Abstract

Introduction

Early in the pandemic, male sex was identified as a risk factor for hospitalization and mortality after SARS-CoV-2 infection (1). While behavioral and socio-economic factors are implied, immunological differences between the sexes may be more important in this disparity. Females typically show a stronger immune response than males and possible factors that could explain this phenomenon are X chromosomal genes, immunomodulatory functions of sex hormones and sex-dependent expression of susceptibility genes (2, 3). Females have better clinical outcomes than males following sepsis and infection (4–7). For instance, males suffer from more severe and intense disease after hepatitis B virus (HBV) or Epstein Barr virus (EBV) infection (7). The immune response differs between males and females in various ways, ranging from innate recognition to downstream adaptive immunity, resulting in distinctly different cytokine responses to infection. Induction of pattern recognition receptors (PRRs), subsequent interferon (IFN) and antibody production is higher in females than males. Sex hormones display immunomodulatory activity at various levels, and immune response to viruses can vary with natural hormonal fluctuations in females. Novel research suggests also an important role for sex chromosomal gene that can modulate differences in the immune response or cause differential expression of disease susceptibility genes such as class I and II MHC glycoproteins (3).

On the other hand, autoimmune diseases are predominantly present in females (4–6, 8). For instance, the incidence of systemic lupus erythematosus (SLE) is almost nine-fold higher in females and male Klinefelter patients (47, XXY), but rarely develops in female Turner patients (45, X0) (9), indicating the importance of the X chromosome in SLE and immunity. The X chromosome is home to many genes directly or indirectly related to immunity and variability in expression of X chromosomal genes is often associated with sex-related immune disparities

The sex-bias observed in the COVID-19 pandemic fits this pattern. An X chromosomal gene of interest is TLR7, identified as a pattern recognition receptor (PRR), recognizing ssRNA viruses such as SARS-CoV-2. It is one of the PRRs involved in type 1 interferon (IFN) production in COVID-19 (10, 11). In the present article, we hypothesize that contribution of beneficial gene expression of the second female X chromosome could partially explain the sex-skewed ICU admission. In particular, this hypothesis and theory article focuses on TLR7 expression and activation and the consequences for males and females in COVID-19.

Subsections

COVID-19 and TLR7

Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) is the single-stranded RNA (ssRNA) virus that causes COVID-19. The spike protein on the viral envelope binds to the angiotensin-converting enzyme receptor 2 (ACE2) on the host cell. ACE2 has a systemic function in lowering blood pressure and is expressed in most organs, but abundantly in lung respiratory epithelium. The spike protein is activated by the serine protease 2 transmembrane protein (TMPRSS2) on the host cell, allowing for viral particles’ internalization. The viral RNA is subsequently released into the host cytoplasm, from where it can move to the host ribosome for production of new virions to infect other cells. Alternatively, it can enter the endosome (2, 12). Various cellular sensors can recognize SARS-CoV-2 ssRNA or other intermediates of viral replication and induce downstream signaling to activate the innate immune response. At the level of the lung epithelium, these sensors are cytoplasmic RIG-I and MDA5 (13, 14). The inflammatory mediators produced by the epithelium are picked up and further amplified by PRRs in innate immune cells (15), such as TLR7 and TLR8.

Toll-Like Receptor 7 (TLR7), and also TLR8, is a PRR encoded on the X chromosome. TLR7 is expressed at the endosomal membrane of plasmacytoid dendritic cells (pDCs) and B cells and to a lesser extent in monocytes and macrophages (16). TLR7 can recognize ssRNA derived from viruses and bacteria in humans, and can also respond to RNA-associated autoantigens (17). Following detection of ssRNA, TLR7 activates the MyD88-dependent pathway, activating both the nuclear factor kappa Beta (NFκB) and the type 1 interferon (IFN-α and IFN-β) pathway in pDCs (18). This process is dependent on the gene product of CXorf21, an X chromosomal gene encoding for the TLR adaptor interacting with endolysosomal SLC15A4 (TASL) protein which stimulates nuclear migration of interferon regulatory factor 5 (IRF5) (19). IFN-α transcription in pDCs is further dependent on the nuclear migration of IRF7 (18), as shown in Figure 1 . pDCs are the most potent IFN producers and while other cells can also produce IFNs, pDCs function as the primary type 1 IFN producers during viral infections (20). In macrophages, TLR7 stimulation can lead to production of pro-inflammatory cytokines (21).

Open in a separate window

Figure 1

Schematic representation of the effect of sex and X chromosome inactivation escape on the TLR7 expression. TLR7 gene (red) on the inactivated X chromosome (Xi) wrapped with Xist (blue) escapes inactivation. Escape from distal X chromosome inactivation in female pDCs may result in more TLR7 mRNA and protein compared male cells, resulting in higher production of type 1 IFNs.

The most important type 1 IFNs are IFN-α and IFN-β. The production of type 1 IFN is essential for the antiviral response, especially in the early stages of infection. Type 1 IFNs have essential functions: activating an antiviral state in infected cells to limit the viral spread, restraining innate immunity, controlling pro-inflammatory pathways, and activating adaptive immunity. They exhibit both direct and indirect antiviral activity. IFNs bind the interferon-alpha/beta receptor (IFNAR) at infected cells, which initiates the JAK-STAT signaling pathway and the following cascade destroys both host and viral RNA. Viral replication and production are directly restricted by the induction of IFN-stimulated genes (ISGs) (22–24).

TLR7 has been implicated as PRR in SARS-CoV-1 and Middle East Respiratory Syndrome Coronavirus (MERS-CoV) infections (25). These related viruses possess a high number of binding motifs for TLR7 and bioinformatic analysis showed that the SARS-CoV-2 genome has even more ssRNA motifs that can potentially interact with TLR7 than its family member SARS-CoV-1 (23). Early in the COVID-19 pandemic, the importance of TLR7 after SARS-CoV-2 infection was first demonstrated in a clinical setting by van der Made et al. (10). After four young males (two brother pairs from unrelated families) without medical history were admitted to the intensive care unit (ICU), a genetic cause was suspected. In all four patients, different TLR7 deleterious variants and concurrent low levels of type 1 IFN were identified. Loss of function was confirmed when experiments with primary PBMCs showed no upregulation of TLR7 and a failure to induce IFN-inducible genes (ISGs) after stimulation with a TLR7 agonist (10). Later, more such cases of deleterious TLR7 variants in severe male patients have been described (11, 26, 27), suggesting that a loss-of-function of TLR7 could explain some cases of severe COVID-19 which result in functional defects of type 1 IFN. Likewise, single nucleotide polymorphisms (SNPs) could alter TLR7’s effectivity.

As pulmonary epithelial cells do not express TLR7, they alone are not sufficient for the defense against SARS-CoV-2; TLR7 deficiency was pathogenic in patients by impairing the production of large amounts of type 1 IFNs by pDCs (11). After infecting lung epithelial cells, innate phagocytic cells such as pDCs are recruited and can phagocytose the infected cells and produce type 1 IFNs after PRR stimulation. The clinical observations of deleterious TLR7 mutations and a poor type 1 IFN response identify TLR7 as an important PRR in the immune response against COVID-19.

Since TLR7 is located on the X chromosome, mutations in TLR7 will affect males more than females, who bear two X chromosomes per cell. TLR7 deficiency has been reported as a genetic mediator for severe COVID-19 especially in younger males, with percentages of 1-2% found across cohorts (10, 11, 26, 27). Therefore, genetic screening for TLR7 primary immunodeficiency was recommended in young males with severe COVID-19 in the absence of other relevant risk factors (11, 26, 27). Despite the observed prevalence in young males, the penetrance of deleterious TLR7 mutations may be worse in older patients, as both the pDC number and functional IFN secretion decreases with age, which was associated with a reduced number of pDCs expressing TLR7 (28).

Evidence for XCI Escape of TLR7

TLR7 escapes XCI in different female immune cells: pDCs (20, 44), monocytes and B cells (44). A study by Souyris et al. in 2018 was the first to demonstrate the XCI escape of TLR7 in primary human cells. Souyris et al. confirmed biallelic mRNA expression of TLR7 in 30% of pDCs, monocytes, and B cells from healthy females and male Klinefelter Syndrome (KS) patients (47, XXY). TLR7 transcription from both X chromosomes was observed using RNA FISH in B cells, confirming XCI escape. The relative abundance of TLR7 mRNA transcripts in female biallelic B cells was up to 50% higher than male monoallelic B cells. Western blot testing confirmed this at the protein level: a 1.38- and 1.31-fold increase in TLR7 protein was found in biallelic peripheral blood mononuclear cells (PBMCs) compared to female monoallelic PBMCs and male cells, respectively. Additionally, female B cells differentiated more efficiently into plasmablasts than male B cells upon TLR7 stimulation. None of these effects were observed upon TLR9 stimulation.

Souyris et al. (44) hypothesized that the increased TLR7 dosage would result in increased TLR7-induced type 1 IFN levels, which was confirmed by Hagen et al. (20). This study confirmed 30% female pDCs with biallelic expression and a similar 50% higher abundance of TLR7 mRNA transcripts. Moreover, Hagen et al. observed that biallelic pDCs showed significantly higher type 1 IFN mRNA and protein in the first 2 hours after TLR7 stimulation with a TLR7 agonist (2020). As visualized in Figure 2 , these results confirmed that human female pDCs produce more type 1 IFN after TLR7 stimulation, showing that TLR7 XCI escape is correlated with the antiviral type 1 IFN response.

Open in a separate window

Figure 2

Schematic representation of a plasmacytoid dendritic cell with viral ssRNA in the endosome. This is recognized by TLR7 on the endosomal membrane. Via the MYD88 pathway, ‘TLR adaptor interacting with SLC15A4 on the lysosome’ (TASL) is activated which is needed for recruitment of Interferon Regulatory Factor 5 (IRF5), a transcription factor involved in transcription of NF-kβ and type 1 interferons. This results in further transcription of IL-1, IL-6 and TNF-α for NF-kβ and interferon stimulated genes for type 1 interferons, respectively.

Before these two publications, it was already established that female pDCs produced more type 1 IFN after TLR7 stimulation and expressed more interferon alpha/beta receptors (IFNAR) than male pDCs (45–48), especially in females after puberty (49). Similarly, Berghöfer et al. found higher induction of IFN-α in female PBMCs stimulated with TLR7 agonists and not by TLR9 agonists, confirming a skewed and beneficial TLR7 response which could be explained by XCI escape. However, this could not be confirmed in immortalized B-cell lines (46). Others did not investigate XCI escape as a cause of their observations and sought other explanations. Estrogens were suggested as a possible cause of increased IFN signaling in females, as in vitro blocking of estrogen receptor (ER) signaling inhibited TLR7-mediated type 1 IFN production by pDCs. However, when female pDCs were transplanted into male mice, an enhanced TLR7-mediated type 1 IFN production was still seen, suggesting that female sex hormones and the number of X chromosomes independently contributed to the enhanced type 1 IFN production (48). Webb et al., who compared genetic deviations with more than one X chromosome, also showed that increased IFN production was dependent on the number of X chromosomes present in the cell (49). A study by Sarmiento et al. found slightly contradictory results: they observed that stimulated female and Klinefelter syndrome (KS) PBMCs had higher TLR7 and IFN-α mRNA levels than male and Turner’s syndrome female PBMCs (45, X0), supporting TLR7 XCI escape, but the KS cells did not show higher IFN-α levels after TLR7 stimulation. In comparison, the female cells showed higher levels compared to the male and TS cells: in the case of the KS cells with two X chromosomes, their higher TLR7 expression by XCI escape did not result in higher IFN, while it did in the female cells with two X chromosomes. Accordingly, the increased IFN production was not dependent on the number of X chromosomes (50), opposite to what others observed (49). One of the explanations provided by Sarmiento et al. is that male KS cells were used, which cannot rule out epigenetic modification of TLR7 by testosterone (50). Altogether, these results strongly suggest that increased type 1 IFN secretion after TLR7 stimulation in female pDCs is linked to biallelic TLR7 expression by XCI escape.

Discussion

COVID-19 is a disease where the immune system is a double-edged sword, as visible in the different clinical stages. In severe patients, IFN levels are low early on and symptoms are moderate but later, IFN levels rise, and a cytokine storm subsequently occurs. Males are significantly more likely than females to experience severe disease and TLR7 XCI escape in females may be partly responsible for this. TLR7 has been observed to escape XCI in 30% of female immune cells, resulting in biallelic TLR7 expression and increased IFN production compared to males. Severe COVID-19 disease has been associated with a delayed IFN response and increased TLR7-mediated IFN production in females may prevent females from progressing to severe disease.

This sex-determined gene-dose effect may have important implications for managing the COVID-19 pandemic. While sex-based immunological differences are not new (2), sex as a biological variable is often neglected in clinical and especially pre-clinical research, especially in inflammatory diseases (107, 108). Unfortunately, COVID-19 research is no exception. A recent preprint established that the vast majority of many registered clinical trials for COVID-19 do not mention sex or gender as a recruitment criterium nor bring up sex or gender in the description of the analysis phase (109). Fortunately, two major vaccine trials (Pfizer, Moderna) did include sex-disaggregated primary outcome data. However, further analysis of sex-disaggregated adverse events and secondary outcomes was lacking, which could have set a reporting benchmark for future medical intervention reporting (108). Sex-disaggregated data are not provided by all countries and clinical reports about infection or mortality rate are often not sex-disaggregated, further complicating sex-specific analyses (110, 111). The Global Health 50/50 research initiative and the International Center for Research on Women have started the Sex, Gender and COVID-19 Project which tracks sex-disaggregated data and aims to further attract attention for this issue (111).

In various articles calling for investigation of TLR7 agonist or IFN administration early in COVID-19, sex is also not named as a factor to be taken into account (22, 24, 60, 78, 98, 99). This trend is especially surprising considering the gendered hospitalization and mortality rates of COVID-19 and many have called to reverse this (1, 108, 110, 112, 113). While several articles noted the possible sex differences in immune response to COVID-19, only few clinical and experimental studies have been conducted to investigate this further (70). While it is no exception, it is unfortunate to see this, especially in the light of a possible role for higher expression of TLR7 and IFN in females, probably allowing for sex-specific treatments. In treatment based on modulating TLR7 or administrating IFNs, the sex-determined TLR7 gene-dose effect by XCI escape may be especially important considering the importance of proper timing in the different disease phases. While many questions remain about sex differences in the TLR7-mediated IFN response, it is well-established that these differences exist. Not taking these into account in clinical studies is counter beneficial for exploring these differences. For example, it could be studied whether males need a higher dose of TLR7 agonist or antagonist, or IFN administration than females. Another question is the possible difference in TLR7 XCI status between ICU admitted and non-ICU admitted females.

Thus far, two recent studies have shown proof of XCI escape of TLR7 (20, 44) or examined its effect on IFN expression (20). Replications of these results would further support these findings, in conjunction with previously reported increased IFN response of female immune cells to TLR7 stimulation (45, 47, 48). In these studies, it seems likely that some degree of TLR7 XCI escape may play a role. However, little is known about the stability and duration of XCI and XCI escape: TLR7 is no exception. Insights in the mechanisms of Xist spreading and Xist maintenance may provide more answers into stability, modulation and timeline of XCI escape (50). A recent study demonstrated that Xist is continually required in adult human B cells to regulate several X-linked immune genes including TLR7 (81). Single-cell transcriptome data of female patients with either systemic lupus erythematosus (SLE) or COVID-19 revealed Xist dysregulation and overexpression of TLR7 in PBMCs and atypical B cells. These results suggest that Xist RNA maintenance is crucial for TLR7 silencing, and that Xist regulation may be subject to change in disease states such as COVID-19 or SLE and that Xist maintenance may alter TLR7 expression from the X chromosomes (81). Recently, methods for manipulating Xist have been developed for mouse pre-implantation embryos (114). However, there is still much to be learned about Xist maintenance and TLR7 transcription regulation alike, which could open up possibilities for further investigation of XCI escape and possible manipulation.

Other mechanisms than XCI escape may increase the protein dosage. Both TLR7 and TLR8 expression are under the influence of single nucleotide polymorphisms (SNPs) which can impact expression. For instance, the TLR7 SNP rs179008 is carried by approximately 30% of females of European descent and lowers TLR7 protein dosage in pDCs via controlling TLR7 mRNA translation. Affected pDCs from both pre and postmenopausal females showed an impaired TLR7-mediated IFN production, but affected male pDCs did not (115, 116). The TLR8 SNP rs3764880 influences expression of TLR8 isoforms in human monocytes and increases the amount of TLR8 protein (21). Additionally, genomic copy number variation (CNV) can also control TLR7 protein dosage, and this CNV has been associated to SLE incidence (117, 118). Thus, the amount of protein dosage is not only under control of epigenetic mechanisms such as XCI escape, but also by CNV and SNPs.

Other X-chromosomal genes may also be implicated in COVID-19, specifically Toll-Like Receptor 8 (TLR8) and TLR adaptor interacting with SLC15A4 on the lysosome (TASL). TLR8 is in many ways related to TLR7: both are encoded on the X chromosome in close proximity to each other. Both are activated by ssRNA and are complementary to each other in the detection of specific ssRNA motifs. TLR8 is absent in pDCs and B cells, but is expressed in myeloid cells such as monocytes, macrophages and neutrophils, where it can drive IFN-β production dependent on IRF5 (21, 119–121). Compared to TLR7, TLR8 is studied to a lesser extent (119, 122). TLR8 has been suggested to escape XCI in mice but this has not been studied yet in humans (123). In SLE, TLR7 dosage as well as TLR8 dosage are important determinants for the observed sex differences (19, 116, 120).

While SARS-CoV-2 ssRNA can trigger a TLR8-dependent inflammatory response from macrophages in vitro (124), it was suggested by others that TLR8 is of lower clinical importance in COVID-19, due to its lack of expression on pDCs and that no deleterious TLR8 variants were found in severe patients (11). TASL interacts with lysosomal SLC15A4 to activate IRF5 after TLR7 activation and is an interferon stimulated gene (ISG) (116). Similar to TLR7, both TLR8 and TASL are involved in the etiology of systemic lupus erythematosus (SLE), a female-biased autoimmune disease (19, 116, 120). TASL has not been associated with COVID-19 directly, but indirectly via its adaptor function in the TLR7-mediated IFN pathway. Despite its location on the X-chromosome, XCI escape of TASL has not been studied, so it is unclear if there are sex-based dosage differences of TASL.

This article proposed a hypothesis that XCI escape of TLR7 and the resulting gene-dose effect in females is beneficial in COVID-19. Unfortunately, it will prove difficult to establish a causal link between this TLR7 gene-dose effect and increased hospitalization and mortality in males. It seems important to first establish interindividual variation in biallelic expression between mild and severe COVID-19 female patients. Recently, software has been developed to include X chromosomal associations, such as XCI escape, into GWAS analyses (30), but this cannot associate gene dosage, only genetic variation such as SNPs. However, it could be a way forward, for instance in identifying the genetic make-up of female patients who experience severe COVID-19. Various methods are available to study XCI escape (43, 125). RNA sequencing (RNAseq) and expression data to identify XCI escape relies on good methods to differentiate alleles, usually SNP frequencies to distinguish alleles at multiple loci. Single-cell hybridization methods such as RNA fluorescence in situ hybridization (RNA-FISH) used by Souyris et al. (44) do not rely on SNPs and can be used to identify cell types within a tissue. Bulk RNAseq methods could be used for larger-scale investigations into quantitative differences in TLR7 expression between males and females, between males and females of various ages, interindividual variability between females and tissue variability.

Age is another risk factor for severe COVID-19. Ageing differentially affects the male and female immune system, with a male trend towards accelerated immune ageing (1, 126). Ageing might also have an effect on XCI escape, which could be established in bulk RNAseq studies. Especially the role of estrogens may be important here, considering that females after menopause experience a significant drop in estrogen levels. Estrogens are known to modulate the female immune response and protect against infection severity via upregulating both the innate and adaptive immune response (2). Estrogens and to a lesser extent progesterone can enhance the TLR7-mediated IFN production by pDCs (45, 127) and blocking estrogen receptor (ER) signaling inhibits this IFN production (48, 127). Treating SARS-CoV-1 infected female mice with ovariectomy or an ER antagonist significantly increased mortality, indicating a protective effect of ER signaling in mice (52). Possibly, this protective effect was mediated by TLR7-mediated IFN signaling, but this was not studied. If ER signaling indeed enhances TLR7 signaling, the menopausal estrogen drop could mean that the protective effect of increased TLR7 signaling in females would be partly lost and the female sex bias in COVID-19 would be diminished.

TLR7 is also expressed in stimulated B cells, where it is involved in B cell receptor activation, antibody production and recognizing foreign nucleic acids (128). TLR7 XCI escape was also demonstrated in B cells (44)and increased B cell stimulation and its effects may also be relevant in COVID-19.

Besides the TLR7 gene-dose effect caused by XCI escape, other biological differences between the sexes may influence the sex bias observed in COVID-19. Other factors are differential ACE2 expression, generally increased T cell activity and immunoglobulin production by B cells in females, other X-encoded immune genes, and activity of sex hormones (1, 51).

Interestingly, ACE2, the entry receptor of SARS-CoV-2, is encoded on the short arm of the X chromosome in sites that are known to escape XCI. Despite this localization, XCI escape of ACE2 has not been experimentally demonstrated in the way that TLR7 has been with RNA fluorescence in situ hybridization (RNA FISH). Male-female expression differences could be used as an indirect proxy measurement for XCI status. Data from a genotype-tissue expression study shows a tissue-specific heterogenous expression pattern which is male-biased in most tissues (40) while XCI escape would show a female-biased expression pattern. However, there are indications that ACE2 expression does not necessarily correlate to ACE2 enzyme activity due to regulation by sex hormones. Therefore, it is necessary to study at the transcriptional, translational and post-translational level (40, 62).

Unexpectedly, ACE2 was recently identified as an ISG (interferon-inducible gene) in respiratory epithelial cells (129). The antiviral type 1 IFN response may allow for viral entry in neighboring cells by upregulating ACE2 expression as SARS-CoV-2 exploits the ACE2-mediated tissue-protective IFN response to gain cellular entry. This raises questions about IFN treatment, because of the possibility of increased cellular entry by increased ACE2 levels. However, patients receiving angiotensin II receptor blockers and ACE inhibitors, widely used antihypertensives that increase ACE2 expression, showed a lower COVID-19 mortality rate compared to patients not receiving these antihypertensives (130). These results imply that ACE2 is important but not sufficient for viral entry, and that other co-receptors such as serine protease 2 transmembrane protein (TMPRSS2) may also be needed. Of interest, IFN-β increases ACE2 expression, while IFNα does not, suggesting that the type 1 IFN subtypes may have distinct functions (130).

Various other immune genes are encoded on the X chromosome and mutations in X-linked immune genes are known to affect males more than females. For instance, a mutation in interleukin-2 receptor subunit gamma (IL2RG) causes severe combined immunodeficiency (SCID). Males with SCID have impaired cellular and humoral immunity, resulting in increased susceptibility for infections, with a symptom onset before the age of one (131). Likewise, other X-linked immune genes may add to the observed sex bias in COVID-19, possibly by escape of XCI. IL-1 receptor-associated kinase-1 (IRAK-1) is the most studied gene regarding the sex bias in inflammatory responses. It escapes XCI, favoring an increased NFκB-dependent gene transcription response in females (132). CD40L and CXCR3, which have an important role in the T cell response, are both encoded at the X chromosome and escape XCI. It is unknown if the sex-differential role of T cells has a role in COVID-19 (51).

Additionally, The X chromosome encodes for 10% of human microRNAs (miRNAs), many of which are associated with modulation of the immune response. Dysregulated expression of miRNAs has been associated with various inflammatory diseases and may also contribute to the sex differences in COVID-19 (54, 132) but there is yet too little evidence to attribute specific sex-typical immune responses to X-linked miRNAs (2).

To conclude, there are clear indications for an X chromosomal TLR7 gene-dose effect in COVID-19. As sex-determined variables are currently understudied, and the effects of other X-related immune genes and TLR7 modulators are unclear, future research should consider to shed more light on the sex bias in COVID-19.

Data Availability Statement

The original contributions presented in the study are included in the article/supplementary material, further inquiries can be directed to the corresponding author.

Author Contributions

Research topic was mutually agreed on by both authors, research was primarily conducted by AS, writing was initiated by AS. AS was supervised by TV during the research period, TV supervised and corrected writing. All authors contributed to the article and approved the submitted version.

Conflict of Interest

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Publisher’s Note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Acknowledgments

References