Characterization of a mouse model of sickle cell trait: parallels to human trait and a novel finding of cutaneous sensitization

Abstract

Sickle cell trait (SCT) has classically been categorized as a benign condition except in rare cases or upon exposure to severe physical conditions. However, several lines of evidence indicate that individuals with SCT are not always asymptomatic, and additional physiological changes and risks may remain unexplored. Here, we utilized mice harbouring one copy of normal human β globin and one copy of sickle human β globin as a model of SCT, to assess haematological, histopathological and somatosensory outcomes. We observed that SCT mice displayed renal and hepatic vascular congestion after exposure to hypoxia. Further, we observed that SCT mice displayed increased cold aversion as well as mechanical and heat sensitivity, though to a lesser degree than homozygous sickle cell disease mice. Notably, mechanical hypersensitivity increased following hypoxia and reoxygenation. Overall our findings suggest that SCT is not entirely benign, and further assessment of pain and cutaneous sensitization is warranted both in animal models and in clinical populations.

Introduction

While sickle cell trait (SCT) has been considered an essentially asymptomatic and benign condition associated with increased survival from malaria infection and normal life expectancy, recent studies suggest that it may be associated with some complications of sickle cell disease (SCD) (Key et al, 2015) while other studies revealed no impact on baseline health and fitness (Liem et al, 2017). Individuals with SCT carry only a single copy of the mutated allele encoding the variant of haemoglobin that is known to cause SCD; the presence of normal β globin subunits within circulating red blood cells (RBCs) is protective against haemoglobin polymerization and RBC sickling. However, severe dehydration, hyperthermia and high altitude, particularly when combined with tissue hypoxia from extreme strenuous exercise, can promote sickling and lead to rhabdomyolysis, splenic rupture or even sudden death (Shephard, 2016; Tsaras et al, 2009; Kark et al, 1987; Nelson et al, 2016). Additional complications in SCT may include renal papillary damage due to the relatively acidotic and hypoxic milieu of the medulla (Tsaras et al, 2009) as well as traumatic hyphaema and increased risk of venous thromboembolic disease (Austin et al, 2007; Folsom et al, 2015).

In contrast to SCT, it is well established that the clinical presentation and complications of homozygous SCD can be quite severe and that acute and chronic pain are major complications that decrease the quality of life (Brandow et al, 2013; Smith et al, 2008; Niscola et al, 2009). Quantitative sensory testing has shown increased cutaneous sensitivity and pain in both children and adults with SCD (Jacob et al, 2014; Brandow et al, 2013). The two prevalent mouse models of SCD, Berkeley (Paszty, 1997) (Berk) and Townes (Wu et al, 2006), both express 100% haemoglobin S in their circulating erythrocytes and model many features of human SCD, including RBC sickling, severe haemolytic anaemia and organ pathology (Manci et al, 2007; Paszty, 1997; Wu et al, 2006). Homozygous sickle mice display prominent hypersensitivity (pain behaviour) to mechanical, cold and heat stimuli (Hillery et al, 2011; Kohli et al, 2010; Cain et al, 2012; Cataldo et al, 2015; Zappia et al, 2014), replicating some of the mechanical and thermal phenotypes in patients.

Due to the absence of acute pain crises in SCT, it has largely been assumed that individuals with SCT do not have sickle trait-associated pain; however, the possibility of less severe pain and cutaneous hypersensitivity in SCT is not known. SCT occurs in approximately 8% of African Americans, resulting in a prevalence of 3–4 million individuals living with SCT in the United States alone, and 43 million individuals globally (Jordan et al, 2011; Global Burden of Disease Study 2013 Collaborators, 2015; Tsaras et al, 2009). Human SCT impacts an understudied population that may or may not be at risk for a subset of clinical complications. As no systematic review of such complications has been performed, a lack of knowledge and subsequent controversy in the field may contribute to stigmatization (Naik & Haywood, 2015). Similarly, there have been few efforts to characterize SCT in animal models. Importantly, suitable animal models are needed to begin to address whether SCT confers pain hypersensitivity.

Berk SS mice are a reliable and widely used model of severe SCD. Several groups have characterized Berk hemizygous mice as having an intermediate sickle-like phenotype (Szczepanek et al, 2012; Wang et al, 2016; Kohli et al, 2010). However, these mice contain chimeric haemoglobin due to heterozygous expression of murine β globin as well as expression of one or two copies of the transgene containing the sickle variant of human β globin [Tg(Hu-miniLCRα1^Gγ^Aγδβ^S) Hba⁰//Hba⁰Hbb⁺//Hbb⁰] (Wang et al, 2016; Kohli et al, 2010; Manci et al, 2007). This chimeric haemoglobin (human α globin paired with mouse β globin) has impaired oxygen affinity, leading to increased tissue hypoxia and renal pathology in these hemizygous mice (Noguchi et al, 2001). Consequently, these Berk hemizygous mice do not model any known human haemoglobinopathy and are not a suitable model of SCT. Thus we generated mice containing one copy of human sickle haemoglobin (hα,hβ^S) and one copy of normal human β globin (hα,hβ^A) to more appropriately model of human SCT [Tg(Hu-miniLCRα1^Gγ^Aγδβ^A)//Tg(Hu-miniLCRα1^Gγ^Aγδβ^S) Hba⁰//Hba⁰Hbb⁰//Hbb⁰] and characterize these mice for the first time here.

Methods

Animal models

In the Berkeley (Berk) model of SCD, transgenic mice are homozygous knockout of all murine α and β haemoglobin genes (Paszty, 1997). To replace these genes, sickle mice express a transgene (Tg β^S: Hu-miniLCRα1^Gγ^Aγδβ^S) allowing these mice to express exclusively human sickle haemoglobin in their circulating erythrocytes (SS mice). Control mice similarly lack murine haemoglobin genes and express a transgene (Tg β^A: Hu-miniLCRα1^Gγ^Aγδβ^A) to express all normal adult human haemoglobin in circulating erythrocytes (AA mice). Mice containing both the HbA and HbS transgenes were first intercrossed for 5 generations. Subsequently, AA mice were generated by mating AA with AA mice. Both SCT mice (AS) and sickle mice (SS) were generated by breeding AS females with SS males. Therefore, all mice have a similar mixed genetic background as the Berkeley SCD mice. All mice used in this study were females aged 6–9 months (mean 7.6) and age-matched between genotypes. All mice were phenotyped by haemoglobin electrophoresis and terminal CBC. For some experiments, animals were exposed to hypoxia (10% F_IO₂) followed by reoxygenation (hypoxia and reoxygenation; HR); 2 h of hypoxic exposure in a chamber maintaining a fraction of inspired oxygen (F_IO₂) of 10% using nitrogen mixed with room air was used for behavioural studies and 3 h of hypoxic exposure for blood and histological experiments. Reoxygenation in ambient air lasted 1–3 h, as noted. All procedures and protocols were approved by the Institutional Animal Care and Use Committee of the Medical College of Wisconsin.

Pain behaviour analysis

Behavioural assays were used to assess three modalities of somatosensation: sensitivity to mechanical (or tactile), heat and cold stimuli. As a measure of mechanical sensitization, 50% paw withdrawal thresholds were obtained by application of a series of calibrated von Frey filaments (0.38–37 mN; Bioseb, Pinellas Park, FL) to the plantar hind paw, determined using the Up-Down method (Chaplan et al, 1994; Dixon, 1980). Heat sensitivity was measured as the latency to paw withdrawal from a focal radiant heat source applied to the plantar hindpaw (Hargreaves et al, 1988). In both assays, data were collected from both hind paws and averaged per mouse. Cold sensitivity was determined using a thermal preference assay (Zappia et al, 2014). Briefly, mice were allowed to freely explore a testing chamber (TECA Corp, Chicago, IL) in which the floor was divided into a warm side (30°C) and a second test plate (either 20°C or 23°C, as described); the percentage of time spent on the cold test plate was measured. The experimenter was blinded to genotype for all assays.

Tissue and Blood Collection and Processing

Whole blood was drawn from the left ventricle of anesthetized mice and collected into trisodium citrate (9:1). Whole blood was gently centrifuged (180 × g for 10 min) followed by separation of the RBC pellet from plasma. Platelet-rich plasma was centrifuged (1500 × g for 10 min) to sediment platelets and clarify plasma; plasma aliquots were stored at −80°C. Prior to harvesting organs, mice were completely perfused with 20 ml phosphate-buffered saline (PBS) at a pressure of 80 mmHg (physiological blood pressure) using gravity to perfuse the PBS from a syringe taped to a height to produce this pressure. PBS was perfused through the left ventricle to the clipped right atrium until the lungs were visibly cleared. After organ mass was measured, the livers, lungs and kidneys were fixed in 10% buffered formalin for 48 h, paraffin-embedded, sectioned at 4 µm and trichrome stained for histological evaluation. Global adjustment of white balance was applied after image capture.

Histopathology

All histological, blood, and plasma analyses were performed by evaluators who were blinded to the genotype of the mice and whether or not they were exposed to HR. Histopathology was graded on a scale of 0 to 5, primarily based on per cent area with vessels filled with blood despite organ perfusion with buffered saline prior to organ harvesting as described above. Vessels were scored as “filled” if the entire vessel was filled with blood. Score of: 0 denotes no congestion; 1 denotes less than 10% vessels/area congested; 2 denotes 15–40% vessels/area congested; 3 denotes 40–60% vessels/area congested; 4 denotes 60–90% vessels/area congested; 5 denotes greater than 90% vessels/area congested. Absent or low levels of vascular congestion (score 0 to 1) were considered within normal limits while levels of vascular congestion scoring greater than 1 (>10% with congestion) were considered “positive.” Other histological findings are represented in the figures and described in the text.

Whole blood, RBC and plasma analyses

From whole blood samples, haemoglobin (Hb), haematocrit (HCT), mean corpuscular volume (MCV), mean cell haemoglobin concentration (MCHC) and white blood cell (WBC) counts were estimated using a veterinary animal complete blood counter (Heska, Fort Collins, CO); measurements were corrected for citrate dilution. Reticulocytes were quantified from whole blood by flow cytometry using thiazole orange to stain reticulocytes within the erythrocyte gate (using antibodies against RBC-specific antigen TER119). Anion-exchange high-performance liquid chromatography (HPLC) was used to identify and quantify the per cent contribution of haemoglobin from a pool of adult AS mouse whole blood collected into EDTA per the Children’s Hospital of Wisconsin clinical laboratory standard protocol.

Plasma levels of soluble vascular cell adhesion molecule-1 (sVCAM1) were determined via enzyme-linked immunosorbent assay, according to the manufacturer’s instructions (Quantikine, R&D Systems, Minneapolis, MN). Similarly, serum alanine aminotransferase (ALT) was measured using ALT Color Endpoint Assays per manufacturer’s instructions (Bioo Scientific, Austin, TX). Lactate dehydrogenase (LDH) was measured using a QuantiChrom Lactate Dehydrogenase kit (BioAssay Systems, Hayward, CA).

Red blood cell adhesion assay

RBC adhesion to purified thrombospondin was quantified for washed RBCs collected from both AA and AS mice. Whole blood was centrifuged at 180 g for 20 min; platelet-rich plasma was removed and the RBC pellet was washed and centrifuged twice with CGS buffer (14.7 mM sodium citrate, 33 mM glucose, and 124 mM NaCl), and resuspended in serum-free M199 (SFM). A 2% haematocrit was prepared in SFM supplemented with 0.2% bovine serum albumin. Adhesion of RBCs to a thrombospondin-coated surface was measured using a VenaFlux platform (Cellix, Dublin, Ireland). Briefly, a Vena8 Fluoro Biochip was coated with thrombospondin (1 µg/75 µl in trisodium phosphate buffer) overnight. RBCs were perfused though the channel for 1 min at a wall shear stress of 0.4 dyne/cm². Images of adherent RBCs were captured at a total of five areas to cover the middle section of the channel. Adherent RBCs were manually quantified from captured images.

Statistical Analysis

Paw withdrawal thresholds were compared using a non-parametric Kruskal-Wallis test of AA, AS and SS mice. For paw withdrawal thresholds compared between all genotypes and exposure to hypoxia/reoxygenation, results were analysed with overall two-way analysis of variance (ANOVA) followed by post hoc comparisons to baseline using Kruskal-Wallis tests and a Bonferroni correction for multiple testing. Behavioural responses to thermal stimuli were analysed using a one-way ANOVA comparing AA, AS and SS animals. Haematological parameters (Table I) were analysed by multiple t tests followed by Sidak-Bonferroni correction for multiple testing. RBC adhesion, sVCAM1, ALT and LDH were compared by student’s t tests. Organ weights and haem concentrations were compared by two-way ANOVA and post-hoc Sidak correction. An α of 0.05 set a priori, and all data were analysed using Prism 6 (GraphPad Software Inc., La Jolla, CA).

Table I.

Haematological parameters

	AA (n = 12)	AS (n = 12)	p value
RBC (×1012/l)	9.5 ± 0.3	9.4 ± 0.3	0.86
Reticulocytes (%)	3.6 ± 0.2	5.0 ± 0.2	0.0001*
Hb (g/l)	115.6 ± 4.5	125.8 ± 4.3	0.11
HCT (%)	37.5 ± 1.4	40.0 ± 1.4	0.23SS
MCV (fl)	39.7 ± 0.4	42.5 ± 0.3	< 0.0001*
MCH (pg)	12.2 ± 0.2	13.4 ± 0.2	0.0002*
MCHC (g/l)	307.8 ± 1.9	315.1 ± 3.9	0.12
RDW (%)	14.4 ± 0.1	14.4 ± 0.1	0.97
WBC (×109/l)	6.4 ± 0.9	11.2 ± 0.9	0.007*

Data presented as mean ± standard error of the mean

^*significant after Bonferroni correction for multiple comparisons (α = 0.0056)

AA: mice that express all normal adult human haemoglobin in circulating erythrocytes; AS: mice displaying human sickle cell trait; Hb: haemoglobin; HCT: haematocrit; MCHC: mean cell haemoglobin concentration; MCV: mean corpuscular volume; RBC: red blood cells; RDW: red cell distribution width; SS: mice that express exclusively human sickle haemoglobin in their circulating erythrocytes; WBC: white blood cells;

Histology statistics: Means were calculated for variables measured on this grading scale with outcomes spread over a large part of the 0-to-5 range, which applied only to the AS mouse groups. These mean values were compared between groups by t tests (AS-normoxia versus AS-HR). Given that most of the outcomes for the AA groups equalled 0 or 1 (less than 10% vascular congestion or normal), in order to compare AA with AS mice, outcomes with scores greater than 1 were characterized as “positive”. The proportion of each group with a positive outcome was analysed with a Chi square (χ²) and post hoc Fisher’s exact tests and a Bonferroni adjustment for multiple comparisons, similar to previous methods (Manci et al, 2007). Means of the positive values also were tabulated.

Results

Blood and plasma characterization

Haematological parameters from the blood of age-matched AA and AS mice are presented in Table I. Compared to AA mice, AS mice had similar haemoglobin, haematocrit and RBC counts, though mildly increased reticulocyte and WBC counts. Blood from neither AA or AS mice contained sickled RBCs on Wright-Giemsa stained peripheral blood smears (data not shown). As measured via HPLC analysis from a clinical laboratory, we identified approximately 77% HbA, 23% HbS and <0.1% HbF in pooled blood from 4 AS mice tested on two separate runs, confirming the RBC phenotype of the SCT mice.

Given that sickle RBCs have increased adhesion to vascular endothelium and matrix proteins (Hillery et al, 1996; Joneckis et al, 1996; Sugihara et al, 1992), we measured the RBC adhesion in subset of AS and AA samples. We observed a non-significant trend towards increased adhesion of AS RBCs compared to AA RBC binding to immobilized thrombospondin under conditions of low shear flow (35% increase; n = 5, p = 0.09) (Fig. 1A).

Figure 1. RBC adhesion and plasma markers of injury.

(A) RBC adhesion to a thrombospondin-coated surface (n = 5 per genotype), expressed as number of adherent cells per high-powered field. (B) Plasma levels of sVCAM were slightly decreased in AS mice (n = 13) compared to AA mice (n = 11). (C) ALT was increased in the plasma of AS mice compared to controls (n = 12 per genotype), but still within normal range. (D) Plasma LDH was mildly increased in AS mice (n = 13) versus AA mice (n = 12), but still within expected normal range. * P < 0.05, ** P < 0.01, and n.s. denotes not significant. AA: mice that express all normal adult human haemoglobin in circulating erythrocytes; ALT: alanine aminotransferase; AS: mice displaying human sickle cell trait;LDH: lactate dehydrogenase; RBC: red blood cell; sVCAM: soluble vascular cell adhesion molecule

Next, we measured several biomarkers of endothelial and organ injury that are known to be dysregulated in SCD and less well studied in SCT. Levels of sVCAM1 were within normal limits for both AA and AS mice, with a slight decrease in sVCAM1 in AS plasma (Fig. 1B). As SS Berk mice often have elevated ALT and liver infarcts at baseline (Zhang et al, 2013), we measured ALT as a measure of liver injury and found that both AA and AS mice had normal values with a small increase in plasma ALT in AS mice compared to AA (Fig. 1C). Similarly, while LDH levels were higher in AS mice compared to AA mice, the levels of LDH were within the expected normal range for both genotypes (Fig. 1D).

Organ pathology

Total body weight was not different between AA and AS mice of similar ages (AA mice 29.6 ± 1.3 g, 7.8 ± 0.2 months; AS mice 28.8 ± 1.3 g, 7.5 ± 0.3 months). Unlike the classically large spleens of homozygous sickle mice, the spleens of AS mice were not heavier than those of AA mice (Table II). Adjusting for total body weight per mouse revealed a subtle increase in the liver weight of AS mice, but no other organ-specific weight differences.

Table II.

Organ weights

	AA (n = 12)	AS (n = 12)	p value
Organ weight (g)
Heart	0.13 ± 0.01	0.12 ± 0.01	> 0.99
Lung	0.14 ± 0.01	0.16 ± 0.01	> 0.99
Liver	1.17 ± 0.05	1.22 ± 0.06	0.87
Kidneys	0.31 ± 0.02	0.29 ± 0.01	0.25
Spleen	0.12 ± 0.01	0.14 ± 0.01	> 0.99
Organ weight (% of body weight)
Heart	0.43 ± 0.01	0.41 ± 0.01	> 0.99
Lung	0.49 ± 0.03	0.57 ± 0.03	0.89
Liver	3.95 ± 0.11	4.19 ± 0.12	0.01*
Kidneys	1.03 ± 0.04	1.02 ± 0.03	> 0.99
Spleen	0.42 ± 0.02	0.50 ± 0.03	0.84

^*significant after Sidak post-hoc correction

AA: mice that express all normal adult human haemoglobin in circulating erythrocytes; AS: mice displaying human sickle cell trait.

Representative sections of kidneys, livers, lungs, and spleens from AA, AS and AS-HR mice are shown in Fig. 2 (AA-HR mice did not differ from AA mice and are not shown). Vascular congestion of lung, liver and kidney tissue sections were scored (Table III). AA mice had minimal to no vascular congestion and normal histopathology both at baseline and after HR. Kidneys from AS mice revealed pronounced vascular congestion, especially in the area of the vasa recta with vessels noticeably engorged with RBCs; to a lesser extent, there were also some excess RBCs in the AS glomeruli. HR induced a trend toward increased apparent congestion in the renal medulla and glomeruli in AS mice (Table III). The percentage of positive kidney sections was significantly higher in AS mice after hypoxia compared to AA mice following hypoxia. In contrast to reports of infarcts, glomerular hypertrophy and increased iron deposition in SS kidneys (Manci et al, 2007), such findings were not observed in the kidneys from AS mice.

Figure 2. Microscopy of kidney, liver, lung and spleen from AA, AS and AS-HR mice.

Histological representative images are presented for AA, AS, and AS-HR mice. Kidneys of AS mice revealed increased vascular congestion in the vasa recta and glomeruli. χ² analysis revealed an effect of condition (genotype × hypoxia pairs) for the lung and liver vascular congestion scores, but no significant comparisons persisted after post hoc analyses. Scale bar: 100 µm. AA: mice that express all normal adult human haemoglobin in circulating erythrocytes; AS: mice displaying human sickle cell trait; HR: hypoxia and reoxygenation.

Table III.

Histological scoring of vascular congestion

		n	Median Score (95% CI)	Average score (± SD)	Per cent Positive (score > 1)	AA normoxia vs AA-HR	AS normoxia vs AA normoxia	AA-HR vs AS-HR	AS-normoxia vs AS-HR	Overall χ2
LUNG	AA normoxia	8	0 (0–2)	0.3 (± 0.8)	12.5%					0.04*
	AA-HR	7	0 (0–2)	0.3 (± 0.8)	14.3%	>0.05
	AS normoxia	6	1 (0–4)	1.4 (± 1.7)	50.0%		>0.05
	AS-HR	12	1.5 (0.5–4)	2.1 (± 1.6)	66.7%			>0.05	>0.05
LIVER	AA normoxia	8	0 (0–0)	0.0 (± 0.0)	0.0%					0.018*
	AA-HR	7	0 (0–0)	0.0 (± 0.0)	0.0%	>0.05
	AS normoxia	6	1 (0–3)	1.0 (± 1.1)	16.7%		>0.05
	AS-HR	12	1.5 (0–3)	1.6 (± 1.4)	50.0%			0.0436	>0.05
KIDNEY	AA normoxia	8	0 (0–0)	0.0 (± 0.0)	0.0%					<0.0001****
	AA-HR	7	0 (0–0)	0.0 (± 0.0)	0.0%	>0.05
	AS normoxia	6	2.75 (0–3)	2.1 (± 1.3)	66.7%		0.015
	AS-HR	12	2.75 (2.5–5)	3.3 (± 1.5)	91.7%			0.0002###	>0.05

^#represents significance following Bonferroni adjustment, with alpha adjusted to 0.0125.

Final column indicates p value of overall χ²per organ.

Preceding four columns show p value of post hoc Fisher’s exact test for indicated comparisons.

AA: mice that express all normal adult human haemoglobin in circulating erythrocytes; AS: mice displaying human sickle cell trait; CI: confidence interval; HR: hypoxia and reoxygenation; SD: standard deviation.

Livers of AS mice revealed moderate sinusoidal congestion, as well as some large-vessel congestion at baseline that subjectively appeared to increase following hypoxia and reoxygenation, though post hoc tests of liver vascular congestion scores did not reveal a significant difference comparing AS at normoxia with AS-HR specifically. Similar to AS kidneys, the AS livers showed otherwise normal morphology, including minimal evidence for ischaemia, infarcts and inflammation that is typical for SS mice (Manci et al, 2007).

Lungs from AS mice also showed only mild pathological changes that were primarily limited to increased vascular congestion of both larger vessels and microvascular lung lace compared to lungs from AA mice, which had minimal to no vascular congestion. Lung congestion in AS mice was mild at baseline with a trend toward increasing pathology after HR. Splenic architecture appeared normal, with organized white and red pulp. Together with normal spleen size, these data suggest only a minimal increase in splenic haematopoiesis, which is in alignment with clinical observations in human SCT.

Sickle cell trait (AS) mice display hypersensitivity to mechanical, cold and heat stimuli

As has been consistently reported (Kohli et al, 2010; Hillery et al, 2011; Cain et al, 2012; Cataldo et al, 2015), SS mice showed substantial mechanical hypersensitivity, as evidenced by decreased paw withdrawal thresholds compared to controls (Fig. 3A). While it has been assumed that sickle trait does not confer pain, we found that AS mice displayed an intermediate level of mechanical hypersensitivity: AS mice had decreased paw withdrawal thresholds compared to control mice, but remained significantly less sensitive than SS mice (Fig. 3A).

Figure 3. Pain behaviours in sickle cell trait mice.

Trait mice show hypersensitivity to both mechanical and thermal stimuli. (A) AS (n = 28) and SS (n = 28) mice had significantly reduced paw withdrawal thresholds compared to AA controls (n = 29), though AS mice were less sensitive than SS mice. (B) Cold aversion in AS and SS mice compared to controls (n = 13 per group) when test plate was set to 23°C. (C) Thermal preference in AA, AS and SS mice (n = 13 per group) when test plate was set to 20°C. ## P < 0.001, #### P < 0.0001 compared to 50% theoretical mean, one sample t test with Bonferroni correction. (D) AS mice (n = 26) and SS mice (n = 24) had decreased paw withdrawal latencies to a heat stimulus compared to AA controls (n = 23). (E) In AS mice, exposure to 2 h of hypoxia and 2 or 3 h of reoxygenation (2H/2R and 2H/3R, respectively) induced a significant reduction in paw withdrawal threshold compared to baseline (BL; § P < 0.05, §§ P < 0.01; n = 12). 2H/3R also reduced paw withdrawal thresholds in SS mice (§§§ P < 0.001; red symbols; n = 12). HR had no effect on withdrawal thresholds in AA mice (n = 12). For panels A, D, and E, the data for SS and AA mice (only) were previously reported (Hillery et al, 2011). Experiments of mechanical and heat sensitivity in normoxia were repeated a second time, and the results are combined here. * P < 0.05, ** P < 0.01, **** P < 0.0001. AA: mice that express all normal adult human haemoglobin in circulating erythrocytes; AS: mice displaying human sickle cell trait;SS: mice that express exclusively human sickle haemoglobin in circulating erythrocytes.

To measure cold pain and aversion, we used a temperature preference assay with floor plates set at 23°C and 30°C. As expected, sickle mice spent significantly less time on the colder (23°C) half of the testing apparatus than control mice, indicating cold aversion (Fig. 3B). As was also observed for mechanical thresholds, AS mice displayed an intermediate level of behavioural cold aversion (Fig. 3B). When animals were allowed to explore the chamber set at 20°C and 30°C, both AS and SS mice spent significantly less time on the cold plate than would be expected in the absence of any cold aversion (50% expected), while AA mice showed only a mild trend for cold aversion at 20°C (Fig. 3C). As such, there was no significant overall effect of genotype on time spend on the 20°C cold plate (p = 0.076). Therefore, testing at 20°C and 30°C (compared to 23°C and 30°C) reduced the discriminatory ability of this assay for these mice.

Both AS and SS mice showed decreased paw withdrawal latencies to a heat stimulus compared to AA controls (Fig. 3D). Again, AS mice had an intermediate heat hypersensitivity phenotype.

Mechanical hypersensitivity following hypoxia-reoxygenation

Just as HR increases pain behaviours in SS mice (Hillery et al, 2011), we observed a time-dependent exacerbation of mechanical hypersensitivity in AS mice following hypoxia and reoxygenation. By 2 h post-reoxygenation, AS mice were significantly more sensitive to mechanical probing of the hindpaw compared to their baseline and reached levels of hypersensitivity similar to the severe SS mouse phenotype (Fig. 3E).

Discussion

Whereas pain is a primary clinical feature of SCD, it is not normally thought to be associated with carrier status of SCT. Similarly, individuals with SCT are thought to be predominantly asymptomatic, portray largely normal haematological parameters and are assumed to be pain free. We sought to investigate the potential for haematological and pain phenotypes and their underlying mechanisms using a mouse model of SCT. Our current findings highlight that SCT mice do, in fact, display chronic hypersensitivity behaviour to mechanical, cold and heat cutaneous stimuli that suggests enhanced pain sensitivity in the SCT mice. Currently, there is very little known about either chronic pain or cutaneous hypersensitivity in individuals with SCT, and we believe our findings warrant further investigation into this possibility in human trait individuals. Further, our results suggest that inclusion of SCT individuals as “controls” in clinical studies of SCD (often involving siblings by either convenience or necessity of including age and race-matched controls) may dilute the effect sizes obtained, particularly when assessing severity of cutaneous sensitivity.

Interestingly, the clear mechanical and thermal hypersensitivity in the SCT mice persisted in the presence of only relatively minor changes in their haematological parameters. We found that AS mice display fairly minimal haematological or other differences compared to control mice. For example, AS mice demonstrated a mild increase in reticulocyte count and mild increase in LDH that may be due to a subclinical increase in haemolytic rate with reduced RBC lifespan. However, the levels of LDH and reticulocytes in both the AS and AA mice were still within the range of normal values expected for these mouse strains and in agreement with other recent data (Charrin et al, 2016). Though Charrin et al (2016) observed no differences between Townes AA and AS reticulocyte counts, they utilized a different method (manual counting after brilliant cresyl blue staining) and reported levels much lower than had been previously reported (Wu et al, 2006), both of which may contribute to the differences in our results compared to those reported by Charrin et al (2016).

In a small sample size, we found a trend towards increased adhesion of AS RBCs to immobilized thrombospondin, suggesting this may contribute to subtle abnormalities in SCT similar to the increased prothrombic tendency that has been recently reported (Folsom et al, 2015). The finding of only ~23% HbS in AS RBCs by HPLC is relatively low compared to humans with SCT, where HbS percentages more typically range ~25–40% (Head et al, 2004). Thus, the vascular phenotype in this AS mouse model may be less severe compared to humans with SCT.

Similar to the elevated WBC counts observed in SCD, we observed a modest increase in WBCs in the AS mice compared to AA controls, although the level of leucocytosis (11.2 ± 0.9 × 10⁹/l) was only borderline compared to the published upper normal level for the strains within the Berk mixed background. It has been hypothesized that both inflammation and oxidative injury contribute to sickle cell pain, though the effects of either can both directly and indirectly lead to sensory neuron sensitization. For this reason, the modest leucocytosis seen here may either directly contribute to (e.g., reactive oxygen species release from neutrophils) or indirectly reflect a mild increase in inflammation that may promote the increased pain behaviours observed in the AS mice.

The levels of ALT and LDH in plasma from AS mice were modestly elevated compared to AA mice. However, both the ALT and LDH were still within the normal range and not close to the marked elevations noted in SS mice (for ALT and LDH) and SCD humans (for LDH) (Kato et al, 2006). Given these subtle differences in ALT, LDH and reticulocytes, we note that using a mouse model that is more genetically homogenous than humans may have permitted us to discover subtle alterations in values that are still within the reported normal range. This may allow us to refine our efforts in studying human SCT. For example, pursuing the slightly shortened RBC survival and mild increase in tissue injury in human studies should further increase our knowledge of the long-term impact of sickle trait on human physiology.

Of particular interest, we found that AS mice display significant behavioural hypersensitivity, potentially reflecting enhanced baseline pain sensitivity even in the absence of hypoxia-reperfusion, as has been reported in hemizygous Berk mice (Kohli et al, 2010). As chronic pain is not typically considered a common outcome of SCT, these findings may inform future clinical follow-up or may alternatively indicate that even Berk AS mice generate an exaggerated sickle-like phenotype, perhaps owing in part to differential regulation of haemoglobin subunits on the transgene compared to undisrupted regulation in naïve mice. Of note, some similar behavioural phenotypes have been reported in heterozygous Townes AS mice, which display normal haemoglobin levels and haematocrits (Wang et al, 2016)(data not shown). Of note, there are subtle differences between the Berk SS and Townes SS mice, including their respective pain phenotypes (Lei et al, 2016). We would thus anticipate that Berk AS and Townes AS mice show similar mild differences in pain behaviours and could both contribute to our understanding of human SCT.

Overall, these findings indicate that a single copy of β^S, even when paired with conspecific β^A, is sufficient to drive some minor haematological and pathological changes, and is also sufficient to drive fairly robust behavioural pain phenotypes in SCT mouse models. In human SCT, there is good data for low grade renal papillary damage and a mild increase in risk for thrombosis and exertional rhabdomyolysis (Key et al, 2015; Nelson et al, 2016; Folsom et al, 2015; Tsaras et al, 2009). RBCs in both SCD and in SCT have altered mechanical properties and are less deformable in response to shear stress (Zheng et al, 2015), resulting in increased fragility. Any of the described mouse and human SCT phenotypes must ultimately be derived from the genetic co inheritance of sickle and normal haemoglobins expressed in circulating RBCs. It is possible that the subtle biomechanical changes of the sickle trait RBC contribute to a shortened RBC survival, activation of the coagulation and inflammatory systems, and changes in endothelial, organ and neuronal biology in SCT. In particular, each of these findings may contribute to the generation and maintenance of pain in SCT. Overall, we have described a novel phenotype of pain that has not yet been well characterized in human SCT, and our findings strongly support further assessment of pain and cutaneous sensitization in individuals with SCT alongside concurrent studies in mouse models of SCT.