 650 cytokines/chemokines in all annotated cells day 2 post beta infection. C) Inflammatory index scores in 651 epithelial cells. D) Inflammatory index scores in dendritic cells. E) Inflammatory index scores in 652 myeloid cells. F) Inflammatory index scores in lymphocytes

. 573 m  e  i  s  t  e  r  ,  R  .  S  a  t  i  j  a  ,  N  o  r  m  a  l  i  z  a  t  i  o  n  a  n  d  v  a  r  i  a  n  c  e  s  t  a  b  i  l  i  z  a  t  i  o  n  o  f  s  i  n  g  l  e  -c  e  l  l  R  N  A  -s  e  q  d  a  t  a  594   u  s  i  n  g  r  e  g  u  l  a  r  i  z  e  d  n  e  g  a  t  i  v  e  b  i  n  o  m  i  a  l  r  e  g  r  e  s  s  i  o  n  .   G  e  n  o  m  e  B  i  o  l   2  0  ,  2  9  6  (  2  0  1  9  ) . 595 596 ACKNOWLEDGMENTS 597