key: cord-0688090-71jpvsrf
authors: Margaroli, C.; Benson, P.; Sharma, N. S.; Madison, M. C.; Robison, S. W.; Arora, N.; Ton, K.; Liang, Y.; Zhang, L.; Patel, R. P.; Gaggar, A.
title: Spatial mapping of SARS-CoV-2 and H1N1 Lung Injury Identifies Differential Transcriptional Signatures
date: 2021-03-23
journal: Cell Rep Med
DOI: 10.1016/j.xcrm.2021.100242
sha: 0be35b4cbdc9a4d94e98b3d324a4ce3a061c880b
doc_id: 688090
cord_uid: 71jpvsrf

Severe SARS-CoV-2 infection often leads to development of acute respiratory distress syndrome (ARDS), with profound pulmonary patho-histological changes post-mortem. It is not clear if ARDS from SARS-CoV-2 is similar to that observed in Influenza H1N1, another common viral cause of lung injury. Here, we analyze specific ARDS regions of interest utilizing a spatial transcriptomic platform on autopsy-derived lung tissue from patients with SARS-CoV-2 (n=3), H1N1 (n=3), and a dual infected individual (n=1). Enhanced gene signatures in alveolar epithelium, vascular tissue, and lung macrophages identify not only increased regional coagulopathy, but also increased extracellular remodeling, alternative macrophage activation, and squamous metaplasia of type II pneumocytes in SARS- CoV-2. Both the H1N1 and dual infected transcriptome demonstrated an enhanced antiviral response compared to SARS-CoV-2. Our results uncover regional transcriptional changes related to tissue damage/remodeling, altered cellular phenotype, and vascular injury active in SARS-CoV-2 and presents therapeutic targets for COVID-19 related ARDS.

We next sought to characterize the alveolar epithelium, a site of notable damage in ARDS 19 2 ( Fig. 3A and 3B ). As observed with the vascular regions, there was little differential gene 3 expression between dual infected and H1N1 infected epithelial regions (Fig. 3C, right panel) . 4

However, there were significant differences in gene expression with SARS-CoV-2 compared to 5 the H1N1 and dual infected epithelial regions (Fig. 3C, left and middle panel) . Gene pathway 6 analyses showed that although immune signaling was observed in all groups, SARS-CoV-2 7 regions demonstrated increased EMT and ECM-related pathways ( Fig. 3D and Fig. S5A) . 8

Further, we noted that SARS-CoV-2 subjects did demonstrate areas of increased alveolar 9 epithelial hyperplasia (Fig. 3E) , a phenotype which has recently been described in COVID-19 10 autopsy specimens 8 . When compared to non-hyperplastic epithelial regions, we observed no 11 significant difference in gene expression (Fig. 3F) . However, when we compared specific genes 12 related to lung epithelial proliferation and squamous metaplasia of type II pneumocytes in the 13 SARS-CoV-2 hyperplastic and normal epithelium with the non-hyperplastic H1N1 epithelium, 14 we observed increased expression of those genes in the SARS-CoV-2 regions (Fig. 3G) . These 15 results suggest that cellular metaplasia is an important feature of SARS-CoV-2 ARDS. 16 17 Macrophages are differentially activated in COVID- 19 18 As macrophages are critical to the immune response to pulmonary viral infections 20 , we next 19 identified macrophages by H&E and CD68 staining ( Fig. 4A and 4B ). We noted that SARS-20

CoV-2 patients displayed two phenotypes for pulmonary macrophages: either within clusters or 21 infiltrative phenotype ( Fig. S6A and S6B) . However, when these groups were compared with 22 J o u r n a l P r e -p r o o f each other, there were only eight genes which were differentially regulated ( Fig. S6C ) 1 suggesting that these groups are of similar lineage. 2 There was little differential gene expression in macrophages between dual infected and H1N1 3 infected (Fig. 4C, right panel) . However, there were significant differences in macrophage gene 4 expression in SARS-CoV-2 compared to H1N1 regions and dual infected compared to SARS-5

CoV-2 (Fig. 4C, left and middle panels) . Gene pathway analyses showed that although immune 6 signaling was observed in all groups, SARS-CoV-2 regions demonstrated enrichment of 7 pathways related to tissue remodeling and differential activation of the innate immune signaling 8 ( Fig. 4D and Fig. S5B ). SARS-CoV-2 macrophages displayed genes consistent with alternative 9 macrophage activation 21 (Fig. 4E) . 10 11

Our study focused on examining transcriptional signatures active in viral ARDS from H1N1 and 13 SARS-CoV-2 patients to better understand fundamental biologic processes related to these 14 critical disease phenotypes. Although prior transcriptomic studies have focused on individual 15 immune cells 22, 23 , isolated epithelial cells 24 , or bulk lung tissue RNA 25 , these approaches do not 16 have the benefit of maintaining tissue architecture to examine gene expression in situ. To date, 17 only the study from Desai et al 12 addresses the heterogenous responses in the SARS-CoV-2 lung 18 while maintaining an intact tissue architecture. In this study, we utilized the spatial 19 transcriptomic approach to focus on ARDS regions in SARS-CoV-2 subjects compared to other 20 forms of virus-induced ARDS, providing the ability to identify regional or cell-specific gene 21 regulation. 22 J o u r n a l P r e -p r o o f Prior studies have largely delineated the expression of altered coagulopathy as a significant gene 1 signature 9, 11, 26 . Although our results validate these findings, using spatial transcriptomics, we 2 also identify a notable enrichment in extracellular matrix associated genes in SARS-CoV-2 3 regions compared to H1N1. Prior data has shown that excessive ECM turnover portends 4 increased mortality in other causes of ARDS 27 . IL-6 has been associated with worse outcomes in 5 patients with SARS-CoV-2 28, 29 . In line with the study by Desai et al 12 , we did not find major 6 differences within the SARS-CoV-2 cohort for IL-6 expression between patients and regions of 7 differential viral presence. It has been previously shown that IL-6 increases TGF signaling by 8 modulating the compartmentalization of the TGF receptor 30 , which is a known pro-fibrotic 9 signaling pathway. Interestingly, in our dataset genes involved in the IL-6 signaling pathway 10 were enriched in H1N1 patients. This differential expression may be due to diverse temporal 11 regulation of tissue responses in these two viral infections, where early high IL-6 signaling in 12 SARS-CoV-2 induces a quicker fibroproliferative cascade leading to more severe cases of lung 13 injury and a higher collagen deposition observed in the terminal cases in this study. 14 Overall, these results challenge conventional wisdom and provide evidence for a more 15 fibroproliferative ARDS phenotype in SARS-CoV-2 infection versus a more exudative 16 inflammatory ARDS phenotype in H1N1 infection. This finding may explain the observation of 17 COVID ARDS patients often displaying extended time requiring ventilator support 31 . As we are 18 biased with only terminal ARDS subjects, it is possible that this gene signature may represent a 19 unique endotype for COVID-19 32 , which progresses to poorer outcomes versus other COVID-19 20 subjects who recover from ARDS. Our results also strongly suggest that for these COVID-19 21 ARDS subjects, directed anti-fibrotic therapies may provide an important therapeutic approach to 22 improve disease-related outcomes. Furthermore, this clear differential regulation of disease 23 J o u r n a l P r e -p r o o f pathology warrants future studies in the temporal regulation of immune responses to airborne 1 viruses. 2

The transcriptional signature from isolated cell populations were also consistent with these larger 3 ARDS regions. ARDS-associated capillaries and arterioles from SARS-CoV-2 infected subjects 4

show a notable upregulation of coagulopathy, complement activation, and platelet aggregation 5 genes, highlighting that regional vessels are actively transcriptionally contributing to the 6 development of COVID-associated vascular injury and microangiopathy 33 . Alveolar epithelial 7 cells from SARS-CoV-2 subjects demonstrated regions of increased hyperplasia, a finding 8 observed in multiple patients in a recent autopsy study 8 . It is interesting that both hyperplastic 9 and non-hyperplastic SARS-CoV-2 alveolar epithelium had similar transcriptome but both had 10 enhanced metaplasia-related gene expression compared to H1N1 alveolar epithelia. A recent 11 manuscript has suggested that EMT may be protective in early SARS-CoV-2 infection by 12 reducing ACE2 expression 34 ; its role in later stage ARDS may be to induce a wound-healing 13 response which has become dysregulated. 14 Macrophages have previously been shown to be central to immune response with SARS-CoV-2 15 and previous data suggests that they are critical to a "cytokine storm" in early SARS-CoV-2 16 infection 35 . Our data confirms that in end-stage ARDS, both SARS-CoV-2 and H1N1 lungs 17 demonstrated increased immune activation pathways but the SARS-CoV-2 regions demonstrated 18 alternative macrophage activation. These data, together with a lack of differences in gene 19 expression observed between ARDS regions characterized with either increased or low viral 20 staining, suggest that the patients included in this study may be part of the low viral load group 21 described in the study by Desai et al 12 . In that study, patients with higher viral loads presented a 22 macrophage phenotype skewed toward an M1-like activation paired with increased interferon 23 J o u r n a l P r e -p r o o f responses. Furthermore, they suggest that a broad, tissue-based transcriptome response in this 1 patient group may not impacted by viral presence during end-stage ARDS in both SARS-CoV-2 2 and H1N1 infected patients. 3

Overall, these results provide further evidence of a more fibroproliferative response to the SARS-4

CoV-2 versus H1N1. Future studies should also work to define the transcriptome and activity of 5 both innate 36 and adaptive 37, 38 immunity in the lungs of patients with persistent viral infection 6 and injury and their relative contributions to the SARS-CoV-2 ARDS phenotype. Furthermore, 7 future studies should also address the mechanisms leading to the discrepant transcriptional 8 response observed in H1N1 and SARS-CoV-2 ARDS, as the autopsy specimens used in this 9 study limit the ability to address the kinetics of differential ARDS development. 10

We also identified a co-infected individual who, upon virus staining, demonstrated a much higher 11 burden of H1N1 compared to SARS-CoV-2 throughout the lungs. This may underscore why this 12 individual's transcriptome more closely related to the H1N1 subjects. It is important to note that 13 this subject had known lung disease with CT-ILD and COPD and was treated with baseline 14 methotrexate and steroids prior to hospitalization, which may have made this individual more 15 susceptible to dual viral infection. A small clinical series of co-infections with SARS-CoV-2 and 16 influenza has been reported 39 , and in these three cases, there was no obvious evidence of 17 baseline immunosuppression noted. Although the current influenza season has seen reduced 18 cases of hospitalized patients, it is imperative to consider that if both viruses are highly prevalent 19 in the future, co-infection in both immunosuppressed and immunocompetent individuals may be 20 a more frequent occurrence. 21

This method of analysis has uncovered pathways which augment tissue injury in SARS-CoV-2 22

individuals and future studies should examine additional critically ill subjects and focus on 23 further defining these critical pathways. It is our hope that these pathways will improve our 1 understanding of mechanisms leading to progressive worsening of gas exchange and increased 2 mortality in SARS-CoV-2 related ARDS. As a result, the potential for new therapeutic targets to 3 alter the fibroproliferative response will present the potential to improve clinical outcomes in 4 patients with progressive lung injury. 5 6

The number of subjects included in this study are relatively limited, even though the SARS-CoV-8 2 and H1N1 subjects were relatively well-matched for demographic features, severity of disease, 9

and co-morbidities. These features strongly suggest that the transcriptional changes observed 10 were due to the viral lung injury, although future studies with additional subjects would certainly 11 be warranted. Furthermore, this study focused only on the lung parenchyma and future studies 12 addressing spatial heterogeneity (larger and smaller airways, lung parenchyma vs lumen) are 13 needed. Lastly, patient samples used in this study represent a discrete patient population, where 14 disease was most severe. Future studies addressing the temporal regulation of disease will help 15 better understand the differences observed in these two viral infections. genes defining a pro-inflammatory (M1) or alternative activated (M2) macrophage phenotype (* 1 indicates significant genes between H1N1 and SARS-CoV-2 shown in the volcano plots). 2 Differential gene expression was defined as p-value of 0.02 and log2 fold change of 0.5. SARS-3

CoV-2 infected patients (n=3), H1N1 (n=3), and SARS-CoV-2/H1N1 (n=1). 4

Lead contact 7

Further information and requests for resources and reagents should be directed to and will be 8 fulfilled by the Lead Contact, Dr. Amit Gaggar (agaggar@uabmc.edu). 9

10

This study did not generate new unique reagents. 12 13

The datasets generated during this study are available on Mendeley DOI: 15 http://dx.doi.org/10.17632/n5dn4xzg7j.1 . 16

Pulmonary autopsy specimens were collected from patients deceased due to ARDS. Three 20 patients were infected with SARS-CoV-2, three patients were infected with influenza A subtype 21 H1N1, and one patient was infected with both SARS-CoV-2 and H1N1. The study was approved 1 by the Institutional Review Board (UAB-IRB 300005258, VA-IRB 1573682) and summary 2 demographic and clinical data are presented in Table S1 . 3 4

Histology 6

Lungs were inflated isobarically with 10% formalin and preserved in paraffin blocks. Sequential 7 tissue sections of 5 m were used for viral staining by immunofluorescence, for RNA analysis, 8 and for hematoxylin and eosin or Masson's trichrome staining to identify ARDS pathological 9 features and collagen deposition respectively. 10 11 Immunofluorescence 12

Paraffin-embedded tissue slides were incubated for 2 hours at 60°C. Deparaffinization and 13 rehydration of the slides was performed with sequential three 5 minutes incubations in xylene 14 (Fisher Scientific), two sequential 5 minutes incubations in 100% denatured ethyl alcohol (Fisher 15 Scientific), two sequential 5 minutes incubations in 95% denatured ethyl alcohol (Fisher 16 Scientific), followed by three sequential 5 minutes incubations in distilled water under 40rpm 17 gentle shaking. Antigen retrieval was performed in pre-warmed Tris-EDTA pH 9 buffer 18 (Abcam) at 70°C in heated steamer for 20 minutes, followed by three washes in distilled water 19 under 40rpm gentle shaking. Tissue sections were blot dry and incubated with PBS for 10 20 minutes at room temperature, followed by blockade of non-specific binding using 3% w/v BSA 21 (Fisher Scientific) for 40 minutes at room temperature. The SARS-CoV-2 only infected slides 22 and the SARS-CoV-2/H1N1 slide were stained for 1 hour at room temperature with anti-SARS-23

CoV-2 nucleocapsid antibody (GeneTex, GTX135361, RRID: AB_2887484) directly labeled 1 with R-Phycoerythrin lightning-link labeling kit (Novus Biologicals) at 1:500 dilution in PBS, 2 3% w/v BSA. Slides were then washed three times for 5 minutes in PBS under gentle agitation. 3

The SARS-CoV-2/H1N1 slide and the H1N1 only infected slides were then fixed for 10 minutes 4 at room temperature with 4% paraformaldehyde and washed three times with PBS, 0.1% Tween-5 20 (PBST) for 5 minutes. Staining for H1N1 was performed with Alexa 647 (Thermofisher) pre-6 labeled anti-influenza A virus nucleoprotein antibody (Abcam, ab20343, RRID :AA5H) at 1:200 7 in PHEM buffer (Goldbio) for 1 hour at room temperature. Tissues were then washed three times 8 in PBST for 5 minutes. Nuclei counter staining for all slides was performed using 300nM DAPI (Biolegend) in PBS for 17 5 minutes, followed by three 5 minutes washes in PBS. Slides were mounted using ProLong 18 Gold antifade mounting media with DAPI (Thermofisher) and stored in the dark until image 19 acquisition. Confocal immunofluorescence images were acquired using the Nikon A1R confocal 20 microscope. 21

GeoMX digital spatial profiling 23

Paraffin embedded tissues were processed and analyzed at NanoString techonology laboratories 1 using a combination of fluorescently labeled antibodies, anti-CD68 (Santa Cruz, sc-20060 2 AF647, RRID: KP1), anti-EpCAM (Abcam, ab213500, RRID:EPR20532-222), anti-smooth 3 muscle actin (Invitrogen, 53-9760-82, RRID:1A4) and the GeoMX COVID-19 Immune 4

Response Atlas gene set with custom probe set specific for SARS-CoV-2 lung infection and 5 tissue responses (see Table S2 for SARS-CoV-2 related gene list and Fig. S1 for workflow), Sequencing through the Nanostring GeoMx platform is performed on the RNA probe tag and not 19 on the transcript itself, providing less sequencing bias and a more accurate transcript count. RNA 20 probe counts used in the analyses were selected following a sequencing QC according to 21

Nanostring protocols, where counts from each area of interest are analyzed and under-sequenced 22 samples are dropped (field of view percentage of 75% and Binding density from 0.1 to 2.25), and 23 a probe QC, where mRNAs are targeted by multiple probes and outlier probes are dropped from 1 downstream data analysis (positive spike-in normalization factor between 0.3 and 3). Then RNA 2 counts were normalized using a signal-based normalization, in which individual counts are 3 normalized against the 75 th percentile of signal from their own area of interest. The final list of 4 detectable genes was then obtained by dropping genes in each specific group (ARDS regions, 5 vascular, epithelium, macrophages) by using a limit of quantification (LOQ) of 20% coverage 6 within replicates. The LOQ was calculated using the geometric mean and geometric standard 7 deviation of negative probes in the dataset. Counts were normalized to log2 and statistical 8 comparisons were performed using a two sample t-test upon normality testing and ComBat 9 correction for batch effect 40, 41 . Comparison of SARS-CoV-2 to H1N1 was performed by 10 averaging the technical replicates and by comparing biological replicates (n=3 per group). 11

Comparison of the double infected patient to the single infection was done using technical 12 replicates (ROIs) as unique samples for statistical reasons. P-value threshold for differential gene 13 expression were set at p=0.02 and log2 fold change of 0.5. All details for the statistical analyses 14 and number of replicates can be found in the figure legends. All analyses for the volcano plots 15 can be found in the Data S1 file. 16 Statistical analysis done for specific gene expression ( Figure S4 ) was performed using Mann-17 Whitney test and data are shown as median and interquartile range (n=3 per group). 

H1N1-associated acute respiratory distress 3 syndrome

A new coronavirus associated with human respiratory disease in China

Clinical Characteristics of 138 Hospitalized Patients With

ICU and Ventilator Mortality Among 12 Critically Ill Adults With Coronavirus Disease

Covid-19 in Critically Ill Patients in the Seattle Region -15 Case Series

Hospitalized patients with 2009 H1N1 influenza 18 in the United States

Comparison of the clinical characteristics and outcomes of 21 hospitalized adult COVID-19 and influenza patients: a prospective observational study

Pulmonary post-mortem findings in a series of COVID-19 cases 25 from northern Italy: a two-centre descriptive study

SARS-CoV-2 induces transcriptional 30 signatures in human lung epithelial cells that promote lung fibrosis

Pulmonary Vascular Endothelialitis, Thrombosis, and 33 Angiogenesis in Covid-19

Temporal and Spatial Heterogeneity of Host Response to SARS-CoV-2 36 Pulmonary Infection. medRxiv. 2020

Co-infections among patients with COVID-19: The need 38 for combination therapy with non-anti-SARS-CoV-2 agents?

The acute respiratory distress syndrome

PDGF-BB 42 regulates the pulmonary vascular tone: impact of prostaglandins, calcium

PI3K/AKT/mTOR signalling and actin polymerisation in pulmonary veins of guinea pigs

Cross-Species Transcriptome Profiling Identifies New Alveolar 4

Epithelial Type I Cell-Specific Genes

Macrophage-specific gene expression: current paradigms and 6 future challenges

Fibrinolytic abnormalities 8 in acute respiratory distress syndrome (ARDS) and versatility of thrombolytic drugs to treat 9 COVID-19

The alveolar-epithelial barrier: a target for potential therapy

Airway immune homeostasis and implications for 13 influenza-induced inflammation

Macrophage Polarization: Different 15

Gene Signatures in M1(LPS+) vs. Classically and M2(LPS-) vs. Alternatively Activated 16 Macrophages

Single-cell landscape of bronchoalveolar immune cells in patients with COVID-19

A single-cell atlas of the peripheral immune response in 22 patients with severe COVID-19

Tobacco Smoking Increases the Lung 24

Gene Expression of ACE2, the Receptor of SARS-CoV-2

Comorbidities Associated with Severe COVID-19

Complement associated microvascular injury and thrombosis in the pathogenesis of 31 severe COVID-19 infection: A report of five cases

Type III procollagen peptide in the 33 adult respiratory distress syndrome. Association of increased peptide levels in bronchoalveolar 34 lavage fluid with increased risk for death

Interleukin-6 as prognosticator in patients with COVID-37 19

Elevated levels of IL-6 and CRP predict the need for mechanical 40 ventilation in COVID-19

Interleukin-6 regulation of transforming growth 42 factor (TGF)-beta receptor compartmentalization and turnover enhances TGF-beta1 signaling

Characteristics, treatment, outcomes and cause of death 2 of invasively ventilated patients with COVID-19 ARDS in

Covid-19, Angiogenesis, and ARDS Endotypes

COVID-19 update: Covid-19-associated coagulopathy

SARS-CoV-2 infection induces EMT-like molecular 9 changes, including ZEB1-mediated repression of the viral receptor ACE2, in lung cancer models

Pathological inflammation in patients with COVID-19: a key role 12 for monocytes and macrophages

Cellular Innate 14 Immunity: An Old Game with New Players

B cell responses to influenza infection and 16 vaccination

Defining 'T cell exhaustion

A Case Series of Patients Coinfected With Influenza and COVID-19

The sva package for removing 25 batch effects and other unwanted variation in high-throughput experiments