Autoantibodies neutralizing GM-CSF in HIV-negative Colombian patients infected with Cryptococcus gattii and C. neoformans

Background: Cryptococcosis is a life-threatening disease caused by Cryptococcus neoformans or C. gattii. Autoantibodies (auto-Abs) neutralizing granulocyte-macrophage colony-stimulating factor (GM-CSF) in otherwise healthy adults with cryptococcal meningitis have been described since 2013. We searched for neutralizing auto-Abs in sera from Colombian patients with non-HIV related cryptococcosis in a retrospective national cohort collected from 1997 to 2016. Methods: We reviewed clinical and laboratory records and assessed the presence of neutralizing auto-Abs in 30 HIV (−) adults presenting cryptococcosis (13 by C. gattii, and 17 by C. neoformans). Results: We detected auto-Abs neutralizing GM-CSF in the plasma of 9 out of 13 (69%) patients infected with C. gattii and 1 out of 17 (6%) patients with C. neoformans. Conclusions: We report ten Colombian patients with cryptococcosis due to auto-Abs neutralizing GM-CSF. Nine of the ten patients were infected with C. gattii, and only one with C. neoformans.


Introduction
Cryptococcosis is initiated by breathing blastoconidium or basidiospores of the yeasts of the species complexes Cryptococcus neoformans and C. gattii [1].After inhaling the propagules from the environment, mainly soil, avian excreta, trees, and decaying wood, cryptococcal infection presents initially as pneumonia, to later disseminate to the central nervous system (CNS), causing meningitis, the most frequent form, or even meningoencephalitis [1].
Despite common environmental exposure to cryptococcal species, cryptococcosis is rare in the healthy population because of high natural resistance.Defects in T cell-mediated immunity, speci cally the decline in number and function of CD4 + lymphocytes, which usually occurs in people infected with the human immunode ciency virus (HIV), remains the main risk factor to acquire infection by C. neoformans [1].Cryptococcosis caused by C. gattii, which is much less common (~ 20%), has traditionally been associated in otherwise healthy individuals, particularly HIV-seronegative, or those with unidenti ed risk factors [2].However, other immunosuppressive and pulmonary diseases are underlying conditions that have been associated with a signi cant higher risk of C. gattii infection [2].Moreover, in the last decade, subtle immunologic alterations, which are host-dependent risk factors, have been detected in most patients with cryptococcosis by C. gattii, placing this species as an opportunistic pathogen [3,4].
Anti-cytokine neutralizing autoantibodies (auto-Abs) are considered to constitute autoimmune phenocopies of Inborn Error of Immunity (IEI) with selective predisposition to infectious diseases [5][6][7].Indeed, by blocking the biological function of their target cytokines, these auto-Abs cause clinical phenotypes mimicking those of IEI of the corresponding cytokines or their receptors [7].Patients affected by different infectious diseases, including fungal infections, have been found to carry such neutralizing auto-Abs [5].For example, chronic mucocutaneous candidiasis is attributed to interleukin (IL)-17A/F auto-Abs, adult-onset susceptibility to mycobacterial disease to interferon-gamma (IFN-γ) auto-Abs, recurrent staphylococcal infection to IL-6 auto-Abs, and more recently, severe SARS-CoV2 infection to type I IFNs auto-Abs [5,7].
Since 2013, high titers of immunoglobulin G (IgG) auto-Abs neutralizing granulocyte-macrophage colonystimulating factor (GM-CSF) have been identi ed in patients with adult-onset isolated idiopathic disseminated diseases, mostly cryptococcosis, almost exclusively by C. gattii [8-10], and other infections such those caused by species of Nocardia and more rarely Aspergillus [9,10].In addition to the infectious phenotype, those auto-Abs against GM-CSF have also been associated with pulmonary alveolar proteinosis (PAP), a severe lung disease characterized by the accumulation of surfactant in the alveoli, with progressive respiratory failure and an increased risk of secondary infections [11].The causality among the presence of neutralizing auto-Abs GM-CSF and these two clinical phenotypes (PAP and cryptococcosis) is not well understood.Nevertheless, patients with such auto-Abs presenting with cryptococcosis have been reported with or without PAP manifestations, and patients with PAP have been described with or without cryptococcosis [8].Finally, the presence of auto-Abs against GM-CSF in those pathologies, suggests an important role of GM-CSF in the correct maturation and function of alveolar macrophages, which constitutes the main cellular component of immunity against Cryptococcus [12,13].
The fact that about 13% of cases of cryptococcosis in Colombia occur in non-HIV patients with no apparent risk factor [14,15], together with the recent identi cation of auto-Abs neutralizing GM-CSF in three Colombian patients with cryptococcal meningitis by Cryptococcus spp.[16], prompted us to hypothesize that other Colombian patients affected by this mycosis and who had been regarded as otherwise healthy individuals, especially given that they were not infected with HIV, have a hidden immune dysfunction.As such, the aim of this study was to establish the presence of auto-Abs against GM-CSF in plasma or serum from 30 Colombian patients, without HIV, who developed cryptococcosis by C. gattii or C. neoformans species complexes, and to correlate our ndings with the patients' clinical data.

Subjects and sera selection
As part of the National Surveillance Program for Cryptococcus and cryptococcosis in Colombia, led by the Instituto Nacional de Salud, in Bogota, Colombia, 1974 surveys from patients with cryptococcosis have been lled out between 1997 and 2016.In summary, the survey contains demographic data, risk factors, clinical manifestations, diagnostic methods, and the patients' initial treatment [14].From these surveys, 392 (19.9%) patients were reported without evident HIV infection (HIV negative), from whom, the etiological agent causing cryptococcosis was identi ed in 343 cases, 292 (85.1%) caused by C. neoformans and 51 caused by C. gattii (14.9%) [14] (See supplementary Annex 1).From all non-HIVrelated cryptococcosis, only 30 patients' serum samples were collected in a 15-year period, from 1997 to 2011, and stored in the sera collection of the Microbiology Group at the Instituto Nacional de Salud.From these available samples, which were used to establish the presence of GM-CSF auto-Abs, 13 sera were from patients with cryptococcosis by C. gattii and 17 sera by C. neoformans [14] (See in supplementary Tables S1, S2 and S3).
As part of the diagnosis of cryptococcosis, which was done by direct visualization of the encapsulated blastoconidia in cerebrospinal uid (CSF) using India ink, some of the studied sera had data on the titter of the cryptococcal antigen (CrAg) detected in this sample.CrAg titter was also established in CSF of some patients.All isolates of C. gattii or C. neoformans causing cryptococcosis in the studied patients were identi ed by routinely used phenotypic methods.Most C. gattii isolates had data on antifungal susceptibility testing, serotype, mating type, molecular type and sequence type (ST), while most C. neoformans isolates had only data on serotype, mating type and molecular type [14,15,17] (Tables S4  and S5).All studied sera had in addition total levels of IgG, IgA and IgM, as well as cryptococcal-speci c IgG, IgA and IgM, as established previously [18].For some analysis, data on speci c levels of serum immunoglobulins against cryptococcal proteins were used from serum from adults without cryptococcosis or any other infectious disease (healthy controls) [18] (See in supplementary Figures S1  and S2 ).

Statistical analysis and data availability
p-value was calculated among groups with a Chi-square test with Fisher's correction (given the low n in some of the cells of the contingency tables).A p-value < 0.05 was considered as statistically signi cant.
All raw and processed data will be made available by the corresponding authors upon request.

Cryptococcosis in Colombian HIV negative patients
Demographic and clinical characteristics of the 30 studied patients are summarized in the supplementary material (Supplementary Tables 1, 2 and 3).From them, 21 (70%) were men and nine (30%) were women, ranging from 1 to 71 years old, with an average age of 40.8 years.While most patients (86.7%) did not have any recognizable predisposing factor, hematological malignancy was registered in two patients (6.7%) as well as systemic lupus erythematosus (SLE) and rheumatoid arthritis in one patient each (3.3%).Signs and symptoms of cryptococcosis were variable among the studied patients, however, headache (70%), mental confusion (46.7%) and nausea (40%) were the most frequent clinical manifestations, considering that most patients (96.7%) were diagnosed with cryptococcal meningitis.
Treatment was mainly amphotericin B deoxycholate alone (46.7%), or in combination with uconazole (30%).Outcome of patients was recorded in 11 cases, from whom eight recovered after treatment while three died of cryptococcosis [14] Detection of neutralizing GM-CSF auto-Abs in patients' sera First, we evaluated the C. gattii patient's cohort; unlike serum from healthy individuals, serum from nine patients (69.2%), incubated with 10 ng/µL of GM-CSF prevented GM-CSF-induced p-STAT5, whereas the level of IL-3-induced p-STAT5 was similar in cells incubated with controls or patients' sera (Fig. 1).
Regarding the C. neoformans cohort, one out of 17 s patients (6%) presented neutralizing GM-CSF auto-Abs (Fig. 2).Altogether, these results strongly suggest that the presence of circulating neutralizing GM-CSF auto-Abs is the risk factor to develop cryptococcosis in 10 of the studied patients.

Clinical and immunological correlation in patients with neutralizing GM-CSF auto-Abs
From the 30 studied patients, nine (69.2%) and one (6%) patient affected with cryptococcosis by C. gattii and C. neoformans, respectively, were positive for neutralizing GM-CSF auto-Abs.The demographic, clinical and microbiological characteristics of both groups (positive and negative for neutralizing GM-CSF auto-Abs) are compared (Table 1),.Clearly, patients with neutralizing GM-CSF auto-Abs are predominantly affected by C. gattii.Sex, age, and clinical presentation did not differ among patients with and without neutralizing GM-CSF auto-Abs.Clinical characteristics of patients with cryptococcosis with and without neutralizing GM-CSF auto-Abs respectively, are summarized in Supplementary Tables 2-3.Total and speci c levels of immunoglobulins were determined, however total levels of IgG, IgM and IgA did not differ between patients with or without neutralizing GM-CSF auto-Abs, nor if they were infected by C. neoformans or C. gattii (Supplementary Fig. 1a-c).When compared with healthy controls, we observed higher levels of total IgG in patients with cryptococcosis, whereas total IgA and IgM were in lower levels.
Nevertheless, sera from patients with cryptococcosis by C. gattii with neutralizing GM-CSF auto-Abs showed higher levels of speci c IgG against cryptococcal proteins, compared to healthy controls, as established elsewhere (Supplementary Figs.1d-e 4 and 5).This means that none of showed resistance or decreased susceptibility to any antifungal drug.In addition, C. gattii isolates recovered from patients with and without neutralizing anti GM-CSF auto-Abs, did not differ regarding the serotype, mating type, molecular type, and ST of the isolates (Supplementary Fig. 1f).Serotype, mating type and molecular type did not differ either among the C. neoformans isolates.

Discussion
Our study reports neutralizing GM-CSF auto-Abs in 10 Colombian patients who developed cryptococcosis, even though these patients were deemed as otherwise healthy, based upon their clinical history at the time of diagnosis of this mycosis [14].We also evidence that this subtle immunologic alteration prevails in patients affected by C. gattii, compared with patients by C. neoformans, which adds to the differences described in the epidemiology of patients with cryptococcosis caused by one species complex or another [3,4].Remarkably, the risk factor of patients with C. gattii infection and 16 with C. neoformans infection, remains hidden, as neutralizing auto-Abs against were not detected.
The epidemiology of cryptococcosis has focused especially on patients with HIV, the major risk factor for infection [19].Though, in recent years, different studies on cryptococcosis have drawn attention to the increase in those apparently otherwise healthy individuals without HIV infection [20][21][22][23].In Colombia, in the cryptococcoses survey performed between 1997-2016, HIV infection was found in 75.4% of patients and non-HIV in 24.6% [14].In Brazil, HIV infection remains the main risk factor for cryptococcosis (82-86%) [24,25], however, in a series of 29 patients without HIV infection and who did not receive transplants, 77.8% of the patients had no apparent risk factor and the majority were infected by C. gattii [21].Similarly, in the French Guiana, HIV infection has been reported as the main risk factor for cryptococcosis (67.4%), but it was precisely in this country where was reported the rst two cases of Latin-American patients with cryptococcosis presenting anti-GM

Conclusions
Our ndings, therefore, support the association between neutralizing GM-CSF auto-Abs and cryptococcosis, which has been described for a decade [34][35][36].These auto-Ab can be totally silent until the patients are diagnosed with cryptococcosis.However, the reason why patients with this predisposing factor are more susceptible to acquire infection by C. gattii than by C. neoformans, remains elusive.Anti-GM-CSF auto-Abs are associated with some cases of pulmonary and meningeal cryptococcosis in otherwise healthy individuals.Our data support the idea that GM-CSF is a critical actor in host defense against Cryptococcus.

Declarations Figures
Neutralizing

Supplementary Files
GM-CSF auto-Abs in patients with cryptococcosis by Cryptococcus gattii.STAT5 phosphorylation (p-STAT5), assessed by ow cytometry, upon the stimulation with recombinant human rhGM-CSF (red) or rhIL-3 (black) of control PBMCs, in the absence of sera, or in the presence of 1:10 dilution of sera from two healthy individuals (HC-1 and HC-2), sera of two individuals previously described carrying GM-CSF auto-Abs (C-1 and C-2) or from thirteen patients with cryptococcosis by C. gattii.NS: non-stimulated with rhGM-CSF or rhIL-3.
2Pulmonary cryptococcoma with positive culture and histopathology3SLE=systemic lupus erythematosus, RA = rheumatoid arthritis3AMBd= amphotericin B deoxycholate, 5FC = 5-ucytosine, FLC = uconazole Phenotypic and genotypic characteristics of cryptococcal isolates do not differ among patients with without neutralizing GM-CSF All isolates causing cryptococcosis in patients with and without neutralizing GM-CSF auto-Abs, which had data on antifungal susceptibility testing, distributed among the wild-type population of the species, per antifungal drug (Supplementary Tables 34,35]2][33]s in otherwise healthy individuals with C. gattii infection[26].These observations indicate that the epidemiology of cryptococcosis is changing with a notable increase of non-HIV-related infection suggesting that additional risk factors may be involved in the susceptibility to cryptococcosis.In Colombia, 1974 patients with cryptococcosis have been identi ed between 1997 and 2016.From them, just 392 patients were HIV-negative, of whom 51 (14.9%) were affected by C. gattii infection.The study of 13 patients with C. gattii and 17 with C. neoformans from this cohort allows us to establish that those auto-Abs collectively account for 68% (n = 9) of C. gattii patients and 6% (n = 1) of C. neoformans, showing that GM auto-Abs underlie cryptococcosis in a signi cant proportion of cases.Nevertheless, cryptococcosis remains unexplained in most HIV-negative patients affected with C. neoformans and onethird of C. gattii.IEI of GM-CSF pathway cytokine or their receptors might potentially explain a proportion of these remaining cases.Similarly, to previous studies, all 10 patients with GM-CSF auto-Abs described here were adults (23 to 67 years old).If those auto-Abs were present before cryptococcal infection and if they remained silent up to cryptococcal infection, remains unknown.Recent studies of auto-Abs against cytokines as a main risk factor for a speci c infectious disease have demonstrated the causality of those auto-Abs[27][28][29].Pre-existing Type I IFN neutralizing auto-Abs is the main clear example as a risk factor for several viral diseases such as life-threatening COVID-19 pneumonia (15-20%), a third of the rare lifethreatening adverse reactions to yellow fever vaccination[28], and about 40% of cases of West Nile virus encephalitis[30].In addition, these auto-Abs showed to be present in around ∼0.3% of the general population under 65 years, whereas this prevalence increases sharply after 70 years to ∼4% [28-30].The actual prevalence of GM-CSF auto-Abs in patients with cryptococcosis or healthy populations remains elusive.Interestingly, one patient described here (P28) presented disseminated cryptococcosis by C. neoformans and pulmonary tuberculosis (Tb) by Mycobacterium tuberculosis (Mtb) one year after the cryptococcal diagnosis.To our knowledge, this is the third adult patient with disseminated cryptococcosis and neutralizing anti-GM-CSF auto-Abs who have developed Tb[8, 16].Furthermore, mice studies and ex vivo studies, and human monocyte-derived macrophages[31][32][33], and our date suggest that intact GM-CSF signaling is crucial for the appropriate alveolar macrophage functions to mediate the immunity to Cryptococcus spp., and possibly against Mtb lung infection in humans.Our ndings support the hypothesis, and the new research front, that auto-Abs against cytokines may explain the susceptibility to Cryptococcus infection in otherwise healthy HIV-negative individuals[7].However, routinely used laboratory technologies do not detect such speci c targets [16,34,35], and ow cytometry and particle-based technology used to screen for the presence of anti-cytokines auto-Abs, are very expensive, not often available in the clinical setting and, so far, have been utilized as research tools rather than as commonly used techniques to help establishing a de nitive diagnosis in patients with