Understanding the pathogenesis of severe babesiosis in canines can not only help us to improve the treatment of canine babesiosis, but also provide insights into the pathophysiology of human babesiosis and the related parasitic disease, severe malaria. Prior knowledge of canine babesiosis pathophysiology was based on clinical observational studies in which duration of infection as well as the timing and size of the inoculum were variable and unknown. To address these sources of clinical heterogeneity, we performed controlled inoculations at two doses in susceptible, purpose-bred laboratory beagles and characterized the infection in terms of clinical manifestations and gene expression changes over time. In doing so, we learned that experimental inoculation with wild B. rossi parasites consistently induces the rapid onset of symptomatic canine babesiosis. Both the rate of onset and the severity of disease were related to the size of the infectious inoculum. Disease severity, reflected through behavior, vital signs, hematology, biochemistry, acid-base status, and hormone levels, increased rapidly with the rising parasitemia and responded to treatment.
To further understand the progression of this hemoprotozoal infection which transpires entirely within the hematological system, we monitored host gene expression in circulating whole blood at frequent intervals. The results of transcriptomic analysis identified differential expression of pathways involving (1) response to hemolysis (including erythropoiesis and iron homeostasis), (2) metabolism, and (3) distinct immunological responses including innate immunity, adaptive immunity, and viral response genes. These highlight host pathways that could be targeted to slow disease progression and prevent life-threatening complications.
Anemia is a hallmark of babesiosis in humans, horses, cattle, and canines (7). In this experimental model, massive hemolysis, as indicated by the rapid rise in plasma hemoglobin, along with inadequate erythropoiesis caused rapid and severe anemia that required transfusion to correct. After an early rise in reticulocyte counts, the reticulocytosis was suppressed as parasitemia climbed and remained inadequate for several days after treatment with an anti-parasite drug, implying that the parasite or the host response to infection was limiting erythropoiesis or blocking the release of reticulocytes from the bone marrow into circulation.
At a transcriptional level, enrichment of erythrocyte-related genes and pathways was prominent during peak parasitemia and also during recovery. During peak infection in both inoculum cohorts, oxygen transport (GO:0015671), iron ion homeostasis (GO:0055072), erythrocytes take up oxygen and release carbon dioxide (R-CFA-1247673), heme biosynthesis (R-CFA-189451), and metabolism of porphyrins (R-CFA-189445) were all among the top 10 most enriched pathways as measured by negative log FDR (Fig S1).
Many genes with functions related to erythopoesis and heme biosynthesis were in cluster 2, which increased in expression as infection progressed and returned to baseline during recovery. This includes KLF1, a key transcription factor regulating erythropoiesis and hemoglobin expression (17,18), ALAS2, which catalyzes the first and rate limiting step in heme synthesis in erythroid cells (19), and UROD, which codes an enzyme in the heme biosynthetic pathway (20). In the same cluster, SPTB and SPTA1 were differentially expressed; these genes are expressed in erythroid precursors and code for a constituent of the erythrocyte cytoskeleton (21). Mutations in these genes are also associated with hereditary spherocytosis (22). The rise and fall in erythroblast transcripts parallels the reticulocyte count, which peaked during peak infection (day 3 in the high inoculum and day 4 in the low inoculum cohort) and began to rise again after treatment.
Other genes related to hemolysis were in cluster 2. This includes HP, the gene that codes for haptoglobin, a protein that binds free hemoglobin from lysed red cells and is a marker for hemolysis (23). Additionally PRDX2, which encodes an antioxidant enzyme that stabilizes hemoglobin under oxidative conditions, was differentially expressed in both cohorts. Mutations in this gene are strongly associated with hemolytic anemia (24). Finally, GCLM, which has been shown to be essential to erythrocyte survival during oxidative stress (25) and whose deficiency is associated with hemolytic anemia (26), was also upregulated during infection.
Though there have not been extensive studies correlating gene expression with hemolysis in clinical malaria pathogenesis (27), the transcriptomic profiles reported here share many features in common with the host response to hemolytic anemia, including in sickle cell disease (18,28–31).
Both high and low dose inoculation with B. rossi was associated with enrichment of the Reactome pathway metabolism (R-CFA-1430728) during peak infection. In the low inoculum, long-chain fatty acid metabolic process (GO:0001676) was enriched on days 4 and 6 and cholesterol metabolic process (GO:0008203) was enriched on days 3 and 4. Both pathways were also enriched in the high inoculum on days 3 and 4 (Fig S1).
Many lipid metabolism-related genes were in cluster 2 in both inoculum cohorts. This includes: PLA2G4A, which encodes a phospholipase involved in membrane lipid remodeling and biosynthesis of lipid mediators of the inflammatory response (32,33); GPR84, which serves as a receptor for free fatty acid (34); CD5L, which encodes a regulator of lipid synthesis that in turn regulates inflammatory response mechanisms and T cell activities (35); and ALOX15, which encodes a lipoxygenase that catalyzes the deoxygenation of polyunsaturated fatty acids and has known roles in red cell maturation and the inflammatory immune response (36). Several lipid metabolism-related genes that were in cluster 2 in the low inoculum cohort were in cluster 3 in the high, indicating that they continued to increase expression levels throughout infection and recovery. This includes both ECHDC3, known to play a major role in fatty acid biosynthesis (37), and FADS1, which encodes a fatty acid desaturase (38). Previous studies implicated the role of lipid droplets as regulators of the immune response to protozoan infection (39). The high prevalence of lipid metabolism genes among differentially expressed genes implicates these pathways in the host response to B. rossi infection. (Fig 5).
Genes related to glucose metabolism were also differentially expressed throughout the infection and recovery. In the high inoculum cohort, glucose metabolic process (GO:0006006) was enriched on days 6 and 8. GATM, which encodes the rate-limiting step in creatine metabolism (40), was in cluster 3 in the low inoculum cohort and cluster 2 in the high. TKT, a key enzyme in the pentose phosphate pathway (41), was in cluster 2 in both cohorts, indicating an increase in glycolytic activity. Hypoglycemia and hyperlactatemia are a feature of both babesia and malaria infections (42,43). Studies in malaria show glycolysis pathway genes prominently activated early during infection in mouse models and in humans (27,44). Thus, canine babesiosis and human malaria appear to share features of both lipid and carbohydrate metabolic pathway activity.
Both the clinical and transcriptomic data indicated a robust immune response to infection in the high and low inoculum cohorts. The inflammatory nature of the disease was reflected by the inoculum-dependent changes in the fever curve, CRP level, as well as the white cell, neutrophil, and band counts – the last of which is known to predict poor outcomes (11). While some cytokines reached a maximum at peak parasitemia (IL-8, IL-10, KC-like) others rose only after the infection was treated (TNF, IL-6, MCP-1). Multiple immune pathways were differentially regulated in response to both high and low inocula. Defense response (GO:0006952) was enriched in the high inoculum on day 1 and both inoculum cohorts on day 3. Immune system/response (R-CFA-168256; GO:0006955) was enriched in the high inoculum on days 1 and 8, and both inoculum cohorts on days 3-6. Innate immune system/response (R-CFA-168249; GO:0006955) was enriched on every experimental day in the high inoculum cohort. Additionally, cytokine signaling in immune system (R-CFA-1280215) was enriched during infection onset in both inoculum cohorts (day 1 in the high and day 3 in the low inoculum) These pathways are composed of hundreds of constituent genes, which primarily followed cluster 2 trajectories. These pathways reflect the differential expression of specific genes involved in innate immunity, adaptive immunity, and viral response. (Figs 4,S1).
Genes associated with positive regulation of interleukin-1 beta (IL-1β) production (GO:0032731) were in cluster 2. This includes: IL1B, which encodes the pro-inflammatory cytokine (45); PSTPIP2, which encodes a protein that regulates IL-1β (46,47); CASP4, which regulates IL-1β synthesis in macrophages (48); and SIGLEC14, which encodes a receptor that enhances inflammasome activation and macrophage IL-1β release. Furthermore, genes associated with NF-kB signaling were also constituents cluster 2, including FABP5, ACOD1, and TNFAIP2. FABP5 encodes a fatty acid binding-protein that promotes the activation of NF-kB signaling (49). ACOD1 serves a multifaceted role in the innate immune response to pathogen infection (bacteria and viruses) and in cytokine signaling (TNF and interferons), and is associated with both Toll-like receptor and NF-kB signaling pathways (50). Expression of TNFAIP2 is upregulated via the NF-kB pathway and mediates the inflammatory response (51,52). Studies of malaria infection in humans also implicate the role of IL-1β signaling and NF-kB pathways in the innate immune response to infection (53,54). Thus, Babesia and Plasmodium induce similar innate immune responses across different host species.
Pathways associated with the production and regulation of interferon-gamma (GO:0032649) were significantly enriched in the host response to B. rossi infection, especially during infection onset (Fig S1). IFN-γ signaling-related genes, including IRGM (IFI1), IFGGC1, SPP1, and APOL2, were in cluster 2 in both inoculum cohorts. IRGM plays a key role in regulating IFN-γ-induced host defense to protozoa (55), while IFGGC1 is a member of the IFN-γ-inducible GTPases that are involved in immune response against pathogens (56). SPP1 codes for a cytokine known to upregulate IFN-γ production (57), and APOL2 which mitigates the cytotoxic effects of IFN-γ (58). IFN-γ is a major pro-inflammatory cytokine that has both direct antiparasitic activity and immunoregulatory roles in the response to parasitic infection (59). IFN-γ has a role in protective immunity against infection by B. microti in mice (60) and is involved in the host immune response to malaria in both mice and humans (27,53,54). Our findings suggest IFN-γ has a similar role in protective immunity against B. rossi infection in canines.
Genes that code the MHC class II protein complex, namely DLA-DOA and DLA-DOB (61), were in cluster 1, indicating that they were significantly downregulated in both inoculum cohorts. In contrast, MHC class I genes DLA-12,DLA-64, and DLA-88 were upregulated during infection (though they did not meet the log2 fold change threshold for significance, excepting the low inoculum cluster 2 gene DLA-12). Additionally, GZMA and PRDX2, genes specifically expressed by CD8+ T cells and associated with their activity (62,63), were cluster 2 DEGs in both cohorts.
CD8+ and CD4+ T cells, whose T-cell receptors recognize antigens presented by MHC-I and MHC-II respectively, have been previously implicated in protective immunity against B. microti infection in mice and B. bovis infection in cattle. Previous studies in mice and humans suggest that both and MHC-I and MHC-II play a role in the host immune response to malaria infection (27,54). However, recent studies in human cerebral malaria suggest that CD8+ T cells migrate to and sequester in the brain and contribute to mortality (64,65). Our data reveal downregulation of genes associated with MHC class II CD4+ T cells and potential induction of MHC class I CD8+ T cells in the host response to B. rossi, raising the question of what cells are presenting antigens and whether the CD8+ and CD4+ responses are immunopathologic or contribute to parasite killing and adaptive immunity.
Interestingly, many viral response genes were differentially regulated during B. rossi infection, particularly during infection onset. On day 1 in the high inoculum cohort, pathways including positive regulation of defense response to virus by host (GO:0002230), antiviral mechanisms by IFN-stimulated genes (R-CFA-1169410), and ISG15 antiviral mechanism (R-CFA-1169408) were enriched. The latter two pathways were also enriched on day 3 in the low inoculum cohort, along with response to interferon-alpha (GO:0035455) and cellular response to interferon-beta (GO:0035458). (Fig S1).
This is also reflected in the significant upregulation of constituent viral response genes throughout the infection in both inoculum groups. All of the following genes were identified as cluster 2 DEGs: RSAD2, ISG15, DDX58, DDX60, OAS1, OAS2, OASL1, MX1, PLSCR1, ICAM1, TRIM10, KLRD1, LGALS9, and LTF. In general these genes are related to type I interferon antiviral signalling and have been associated with a range of viruses including influenza, west Nile virus, hepatitis C, and dengue virus (66–80). Interestingly, previous studies in malaria and leishmaniasis models suggest that type I interferons suppress anti-parasitic immunity (81–84). CD8+ T cells are also associated with antiviral activity (62,63,85,86). Together, these transcriptomic data implicate the role of viral response genes and pathways in the host immune response to B. rossi infection.
The inoculum of 109 parasites resulted in an acute and severe disease course. This rapid disease course may be more representative of transfusion-associated babesiosis than natural tick-transmitted infections which would evolve more slowly. The low-mapped reads on day 4 in the high and low inoculum cohorts and day 3 in the high appear to be associated with the high levels of parasitemia recorded on those days. The lower percentage of mapped reads indicates lower overall sequencing accuracy and the potential presence of contaminating DNA, thus increasing the likelihood of spurious results on these days. Additionally, one animal in the high inoculum died on the morning of day 4, which may have affected the expression means reported on days 4, 6, and 8. Transcriptomic and clinical results reported on days 6 and 8 may have been further affected by the administration of day 4 interventions including blood transfusions and anti-parasite drug. Furthermore, both cohorts were small (low n = 2; high n = 3); however, there was a high degree of consistency between subjects of the same cohort with regard to clinical observations and transcriptomic data.