Cells and viruses

MEF, HEK293T, Vero E6, HeLa, Huh-7, and HepG2 cells were cultured at 37 °C and 5% CO₂ in DMEM supplemented with 10% fetal bovine serum (FBS) and 1% penicillin–streptomycin. THP-1 (ATCC) cells were cultured in RPMI 1640 medium supplemented with 10% FBS and 1% penicillin–streptomycin. All cell lines were confirmed to be free of mycoplasma.

SFTSV (AH001) was isolated from a patient with severe fever with thrombocytopenia syndrome (SFTS) and propagated in Vero E6 cells. The virus was sequenced, and the sequences were deposited in BankIt under accession numbers PQ619853, PQ619852, and PQ619854.

Genome-wide knockout screening and data analysis

The Mouse-GeCKOv2 Library B (Addgene), comprised of 62,804 sgRNAs targeting 20,611 genes, was originally generated by the Zhang laboratory⁴⁵ and obtained from Azenta. To prepare the virus library, HEK293T cells (in 15-cm dishes) were transfected with 20 μg of library plasmid together with 12 μg of psPAX2 and 8 μg of pMD2.G. After 8 h, the medium was changed to DMEM containing 10% FBS. The supernatant was collected 48 h post transfection, centrifuged at 3000 × g and 4 °C for 10 min, and then filtered through a 0.22 μm filter. The aliquots were stored at −80 °C. The virus titer was quantified with the Lenti-X qRT‒PCR Titration Kit (Takara).

To perform the screen, MEFs stably expressing Cas9 were seeded in 15-cm dishes at 6.5 × 10⁶ cells per dish. In total, 1.3 × 10⁸ cells were infected with the sgRNA lentivirus library at an MOI of 0.3 to achieve library coverage >600⁴⁶. Forty-eight hours after infection, the cells were reseeded and selected with 2 μg/ml puromycin for 2 days. The cells were then reseeded and allowed to recover in DMEM containing 6% FBS (without puromycin) for 2 days. Next, 2.2 × 10⁸ untreated library cells were collected as the control sample, and another 2.2 × 10⁸ library cells were infected with SFTSV at an MOI of 1 as the screen sample. A plate of MEF-Cas9 cells without the sgRNA library was used as an infection control. Five days postinfection, when the MEF-Cas9 cells (infected control) had detached due to the severe cytopathic effect (CPE), the remaining library cells in the screen sample were reseeded and allowed to recover for 3 days. These reseeded library cells were then rechallenged with SFTSV at an MOI of 1, and the surviving cells were collected 10 days post-infection for DNA extraction⁴⁷. Two parallel screens were conducted.

The gRNAs were amplified by a two-step PCR method with NEBNext® Ultra™ II Q5® Master Mix (New England Biolabs, NEB).

In the first step, twenty 100 μl PCR reactions (each containing 1.5 μg of genomic DNA) were performed with the forward primer NGS-F1(5′-CTCTTTCCCTACACG ACGCTCTTCCGATCTCTTGTGGAAAGGACGAAACA-3′) and the reverse primer NGS-R1 (5′-GTGACTGGAGTTCAGACGTGTGCTCTTCC GATCTCCGACTCGGTGCCACTTTTTCAA-3′).

The first-step PCR conditions were as follows: initial denaturation at 98 °C for 30 s; followed by 25 cycles of 98 °C for 10 s, 65 °C for 45 s, and 72 °C for 30 s; and a final extension at 72 °C for 2 minutes. Products from all first-step PCRs were pooled and used as the template for the second-step PCRs. For the second step, twenty 100 μl PCRs (each containing 6 μl of the pooled first-step PCR product) were performed using different primers each with unique barcodes (underlined) for each sample. Specifically:

• Control sample 1: forward primer NGS-F2-index1 (5′-AATGATACGGCGACCACCGAGATCTACACAGCGCTAGACACTCTTTCCCTACACG-3′) and reverse primer NGS-R2-index1 (5′-CAAGCAGAAGACGGCATACGAGATAACCGCGGGTGACTGGAGTTC-3′)

• Control sample 2: forward primer NGS-F2-index2 (5′-AATGATACGGCGACCACCGAGATCTACACGATATCGAACACTCTTTCCCTACACG-3′) and reverse primer NGS-R2-index2 (5′-CAAGCAGAAGACGGCATACGAGATGGTTATAAGTGACTGGAGTTC-3′)

• Screen sample 1: forward primer NGS-F2-index3 (5′-AATGATACGGCGACCACCGAGATCTACACCGCAGACGACACTCTTTCCCTACACG-3′) and reverse primer NGS-R2-index3 (5′-CAAGCAGAAGACGGCATACGAGATCCAAGTCCGTGACTGGAGTTC-3′)

• Screen sample 2: forward primer NGS-F2-index4 (5′-AATGATACGGCGACCACCGAGATCTACACTATGAGTAACACTCTTTCCCTACACG-3′) and reverse primer NGS-R2-index4 (5′-CAAGCAGAAGACGGCATACGAGATTTGGACTTGTGACTGGAGTTC-3′)

The second-step PCR conditions were as follows: initial denaturation at 98 °C for 30 s; followed by 13 cycles of 98 °C for 10 s, 65 °C for 45 s, and 72 °C for 30 s; and a final extension at 72 °C for 2 minutes. The resulting ~247-bp PCR products were subjected to electrophoresis on a 1.5% agarose gel, excised, and purified. Finally, the purified amplicons were sequenced on a NovaSeq 6000 platform (Illumina).

The quality of the paired-end FASTQ files obtained from sequencing was checked with FastQC (v0.11.9). The MAGeCK (v0.5) workflow was then used for sgRNA quality control and quantification. First, sgRNAs were quantified with the MAGeCK count command and read counts from different samples were median-normalized to adjust for library size and read count distribution effects. Next, on the basis of a negative binomial (NB) model, the MAGeCK test command was used to determine whether there was a significant difference in sgRNA abundance between the virus-infected and mock control groups. sgRNAs were ranked according to P values from the NB model. The fold change values for each sgRNA were mapped to their corresponding genes in MAGeCK, facilitating the determination of selection patterns at the gene level. In the design matrix of MAGeCK, samples are divided into experimental and control groups on the basis of treatment. The maximum-likelihood estimation (MLE) algorithm was then applied to estimate the beta effect size.

Propidium iodide (PI) staining

A total of 5 × 10⁵ MEFs per well were seeded in 12-well plates. At 12 h post seeding, the cells were infected with SFTSV at an MOI of 1. Propidium iodide (PI) (Sangon) staining was performed at 0 (mock), 24, 48, and 72 h post infection. To do this, a final concentration of 2 μg/ml PI was added to the cells, which were subsequently incubated at 37 °C for 15 min. After being washed twice with PBS, the cells were photographed to document PI staining.

RNA interference

A total of 1.5 × 10⁵ MEFs were seeded in 6-well plates. Twelve hours later, a 100 nM final concentration of small interfering RNA (siRNA, GenePharma) targeting lrp1 (GeneID: 16971) was transfected into each well with jetPRIME (Polyplus). The siRNA sequences used were as follows:

• Lrp1-siRNA1: 5′-CCACCUGCUAUGAGUUUAATT-3′

• Lrp1-siRNA2: 5′-GCCCAUUGGAUGAGUUUCATT-3′

• Lrp1-siRNA3: 5′-GCGUGGUGUUCUGGUAUAATT-3′

A 21-mer nontargeting siRNA was included as a negative control:

The knockdown efficiency was measured by RT‒qPCR and Western blotting. One day after siRNA transfection, the MEFs were infected with SFTSV at an MOI of 1. Viral RNA and protein were assessed at the indicated time points by RT‒qPCR, Western blotting, and the TCID₅₀ assays.

RNA extraction and RT‒qPCR

Total RNA was extracted with RNAiso Easy (Takara) following the manufacturer’s instructions, and reverse transcription was performed with Evo M-MLV RT Premix (Accurate Biotechnology #AG11706). Quantitative real-time PCR (RT‒qPCR) was conducted with SYBR® Green Premix Pro Taq HS (AG11701, Accurate Biotechnology) on a Roche LightCycler® 96.

• Lrp1-Fwd: 5′-CAGCCGCAGTTGCATTCATT-3′

• Lrp1-Rev: 5′-GGCTTGCACAAGGTGTTGTT-3′

• SFTSV-S -Fwd: 5′-TCCCCAATTCTGGATGTGGC-3′

• SFTSV-S -Rev: 5′-AGAGGCTTACGCAATGAGGA -3′

• GAPDH-Fwd: 5′-GTGTTCCTACCCCCAATGTG-3′

• GAPDH-Rev: 5′-GGTCCTCAGTGTAGCCCAAG-3′

• Ccr2-Fwd: 5′-GGGAATGAGTAACTGTGTGATTG-3′

• Ccr2-Rev: 5′-ACAACCCAACCGAGACCTCT-3′

• Lman2-Fwd: 5′-GTTCCAGTTCCATGCCGCTC-3′

• Lman2-Rev: 5′-GTGGTTCCAGATCGAGCCCT-3′

• Tmem90b-Fwd: 5′-CACAGTAAGATCAGTGATGC-3′

• Tmem90b -Rev: 5′-GCACTGGCCACTAGCCTGTGGC-3′

• Tmem243-Fwd: 5′-CACTGATCAGTGCTTTTGTG-3′

• Tmem243-Rev: 5′-AAGTATGCAGGCAGTAATAC-3′

• Islr2 -Fwd: 5′-CTAGGAGTGGTCAGAGCGTG-3′

• Islr2 -Rev: 5′ GACAGACTAAGCGTGGTCACG-3′

Absolute quantification of SFTSV RNA in mouse samples was performed by RT–qPCR as previously reported⁴⁸. Briefly, the T7 promoter was tagged to a 480 bp L fragment by PCR and subsequently subjected to RNA in vitro transcription (NEB) and purification (NEB). The L RNA fragment was quantified by a Nanodrop and serially diluted to obtain RNA standards. The RT–qPCR standard curve was carried out simultaneously with the experiment RNA samples to determine the absolute quantity of SFTSV RNA copies. The following primers were used for absolute quantification:

• L-probe: HEX-CAATGACAGACGCCTTCCATGGTAATAGGG-BHQ1

• Forward (Fwd): 5′-AGTCTAGGTCATCTGATCCATTTAG-3′

• Reverse (Rev): 5′-TGTCTCTGTCTTTATGTAAGTT-3′

Western blotting

The cells were lysed in RIPA lysis buffer (Biosharp) on ice for 20 min, and the lysates were subjected to centrifugation. Proteins were separated by SDS–PAGE and transferred onto a polyvinylidene difluoride (PVDF) membrane. All the membranes were blocked with 5% (w/v) nonfat dry milk in TBST for 1 h at room temperature. Primary antibodies against the following proteins were used: SFTSV NS (rabbit anti-NS, a gift from Professor Mifang Liang, Chinese Center for Disease Control and Prevention), SFTSV Gn (mouse anti-Gn, self-prepared), SFTSV Gc (mouse anti-Gc, self-prepared), GAPDH (rabbit anti-GAPDH, Proteintech #10494-1-AP; mouse anti-GAPDH, Proteintech#60004-1-Ig;), LRP1(rabbit anti-LRP1, CST #64099), Strep-tag (mouse anti-Strep-tag, Abbkine #ABT2230), HA (rabbit anti-HA, Proteintech #51064-2-AP). After incubation with the primary antibody overnight at 4 °C, the membranes were washed with TBST and then incubated for 1 h at room temperature with the appropriate horseradish peroxidase (HRP)-conjugated secondary antibody (Apexbio).

CRISPR/Cas9 knockout

Single guide RNAs (sgRNAs) targeting Lrp1 (Gene ID: 16971) were designed with chopchop⁴⁹ and cloned and inserted into a lenti-CRISPRv2 plasmid (Addgen#52961). Lentivirus was collected from the supernatant of HEK293T cells transfected with lenti-CRISPRv2, psPAX2 (Addgen #12260) and pMD2.G (Addgen #12259). LRP1 knockout (KO) clones were generated by lentiviral transduction with two guides, selected with 2 μg/mL puromycin for 2 days, and further recovered for 5‒10 days without puromycin. The resulting LRP1 KO pool was seeded in 96-well plates at a density of 1 cell per well to obtain single-cell clones. Each clone was validated by Western blot or Sanger sequencing.

To confirm the gene editing events, genomic DNA was extracted with the Qiagen DNeasy Blood & Tissue Kit (Qiagen) and used as a template for amplifying an approximately 400–800 bp region surrounding the PAM site. The following sgRNAs were used in this study:

• Lrp1-sgRNA1: 5′-CGAAGCTCACAGCCCCATCT-3′

• Lrp1-sgRNA2: 5′-GGTTATCAAGGGTAACATGT-3′

• Lrp1-sgRNA3: 5′- ACAAGAAGTAACAGGACCAG -3′

• Lrp1-sgRNA4: 5′- TCTGATTACACCACTTATTG -3′

Lrp1-sgRNA1 and Lrp1-sgRNA4 were also used to delete exon 1 and exon 2 of Lrp1 in human (Gene ID: 4035) and African green monkey (Gene ID:103238555) cells with a single or double nucleotide (underlined) modification in the Lrp1-sgRNA4 sequence:

TCID₅₀ assay

Monolayer Vero cells were seeded in 96-well plates and incubated overnight before virus infection. The virus samples were serially diluted from 1:10 to 1:10⁸ in DMEM. Each dilution was used to infect Vero cells in quintuplicate and then incubated at 37 °C with 5% CO₂ for 3 days. After incubation, the cells were fixed with 4% (w/v) paraformaldehyde and subjected to immunostaining with a primary mouse anti-SFTSV NP antibody (1:600, self-made) and a secondary goat anti-mouse IgG-FITC antibody (Invitrogen #F-2761). SFTSV NP-positive cells were counted under a fluorescence microscope, and viral titers were calculated according to the Reed ‒ Muench method as previously decribed^50,51.

Immunostaining assay

Monolayer cells were fixed with 4% (w/v) paraformaldehyde for 30 min, followed by three additional PBS washes. The cells were then permeabilized with 0.5% Triton X-100 for 10 min, washed with three additional PBS washes, and blocked with 5% (w/v) bovine serum albumin (BSA) for 1 h at room temperature. The cells were subsequently incubated with primary antibody overnight at 4 °C, followed by three additional PBS washes. Next, the samples were incubated with the secondary antibody for 1 h at room temperature, stained with DAPI, and washed three more times. Finally, the samples were mounted and observed with the THUNDER Imager 3D assay (Leica) or confocal microscopy (Zeiss). The primary antibodies used for immunostaining included mouse anti-SFTSV NP (1:600, self-prepared), rabbit anti-SFTSV (1:200, a gift from Professor Mifang Liang, Chinese Center for Disease Control and Prevention), Syrian hamster anti-PDPN (1:200, Invitrogen #MA5-16113). The secondary antibodies used for immunostaining included the following: goat anti-mouse IgG-FITC antibody (Invitrogen #F-2761), goat anti-rabbit IgG 594 (1:200, Bioss #BB07081455), and goat anti-Syrian hamster IgG 488 (1:200, Invitrogen #A78958).

Virion binding assay

To visualize virus binding, WT, and LRP1 KO MEFs were incubated with SFTSV in precooled DMEM at an MOI of 50 or an MOI of 10 for 1 h at 4 °C. The cells were subjected to three washes with prechilled phosphate-buffered saline (PBS) containing 2% BSA. The cells were subsequently fixed with 4% paraformaldehyde and labeled with rabbit anti-SFTSV (1:200) and Syrian hamster anti-PDPN (1:200, Invitrogen #MA5-16113) antibodies for subsequent microscopy. The numbers of DAPI-stained cells and adsorbed virions from three independent views belonging to three independent wells were quantified automatically with ImageJ software.

Pseudotyped virus production and infection

SFTSV glycoprotein–pseudotyped viruses were generated as previously described⁵².

Briefly, the SFTSV M protein (YP_006504094.1) was human codon-optimized and cloned and inserted into pcDNA3.1 between the BamHI and XhoI restriction sites, resulting in the pcDNA3.1-SFTSV-G plasmid, which expresses the SFTSV glycoprotein. To produce the pseudotyped virus, HEK293T cells were cotransfected with pLenti-CMV-GFP, psPAX2, and pcDNA3.1-SFTSV-G. The pseudoviruses were harvested 48 h post transfection and titrated with the Lenti-X qRT‒PCR Titration Kit (Takara). pMD2.G, which encodes the vesicular stomatitis virus glycoprotein (VSV-G), was used as a positive control.

For pseudovirus infection, WT MEFs and LRP1 KO MEFs were infected with the SFTSV pseudoviruses at an MOI of 5 for 48 h. The GFP intensity in the infected cells was then measured by microscopy. For the neutralization assay, cells were pretreated with the corresponding neutralizing antibody or inhibitor before being subjected to SFTSV pseudovirus infection, as described above.

SFTSV binding and internalization assays

The cells were seeded in 12-well plates 12 h before infection. For the binding assay, the cells were incubated with SFTSV at an MOI of 10 in precooled DMEM for 1 h at 4 °C. After incubation, the cells were washed three times with precooled PBS supplemented with 2% bovine serum albumin. The washed cells were then collected for RNA extraction, and the relative level of bound virions was quantified by RT‒qPCR. For the internalization assay, following the binding step, the cells were further incubated with prewarmed DMEM containing 2% FBS and 50 mM NH₄Cl at 37 °C for 3 h. The cells were then washed again with PBS and collected for RNA extraction. The relative level of internalized virions was subsequently quantified by RT‒qPCR.

Neutralization assays

MEFs were pretreated with serially diluted concentrations of anti-LRP1 antibody (Invitrogen #MA1-27198), LRP1-CLII-Fc (RD #2368-L2), and LRPAP (Sino Biological #50281-M08H) or with a negative IgG control (Proteintech #66002-1-Ig). After 4 h, the cells were infected with SFTSV at an MOI of 1. The cells were harvested 24 h post infection and analyzed for intracellular viral RNA levels using RT‒qPCR. The viral titer in the supernatant was measured with TCID₅₀.

Coimmunoprecipitation

HEK293T cells were cotransfected with Strep-tagged SFTSV Gn and HA-tagged truncated LRP1 for 48 h and were subsequently lysed in 1% NP-40 supplemented with an EDTA-free protease inhibitor cocktail (Apexbio #K1007), 0.1 mM PMSF, and 5 mM NaF. Protein concentrations were measured by a BCA assay. A total of 600 μg of whole cell lysate was incubated with 30 μL of Strep-Tactin® Sepharose® resin (Iba#2-1201-010) or anti-HA nanobody agarose beads (AlpaLifeBio #KTSM1305) overnight at 4 °C. Subsequent wash steps were performed following the manufacturer’s instructions. The beads were subsequently eluted with 2× protein loading buffer (0.1 M Tris, pH 6.8, 4% SDS, 20% glycerol, 10% β-mercaptoethanol, 0.2% bromophenol blue) and denatured for 10 minutes at 95 °C. The eluate was further assessed by Western blotting.

Protein expression and purification

The ectodomain of the SFTSV Gn glycoprotein (amino acids 20–453, YP_006504094) was prokaryotically codon-optimized and cloned and inserted into the pET-30a (+) vector between the NdeI and BamHI restriction sites to obtain the expression plasmid pET-30a-Gn-his. The expression plasmid was subsequently transformed into BL21(DE3) Escherichia coli cells. A single colony was cultured in 4 mL of Luria broth (LB) media at 37 °C until an OD600 of 0.6 was reached, which was then inoculated into 200 mL of LB media supplemented with 1 mM isopropyl-β-D-thiogalactoside (IPTG) to induce protein expression at 18 °C for 14 h. The cells were then collected and resuspended in 30 mL of lysis buffer (50 mM Tris-HCl, 500 mM NaCl(pH 7.5), 5% glycerol [v/v], 1 mM TCEP, 0.5 mM PMSF, 0.25 mg/mL lysozyme) and sonicated on ice for 10 minutes. The lysate was subsequently centrifuged at 10,000 × g and 4 °C for 30 min. The inclusion bodies in the pellet were isolated with 20 mL of cold binding/wash buffer (8 M urea, 20 mM Tris-HCl, 500 mM NaCl and 5 mM imidazole, pH 8.0). The Gn-his protein was purified with Ni-NTA Sepharose Resin (Sangon #C600791) and eluted with 3 mL of elution buffer (20 mM Tris-HCl, 500 mM NaCl, and 500 mM imidazole, pH 8.0). The Gn-his protein was further refolded by dialysis with a gradually decreasing urea concentration gradient. The refolded protein was subsequently concentrated with a centrifugal filter (Amicon #30 kDa). The protein was further purified with size-exclusion columns, and Coomassie blue staining was performed to determine its purity.

SPR assay

Surface plasmon resonance (SPR) assays were conducted with the GE Healthcare Biacore T200. The purified ectodomain of SFTSV Gn was immobilized on the CM5 chip at a concentration of 20 μg/mL in HBS buffer. Recombinant human LRP1 CLII-Fc (RD #2368-L2), LRP1 CLIII-Fc (RD #4824-L3), LRP1 CLIV-Fc (RD #5395-L4) or recombinant human IgG1 Fc (RD #110-HG-100) was loaded at serial concentrations (0, 1.95, 3.9, 7.8125, 15.625 and 31.25 nM) in HBS buffer. The experiments were conducted with the following settings: contact time of 180 s, flow rate of 30 μL/min, and dissociation time of 300 s. Data were baseline-corrected by subtracting the buffer-only controls.

Mouse experiments

Ten-week-old male C57BL6J mice were used in this study. To render the mice susceptible to SFTSV infection, mitomycin C (0.03 mg/mouse) were administered intraperitoneally 3 days before infection. Thirty micrograms /mouse of anti-LRP1 antibody (Invitrogen #MA1-27198) or a negative control IgG (Proteintech #66002-1-Ig) was intraperitoneally administered 4 h before the challenge. SFTSV was intraperitoneally inoculated at 10⁵ TCID₅₀ per mouse, and mitomycin C (0.003 mg/mouse) was intraperitoneally administered for three days following infection²⁸. Body weight and temperature were monitored daily. The mice were euthanized when they exhibited no response to touch stimulation. Independent experiments were performed to collect tissue samples. A subset of mice was marked at the beginning and the mice were euthanized six days postinfection. Serum, liver, spleen, lung, heart, and kidney samples were collected for further pathological analysis.

The Animal Ethics Committee of the Anhui Medical University approved the research (LLSL20242191). The mice were housed under controlled environmental conditions with a 12-h light/dark cycle and observed daily. Ambient temperature was maintained at 22–25 °C and relative humidity ranged between 40% and 80% throughout the experimental period. The animal experiments were conducted in the BSL-3 animal facility of the Second Military Medical University.

H&E and immunohistochemical staining

The collected tissue samples were immersed in 4% paraformaldehyde and fixed at room temperature for 24 h to preserve tissue architecture. The tissues were subsequently dehydrated through a series of alcohol gradients (50%, 70%, 80%, 95%, and 100%) followed by clearing in xylene. The dehydrated tissues were then embedded in paraffin and sectioned into 4–5 µm thin slices with a microtome, which were subsequently mounted onto glass slides.

For H&E staining, the slices were dewaxed and the sections were stained with hematoxylin for 5–10 minutes to render the nuclei blue, followed by differentiation and bluing steps. Next, the sections were counterstained with eosin for 10–30 seconds, which imparted a pink hue to the cytoplasm. Finally, the sections were dehydrated through an alcohol gradient, cleared in xylene, and mounted with a coverslip.

For immunohistochemical staining, tissue sections were dewaxed and subjected to antigen retrieval by immersion in citrate buffer (pH 6.0), heated in a microwave for 10–15 min until boiling, and then allowed to cool to room temperature. After three washes with PBS, endogenous peroxidase activity was blocked with an endogenous peroxidase blocker (Beyotimet #P0100B) to reduce background interference. The sections were subsequently washed three times with PBS and blocked with 5% BSA for 1 h at room temperature to prevent nonspecific binding. A primary antibody (rabbit anti-SFTSV, 1:200) was applied, and the samples were incubated overnight at 4 °C. The next day, the slides were returned to room temperature for 40 min, washed three times with PBS, and then incubated with a secondary antibody for 1 h at room temperature. Following three additional PBS washes, a DAB substrate was used for color development for 30–60 s, which resulted in a brown precipitate. The sections were counterstained with hematoxylin for 2–5 min, blueing was achieved with lithium carbonate, and finally, the slides were cleared and mounted with coverslips.

Statistical analysis

All experiments were performed with n ≥ 3 biological replicates (as indicated). All the statistical analyses were performed with GraphPad Prism 8 software. Two-tailed unpaired Student’s t-test was used to determine the significance of variability between two groups. Survival curves were generated by the log-rank (Mantel‒Cox) test. The data were presented as means ± SD. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.