Streptococcus pneumoniae Nasopharyngeal Carriage among PCV-10-Vaccinated HIV-1-Infected Children with Maintained Serological Memory in Ethiopia

Streptococcus pneumoniae (S. pneumoniae) vaccines have substantially reduced the burden of invasive pneumococcal diseases (IPDs) worldwide. Despite high coverage with S. pneumoniae vaccination, upper-respiratory-tract colonization by S. pneumoniae is still common. We assessed maintenance of serological responses to S. pneumoniae serotypes included in PCV-10 by ELISA in HIV-1-infected children (n = 50) and age-matched controls (n = 50) in Ethiopia. We isolated S. pneumoniae in nasopharyngeal swabs and determined S. pneumoniae serotype by whole genome sequencing (WGS). Comparable levels of S. pneumoniae serotype-specific IgG concentrations were detected in plasma of HIV-1-infected children and matched controls, with geometric mean concentrations (GMCs) consistently higher than the protective threshold for PCV-10 serotypes of 0.35 µg/mL. We isolated S. pneumoniae from 38 (out of 97) nasopharyngeal swabs, 25 from HIV-1-infected children and 13 from controls. WGS based serotyping revealed 22 known S. pneumoniae serotypes and 2 nontypeable (NT) isolates. Non-PCV-10 serotypes represented >90% of isolates. We showed that HIV-1-infected children and matched controls in Ethiopia carry a level of maintained serological memory to PCV-10 considered protective for IPDs. We identified a higher proportion of nasopharyngeal carriage with highly pathogenic S. pneumoniae non-PCV strains among HIV-1-infected children compared to controls.


Introduction
Streptococcus pneumoniae (S. pneumoniae), although part of the normal flora in the upper respiratory tract, can cause severe noninvasive (pneumonia, sinusitis, otitis) and invasive (bacteremia and meningitis) diseases [1,2]. Based on the variability of capsular antigens, more than 90 serotypes have been described [3]; among these are opportunistic pathogens which can cause different noninvasive and invasive pneumococcal diseases (IPD). Colonization of the upper respiratory tract is the initial step for all forms of pneumococcal disease, and the rate of colonization is higher at an early age [4]. Asymptomatic carriers can easily transmit the bacteria through respiratory droplets to healthy individuals [2]. Age, genetics, comorbidity, socio-economic status and seasons are some of the risk factors for pneumococcal Healthy individuals have been reported to exhibit higher S. pneumoniae serotype-specific IgG and IgM levels compared to HIV-1-infected individuals up to one year postvaccination with PPV-23 and PCV vaccines [29,30]. In contrast, other studies reported a comparable level of specific IgG production in both HIV-1-infected and -uninfected individuals after PCV vaccination, although with a significant difference in the effectiveness of the produced antibodies against IPD [31,32]. Nasopharyngeal carriage of nonvaccine serotypes is still a common phenomenon in PCV-vaccinated HIV-1-infected children in Africa [33]. In this study, we assessed the maintenance of humoral responses to PCV-10 in HIV-1-infected children and age-matched controls and studied nasopharyngeal carriage in these individuals by determining the serotype of S. pneumoniae in nasopharyngeal swabs with whole-genome sequencing (WGS).

S. pneumoniae Serotype-Specific Antibody Responses
Persistence of S. pneumoniae serotype-specific serological memory was detected in HIV-1-infected children and age-matched controls after a median time period of 4.9 years from PCV-10 vaccination; the clinical characteristics of patients and controls included in the study are presented in Table 1. A comparable level of serotype-specific IgG concentrations was detected in both HIV-1-infected children and age-matched controls ( Figure 1).
The IgG geometric mean concentration (GMC) to all serotypes incorporated in PCV-10 was shown to be consistently higher than the conventional threshold of 0.35 µg/mL, which is the protective threshold antibody level for IPD (Table 2), with the highest IgG GMC detected for serotype 19F (3.59 µg/mL in controls and 3.36 µg/mL in HIV-1-infected children) and the lowest for serotype 7F (0.63 µg/mL in controls) and serotype 4 (0.56 µg/mL in HIV-1-infected children). The IgG geometric mean concentration (GMC) to all serotypes incorporated in PCV-10 was shown to be consistently higher than the conventional threshold of 0.35 μg/mL, which is the protective threshold antibody level for IPD (Table 2), with the highest IgG GMC detected for serotype 19F (3.59 μg/mL in controls and 3.36 μg/mL in HIV-1-infected children) and the lowest for serotype 7F (0.63 μg/mL in controls) and serotype 4 (0.56 μg/mL in HIV-1-infected children). The percentage of individuals with antibody level ≥ 0.35 μg/mL lies between 75% (serotype 4) and 100% (serotype 1 and 19F; Table 2), with all other serotypes in between these two values. There was no statistically significant difference between the GMCs of the two studied groups and the percentage of individuals with ≥ 0.35μg/mL for the 10 serotypes. However, a lower percentage of  The percentage of individuals with antibody level ≥ 0.35 µg/mL lies between 75% (serotype 4) and 100% (serotype 1 and 19F; Table 2), with all other serotypes in between these two values. There was no statistically significant difference between the GMCs of the two studied groups and the percentage of individuals with ≥ 0.35µg/mL for the 10 serotypes. However, a lower percentage of individuals with ≥ 0.35 µg/mL was detected for serotype 7F (76%) in the controls and serotype 4 (68%) in the HIV-1-infected group compared to the other serotypes (> 80% of protected individuals) ( Table 2). A high proportion of individuals (58% HIV-1-infected and 60% controls) had protective antibody levels to all 10 serotypes, and 2% of children from each group showed a protective level of antibody only for 3 serotypes included in the vaccine (results not shown).
A significant positive correlation was observed between the age of the controls and serotype-specific IgG concentrations for serotype 14 (p < 0.01) and 18C (p < 0.05) ( Figure 2). In HIV-1-infected children, on the other hand, no significant correlation was found between age and serotype-specific IgG concentration, and no differences in GMCs were observed between viremic and aviremic children (data not shown).
Pathogens 2020, 9, 159 5 of 16 individuals with ≥ 0.35 μg/mL was detected for serotype 7F (76%) in the controls and serotype 4 (68%) in the HIV-1-infected group compared to the other serotypes (> 80% of protected individuals) ( Table  2). A high proportion of individuals (58% HIV-1-infected and 60% controls) had protective antibody levels to all 10 serotypes, and 2% of children from each group showed a protective level of antibody only for 3 serotypes included in the vaccine (results not shown). A significant positive correlation was observed between the age of the controls and serotypespecific IgG concentrations for serotype 14 (p < 0.01) and 18C (p < 0.05) ( Figure 2). In HIV-1-infected children, on the other hand, no significant correlation was found between age and serotype-specific IgG concentration, and no differences in GMCs were observed between viremic and aviremic children (data not shown).
A higher percentage of vaccine serotypes were detected in HIV-1-infected children compared to age-matched healthy controls ( Figure 3). PCV-10 serotypes are responsible for 8% of the carriage in HIV-1-infected children (which included a total of 25 isolates) and were not detected in the serotypes isolated (n = 13) from age-matched controls. PCV-13 serotypes 6A and 19A contributed 12% and 7.6% of pneumococcal carriage in HIV-1-infected children and controls, respectively. On the other hand, the carriage rate of vaccine serotypes incorporated in PPV23 is two times higher in controls (31%) than in HIV-1-infected children (16%). Despite the higher rate of PCVs serotypes carriage in HIV-1-infected children with history of PCV-10, 80% of carriage was caused by nonvaccine serotypes which are not incorporated in either PCV10 or PCV13.
Serotype 34 was the most common serotype isolated from the nasopharyngeal swabs (13.1%). Serotypes 15A and 15B each contribute to 7.8% of the nasopharyngeal carriage isolates followed by 6A, 6C, 19A, 20 and 23A (5.2% each), as shown in Figure 3. There was no statistically significant difference in the GMCs against S. pneumoniae serotypes between children with nasopharyngeal carriage and noncarriers for both groups except for serotype 14 in controls, where the noncarriers had a higher GMC than the carriers (p = 0.01). A higher percentage of vaccine serotypes were detected in HIV-1-infected children compared to age-matched healthy controls ( Figure 3). PCV-10 serotypes are responsible for 8% of the carriage in HIV-1-infected children (which included a total of 25 isolates) and were not detected in the serotypes isolated (n = 13) from age-matched controls. PCV-13 serotypes 6A and 19A contributed 12% and 7.6% of pneumococcal carriage in HIV-1-infected children and controls, respectively. On the other hand, the carriage rate of vaccine serotypes incorporated in PPV23 is two times higher in controls (31%) than in HIV-1-infected children (16%). Despite the higher rate of PCVs serotypes carriage in HIV-1infected children with history of PCV-10, 80% of carriage was caused by nonvaccine serotypes which are not incorporated in either PCV10 or PCV13.
Serotype 34 was the most common serotype isolated from the nasopharyngeal swabs (13.1%). Serotypes 15A and 15B each contribute to 7.8% of the nasopharyngeal carriage isolates followed by 6A, 6C, 19A, 20 and 23A (5.2% each), as shown in Figure 3. There was no statistically significant difference in the GMCs against S. pneumoniae serotypes between children with nasopharyngeal carriage and noncarriers for both groups except for serotype 14 in controls, where the noncarriers had a higher GMC than the carriers (p = 0.01).
Using multi-locus sequence typing (MLST) analysis (Table 3), which, through sequencing, identifies seven alleles in the S. pneumoniae sequences, 34 S. pneumoniae sequence types (ST) were identified; among these, 12 are novel types, suggesting that S. pneumoniae strains from Ethiopia may be underrepresented in the MLST database. ST11162 is the most frequent sequence type, followed by ST344 and ST156. In addition, various sequence types were also comprised in the same serotype: serotype 34 (ST11546, ST5934 and three novel STs), serotype 15A (ST991, ST2318 and one novel ST), serotype 23A (ST4168 and one novel ST) and serotype 20 (ST13704 and one novel ST). A graphical representation of the MLST findings is presented in Figure 4 showing a network tree where the phylogenetic relationship of the alleles characterized from the different S. pneumoniae isolates is shown. As shown in Figure 4, some novel STs are single-or double-loci variants, but for other STs, several loci variations were identified. Using multi-locus sequence typing (MLST) analysis (Table 3), which, through sequencing, identifies seven alleles in the S. pneumoniae sequences, 34 S. pneumoniae sequence types (ST) were identified; among these, 12 are novel types, suggesting that S. pneumoniae strains from Ethiopia may be underrepresented in the MLST database. ST11162 is the most frequent sequence type, followed by ST344 and ST156. In addition, various sequence types were also comprised in the same serotype: serotype 34 (ST11546, ST5934 and three novel STs), serotype 15A (ST991, ST2318 and one novel ST), serotype 23A (ST4168 and one novel ST) and serotype 20 (ST13704 and one novel ST). A graphical representation of the MLST findings is presented in Figure 4 showing a network tree where the phylogenetic relationship of the alleles characterized from the different S. pneumoniae isolates is shown. As shown in Figure 4, some novel STs are single-or double-loci variants, but for other STs, several loci variations were identified.

S. pneumoniae Serotype-Specific Antibody Responses in Relation to Nasopharyngeal Carriage
It has been previously suggested that protection from S. pneumoniae carriage should be evaluated considering different IgG thresholds for the different serotypes included in PCV-10 or PCV-13 [35]. We took into consideration these different thresholds and calculated the frequency of individuals who showed levels of protective antibody responses against carriage with the different serotypes. The results are presented in Table 4. The largest frequency of individuals showing protective levels was detected for serotypes 1 (94%) and 6B (93%) and the lowest for serotypes 4 (26%) and 7F (13%), and these values were comparable in both groups of healthy controls and HIV-1-infected children. These results should be taken into consideration when evaluating the complex picture of immunological interactions ensuring protection against S. pneumoniae carriage. It is interesting and puzzling that the two HIV-1-infected children displaying nasopharyngeal carriage for PCV-10 serotypes 7F and 23F had protective levels against the two serotypes substantially above (9.89 and 2.1 μg/mL, respectively) the suggested threshold value suggested for 7F and 23F serotypes in plasma.

S. pneumoniae Serotype-Specific Antibody Responses in Relation to Nasopharyngeal Carriage
It has been previously suggested that protection from S. pneumoniae carriage should be evaluated considering different IgG thresholds for the different serotypes included in PCV-10 or PCV-13 [35]. We took into consideration these different thresholds and calculated the frequency of individuals who showed levels of protective antibody responses against carriage with the different serotypes. The results are presented in Table 4. The largest frequency of individuals showing protective levels was detected for serotypes 1 (94%) and 6B (93%) and the lowest for serotypes 4 (26%) and 7F (13%), and these values were comparable in both groups of healthy controls and HIV-1-infected children. These results should be taken into consideration when evaluating the complex picture of immunological interactions ensuring protection against S. pneumoniae carriage. It is interesting and puzzling that the two HIV-1-infected children displaying nasopharyngeal carriage for PCV-10 serotypes 7F and 23F had protective levels against the two serotypes substantially above (9.89 and 2.1 µg/mL, respectively) the suggested threshold value suggested for 7F and 23F serotypes in plasma.

Discussion
This study is the first study to report, in Ethiopia, the maintenance of S. pneumoniae serotype-specific antibody responses in ART-treated HIV-1-infected children and age-matched controls previously vaccinated with PCV-10. The children's median age was 4.9 years, and they had all received a full dose of PCV-10 at 6, 10 and 14 weeks of age. In the study, similar percentages of HIV-1-infected children and age-matched controls were found to carry antibody concentrations considered to be protective against IPD (≥0.35 µg/mL) for all serotypes included in PCV-10. A study conducted by Madhi and collaborators [36] reported a similar result among HIV-1-infected and noninfected children who had received PCV10 (three doses plus one booster) with samples collected 14 months after the booster. Furthermore, the persistence of serotype-specific IgG responses in non-HIV-1-infected children reported by other studies after 3, 4 and 5 years of completed PCV-10 vaccination is comparable with our results [37][38][39]. These findings confirm the ability of PCV-10 to induce a long-lasting serotype-specific IgG response. The maintenance of high protective levels of S. pneumoniae antibodies in plasma after two to six years of vaccination may, on the other hand, also reflect natural exposure to S. pneumoniae serotypes included in PCV10, which may reinvigorate immunological memory elicited during vaccination.
It has previously been reported that, in the current era of PCVs, there is a replacement of vaccine by nonvaccine S. pneumoniae serotypes, with the latter isolated from nasopharyngeal carriage and IPD patients, particularly in individuals with underlying conditions including HIV-1 infection [20,21,28,40]. In this study, nasopharyngeal carriage was detected in 39.1% of children who had received the full dose of PCV-10, with over 90% of the carriage caused by nonvaccine serotypes not included in PCV-10. The carriage rate in the vaccinated children included in the present study is higher than that previously reported in Ethiopia (21.5%) [20].
In contrast to the comparable amount of specific plasma IgG levels to the serotypes included in PCV-10, we detected that the carriage rate in HIV-1-infected children (52%) was twice the carriage rate in age-matched controls (26.5%); carriage of vaccine serotypes 7F and 23F was only detected in the HIV-1-infected group. This result indicates the need for immunological studies assessing the mechanisms of immune responses against pneumococcal nasopharyngeal carriage in PCV-10-vaccinated individuals, particularly in HIV-1-infected children, in areas with a high burden of pneumococcal diseases. Apart from 7F and 23F, nasopharyngeal carriage by other serotypes included in PCV-10, serotypes 1, 4, 14 and 19F, was previously reported in Ethiopia [20,21]. PCV13 serotypes 6A and 19A each contributed 5.2% of pneumococcal carriage in our study; this result is significantly lower than what previously shown by Negash et al. [20] for 19A (27%) but comparable to what was shown for 6A (4.9%) in children with CAP after 5 years of PCV-10 introduction. All S. pneumoniae serotypes reported in this study, except 28A, were previously reported in nasopharyngeal carriers in Ethiopia [20,21]. The MLST analyses, which characterize microorganisms by their allelic profiles, suggested that 12 of the 38 S. pneumoniae isolates in our study carry novel combinations of alleles; a large scale study should be performed in Ethiopia to further characterize the genomic profile of S. pneumoniae and the novelty degree of the alleles identified by us. Most carriage serotypes isolated and characterized in our study have previously been associated with complications of S. pneumoniae infection, from otitis media to IPD cases, including pneumococcal meningitis, in different parts of the world [40][41][42][43]. Therefore, our findings on the carriage in HIV-1-infected children are worrisome as these individuals, despite ART treatment and PCV-10 vaccination, remain at high risk for S. pneumoniae-related complications.
The immunological mechanisms responsible for protection against S. pneumoniae carriage remain elusive, and arguments have previously been presented for innate and adaptive immune responses playing a role in this context. MZ B cells, which are found in the spleen MZ [44] and to some extent also in blood, play an important role in T-cell-independent humoral immunity to encapsulated bacteria rich in polysaccharides antigens, including S. pneumoniae [45]. MZ B cells are fully developed in humans around two years of age [46]. A decline in frequency of mature MZ B cells has been reported in the blood of HIV-1-infected children, probably due to impaired migratory capacity of these cells [47]. HIV-1 infection leads to the dysregulation of several B cell subpopulations, including memory B cells [48,49], which, together with MZ B cells, are associated with maintenance of serological memory to PCV vaccination. Inflammation and impaired control of human innate immune responses have recently emerged as important pathogenic players in establishment of S. pneumoniae infection and associated clinical conditions [50]; respiratory viruses may play a negative role in carriage establishment by leading to increased inflammation.
It is still unclear whether protection from nasopharyngeal colonization by vaccine serotypes is due to IgG and/or IgA produced locally or by the leakage of circulating IgG into the mucosa. Despite the observation that PCVs induce comparable amounts of serotype-specific S. pneumoniae antibodies in HIV-1-infected and noninfected individuals, the functional capacity of these antibodies to neutralize S. pneumoniae, as measured by opsonophagocytic assay, was lower in the HIV-1-infected group [51,52]. According to an analysis conducted by Voysey M. et al. [35], the antibody concentration required for protection against pneumococcal carriage varies for each serotype, and a higher antibody concentration is needed for carriage protection in developing countries. In our study, there was no significant difference in the S. pneumoniae serotype-specific antibody concentration between HIV-1-infected and control children using the reference reported by Voysey M. et al. [35]. Voysey M. [35] also suggested that pneumococcal antibody thresholds of protection may be different in low-and high-income countries [35]; further studies conducted in African countries may, however, be needed to create reference values for protection against pneumococcal carriage in this continent. The balance between Th17 and Treg cells in nasopharynx-associated lymphoid tissue has been reported to play a role for clearance of S. pneumoniae in this compartment; Th17 cell frequencies increase with age [53]. The lack of knowledge on the CD4+ T cell counts is a limitation of the present study. Longitudinal studies in HIV-1-infected and noninfected children, regularly assessing memory and MZ B cells before and after PCV-10 vaccination in parallel with determination of T cells and antibody responses to S. pneumoniae serotypes and carriage, may lead to significant information on the relation between cellular and serological components relevant to protection from S. pneumoniae infection and carriage.

Study Population
A cross-sectional study was conducted between May 2018 and March 2019, which included 50 ART-treated HIV-1-infected children (median age in months and range 53.5 (24-89)) and 50 age-matched controls (age and range 60 (43-84)) who had received 3 primary doses of 10-valent pneumococcal vaccine (PCV-10) according to the Ethiopian childhood immunization program. Vaccination history was confirmed using a vaccination history record. All HIV-1-infected children were recruited from ALERT Hospital, Zewditu Memorial Hospital and Yekatit 12 Hospital ART clinics and control subjects from Woreda 03 Health Center, Addis Ababa.
HIV-1 RNA copies were quantified in plasma of HIV-1-infected children using an automated m2000sp Abbott Real-Time HIV-1 assay system following the manufacturer's protocol (Abbott Laboratories, Abbott Park, IL, USA) ( Table 1). The lower detection limit of this assay was 40 copies/mL. Viremia was detected in 40% of HIV-1-infected children with a median viral load of 42,095 copies/mL (range 266-1,882,592). There was no correlation between the viremic status, age and length of treatment.
Blood samples were collected from all study participants; plasma was isolated and stored at −80 • C. Nasopharyngeal swab specimens (n = 97) were obtained using a flexible calcium alginate-tipped swab which was stored in skim milk-tryptone-glucose glycerin medium (STGG) at −80 • C. Samples were transported frozen to Karolinska Institutet for further analyses.

Measurement of S. pneumoniae Serotype Specific IgG in Plasma
The plasma concentration of IgG specific for the 10 individual S. pneumoniae serotypes included in PCV-10 was measured by ELISA using purified pneumococcal polysaccharide antigens donated from SSI Diagnostica (Hillerød, Denmark) and 007SP as a standard reference serum (gift from Dr. Mustafa Akkoyunlu, FDA, USA). The WHO guideline for pneumococcal ELISA was followed with few modifications [54]. Briefly, ELISA plates were coated using 5 µg/mL of purified pneumococcal polysaccharide. The coated plates were incubated for 5 h at 37 • C and stored overnight at +4 • C. Thereafter, they were washed using 0.05% tween in PBS and blocked using 1% BSA. To avoid nonspecific binding, standard reference serum and serum samples were adsorbed using an adsorption solution which contains 10 µg/mL of pneumococcal cell wall polysaccharide and S. pneumoniae serotype 22F for 1 h at 37 • C. Following adsorption, 100 µL of standard and serum samples were added to the plates in duplicate. After 2 h of incubation, 100 µL of anti-human IgG secondary antibody conjugated with alkaline phosphatase were added, followed by substrate (p-nitrophenyl phosphate) and 3M NaOH stopping solution after 1 h incubation at room temperature. All samples and standards were tested in duplicate. Interpolation of concentration for the samples was done in GraphPad Prism 8. Children with IgG level of ≥ 0.35 µg/mL are considered to have IPD-protective antibody levels [19].

Identification of S. pneumoniae from Nasopharynx Swabs and DNA Isolation
Culture-based identification of S. pneumoniae from nasopharyngeal swabs was conducted using the 2016 CDC guidelines [55]. Each swab was plated into trypticase soy agar (TSA) plates supplemented with 5% sheep blood containing 5 µg/mL gentamicin. Identification of S. pneumoniae was performed based on colony morphology and Optochin susceptibility. Pure α-hemolytic optochin susceptible S. pneumoniae isolates were stored at −80 • C in STGG. QIAamp®BiOstic®Bacteremia DNA Kit (Qiagen, Hilden, Germany) was used for DNA isolation.

Whole Genome Sequence-Based S. pneumoniae Serotyping
Whole-genome sequencing was conducted at Scilife Laboratory (Stockholm, Sweden). Genomic DNA was quantified using Quant-iT TM dsDNA High-Sensitivity Assay Kit and FLUOstar Omega plate reader. Samples were normalized to 2 ng/µL; 1 µL of normalized genomic DNA was used in the tagmentation reaction using Nextera chemistry (Illumina, San Diego, CA, USA) to yield fragments of approximately 410 bp. Amplification of tagmented library was carried out using 11 cycles of PCR with Unique Dual Index (UDI) (IDT Technologies, Coralville, IA, USA). The resulting PCR products were purified using Sera-Mag beads (GE Healthcare Life Sciences, Uppsala Sweden). The obtained library was quantified using Quant-iT TM dsDNA High-Sensitivity Assay Kit and FLUOstar Omega plate reader. Library was pair-end (2x151 bp) sequenced to minimum 3 M read pairs on NovaSeq 6000 (Illumina, San Diego, CA, USA). Base-calling and demultiplexing was done using bcl2fastq v2.20.0.422, allowing one mismatch in the index sequence.
The serotypes of the strains were predicted using SeroBA v. 1.0.1 [34]. The k-mer counts database was built through SeroBA using the recommended setting for the kmer size (71 bp). Default settings were utilized for the analysis except that, in order to include all serotypes in the ARIBA-compatible database, the maximum allowed length in nucleotides of non-coding sequences was set to 25,000 bp.
The results of the MLST and serotype analyses are visualized in a minimum spanning tree using the BioNumerics software (Version 7.6, Applied Maths NV, Sint-Martens-Latem, Belgium).

Statistical Analysis
Demographic and clinical data were analyzed using Graph pad prism 8 (La Jolla, CA, USA). The normal distribution of ELISA data was assessed using Kolmogorov-Smirnov test. Unpaired T-test was used to assess the difference between groups. Fisher exact test was used to compare frequency between groups. Spearman correlation was applied to determine the relation of variables. A p-value < 0.05 was considered statistically significant.

Ethical Statement
Ethical clearance to conduct the current study was obtained from the ethical clearance committee of ALERT/AHRI ethics review committee (protocol number P0/06/17) and Ethiopian national ethical review committee (NERC; reference n. 3.10/71/2018). Informed written consent was collected from parents or legal guardians of children in the study. The study was conducted in accordance with the Declaration of Helsinki. The ethical committee at Karolinska Institutet approved the laboratory studies of the collected specimens (approval n 2016/485-32).