Emerging Microbes & Infections

2020, VOL. 9
https://doi.org/10.1080/22221751.2020.1762515

Kinetics of SARS-CoV-2 specific IgM and IgG responses in COVID-19 patients

Baoqing Suna†, Ying Fenga†, Xiaoneng Mob†, Peiyan Zhenga, Qian Wangc, Pingchao Lic, Ping Pengb,
Xiaoqing Liua, Zhilong Chenc, Huimin Huanga, Fan Zhangc, Wenting Luoa, Xuefeng Niua, Peiyu Huc,
Longyu Wangc, Hui Pengb, Zhifeng Huanga, Liqiang Fengc, Feng Lib, Fuchun Zhangb, Fang Lib,
Nanshan Zhonga and Ling Chena,b,c
a
State Key Laboratory of Respiratory Diseases, National Clinical Research Center of Respiratory Disease, Guangzhou Institute of Respiratory
Health, The First Affiliated Hospital of Guangzhou Medical University, Guangzhou, People’s Republic of China; bInstitute of Infectious
Diseases, Guangzhou Eighth people’s Hospital, Guangzhou Medical University, Guangzhou, People’s Republic of China; cGuangzhou
Regenerative Medicine and Health-Guangdong Laboratory (GRMH-GDL), Guangzhou Institutes of Biomedicine and Health, Chinese
Academy of Sciences, Guangzhou, People’s Republic of China

ABSTRACT
The emerging COVID-19 caused by SARS-CoV-2 infection poses severe challenges to global public health. Serum antibody
testing is becoming one of the critical methods for the diagnosis of COVID-19 patients. We investigated IgM and IgG
responses against SARS-CoV-2 nucleocapsid (N) and spike (S) protein after symptom onset in the intensive care unit
(ICU) and non-ICU patients. 130 blood samples from 38 COVID-19 patients were collected. The levels of IgM and IgG
specific to N and S protein were detected by ELISA. A series of blood samples were collected along the disease course
from the same patient, including 11 ICU patients and 27 non-ICU patients for longitudinal analysis. N and S specific
IgM and IgG (N-IgM, N-IgG, S-IgM, S-IgG) in non-ICU patients increased after symptom onset. N-IgM and S-IgM in some
non-ICU patients reached a peak in the second week, while N-IgG and S-IgG continued to increase in the third week.
The combined detection of N and S specific IgM and IgG could identify up to 75% of SARS-CoV-2 infected patients in
the first week. S-IgG was significantly higher in non-ICU patients than in ICU patients in the third week. In contrast, N-
IgG was significantly higher in ICU patients than in non-ICU patients. The increase of S-IgG positively correlated with
the decrease of C-reactive protein (CRP) in non-ICU patients. N and S specific IgM and IgG increased gradually after
symptom onset and can be used for detection of SARS-CoV-2 infection. Analysis of the dynamics of S-IgG may help to
predict prognosis.

ARTICLE HISTORY Received 5 April 2020; Revised 21 April 2020; Accepted 23 April 2020

KEYWORDS COVID-19; SARS-CoV-2; IgM; IgG; C-reactive protein

Introduction
to human health (https://www.who.int/emergencies/
Since December 2019, cases of unexplained pneumonia diseases/novel-coronavirus-2019/situation-reports).
have occurred in Wuhan City, Hubei Province, sub- The four major structural proteins of coronavirus
sequent virus isolation and whole-genome sequencing are the spike surface glycoprotein (S), small envelope
(accession#: MN908947) confirmed that it is an acute protein (E), matrix protein (M), and nucleocapsid
respiratory infection caused by new severe acute respir- protein (N). The spike protein (S) of coronavirus is a
atory syndrome coronavirus 2 (SARS-CoV-2) [1,2]. type I transmembrane glycoprotein and mediates the
Coronaviruses are enveloped, non-segmented, single- entrance to human respiratory epithelial cells by inter-
positive-stranded RNA viruses with round or oval acting with cell surface receptor angiotensin-convert-
particles and a diameter of 50–200 nm. Coronavirus ing enzyme 2 (ACE2) [3], the S protein contains
subfamily is divided into four genera: α, β, γ and δ distinct functional domains near the amino (S1) and
according to serotype and genomic characteristics. carboxy (S2) termini, the peripheral S1 portion can
The SARS-CoV-2 belongs to the genus β which has independently bind cellular receptors while the integral
been confirmed to be highly infectious by research. membrane S2 portion is required to mediate fusion of
As of April 20, 2020, SARS-CoV-2 has caused more viral and cellular membranes [4–6]. The nucleocapsid
than 2446840 laboratory-confirmed human infec- protein (N) forms complexes with genomic RNA,
tions, including 170993 deaths, posing a serious threat interacts with the viral membrane protein during

CONTACT Ling Chen [email protected]; Nanshan Zhong [email protected]; Fang Li [email protected]

†
These authors contributed equally to this work.
Ling Chen, Nanshan Zhong and Fang Li are the senior authors.
Supplemental data for this article can be accessed https://doi.org/10.1080/22221751.2020.1762515
© 2020 The Author(s). Published by Informa UK Limited, trading as Taylor & Francis Group, on behalf of Shanghai Shangyixun Cultural Communication Co., Ltd
This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted
use, distribution, and reproduction in any medium, provided the original work is properly cited.
 Emerging Microbes & Infections 941

virion assembly and plays a critical role in enhancing adenovirus, and human coronaviruses OC43 and
the efficiency of virus transcription and assembly [7–9]. HKU1) infected patients (Supplementary Fig. 1).
The SARS-CoV-2 has human-to-human trans- Microtiter plates were coated with 50 ng/well of target
mission characteristics and a high fatality in critically protein overnight at 4 °C. Plates were then blocked for
ill patients. Compared with non-ICU patients, ICU 2 h at 37°C using 200 μL of 5% non-fat milk in 1 x
patients had higher plasma levels of IL2, IL6, IL7, phosphate buffered saline (PBS). Serum samples were
IL10, GSCF, IP10, MCP1, MIP1A, TNFα, lactate dehy- then diluted 1:50 in 1X PBS and 100 μL of each sample
drogenase (LDH), ferritin and D-dimer. The number was applied to coated ELISA plate and incubated for
of lymphocytes was significantly reduced and C-reac- 2 h at 37 °C. Plates were then washed and incubated
tive protein (CRP) was significantly increased in severe with HRP-labeled anti-human IgM and IgG (Sigma
cases [10–13]. During the submission of this paper, Aldrich, MI, USA), diluted to 1:2000 in 5% non-fat
several publications have also reported the analysis of milk in 1 x PBS. After incubation for another 1 h at
antibody responses to N protein, S protein, and recep- room temperature, the plates were washed and devel-
tor-binding domain (RBD) on S protein in COVID-19 oped with TMB/E substrate (Merck Millipore, MA,
patients [14–18]. However, the seropositive rate of both USA). Finally, the reaction was stopped with 1M
IgM and IgG responses within one week after onset and H2SO4, and the OD450 nm values were read. Negative
in the context of both N protein and S protein has not serum control was run each time the assay was per-
been clarified. The kinetics of antibody responses in formed. The cut-off value for seropositivity samples
critical cases or ICU patients has not been reported, was set as the mean value at optical density 450 (at a
and no studies have suggested whether antibody 1:50 dilution) for the 16 negative serum samples plus
response is associated with disease prognosis. Here, 3 standard deviations (SDs).
we systemically investigated the kinetics of IgG and
IgM responses to both N and S proteins in the first 4
weeks after the symptom onset in ICU and non-ICU Ethics approval
patients. Our study can help to facilitate serologically This study was approved by the First Affiliated Hospi-
based diagnosis and prediction of disease prognosis. tal of Guangzhou Medical University and Guangzhou
Eighth people’s Hospital. Written informed consent
was waived for in the light of this emerging infectious
Materials and methods disease of high clinical relevance. All healthy control
Source of serum samples subjects signed written informed consent before the
collection of peripheral blood.
One hundred thirty blood samples from 38 patients
were collected between 3 and 28 days after symptom
onset. Blood samples from non-ICU patients with Statistical analysis
confirmed SARS-CoV-2 infection were collected from
Statistical analyses and graphical presentations were
27 non-ICU patients from the Guangzhou Eighth
conducted with GraphPad Prism version 7.0 (Graph-
People’s Hospital. Blood samples from 11 ICU patients
Pad Software, Inc., CA, USA). We compared categori-
were collected from the First Affiliated Hospital of
cal variables of basic clinical characteristics of ICU and
Guangzhou Medical University. 16 negative serum
non-ICU patients using Fisher’s exact test. Differences
samples were collected from healthy volunteers. The
of antibody responses between ICU and non-ICU
serum samples were separated after centrifugation at
patient groups were determined by Student’s t test.
3,000 rpm for 10 min, and then inactivated at 56°C
Throughout the text, figures, and legends, the following
for 1.5 h.
terminology is used to show statistical significance: *, P
< 0.05; **, P < 0.01; and ***, P < 0.001.
Enzyme-linked immunosorbent assay (ELISA)
SARS-CoV-2 N protein and S protein-specific binding Results
antibodies were analyzed by ELISA as described pre-
N and S specific IgM and IgG were detectable in
viously [19]. N protein (residue 1–419) was produced
75% non-ICU patients in the first week after
from Baculovirus-Insect Cells (Cat. # 40588-V08B,
symptom onset
Sino biological, Beijing, China). S protein (residue
16–685) was produced from HEK293 Cells (Cat. The basic information and clinical symptoms of 27
#40591-V08H, Sino biological, Beijing, China). The non-ICU patients (14 male and 13 female) and 11
specificity of SARS-CoV-2 N and S proteins were ver- ICU patients (10 male and 1 female) are summarized
ified by Western blot analysis using serum samples (Table 1). The non-ICU patients had a median age
from convalescent COVID-19 patients and from 44.0 (interquartile range, IQR: 32.0–56.0), SARS-
other respiratory pathogens (influenza virus, COV-2 nucleic acid positive days of 13.0 (IQR: 12.0–
942 B. Sun et al.

Table 1. Basic information of COVID-19 patients.

Group A: non-ICU patients (n = 27) Group B: ICU patients (n = 11) P value
Median ages (IQR) 44.0 (32.0–56.0) 58.0 (49.0–69.5) 0.05
Gender
Female 13(48%) 1(9%) 0.03
Male 14(52%) 10(91%) ..
Median days of admission after symptom onset (IQR)
4(3.75–7) 5(2–10.5) 0.35
Median hospital Stay days (IQR)
19.0 (14.3–22.5) 31.0 (30.0–33.5) <0.001
Median days of SARS-COV-2 Nucleic Acid negative after symptom onset (IQR)
13.0 (12.0–16.3) 31.0 (22.5–32.0) <0.001
Presenting symptoms
Fever 26(96%) 10(91%) 0.5
Cough 22(81%) 11(100%) 0.29
Shortness of breath 5(19%) 8(73%) 0.003
Anorexia 8(30%) 10(91%) <0.001
Underlying medical disorders
None 14(52%) 5(45%) >0.99
Hypertension 5(19%) 4(36%) 0.4
Chronic pulmonary disease 0(0%) 2(18%) 0.08
Coronary heart disease 1(4%) 0(0%) >0.99
Chronic gastritis 2(7%) 0(0%) >0.99
Liver cyst 1(4%) 0(0%) >0.99
Tuberculosis 1(4%) 0(0%) >0.99
Hyperlipidemia 1(4%) 0(0%) >0.99
Fatty liver 1(4%) 0(0%) >0.99
Thalassemia 1(4%) 0(0%) >0.99
Colon cancer 1(4%) 0(0%) >0.99
Diabetes 0(0%) 5(45%) <0.001
IQR: Interquartile range.

16.3), and median hospitalization days of 19.0 (IQR: rates were 73.7% for N-IgM, 68.4% for S-IgM, 84.2%
14.3–22.5). The ICU patients had a median age 58.0 for N-IgG, and 78.9% for S-IgG. The seropositive rate
(IQR: 49.0–69.5), SARS-CoV-2 nucleic acid positive of N-IgM + S-IgM was 84.2%, while the seropositive
days of 31.0 (IQR: 22.5–32.0) or longer, and median rate of N-IgM + N-IgG, N-IgG + S-IgG reached
hospitalization days of 31.0 (IQR: 30.0–33.5) or longer. 94.7%. In the third weeks after symptom onset, the ser-
The levels of N-IgM, N-IgG, S-IgM, S-IgG were opositive rates of either N-IgM or S-IgM maintained at
measured by ELISA. Serum samples from 16 healthy 73.7%, while the seropositive rates of N-IgG and S-IgG
people were used as negative controls. The cut-off reached 100% (Table 2). This result showed that the
value for seropositivity samples was set as the mean seropositive rates of N-IgM, S-IgM, N-IgG, and S-
value at optical density 450 (at a 1:50 dilution) for IgG responses increased with disease course in non-
the 16 negative serum samples plus 3 standard devi- ICU patients (Figure 1A, B).
ations (SDs), which were 0.394, 0.291, 0.284 and
0.170 for N-IgM, N-IgG, S-IgM and S-IgG,
Kinetics of N-IgM, S-IgM, N-IgG and S-IgG had
respectively.
different patterns in non-ICU and ICU patients
The results showed that within one week after the
symptom onset, the seropositive rates of N-IgM, N- In most non-ICU patients, N-IgM and S-IgM reached a
IgG and S-IgM in non-ICU patients were 41.7%, and peak in the second week after symptom onset (Figure
the seropositive rate of S-IgG was 58.3%. The seroposi- 2A, 2C, Supplementary Fig.2). Longitudinal analysis
tive rate of N-IgM + N-IgG, N-IgM + S-IgM were showed a decline for N-IgM and S-IgM in the third
58.3%, while the seropositive rate of S-IgM + S-IgG, week after the onset in some non-ICU patients (Figure
N-IgG + S-IgG reached 66.7%. The seropositive rate 2A, 2C, Supplementary Fig.2). In the first week after
of N-IgM + S-IgM + N-IgG + S-IgG reached 75.0% onset, the levels of N-IgM and N-IgG, S-IgM and S-
(Table 2). This result indicated that the combined IgG were similar. N-IgG had a parallel or similar
detection of N and S specific IgM and IgG can be useful dynamic pattern as N-IgM in the first two weeks for
for early detection of SARS-CoV-2 infection. In the the same patient. However, in some patients, N-IgM
second weeks after symptom onset, the seropositive showed plateau or declined in the third week while

Table 2. Seropositive rate (%).

Weeks N-IgM N-IgG S-IgM S-IgG N-IgM + N-IgG S-IgM + S-IgG N-IgM + S-IgM N-IgG + S-IgG N-IgM + S-IgM + N-IgG + S-IgG
1 41.7 41.7 41.7 58.3 58.3 66.7 58.3 66.7 75.0
2 73.7 84.2 68.4 78.9 94.7 89.5 84.2 94.7 94.7
3 73.7 100.0 73.7 100.0 100.0 100.0 89.5 100.0 100.0
N-IgM: N protein specific IgM; N-IgG: N protein specific IgG; S-IgM: S protein specific IgM; S-IgG: S protein specific IgG.
 Emerging Microbes & Infections 943

Figure 1. The seropositive rates of N and S specific IgM and IgG antibody responses in non-ICU patients after symptom onset. A. The
changes in seropositive rates of N-IgM, N-IgG, S-IgM and S-IgG in 27 non-ICU patients. B. The changes in seropositive rates of N-IgM
+ N-IgG, S-IgM + S-IgG, N-IgM + S-IgM, N-IgG + S-IgG, N-IgM + S-IgM + N-IgG + S-IgG in 27 non-ICU patients.

N-IgG continued to increase. The level of N-IgG sur- patients (correlation coefficient r = 0.692, P = 0.0001)
passed N-IgM in the second and third week after than in ICU patients (correlation coefficient r = 0.377,
onset (Table 3). S-IgG also had a parallel or similar P = 0.01) (Supplementary Fig.4B, D).
dynamic pattern as S-IgM for the same person in the
first two weeks for non-ICU patients. In the third
The increase of S-IgG positively correlated with
week, the level of S-IgG continued to increase and sur-
the decrease of CRP in non-ICU patients
passed the level of S-IgM in the same patient (Figure
2A, B, C, D, Supplementary Fig.2), suggesting that C-reactive protein (CRP) is an acute protein that rises
there was an IgM to IgG class-switch from in most sharply in the plasma when the body is infected or the
non-ICU patients. tissue is damaged. It is a non-specific marker of inflam-
In ICU patients, the dynamic patterns of N and S mation and directly participates in the host defence
IgM and IgG were more chaotic. N-IgM in 63.6% of against infection. The levels of N and S specific IgM
ICU patients appeared to remain at low and static and IgG were evaluated for correlations with CRP
levels, while in 36.3% of ICU patients N-IgM had the levels in non-ICU patients. As the disease progressed,
high but static level for at least 4 weeks (Figure 2E). the increase of S-IgG positively correlated with the
N-IgG levels in all ICU patients reached high levels decrease of CRP in non-ICU patients (Figure 4B), the
(OD450 > 2.0) within 2 weeks after symptom onset correlation coefficients r were 0.9 (P = 0.001). However,
(Figure 2F). In 81.8% of ICU patients, N-IgG exceeded the changes of N-IgG showed no correlation with the
N-IgM levels in the same patient by 2 weeks after changes of CRP in non-ICU patients (Figure 4A).
symptom onset (Figure 2F, Supplementary Fig. 3, A- The changes of N-IgM, and S-IgM also showed no sig-
E, G, I, J, K). N-IgG was significantly higher than N- nificant correlations with CRP in non-ICU patients
IgM in the second and third week after onset in ICU (Figure 4C, D).
patients (Table 3, Supplementary Fig. 3). S-IgM had
either poor responses or maintained a static but high
S-IgG was significantly lower in ICU patients
level in ICU patients (Figure 2G, Table 3, Supplemen-
than in non-ICU patients in the third weeks after
tary Fig. 3). S-IgG appeared to increase slowly as com-
symptom onset
pared to the increase of N-IgG (Figure 2H, Table 3,
Supplementary Fig. 3). In the third week after onset, In the second and third week after symptom onset, N-
S-IgG was higher than S-IgM in most ICU patients IgM was significantly higher in ICU patients than in
(Table 3, Supplementary Fig. 3). non-ICU patients (Figure 5A, Table 3, P < 0.001). N-
The correlation between the corresponding S-IgM, IgG was significantly higher in ICU patients than in
S-IgG, N-IgM, and N-IgG levels in each patient were non-ICU patients after onset (Figure 5B, Table 3, P <
analyzed (Figure 3). In non-ICU patients, there was a 0.05). S-IgM was significantly higher in ICU patients
strong correlation between the S-IgG with S-IgM levels, than non-ICU patients only in the second weeks after
whereas there was no correlation between N-IgM with symptom onset (Figure 5C, Table 3, P < 0.05). In con-
N-IgG levels. In ICU patients, there were no corre- trast, S-IgG was significantly lower in ICU patients
lations either between S-IgG with S-IgM or between than in non-ICU patients in the third weeks after
N-IgG with N-IgM levels. The S-IgG levels showed a symptom onset (Figure 5D, Table 3 P < 0.05). More-
higher correlation with N-IgG levels in non-ICU over, N-IgG/S-IgG ratio was significantly higher in
944 B. Sun et al.

Figure 2. Kinetics of N and S specific IgM and IgG responses in non-ICU patients and ICU patients. (A) N-IgM, (B) N-IgG, (C) S-IgM, (D)
S-IgG responses in 7 non-ICU patients; (E) N-IgM, (F) N-IgG, (G) S-IgM, (H) S-IgG antibodies response in 11 ICU patients.

Table 3. SARS-COV-2 N and S specific IgM and IgG responses (OD450nm: mean + SD).
N S
Weeks after onset Patients IgM IgG P value IgM IgG P value The ratio of N-IgG/S-IgG
Week1 non-ICU (n = 14) 0.44 + 0.41 0.51 + 0.53 0.358 0.36 + 0.32 0.24 + 0.20 0.132 2.56 + 3.09
ICU (n = 6) 0.99 + 1.03 1.31 + 1.0 0.316 0.36 + 0.21 0.23 + 0.14 0.145 6.04 + 3.50
P value 0.061 0.020 0.500 0.481 0.025
Week2 non-ICU (n = 19) 0.51 + 0.28 1.0 + 0.69 0.004 0.42 + 0.27 0.58 + 0.49 0.112 2.87 + 3.45
ICU (n = 15) 1.51 + 1.12 2.38 + 0.86 0.014 0.69 + 0.48 0.57 + 0.28 0.216 5.98 + 5.10
P value 0.000 0.000 0.026 0.467 0.024
Week3 non-ICU (n = 20) 0.6 + 0.38 1.93 + 0.73 0.001 0.64 + 0.52 1.25 + 0.62 0.001 2.11 + 1.67
ICU (n = 25) 1.50 + 1.01 2.92 + 0.52 <0.001 0.70 + 0.43 1.01 + 0.36 0.025 3.38 + 1.67
P value <0.001 <0.001 0.335 0.028 0.011
ICU: intensive care unit. N-IgG: N protein specific IgG; S-IgG: S protein specific IgG.
 Emerging Microbes & Infections 945

Figure 3. The correlation between N and S specific IgM and IgG responses in non-ICU patients and ICU patients. A. The correlation
between S-IgG and S-IgM in non-ICU patients; B. The correlation between N-IgG and N-IgM in non-ICU patients; C. The correlation
between S-IgG and S-IgM in ICU patients; D. The correlation between N-IgG and N-IgM in ICU patients. The Pearson correlation
coefficient was used to measure the strength of the correlation between IgM and IgG antibodies. The correlation coefficient
was calculated using Student’s t-test, a P-value < 0.05 was considered statistically significant. *, P < 0.05; **, P < 0.01; ***, P < 0.001.

Figure 4. The correlation between N and S specific IgM and IgG responses with CRP in non-ICU patients. A. The correlation between
N-IgG and the reduction of CRP; B. The correlation between S-IgG and the reduction of CRP; C. The correlation between N-IgM and
CRP; D. The correlation between S-IgM and CRP. The Pearson correlation coefficient was used to measure the strength of the cor-
relation between CPR and IgM or IgG antibodies.
946 B. Sun et al.

Figure 5. The N and S specific IgM and IgG responses in non-ICU patients and ICU patients. A. Comparison of N-IgM responses
between non-ICU and ICU patients; B. Comparison of N-IgG responses between non-ICU and ICU patients; C. Comparison of S-
IgM responses between non-ICU and ICU patients; D. Comparison of S-IgG responses between non-ICU and ICU
patients. E. Comparison of N-IgG/S-IgG ratio between non-ICU and ICU patients. Correlation coefficient was calculated using Stu-
dent’s t test, a P-value < 0.05 was considered statistically significant. *, P < 0.05; **, P < 0.01; ***, P < 0.001.

ICU patients that non-ICU patients throughout the and S-IgG antibody responses in non-ICU patients
disease course (Figure 5E, P < 0.05). ICU patients gradually increased within 1–3 weeks after the onset.
tended to produce more N-IgM and N-IgG than N-IgM and S-IgM reached a peak in the second week,
non-ICU patients. Non-ICU patients tended to have while N-IgG and S-IgG antibodies continued to increase
faster and higher IgM to IgG class switch than ICU in the third week. Joint detection of N-IgM, N-IgG, S-
patients (Table 3, Supplementary Fig.2, Supplementary IgM, and S-IgG antibodies, could detect up to 75% of
Fig.3). This result suggested that the class switch of S- infections in the first week. Joint detection of N-IgM
IgM to S-IgG is vital for clearing the viruses and can be + N-IgG, or N-IgG + S-IgG could detect up to 94.7%
used as a prognosis indicator to predict the outcome of of infections in the second week. In the third weeks
COVID-19 disease. after symptom onset, seropositive rates for N-IgG and
S-IgG reached 100%. In contrast, seropositive rates for
N-IgM and S-IgM remained the same as some patients
Discussion
started to decline as the result of IgM to IgG isotype
This study investigated the kinetics of N and S specific switch, which may help to generate more effective anti-
IgM and IgG responses in COVID-19 patients after bodies that can inhibit virus infection.
symptom onset. A total of 130 blood samples from 38 The effective method to control the spread of the
COVID-19 patients were analyzed. Our study showed virus is the early diagnosis and early isolation of
that the seropositive rates of N-IgM, N-IgG, S-IgM patients. However, the incubation period of the SARS-
 Emerging Microbes & Infections 947

CoV-2 and the limitation of Q-PCR for nucleic acid has not been reported before. Notably, in the third
detection affect the positive rate of early diagnosis. It week after symptom onset, N-IgG was significantly
has been reported that the serum antibody ELISA and higher while S-IgG is significantly lower in ICU
Q-PCR combined detection may increase the positive patients than in non-ICU patients. Non-ICU patients
rate for the early diagnosis of COVID-19 infection tend to produce S-IgG antibodies, while ICU patients
[20–23]. During the preparation of this manuscript, a tend to produce N-IgG antibodies. Interestingly, S-
paper reported antibody detection in 23 COVID-19 IgG had a parallel or similar dynamic pattern as S-
patients. The seropositivity rates were 94% for anti- IgM in the first two weeks, but S-IgG continued to
NP IgG, 88% for anti-NP IgM, 100% for anti-RBD increase in the third week while S-IgM in some patients
IgG, and 94% for anti-RBD IgM [21]. However, the showed plateau or decline in some patients. The similar
blood samples were collected at 14 days or later after pattern also occurred in non-ICU patients, but not in
symptom onset, but not in the first week as our study, ICU patients. This result suggested that the early
which can be more useful for diagnostic purpose. It is class switching of IgM to IgG may help predict a better
also important to note that RBD only represents a outcome of COVID-19 disease.
small part of S protein (237 amino acids in RBD as com- It has been recognized that S-specific antibodies can
pared to1273 amino acids in S protein). Therefore, the block the binding of S protein to cellular receptor
anti-RBD IgM and anti-RBD IgG response may not hACE2 that mediates SARS-CoV-2 binding and entry
represent the antibody response to S protein. Based to target cells. There was no evidence that N-specific
on our study, we proposed that the combined detection antibodies can block virus infection. N protein is a suit-
of both N and S specific IgM and IgG may improve the able candidate for early diagnosis of infection, due to its
serological detection rate SARS-CoV-2 infection in the high immunogenicity and intracellular accumulation
early stage. While the decline of IgM/IgG ratio may help before the packaging of the virus [28–30]. A previous
to identify the post-infected people, although it is still study on SARS-CoV infection indicated that IgG
too early to know when IgM will wane over time. response is directed most frequently and predomi-
There may be a concern that potential cross-reactivity nantly at the N protein (89%), but not S protein
of SARS-CoV-2 N proteins with other human corona- (63%) [31]. In this study, we found that most ICU
viruses, which may affect the seropositive rate of patients had higher N-IgG than S-IgG after the symp-
SARS-COV-2. An earlier analysis showed that there tom onset, which may be caused by longer and a large
were no cross-reactivity of SARS-CoV-2 N protein amount of virus exposure in the early infections of ICU
with human plasma positive for IgG antibodies against patients. It is important to note that ICU patients had
human coronaviruses NL63, 229E, OC43, and HKU1. SARS-CoV-2 nucleic acid positive days of 31.0,
There is a cross-reactivity between SARS-CoV positive whereas non-ICU patients had SARS-CoV-2 nucleic
human plasma and SARS-CoV-2 N protein [20]. The acid positive days of 13. Therefore, a continuous
patients in this study have not been infected with increase of N-IgG may indicate disease progression
SARS-CoV. We also confirmed the specificity of N towards more severe illness. In contrast, S-IgG
and S antigens that we used in our study by checking increased slowly in ICU patients. Research about the
cross-reactivity with serum samples from people contributions of the structural proteins of SARS-CoV
infected with other human coronaviruses and human to protective immunity indicated that only S protein
adenovirus, from people vaccinated with influenza induced a high titre of SARS-CoV-neutralizing anti-
virus vaccine and from early collections. There was bodies and protective efficacy in hamsters, but not N
also a report of serum detection of N specific IgM, protein, matrix M and small envelope E proteins
IgG and IgA in 135 patients of lower respiratory infec- [32]. Homology modelling and structural evidence
tion and 150 healthy individuals, all showed no reactiv- revealed that SARS-CoV-2 had a similar receptor-
ity with SARS-COV-2 N proteins [20]. binding domain structure to that of SARS-CoV, despite
The clinical symptoms, characteristics, and pro- amino acid variation at some key residues [2,3]. In this
gression of the 27 non-ICU patients and 11 ICU study, we found that S-IgG in ICU patients was signifi-
patients in this study were similar to those of the pre- cantly lower than non-ICU patients by 2 weeks after
viously reported COVID-19 patients [10,13,24–26]. the onset, which may explain the longer hospital
Recently, a retrospective study found that the critically stays and nucleic acid positive days in ICU patients.
ill patient’s fatality rate reached 61.5% within 28 days, Therefore, monitoring the kinetics of S-IgG should
and the median time from ICU to death was 7 days help to predict prognosis.
[27]. A case of COVID-19 death study found that the
CRP increased significantly after the onset of the
patient, and lasted for more than 14 days [11]. In our Acknowledgements
study, CRP also increased significantly in most We thank all 38 patients and 16 healthy researchers who pro-
patients. The increase of S-IgG in non-ICU patients vided blood samples to support scientific research on SARS-
positively correlated with the decrease of CRP, which CoV-2 pandemic control.
948 B. Sun et al.

Disclosure statement [16] Liu W, Lu Y, Zhang J, et al. Viral kinetics and antibody
responses in patients with COVID-19. medRxiv pre-
No potential conflict of interest was reported by the authors. print. 2020. doi:10.1101/2020.03.24.20042382
[17] Okba NMA, Müller MA, Li W, et al. Severe acute res-
piratory syndrome coronavirus 2−specific antibody
Funding responses in coronavirus disease 2019 patients.
Emerg Infect Dis. 2020. doi:10.3201/eid2607.200841
This work was supported by the National Natural Science
[18] Xiang F, Wang X, He X, et al. Antibody detection and
Foundation of China for SARS-CoV-2 (82041014), Guangz-
dynamic characteristics in patients with COVID-19.
hou Health Care and Cooperative Innovation Major Project
Clin Infect Dis. 2020. doi:10.1093/cid/ciaa461
(201508020252), Zhejiang University special scientific
[19] Feng Y, Li C, Hu P, et al. An adenovirus serotype 2-
research funding for COVID-19 prevention and control
vectored ebolavirus vaccine generates robust antibody
(2020XGZX001, 2020XGZX025).
and cell-mediated immune responses in mice and rhe-
sus macaques. Emerg Microbes Infect. 2018;7(1):101.
[20] Guo L, Ren L, Yang S, et al. Profiling early humoral
References response to diagnose novel coronavirus disease
[1] Wu A, Peng Y, Huang B, et al. Genome composition (COVID-19). Clin Infect Dis. 2020. doi:10.1093/cid/
and divergence of the novel coronavirus (2019- ciaa310
nCoV) originating in China. Cell Host Microbe. [21] To KK-W, Tsang OT-Y, Leung W-S, et al. Temporal
2020;27(3):325–328. profiles of viral load in posterior oropharyngeal saliva
[2] Lu R, Zhao X, Li J, et al. Genomic characterisation and samples and serum antibody responses during infection
epidemiology of 2019 novel coronavirus: implications by SARS-CoV-2: an observational cohort study. Lancet
for virus origins and receptor binding. The Lancet. Infect Dis. 2020. doi:10.1016/S1473-3099(20)30196-1
2020;395(10224):565–574. [22] Zhao J, Yuan Q, Wang H, et al. Antibody responses to
[3] Wrapp D, Wang N, Corbett KS, et al. Cryo-EM struc- SARS-CoV-2 in patients of novel coronavirus disease
ture of the 2019-nCoV spike in the prefusion confor- 2019. Clin Infect Dis. 2020. doi:10.1093/cid/ciaa344
mation. Science. 2020;10:1126. [23] Zhang W, Du RH, Li B, et al. Molecular and serological
[4] Gallagher TM, Buchmeier MJ. Coronavirus spike pro- investigation of 2019-nCoV infected patients: impli-
teins in viral entry and pathogenesis. Virology. cation of multiple shedding routes. Emerg Microbes
2001;279(2):371–374. Infect. 2020;9(1):386–389.
[5] Li F. Structure, function, and evolution of coronavirus [24] Chen N, Zhou M, Dong X, et al. Epidemiological and
spike proteins. Annu Rev Virol. 2016;3(1):237–261. clinical characteristics of 99 cases of 2019 novel coro-
[6] Zhu Z, Zhang Z, Chen W, et al. Predicting the recep- navirus pneumonia in Wuhan, China: a descriptive
tor-binding domain usage of the coronavirus based study. The Lancet. 2020;395(10223):507–513.
on kmer frequency on spike protein. Infect Genet [25] Wang D, Hu B, Hu C, et al. Clinical characteristics of
Evol. 2018;61:183–184. 138 hospitalized patients with 2019 novel corona-
[7] McBride R, van Zyl M, Fielding BC. The coronavirus virus-infected pneumonia in Wuhan, China. JAMA.
nucleocapsid is a multifunctional protein. Viruses. 2020;323(11):1061–1069.
2014;6(8):2991–3018. [26] Guan WJ, Ni ZY, Hu Y, et al. Clinical characteristics
[8] Verheije MH, Hagemeijer MC, Ulasli M, et al. The cor- of coronavirus disease 2019 in China. N Engl J Med.
onavirus nucleocapsid protein is dynamically associ- 2020;382(18):1708–1720.
ated with the replication-transcription complexes. J [27] Yang X, Yu Y, Xu J, et al. Clinical course and outcomes
Virol. 2010;84(21):11575–9. of critically ill patients with SARS-CoV-2 pneumonia
[9] Chang CK, Hou MH, Chang CF, et al. The SARS cor- in Wuhan, China: a single-centered, retrospective,
onavirus nucleocapsid protein–forms and functions. observational study. Lancet Resp Med. 2020. doi:10.
Antiviral Res. 2014;103:39–50. 1016/S2213-2600(20)30079-5
[10] Chang D, Lin M, Wei L, et al. Epidemiologic and clini- [28] Garoff H, Hewson R, Opstelten DJ, et al. Virus matu-
cal characteristics of novel coronavirus infections ration by budding. Microbiol Mol Biol Rev. 1998
involving 13 patients outside Wuhan, China. JAMA. Dec;62(4):1171–1190.
2020;323(11):1092–1093. [29] Chan PK, Liu EY, Leung DT, et al. Evaluation of a
[11] Xu Z, Shi L, Wang Y, et al. Pathological findings of recombinant nucleocapsid protein-based assay for
COVID-19 associated with acute respiratory distress anti-SARS-CoV IgG detection. J Med Virol. 2005;75
syndrome. Lancet Resp Med. 2020;8(4):420–422. (2):181–184.
[12] Liu T, Zhang J, Yang Y, et al. The potential role of [30] Liang FY, Lin LC, Ying TH, et al. Immunoreactivity
IL-6 in monitoring severe case of coronavirus disease characterisation of the three structural regions of the
2019. medRxiv preprint. 2020. doi:10.1101/2020.03. human coronavirus OC43 nucleocapsid protein by
01.20029769 Western blot: implications for the diagnosis of
[13] Huang C, Wang Y, Li X, et al. Clinical features of coronavirus infection. J Virol Methods. 2013;187
patients infected with 2019 novel coronavirus in (2):413–420.
Wuhan, China. The Lancet. 2020;395(10223):497–506. [31] Leung DT, Tam FC, Ma CH, et al. Antibody response
[14] Wölfel R, Victor MC, Wolfgang G, et al. Virological of patients with severe acute respiratory syndrome
assessment of hospitalized patients with COVID- (SARS) targets the viral nucleocapsid. J Infect Dis.
2019. Nature. doi:10.1038/s41586-020-2196-x (2020). 2004;190(2):379–386.
[15] Liu W, Liu L, Kou G, et al. Evaluation of nucleocapsid [32] Buchholz UJ, Bukreyev A, Yang L, et al. Contributions
and spike protein-based ELISAs for detecting anti- of the structural proteins of severe acute respiratory
bodies against SARS-CoV-2. J Clin Microbiol. 2020. syndrome coronavirus to protective immunity. Proc
doi:10.1128/JCM.00461-20 Natl Acad Sci U.S.A.. 2004;101(26):9804–9809.