Anti-SARS-CoV-2 Spike Protein and Anti-Platelet Factor 4 Antibody Responses Induced by COVID-19 Disease and ChAdOx1 nCov-19 vaccination

This study explores a potential mechanism for the rare thrombotic events observed following vaccination with AstraZeneca’s COVID-19 vaccine (AZD1222). The researchers conclude that antibodies against SARS-CoV-2 spike protein do not cross-react with platelet factor 4, the protein targeted in this rare disorder. At the time of this posting, the World Health Organization has stated that “a causal relationship between the vaccine and the occurrence of blood clots with low platelets is considered plausible but is not conrmed.” Some of the authors disclose relevant conicts of interest. Abstract Background: Some recipients of ChAdOx1 nCoV-19 COVID-19 Vaccine AstraZeneca develop antibody-mediated vaccine-induced thrombotic thrombocytopenia (VITT), associated with cerebral venous and other unusual thrombosis resembling autoimmune heparin-induced thrombocytopenia. A prothrombotic predisposition is also observed in Covid-19. We explored whether antibodies against the SARS-CoV-2 spike protein induced by Covid-19 cross-react with platelet factor 4 (PF4/CXLC4), the protein targeted in both VITT and autoimmune heparin-induced thrombocytopenia. Methods: Immunogenic epitopes of PF4 and SARS-CoV-2 spike protein were compared via prediction tools and 3D modelling software (IMED, SIM, MacMYPOL). Sera from 222 PCR-conrmed Covid-19 patients from ve European centers were tested by PF4/heparin ELISA, heparin-dependent and PF4-dependent platelet activation assays. Immunogenic reactivity of puried anti-PF4 and anti-PF4/heparin antibodies from patients with VITT were tested against recombinant SARS-CoV-2 spike protein. Results: Three motifs within the spike protein sequence share a potential immunogenic epitope with PF4. Nineteen of 222 (8.6%) Covid-19 patient sera tested positive in the IgG-specic PF4/heparin ELISA, none of which showed platelet activation in the heparin-dependent activation assay, including 10 (4.5%) of the 222 Covid-19 patients who developed thromboembolic complications. Puried anti-PF4 and anti-PF4/heparin antibodies from two VITT patients did not show cross-reactivity to recombinant SARS-CoV-2 spike protein. Conclusions: The antibody responses to PF4 in SARS-CoV-2 infection and after vaccination with COVID-19 Vaccine AstraZeneca differ. Antibodies against SARS-CoV-2 spike protein do not cross-react with PF4

Patients often show laboratory signs of disseminated intravascular coagulation with severe thrombocytopenia without preceding heparin exposure. These features mimic the severe prothrombotic disorder, autoimmune heparin-induced thrombocytopenia (aHIT), which unlike classic HIT features heparin-independent platelet-activating antibodies. 10,11 Some of these clinical features of thrombocytopenia and disseminated intravascular coagulation are also observed in Covid-19 patients. In addition, two recent studies showed FcγIIa receptor-dependent platelet activation by sera of some Covid-19 patients. 12,13 Further, patients with Covid-19 antibodies reacting strongly in the PF4/heparin ELISA have also been described, but these sera did not activate platelets in the presence of heparin. 14 These antibodies were considered likely to represent an epiphenomenon attributed to the strong systemic in ammatory response of Covid-19. However, given that unusual thromboses such as CVST have been observed in Covid-19 as well as rarely in otherwise healthy individuals receiving the COVID-19 Vaccine AstraZeneca, the question arises as to whether the immune response against the spike protein could induce antibodies that cross-react with immunogenic antigens shared between spike protein and PF4. Accordingly, we sought to identify structural similarities between the SARS-CoV-2 spike protein and PF4. Further, we assessed binding characteristics of anti-PF4 antibodies obtained from patients with the newly recognized disorder, VITT, who developed unusual thrombotic complications. We compared their pattern of PF4-dependent platelet activation with those of sera obtained from a large series of Covid-19 patients from different independent cohorts. Our overall aim was to differentiate whether the reactivity patterns of antibodies in these two patient cohorts, Covid-19 patients immunized by the virus, and patients with VITT, indicate cross-reactivity between SARS-CoV-2 spike protein and PF4, or whether they are distinct entities. Patients between the age of 4 months to 88 years with available serum and positive PCR testing of SARS-CoV-2 in nasopharyngeal swabs were enrolled. Registries began recruiting patients at varying start dates ranging from February 2020 until October 2020. Patient characteristics are summarized in Table 1; registries are described in detail in the Supplementary Material 1.

Covid-19 Patient Cohorts
Sera from 6 VITT patients presenting with thrombocytopenia and thromboembolic events approximately 5-20 days after COVID-19 Vaccine AstraZeneca vaccination were available. 9 Identi cation of immunogenic epitopes and homologies of human PF4 and SARS-CoV-2 spike protein and the comparative analysis of their 3D structures The protein sequence for human PF4/CXCL4 was retrieved from the ENSEMBL gene data base (ENSG00000163737). 15 Similarly, the protein sequence of the SARS-CoV-2 spike protein (1273 amino acids) was retrieved from publicly available data bases (NCBI: Gene ID 43740568). 16 Using the online prediction tool of the University of Madrid, Spain (http://imed.med.ucm.es/Tools/antigenic.pl), 17 we identi ed potential immunogenic peptide sequences (epitopes) in both protein sequences. In the SIM Alignment online Tool (https://web.expasy.org/sim/), 18 the following default setting parameters were applied (comparison matrix BLOSUM62, gap opening penalty=12 and gap extension penalty=4). For 3D analysis, the MacMYPOL program (https://pymol.org/2/) 19 was used together with the les 6vxx.pbd and 4r9w.pbd available for the PBD database (http://www.rcsb.org) 20 to compare the epitopes on the published structures of the proteins. Testing for PF4/heparin-reactive and platelet-activating immunoglobulin G antibodies For screening of all sera of the Covid-19 cohorts and the patients with VITT, we used an IgG-speci c anti-PF4/heparin ELISA, with antibody binding measured using a secondary antihuman IgG antibody, as described. 22 Optical density (OD) results <0.5 units were considered negative, ≥0.5<1.0 weak-positive, and OD≥1.0 strong-positive.
We performed platelet activation assays using puri ed, washed platelets from healthy volunteers, as described, 9 using patient sera, or the respective puri ed anti-PF4/heparin IgG fractions with and without addition of PF4 (10 µg/mL) (Chromatec, Greifswald, Germany). Unfractionated heparin (100 IU/mL, nal) was added to evaluate inhibition of antibody-and PF4-dependent platelet activation. Platelet activation was judged positive if at least two of 3 donor cells aggregated within 30 minutes. 23,24 A nity puri cation of PF4 and PF4/heparin IgG antibodies Biotinylated PF4 (biotin-PF4) (Chromatec, Greifswald, Germany) and biotin-PF4/heparin complexes were coupled to streptavidin-conjugated paramagnetic microbeads (Dynabeads MyOne Streptavidin T1, Invitrogen). Beads were incubated with the serum, unbound antibodies and plasma removed by washing, and the IgG fractions were eluted (details in Supplementary Material 2).
Binding studies of a nity puri ed anti-PF4 and anti-PF4/heparin IgG to SARS-CoV-2 S-1 domain, receptor-binding domain, full-length spike protein, PF4 and PF4/heparin complexes We identi ed sera testing positive for anti-PF4/heparin antibodies from two patient groups, (a) patients with Covid-19 disease (only a minority tested positive), and (b) patients with VITT (all tested positive). These sera were assessed for anti-spike protein antibodies using the SARS-CoV-2 full-length spike protein, the receptor-binding domain (RDB) using in-house ELISAs, and a commercially-available CoV-2 ELISA (recombinant S1-domain; EI 2606-9620 G; EUROIMMUN Medizinische Labordiagnostika AG, Lübeck, Germany). Anti-PF4 and anti-PF4/heparin a nity-puri ed IgG fractions of two VITT patients with documented thromboembolic events were used in a 1:20 dilution (detailed description in Supplementary Material 2).

Results
Identi cation of immunogenic peptide sequences (epitopes) in human PF4 and the SARS-CoV-2 spike protein A total of three and 63 potential immunogenic epitopes, respectively, were identi ed within the 70 amino acid sequence of PF4 (6-21, 23-43 and 49-66) and the 1273 amino acid long sequence of the SARS-CoV-2 spike protein (bold letters in the amino acid sequences, Supplementary Table S1A). The spike protein variants ∆H69∆V70, E484K and N501Y (present in B.1.1.7, B.1.351, and P.1, respectively) showed the same immunogenic pro le (data not shown).
Both proteins showed sequence homologies between 23.5% and 66.7% (Supplementary Table S1B). Overlapping homologous sequences varied between 5 and 22 residues in length. In addition, sequence identities were manually investigated for identical amino acids that are spaced by 2-3 amino acids to be localized on the same side of a particular motif. Restricting the search to motifs longer than 10 amino acids identi ed three motifs within the spike protein sequence (145-155, 323-335, and 677-694) that shared a potential immunogenic epitope with PF4. One of them is located in the structure le (6vxx.pbd: 323-335) and displays high similarity to two consecutive epitopes within PF4 (6-21/23-43).
For 3D comparison analysis, we used the following pbd-les: 6vxx.pdf for the trimeric spike protein and 4r9w.pbd for dimeric PF4 bound to fondaparinux. Both the "15-27"-and the "323-335"-sequences display a ß-sheet-exible loop structure. While the spike epitope resembles a planar con guration, the PF4 structure is more of a pleated sheet (see Figure 1A). Of interest, the same motif in PF4 is involved in binding fondaparinux and heparin (see Figure 1B). 25 In Figure 1C, the surface epitope "323-335" is shown in one subunit of the trimeric SARS-CoV-2 spike protein (left), and the identi ed epitope is enlarged again in the magni ed inset on the right. We assume that part of this epitope is similar in structure and shape surrounding the central Valine-Arginine motif between the spike protein and PF4. Binding of an antibody to this epitope may induce small conformational changes in PF4, similar to what has been observed by heparin binding to PF4.

Covid-19 patient cohorts
From the ve patient cohorts, a total of 222 patients (125 males, 97 females; median age, 55 years [range, 4 months to 88 years]) were evaluated in the IgG-speci c PF4/heparin ELISA. Nineteen of 222 (8.6%) patients tested positive (above the 0.500 optical density threshold), with 13 testing in a range between OD 0.500 and <1.000 and 6 testing between OD 1.000 to <2.000) ( Table 1). We did not observe differences in reactivity among the ve different patient cohorts when analyzed per participating center, excluding preanalytical problems or a batch effect (data not shown).
Sera from all 19 patients who tested positive in the anti-PF4/heparin ELISA were tested in the platelet activation assay in the presence of heparin and of PF4, respectively, to judge heparin-and PF4-dependent platelet activation. Under reaction conditions previously shown to result in typically strong serum-induced platelet activation of COVID-19 Vaccine AstraZeneca vaccinated VITT patients (PF4, 10 µg/mL), we found that 4/19 sera showed weak to moderate platelet activation in the presence of PF4 (lag time, median 15 minutes, range 10 to >30 min [for non-reacting platelets], cut off 30 min); in contrast, none of these sera showed platelet activation in the presence of 0.2 anti-factor Xa U/mL low-molecular-weight heparin. For 10/222 patients, thromboembolic complications were reported (six patients with pulmonary embolism, one patient with stroke, two patients with portal vein thrombosis, one thrombosis of unknown localization). Nine of these 10 patients tested negative by PF4/heparin ELISA. Only one serum was reactive with OD>1.0; for this patient a pulmonary embolism was reported. None of these ten sera, including from the patient with pulmonary embolism, induced platelet aggregation in the functional test, regardless of whether heparin or PF4 was added.
No serological cross-reactivity of puri ed anti-PF4 and anti-PF4/heparin antibodies from VITT patient serum with recombinant SARS-CoV-2 spike protein As expected with an early primary immune response, the sera of (recently-vaccinated) VITT patients contained weakly to moderately binding IgG to the S1 sequence and the RBD sequence of the spike protein, with somewhat higher levels of optical density (OD) values to the full-length spike protein. In contrast, all VITT sera showed very strong binding to PF4 and PF4/heparin complexes (OD>3.0; Figure 2). Antibodies a nity puri ed using PF4 or PF4/heparin from two VITT sera also reacted strongly in the PF4 and PF4/heparin ELISA and strongly activated platelets in the presence of PF4 (data not shown), but did not bind to any of the SARS-CoV-2 spike protein constructs.

Discussion
The mechanism by which ChAdOx1 nCov-19 vaccination rarely induces antibodies that cause marked PF4-dependent platelet activation with resulting thrombocytopenia and unusual thromboses is unresolved. One potential mechanism is a general response of the immune system triggered either by vaccination or by the proin ammatory state of severe acute Covid-19. Clinical observations show unusually strong proin ammatory symptoms in the majority of individuals starting about eight to twelve hours post-vaccination, lasting for 12-24 hours. Potentially, this in ammatory response in certain individuals (e.g. by differences in their genetics, HLA type, or proin ammatory conditions) may have led to the observed occurrences of severe VITT (n=31 patients reported in Germany at the time of writing). Reducing the vaccine dose of the COVID-19 Vaccine AstraZeneca might reduce in ammatory reactions.
The published report from the rst phase I/III study where a reduced dose of 2.5 x 10 10 viral particles was administered in the rst vaccination shot indicates no marked difference in antibody response compared to the currently used dose. 26 Another possibility is that the immune response induced by infection with SARS-CoV-2 results in antibodies against the spike protein that also cross-react with PF4. In this scenario, Covid-19 vaccination could potentially trigger formation of especially strong anti-spike protein antibodies, cross-reacting with PF4 and thereby becoming highly pathogenic through anti-PF4-mediated platelet activation. Indeed, structural analysis of both the spike protein and PF4 indicated potential cross-reactive epitopes. However, by using puri ed recombinant spike protein, puri ed PF4, and a nity puri ed anti-PF4 antibodies from sera of VITT patients, we found no evidence for cross-reactivity. The platelet-activating anti-PF4 antibodies obtained from individuals with VITT post-vaccination with COVID-19 Vaccine AstraZeneca, did not cross-react with the spike protein on SARS-CoV-2. Of particular interest are the different magnitudes of antibody response against both proteins, indicating two different immune responses. 9 The VITT patients showed strong antibody reactivity against PF4 within 5-14 days post-vaccination presumably re ecting a secondary immune response. A primary immune response is extremely unlikely to yield such high IgG reactivity (titers >1:3,000; data not shown). Precedence for this concept is found in the heparininduced thrombocytopenia literature: patients who develop this complication with their rst heparin exposure develop a strong IgG immune response beginning as early as 4-5 days post-immunizing heparin exposure , 27,28 consistent with prior presensitization through naturally-occurring polyanions. [29][30][31] When we performed a combined analysis of ve patient cohorts comprising 222 Covid-19 patients with variable clinical disease severity, we found no evidence for an association between anti-PF4/heparin IgG and thromboembolic complications in Covid-19 patients. The frequency of anti-PF4/heparin IgG detectable by ELISA was 8.6%. This number was even lower than that observed in a prospective study in non-Covid-19 intensive care unit patients, in whom we found 17.2% anti-PF4/heparin IgG detected by ELISA and 5.5% testing positive by platelet activation test 10 days after admission without signs of heparin-induced thrombocytopenia. 32 None of the Covid-19 patients showed heparin-dependent plateletactivating antibodies, while the frequency of PF4-dependent platelet-activating antibodies was only 1.9% (4/222). Moreover, the reactivities of the 4 Covid-19 patient sera were all weak versus the generally strong reactivities seen with VITT sera (lag times, median 15 minutes versus <2-5 minutes, respectively).
Overall, Covid-19 patients with anti-PF4/heparin antibodies and PF4-dependent platelet-activating properties showed clinical characteristics similar to that of the Covid-19 patients without anti-PF4/heparin antibodies; in particular, none of them developed thrombosis. In our patient cohort, thromboembolic events occurred in 4.5% of patients, with no CVST detected. Only one patient with thrombosis was reactive in the PF4/heparin ELISA but that patient's serum did not activate platelets, either in the presence of heparin or PF4. This indicates that thrombotic events in Covid-19 patients are not associated with the presence of the same anti-PF4 platelet-activating antibodies identi ed in vaccinated people who develop VITT.
Patients with Covid-19 and most individuals after SARS-CoV-2 vaccination express antibodies against the spike protein. [33][34][35] However, the a nity-puri ed anti-PF4 and anti-PF4/heparin antibodies from sera of VITT patients did not bind to full-length spike protein, the S1 domain, or the RBD domain, but strongly bound in the PF4/heparin ELISA, and induced strong PF4-dependent platelet activation. In contrast, tested sera from Covid-19 patients strongly bound to the spike protein ( Figure 2). This further indicates that the immune responses to both proteins are independent of each other.
Taken together, our ndings make it unlikely that cross-reacting antibodies recognizing similar antigenic epitopes on SARS-CoV-2 spike protein and PF4 induced by vaccination are the reason for the severe thrombotic complications post-vaccination with COVID-19 Vaccine AstraZeneca. Our results also make it unlikely that anti-SARS-CoV-2 spike protein antibodies are responsible for thrombotic complications in most Covid-19 patients. This information is critical for further risk-bene t assessment of the ongoing large vaccination programs as our ndings make it unlikely that the intended immune response against the SARS-CoV-2 spike protein itself induces severe VITT. Elucidating the underlying mechanism by which vaccination against SARS-CoV-2 spike protein rarely induces anti-PF4 antibodies causing VITT is urgently warranted. However, our study indicates there is no apparent need to change the SARS-CoV-2 antigen target for the vaccination strategy.

Declarations
All COI information is given in the Supplementary Files.  Table   Table 1 Comparison of the 3D-structures of PF4 and SARS-CoV-2 Spike A. The identi ed epitopes TESTVRFPNITNL (Spike) and TTSQVRPRHITSL (PF4 are shown in their secondary structure. Both identi ed linear epitopes share a ß-sheet-exible loop structure that could initiate an unintended crossreactivity of antibodies. B. The structure of dimer PF4 with bound fondaparinux is shown. The identi ed epitope makes part of the binding pocket for fondaparinux. Both identi ed linear epitopes share a ß-sheet exible loop structure that could initiate an unintended cross-reactivity of antibodies. C. The homologous epitope (colored spheres) is shown for one (grey structure) of the three subunits of the spike trimer which are displayed on the left. The same epitope is highlighted in the magni ed inset on the right.