Arterial Calcification In Diabetes: Preclinical Models And Translational Implications

Abstract

Diabetes increasingly afflicts our aging and dysmetabolic population. Type 2 diabetes (T2D) and the antecedent metabolic syndrome (MetS) represent the vast majority of the disease burden –increasingly prevalent in children as well as older adults. However, type 1 diabetes (T1D) is also advancing in preadolescent children. As such, a crushing wave of cardiometabolic disease burden now faces our society. Arteriosclerotic calcification is increased in MetS, T2D, and T1D – impairing conduit vessel compliance and function, thereby increasing the risk for dementia, stroke, heart attack, limb ischemia, renal insufficiency and lower extremity amputation. Preclinical models of these dysmetabolic settings have provided insights into the pathobiology of arterial calcification. Osteochondrogenic morphogens in the BMP-Wnt signaling relay and transcriptional regulatory programs driven by Msx and Runx gene families are entrained to innate immune responses – responses activated by the dysmetabolic state – to direct arterial matrix deposition and mineralization. Recent studies implicate the endothelial-mesenchymal transition (EndMT) in contributing to the phenotypic drift of mineralizing vascular progenitors. In this brief overview, we discuss preclinical disease models that provide mechanistic insights – and point to the emerging potential for translating these insights into new therapeutic strategies for our patients challenged with diabetes and its arteriosclerotic complications.

I. Introduction

In westernized societies, the arterial macrovasculature is the second most extensively calcified vertebrate tissue beyond the skeleton^{1, 2}. In response to diabetes, advanced age, dyslipidemia, uremia, hypertension – and certain primary inflammatory disorders – the arterial vasculature accrues a substantial mineral load, impairing valve and arterial compliance necessary for smooth distal tissue perfusion^{1, 2}. Once considered only a passive process of dead and dying cells, arterial calcification has now emerged as an actively regulated form of tissue biomineralization. Primary and/or secondary vascular inflammatory signals are common to all type of arterial mineral deposition, evident in both genetic and metabolic causes of disease^{1, 2}. Reflecting upon this, Demer and colleagues were the first to note that this likely represents an important role for tissue calcification in certain forms of innate immunity – such as the calcified granuloma that wall off mycobacterial or fungal infections in the lung². However, arterial mineralization has devastating consequences, increasing myocardial workload, decreasing diastolic perfusion, and compromising perfusion thereby increasing risk of stroke, dementia, myocardial infarction, heart failure, renal failure, and lower extremity amputation^1–9. Currently, no effective pharmacotherapies exist to reduce arterial calcification and restore conduit vessel compliance – although useful insights may be forthcoming from the biology of progeria^{10, 11}. Robust preclinical models of disease are necessary to better understand disease biology and validate and vet potential therapeutic approaches that might mitigate disease in our increasingly aging and dysmetabolic population.

Type 2 diabetes (T2D), Type 1 diabetes (T1D), and the metabolic syndrome (MetS) have emerged as particularly important causes of arterial calcification^12–14 (Figure 1). Indeed, at every level of declining renal function – another key contributor to vascular mineral load³ – glycemic control augments the risk for the severity and extent of arterial calcification¹⁵. In this overview, we consider the preclinical disease models of T1D and the T2D-MetS spectrum that help inform our understanding of vascular mineral metabolism in these clinical settings. Strengths, mechanistic insights, and potential shortcomings of the models are briefly discussed to help frame future studies in ways that might lead to new therapeutic approaches for our patients afflicted with arteriosclerotic disease.

Figure 1. Overlapping yet distinct metabolic milieus of T1D and the T2D-MetS spectrum shape cardiovascular disease.

T1D and T2D share many cardiometabolic risk factors – including elevated levels of multiple ligands^169–174 capable of activating the RAGE/AGER and toll like receptors (TLRs) of the innate immune system that can drive arterial calcification^{100, 101, 175} along with prototypic inflammatory cytokines (TNF, IL1beta, etc.)^{70, 90, 176, 177}. Note that expanded visceral and perivascular (aortic, coronary) fat depots^{1, 178–182} and hepatic steatosis^183–186 – harbingers of valve and vascular calcification – are key features of the “diabesity” phenotype along the T2D-MetS spectrum^{12, 187–189}. T2D also impairs reverse cholesterol transport¹⁹⁰, and VLDL metabolism is an independent contributor to arterial calcification risk¹⁹¹. AGE, advanced glycation end product; BMP, bone morphogenetic protein; BP, blood pressure; HMGB1, high mobility group protein B1; oxLDL, oxidized LDL cholesterol.

II. Rodent Models Of Arterial Calcification In Type I Diabetes

While much less common than T2D, the cardiovascular consequences of T1D are protean, and most definitely encompass arterial calcification^16–18. It’s interesting to note, however, that when insulin resistance¹⁹ and hypertension²⁰ are quantified in the setting of T1D that the severity and extent of coronary artery calcification (CAC) is in fact further increased¹⁹. This points to common cardiometabolic underpinnings of arteriosclerotic calcification in T1D and T2D. In this section, we consider the better-studied preclinical models of arteriosclerotic calcification and vascular stiffening in T1D.

II.1. Streptozotocin models

Streptozotocin (STZ), the β-cell toxin, is the most frequently utilized method for inducing T1D in rodent models(Table 1). Its relative tissue selectivity relates to uptake by Glut2 with subsequent DNA alkylation, poly ADP-ribosylation, and depletion of cellular NAD and ATP²¹. While known for decades, only recently has the capacity of STZ-induced T1D been utilized in detailed studies of arterial calcification and vascular stiffness. Chen and colleagues established that the 5 day low-dose STZ model is sufficient to induce diabetes, upregulate arterial Akt-Runx2 osteochondrogenic signaling, enhance arterial mineral deposition, and increase arterial stiffness in C57BL/6 mice²². Additional studies comparing the arterial mineralized matrix ultrastructure in this model vs. other diabetes models promises to be enlightening. It’s important to note that cardiovascular sclerosis often relates more to changes in Runx2 and osteogenic protein accumulation that mRNA accumulation²²; indeed, both Runx2^23–25 and type I collagen biogenesis²⁶ are prodigiously regulated at the post-transcriptional level. Using a rat model, Moreau and colleagues^{27, 28} implemented a combination of STZ with warfarin – the latter an inhibitor of vitamin K – dependent protein γ-carboxylation that impairs the anti-calcific actions of matrix Gla protein (MGP) partially mediated by BMP2/4 inhibition^{29, 30}. In both models, dysglycemic components common to the clinical setting are important, with intracellular (non-enzymatic protein glycation and enzymatic O-GlcNacylation) and extracellular (advanced glycation end product / AGE formation and inflammatory cytokine) signals contributing to calcific vascular disease. Osteochondrogenic differentiation programs dependent upon the transcription factor Runx2^{31, 32} are recruited to direct VSM mineralization. Importantly, protein-protein interactions between Runx2 and BMP-regulated osteogenic Smads figure prominently in orthotopic mineralization as well³³, dependent upon the stable sub-nuclear targeting of Runx2 complexes in the nuclear matrix³⁴. Whether sub-nuclear targeting and turnover of Runx2 in VSM might offer therapeutic approaches has yet to be determined.

Table 1.

Preclinical Models of Arterial Calcification in T1D and the T2D-MetS Spectrum

Model	Comments	References
C57BL/6J Mouse with Streptozotocin (STZ) Treatment	T1D. Aortic calcification and stiffness both quantified and increased. Osteogenic programs identified in other models activated. Comparison of mineralized vascular matrix ultrastructure with other models promises to be enlightening.	Ref. 22
Wistar rat with STZ Treatment + Co-inducer - Warfarin+vitamin K (WVK) - Vitamin D and nicotine (VDN)	T1D. Femoral artery calcification noted as well as aortic calcification. Peripheral artery calcification often assessed in preclinical models but relevant to peripheral arterial disease. Disease responsive to modulation of RAGE/AGER and NADPH oxidase / Nox signaling.	Ref. 27, 28, 39, 40, 196
ApoE−/− mouse with STZ treatment	T1D. Severe hypercholesterolemia Robust arterial calcification responses are observed. ApoE-null VSM cells are given to chondroid metaplasia. S100A12 transgene accelerates calcification in ApoE-null mice.	Ref. 45, 47, 100
Ins2Akita/+ Mouse	T1D. Streptozotocin administration is not required to induce arterial calcification in this T1D model. Aortic calcification increased, stiffness to be characterized. Neuropathy is also induced. On LDLR−/− background diet-induced diabetes, hypercholesterolemia, & hypertriglyceridemia severe in both male & female animals, but arteriosclerosis has yet to be quantified in Ins2Akita/+;LDLR−/− mice.	Ref.30, 66, 197
Male LDLR−/− Mouse with STZ Treatment	T1D. Atherosclerosis lesion more extensive after 2 weeks of diabetes. Aortic calcification and stiffness remain to be characterized.	Ref.198, 199
Golden Syrian Hamster with STZ + High Fat Diet (HFD)	T1D. Nephropathy also arises in this model. Few publications, little physiological phenotypic data available.	Ref.200, 201
Male LDLR−/− Mouse Fed High Fat Diet (HFD)	T2D – MetS spectrum. Severe hypercholesterolemia, modest fasting hypertriglyceridemia (~200 mg/dL), elevated free fatty acids, increased visceral and periaortic fat accumulation, hyperglycemia & hyperinsulinemia, increased HOMA-IR, and oxidative stress signaling regulated by OPN and Nox activity. Induced by high fat Western diet (Teklad 88137; 42% calories from fat). Aortic calcification and stiffness both quantified and increased. Regulated by canononical and noncanonical Wnt signals. Disease is induced in response to clinically relevant stimulus that promote VSM chondroid metaplasia and RAGE/AGER co-localization. Chondroid metaplasia is restrained by VSM OPN. Renal fibrosis induced via oxidized LDL signals. Modifiable with 5/6 nephrectomy to phenocopy arteriosclerotic disease in setting of T2D + renal insufficiency (chronic kidney disease – mineral and bone disorder, CKD-MBD). Latter is sensitive to renal and dietary phosphate metabolism. S100A/calgranulin proteins, key RAGE/AGER ligands that drive ectopic calcification, are increased by high fat diet.	Ref. 42, 67–69, 71, 72, 86, 89, 100, 101, 113, 121, 137, 202–207
LDLR−/−;ApoB100/100;RIP1-IGF2Tg mouse	T2D-MetS spectrum. Similar to male LDLR−/− but more significant fasting hyperglycemia in both male and female mice, Western diet challenge performed in old animals. Severe hypercholesterolemia arises since all ApoB containing lipoproteins are now B100 only – and the LDLR is missing. Transgenic IGF2 expression in pancreatic beta cell from rat insulin promoter (RIP1) accelerates T2D with beta-cell failure upon dietary challenge. Extent of aortic calcification appears to track insulin resistance > hypercholesterolemia, fasting hyperglycemia.	Ref.107, 112
db/db Mouse	T2D-MetS spectrum. Modest hypercholesterolemia, severe hypertriglyceridemia and visceral fat accumulation on Teklad rodent diet 8604 (14% calories from fat). In addition to nephropathy these leptin receptor –deficient mice also develop neuropathy (either on the C57BKS background, or on C57BL/6J with high fat diet to sustain hyperglycemia). Intriguing recent report of arterial stiffening without calcification in aged diabetic db/db mouse; genetic dietary conditions, composition, and basic metabolic profiling important to report to enable comparisons.	Ref.30, 63, 64, 126
Ossabaw Island Hog	T2D – MetS spectrum. Modest hypercholesterolemia, hypertriglyceridemia, increased visceral fat accumulation, impaired glucose tolerance, and insulin-resistant. Early stages of coronary artery microcalcification by histology co-registered with 18F - PET/CT and 41Ca uptake. Epicardial fat surgically demonstrated to impact coronary disease processes.	Ref. 130, 131, 208–211
Male Rhesus Macaque on High-Fat High-Sucrose Diet	T2D – MetS spectrum. Modest hypercholesterolemia, hypertriglyceridemia, increased visceral fat, hyperglycemia & hyperinsulinemia, increased HOMA-IR. Arterial calcification and stiffness are both increased, but arterial calcification assessed only histologically. Resveratrol supplementation over 2 years protective at doses of 80 mg to 480 mg daily (glass of red wine contains ~ 1 mg of resveratrol).	Ref. 134
Watanabe (WHHL) Rabbit Fed High-Fructose High-Fat Diet	T2D – MetS spectrum. Severe hypercholesterolemia   (in-frame LDLR deletion). Metabolic syndrome features similar to the male LDLR−/− model, e.g. hypertriglyceridemia, increased visceral fat, impaired glucose tolerance / increased HOMA-IR. While calcification is histologically increased, not rigorously quantified.	Ref. 212

Of note, as a DNA alkylating agent, STZ chemically induces the DNA damage response in multiple cell and tissue types³⁵. In both osteoblasts and vascular smooth muscle, DNA damage is a potent stimulus for the osteochondrogenic program when elicited by either chemical means, nuclear senescence³⁶, or irradiation^{37, 38}. Thus, while not perfectly mimicking the pathogenesis of human T1D and its complications, the STZ models can rapidly elicit vascular injury and biomineralization responses that reproducibly recapitulate key features of human disease. The cellular toxicities of STZ above and beyond β-cell death may contribute to the vascular disease process in these important preclinical models, not unlike the addition of nicotine to rat model of vitamin D – induced vasculopathy discussed below.

The STZ-treated Wistar rat has been used for studies of arterial calcification that arises in T1D. In this model, calcific vasculopathy is accentuated by either (a) warfarin treatment with vitamin K-mediated, hepatocyte-selective rescue of coagulation factor γ-carboxylation (WVK)²⁷; or (b) vitamin D3 intoxication with nicotine (VDN) administration^{39, 40} (Table 1). Elastinolysis is a prominent histopathologic contributor to the accentuated vascular toxicity of these secondary treatments that drive arterial mineralization with STZ. Many of the same signaling cascades observed in the STZ+VDN model – including hyperphosphatemia-driven by vitamin D3 intoxication, nicotine-induced elastin fragmentation, and advanced glycation end product receptor (RAGE/AGER) signaling – are observed in other preclinical arterial calcification models that more closely reflect common clinical settings^{1, 41, 42}. Pioglitazone-mediated activation of PPAR-ϒ, an inhibitor of non-canonical Wnt signaling⁴³, reduces arteriosclerotic calcification in VDN models⁴⁴. In the STZ-WVK model, RAGE/AGER signaling is also increased and very important; inhibition of AGE ligand formation with pyridoxamine or AGE cross-link destruction with alagebrium inhibited femoral artery calcification in the STZ-WVK model (primarily medial calcification)²⁸. The adventitial inflammation including TNF expression²⁷ – so characteristic of human diabetic macrovascular disease⁴¹ – is also observed in STZ models²⁷. The rapidity with which robust disease arises with STZ-induced T1D and its co-modifiers has lead to its increasing use in preclinical models of arteriosclerotic calcification. The combination of STZ and high-fat diet induced hypercholesterolemia in the apolipoprotein E −/− (ApoE−/−) mouse appears to be particularly effective in inducing arterial calcification⁴⁵, accentuating the propensity for ApoE-null VSM to undergo chondroid metaplasia^{46, 47} (Table 1).

II.2. The Ins2^Akita Mouse

Recently, Bostrom, Yao, and colleagues⁴⁸ addressed one of the potential shortcomings of the STZ mouse model by characterizing arterial mineralization in the Akita mouse (Ins2^Akita heterozygous mice) ⁴⁸. Due to an insulin codon mutation that impairs secretion and induces β-cell endoplasmic reticulum stress, the Akita mouse develops an age-dependent evolution of T1D. In this model, arterial calcium deposition is increased with time, driven by BMP2/4 –Smad signal relays – and mediated part via endothelial dysfunction that promote the endothelial-mesenchymal transition (EndMT)^{30, 48}. In this process, endothelial cells (EC) lose their normal differentiated EC barrier phenotype, and adopt a mesenchymal phenotype capable of osteogenic differentiation with mineralized matrix deposition. EndMT processes have emerged as one potential source of arterial osteogenic cells in the diseased vasculature⁴⁸ as well as in heterotopic bone formation⁴⁹. However, the commonly used “endothelial-specific” transgenic Cre lines used to track the EC lineage also express within myeloid lineage – cells can adopt the EC fate⁵⁰ and contribute to arterial calcification in T2D^{51, 52}. Moreover, co-culture of endothelial progenitors with mesenchymal progenitors have pointed to paracrine interactions between multipotent Sox2-positive populations in both lineages that promote osteogenic mineralization⁵³. Thus, a mechanistically more important consequence of EC de-differentiation during the EndMT likely relates to alterations in paracrine endothelial –mesenchymal signals that program an osteogenic fate of adjacent VSM and mesenchymal progenitors.⁵⁴ Nevertheless, lineage tracing studies suggest that the sources of osteogenic cells in this T1D model quantitatively differ from those of pure atherosclerotic models such as the ApoE – null mouse; in the latter, approximately 80% of vascular osteogenic cells appear to arise from the vascular smooth muscle (VSM) lineage^{55, 56}. The Akita mouse model has also revealed a critical role of Sox2-regulated serine proteases in the EndoMT⁴⁸. The latter has important translational implications, since recombinant serine protease inhibitors (serpins) have been successfully deployed in other clinical settings to ameliorate disease⁵⁷. While predicted to be altered, the arterial compliance and conduit vessel function of the Akita mouse has not been characterized (Table 1).

Of note, neuropathy is emerging as once key contributor to arterial calcification in humans⁵⁸. Indeed, when neurectomy had been utilized in the past to address intractable ischemia rest pain in individuals afflicted with peripheral arterial disease (PAD), profound ipsilateral medial artery calcification blossoms in those so treated⁵⁹. Intriguingly, models of heterotopic ossification have suggested neural progenitors might be capable of adopting osteogenic fates in response to injury⁶⁰ – and, as such, resemble the mineralizing phenotypic plasticity of Sox2-positive neural crest cells^{61, 62}. Because of its broader general toxicity, higher-dose STZ administration is effective in simultaneously inducing neuropathy⁶³. However, the less-toxic, low-dose regimens that are more β-cell selective do not appear to induce neuropathic disease⁶⁴. Neurogenic contributions to ectopic mineralization in other settings are just now being investigated^{60, 65}, and this is an important pathogenic feature to consider moving forward. Importantly, the Akita mouse develops impairment in nerve conduction velocity (NCV)⁶³ as well as abnormal parasympathetic control of heart rate and sinoatrial node function⁶⁶.

III. Preclinical Models Of Arterial Calcification In The Setting Of Type II Diabetes and the Metabolic Syndrome

Obesity is epidemic in westernized societies, driving a crushing wave of metabolic syndrome (MetS) and type II diabetes (T2D)⁴. Preclinical model of diet-induced obesity, insulin resistant diabetes, and arteriosclerotic disease^67–71 predicted the relationships between MetS, T2D, and valve and vascular calcification as subsequently established in MESA (Multiethnic Study of Atherosclerosis)^{12, 13}. Numerous preclinical studies detail arterial atherosis – primarily lipid-laden lesion area. However, by comparison, relatively little attention has been given to quantifying the sclerotic component – extracellular matrix metabolism, calcification, and stiffness - of cardiovascular disease in MetS and T2D models. In this section, we focus upon those models where sclerosis – calcification, fibrosis, and/or vascular stiffening – has been characterized as related to conduit artery disease and dysfunction (Figure 2) in this clinically important setting.

Figure 2. Consequences of arterial stiffening and impaired Windkessel physiology.

During systole, some kinetic energy is stored as potential energy in elastic conduit arteries. This stored energy permits smooth distal tissue perfusion during diastole (blue tracing). With arteriosclerotic stiffening (red tracing), less potential energy is stored during systole, giving rise to more erratic and pulsatile flow during diastole. Diabetes and aging significantly increase arteriosclerotic conduit vessel stiffening¹⁹². Reproduced from reference¹ with permission. See also references^{193, 194,158, 195} for detailed discussion.

III.1. Male LDLR−/− mice fed high fat diets

When fed high fat diets characteristic of westernized societies, the male LDLR−/− mouse develops hypercholesterolemia, hypertriglyceridemia, and insulin-resistant diabetes with profound accumulation of visceral fat^{67, 68, 71}. LeBoeuf and colleagues⁷² identified that the weight gain and hyper-leptinemia of LDLR-deficient mice was greater than that of ApoE-deficient mice, tracking the evolution of insulin-resistant diabetes in the former and not the latter. As LeBoeuf highlights, male LDLR−/− mice – particularly on the C57Bl/6 background⁷³ - are relatively susceptible to diet-induced obesity⁷³. Male LDLR−/− mice also develop more severe hypertriglyceridemia⁷⁴ and hepatic steatosis when challenged with high fat diets⁷⁵; whether these differences^72–74 are responsible for gender-specific responses to atheroma modulation by PPAR-gamma agonists remains to be determined⁷⁴. The enhanced myeloid adipo-inflammatory response of male mice to high fat diet feeding may be responsible, and these sexually dimorphic metabolic responses have been recently reviewed in detail⁷⁶.

Of note, in the absence of hyperglycemia, androgen treatment increases arterial calcification in both male and female ApoE-null mice⁷⁷ – and conditional deletion of the androgen receptor reduces VSM calcification in vitro⁷⁸. However, the effects of androgens on arterial calcification differ between the aortic sinus (unresponsive) and the innominate artery (responsive/worsened)⁷⁷. Moreover, differences between male and female ApoE-null mice with respect to arterial calcification have been observed following genetic challenge⁷⁹. While osteopontin (OPN)-deficiency decreases atheroma size (atherosis) in female ApoE−/− mice with minimal calcification, OPN-deficiency triples lesion calcification area (sclerosis) in male ApoE−/− mice⁷⁹. We have noted that following high fat diet challenge, male LDLR−/− mice develop more significant arterial calcification and lipofuscin accumulation (unpublished). Likewise, male mice transgenic for VSM TNAP (bone alkaline phosphatase) are also more susceptible to arterial calcification and early lethality as compared to their female TNAP transgenic siblings⁸⁰.

In the male LDLR−/− model, high fat diets (HFD) upregulate aortic expression of Msx1 and Msx2⁶⁷, two osteogenic transcription factors important in mineralization of the craniofacial skeleton^{81, 82}. Bone morphogenetic protein 2 (BMP2) and osteopontin (OPN) are also upregulated by diet-induced disease⁶⁷, reflecting contributions first identified in human calcified vessels⁸³ and valves⁸⁴. The pivotal role of these osteogenic programs in the arteriosclerotic calcification arising in diet-induced diabetes was recently demonstrated. Deletion of Msx1 and Msx2 within the VSM lineage (SM22-Cre transgene⁸⁵) reduced arterial calcification and vascular stiffness without altering diet-induced obesity, diabetes, or dyslipidemia in the LDLR-deficient mouse⁸⁶. Importantly, pro-calcific canonical and noncanonical Wnt signaling cascades entrained to Msx2⁶⁹ were also reduced in VSM with Msx gene deletion or knockdown⁸⁶. Since Msx1 and Msx2 are genomic targets and secondary mediators of BMP2⁸⁷, their actions may serve to link paracrine BMP – and Wnt- dependent osteochondrogenic programming of mesenchymal progenitors in the diseased vessel wall⁸⁸ as well as in the craniofacial skeleton⁸⁷.

Additional studies in LDLR−/− mouse model have revealed that inflammatory cues, provided in part by the prototypic inflammatory cytokine TNF^{70, 89, 90}, are proximal signals in the initiation of diabetic arteriosclerosis. TNF-dependent mural (intima-media and adventitial) inflammation upregulates aortic BMP2, Msx2, OPN, and Wnt signaling programs in vivo, and VSM-specific augmentation of TNF expression is sufficient to mimic early phases of diet-induced disease^{70, 90}. However, other inflammatory signals initiated by (a) IL1β^{90, 91}; (b) agonists of the overlapping yet distinct damage/danger – and pathogen- associated molecular pathogen pathways⁹² (viz. S100A/calgranulin family⁹³, HMGB1, AGEs, oxLDL); and (c) hyperphosphatemia are also capable of driving these programs^94–97. As Demer, Tintut and colleagues have pointed out², oxidatively damaged molecules such as oxidized LDL⁹² (oxLDL) likely mimic the foreign lipidaceous signatures of mycobacterial and fungal pathogens that are controlled in part via sclerotic innate immune responses that contain organisms in calcified granuloma. When activated in the elastin-rich arterial wall with diabetes and dyslipidemia, these sclerotic responses are maladaptive for conduit vessel functions.

Importantly, diabetes – with our without uremia – increases the expression of S100A12⁹⁸ (a.k.a. EN-RAGE), a powerful marker of cardiovascular risk in humans⁹⁹. The relevance of these findings with respect to diabetic arterial calcification becomes readily apparent when considering the significant impact of altering S100A12 signaling tone in atherosclerosis models independent of glucose homeostasis. In a series of elegant studies by Hofmann Bowman and colleagues, a “humanized” transgenic mouse was created wherein VSM expression of S100A12 was shown to augment arterial calcification in atherosclerotic ApoE-null background¹⁰⁰ and in mice with renal failure due to ureteral ligation¹⁰¹. Importantly, administration of paquinimod, a S100A neutralizing antagonist¹⁰², mitigated arterial calcification in the S100A12Tg;ApoE−/− mouse¹⁰³. More recently, the Hofmann Bowman lab has created an hBAC transgenic line for the entire human S100A gene cluster (S100A8/S100A9/S100A12); myeloid expression of the hS100A cluster accentuated valve and vascular calcification in the ureteral ligation model of CKD¹⁰⁴. Since S100A12 levels strongly correlate with cardiovascular outcome in CKD – outperforming soluble RAGE prognostication^{105, 106} – strategies targeting S100A signaling in diabetic arteriosclerosis promise to be extremely fruitful (Figure 1).

In 2007, Heinonen et al¹⁰⁷ interbred ApoB100-only;LDLR−/− mice¹⁰⁸ and the rat insulin promoter – IGF2 transgenic mouse¹⁰⁹ to create a new model of arterial calcification in the setting of T2D (Table 1). The IGF2 transgene engenders the beta-cell ER stress response and functional decline that phenocopies an accelerated T2D progression¹¹⁰. This model may prove useful for studies of impaired ventricular-vascular coupling¹¹¹ in diabetic arteriosclerosis¹¹² – similar to the S100A8/S100A9/S100A12 transgenic mouse model of Hofmann Bowman¹⁰⁴.

Hruska and colleagues significantly advanced our understanding of arterial calcification of chronic kidney disease – mineral and bone disorder (CKD-MBD) by introducing 5/6 nephrectomy onto the LDLR−/− model^{113, 114} (Table 1). CKD-MBD denotes the global perturbations in mineral metabolism that convey cardiovascular risk and skeletal fragility¹¹⁵, most frequently arising in the setting of the T2D-MetS spectrum. They demonstrated that low turnover bone disease arises along with arterial calcification in the nephrectomized LDLR−/− model – and that stimulating bone formation with the unique BMP7 ligand mitigated both bone disease and vascular calcification^{113, 114}. This is highly relevant to the human setting; London and colleagues demonstrated that low turnover bone disease arising from either parathyroid hormone (PTH) resistance or deficiency correlates with vascular calcium load¹¹⁶ and arterial stiffness or PAD¹¹⁷ in CKD-MBD. Serum phosphate levels at any level of renal function convey cardiovascular risk, and exert pro-inflammatory actions upon VSM that promote vesicle-mediated vascular mineralization^{118, 119}. It has been proposed that the skeleton represent an important “buffer” for phosphate in CKD¹²⁰, and that orthotopic osteoblast –mediated calcium phosphate (as hydroxyapatite) deposition in bone may be responsible for the vascular benefits of normal bone formation. As compared to ApoE-null models, this model has relatively modest arterial calcium loads. As such, it is highly sensitive to physiological changes in serum phosphate¹²¹, and even modest drug-related improvements in renal function might improve vasculopathy. However, both BMP7 and PTH have direct salutary actions on the VSM^{68, 69, 71, 122}. Moreover, T2D reduces blood flow to bone¹²³ via a process regulated in part by PTH signaling at the principal nutrient artery¹²⁴. The relative contributions of skeletal vs. vascular osteotropic hormone signals with respect to cardiovascular health have yet to be elucidated.

III.2. The male db/db mouse

Bostrom, Yao, and colleagues demonstrated that the db/db mouse model of T2D also accrues arterial calcium with progression of obesity and diabetes arising from leptin resistance³⁰. They demonstrated that activation of the BMP2/4 signaling cascade was a key pathogenic feature of arteriosclerotic calcification³⁰ as in the LDLR−/− mouse⁸⁸. The db/db mouse has an additional advantage in that progressive nephropathy¹²⁵ and neuropathy⁶⁴, contributing features to diabetic arteriosclerosis in both T1D and T2D (Table 1). Dependent upon the strain background (BKS vs. B6), high fat diet feeding may be required to sustain hyperglycemia, however⁶⁴. Using this model, studies from Tsao and colleagues¹²⁶ have highlighted that Runx2-dependent arterial fibrosis and stiffening can progress somewhat independent of calcium deposition. Although the dietary intervention was not specified, this finding has important implications. Extracellular matrix stiffness is a “mechanocrine” regulator of osteogenic potential of mesenchymal progenitors, with a ~25 kPa stiffness optimal for osteogenic differentiation¹²⁷; thus, a stiffening matrix environment arising from non-enzymatic glycated cross-linking, elastolysis, or fibrosis may impact subsequent osteogenic calcium deposition. Since calcium phosphate is proinflammatory^128,52, feed-forward mechanisms are likely to offer strategies that differentially target disease during initiation and progression. Since dysglycemia is present for an estimated 6 years prior to T2D diagnosis in humans¹²⁹, time-dependent vascular matrix remodeling may be rate-limiting in strategies that prove successful in reversing arteriosclerotic calcification and vascular stiffness in patients with T2D.

III.3. Insights from larger animal models of the T2D-MetS spectrum

A variety of large animal models have been recently implemented to study the T2D-MetS spectrum and its relationship to arteriosclerotic calcification¹³⁰. Studies using the Ossabaw Island Hog (Table 1) have permitted characterization of calcium – rich diets on arteriosclerotic calcification, and demonstrated that deposition is not significantly influenced by calcium intake in the absence of renal dysfunction¹³¹; whether this differs in the setting of CKD¹³² remains to be assessed. A high fructose high salt diet induces arteriosclerotic calcification with insulin resistance and elevated oxLDL in the LDLR(R94/C94) Heterozygous Hog, but the model has yet to be completely characterized¹³³. One of the more interesting preclinical models recently reported is the male rhesus macaque fed a high fat high sucrose diet¹³⁴. In this nonhuman primate model, histological evidence of arteriosclerotic calcification was co-registered with non-invasive assessment of vascular stiffness by Doppler-assessed carotid-femoral pulse wave velocity¹³⁴. It’s interesting to note that β-catenin and TGM2 (transglutaminase 2, a ligand for activation of Wnt signaling cascades in VSM^{135, 136}), USF1 (a novel mediator of noncanonical Wnt signaling during VSM calcification¹³⁷), and Gas6 (a vitamin K –dependent γ-carboxylated protein that inhibits osteogenic differentiation of vascular pericytes^{138, 139}) were all upregulated with diet-induced disease, and downregulated with restoration of vascular compliance with high dose resveratrol¹³⁴. Since fructose derived from diet or dietary sucrose metabolism induces insulin resistance with de novo lipogenesis in multiple species – including humans¹⁴⁰ – findings from such diet-induced disease models promise to faithfully recapitulate the pathogenic features of arteriosclerotic calcification most important to address in therapeutic translation to patients afflicted with T2D or MetS.

IV. Translational Implications and Future Directions

In this overview, we’ve very briefly touched upon the preclinical models of arteriosclerosis that are beginning to inform the metabolic origins of arterial calcification in the setting of diabetes (Table 1). As of yet, no single model has emerged as perfectly recapitulating all aspects of human disease in clinically relevant settings. This arises, in part, because of (a) the multifactorial contributions – hyperglycemia, insulin resistance, hypertension, dyslipidemia, neuropathy – to arterial disease biology (Figure 1); (b) an imperfect understanding of bone-vascular interactions in cardiometabolic disease; and (c) the absence of therapeutic tools validated in preclinical models that have been successfully assessed in humans with robust clinical and/or physiological phenotypic responses. These limitations, in many ways, mirror obstacles that were eventually overcome during the nearly century-long process required to prove Anitschkow’s lipid hypothesis¹⁴¹

To date, the only drugs clearly impacting arterial calcification risk have been aminobisphosphonates¹⁴², phosphate binders (calcium vs. non-calcium based^143–145), and statins¹⁴⁶ – and with impact highly dependent upon subject age, metabolic milieu, and duration of treatment. For example, in MESA (Multiethnic Study of Atherosclerosis), aminobisphosphonate use was associated with reduced vascular calcification in elderly woman – but increased vascular calcium load in younger women¹⁴². However, in the setting of metabolic pyrophosphate deficiency, bisphosphonates can inhibit vascular mineralization^{147, 148} and prolong life. Statins – like bisphosphonates – inhibit bone resorbing osteoclast-like cell function¹⁴⁹ while additionally limiting inflammatory osteogenic programs in VSM¹⁵⁰. The former may explain why longer-term statin use (healthy survivors) may accrue increased coronary calcification with time¹⁴⁶, while the latter has been used to argue for early intervention in preclinical porcine disease models¹⁵¹. Thus, physiological context and metabolic milieu significantly impacts the net impact of any intervention that directly or indirectly targets vascular mineralization. The only randomized trial of bisphosphonate treatment in arterial calcification was very small, very short (18 months), and emphasized CKD3-CKD4¹⁵²; vascular calcium load was unchanged and only a non-significant trend (p = 0.07) for improved arterial compliance was observed. Notably, none of the clinical data currently available for consideration prospectively interrogate the impact of intervention as a function of T1D, T2D, MetS contributions, glycemic control and duration, renal function, baseline arterial disease, arterial anatomic venue, or gender. The recent NIH-AARP Diet and Health Study reinforces the lesson that gender-specific outcomes must be considered when assessing the cardiovascular consequences of manipulating calcium metabolism¹⁵³ – even if only by dietary supplements. The elegant genetic studies of St. Hilaire and colleagues teach us that lower extremity peripheral artery calcification is genetically distinct from coronary, aortic, and valvular calcification¹⁵⁴. Additionally, it has become abundantly clear that advanced aortic valve calcification, whether arising in the presence or absence of diabetes, is physiologically and pharmacologically unique^{155, 156}.

Moving forward, it will be important to systematically phenotype (i.e., in both preclinical and clinical models) not only conduit vessel compliance but also end-organ perfusion and physiology¹⁵⁷ –- an integrated function of the arterial Windkessel (Figure 2), vessel patency, resistance arteriole neuroregulation, microvascular integrity and density, and efferent vessel anatomy^{7, 121, 158, 159}. It is reassuring that many of the molecular signatures identified as mechanistically important in pre-clinical models of arteriosclerosis are also manifest in humans^{83, 160–164}. The development of preclinical models that are “humanized” for clinically relevant molecular mediators, dietary compositions, and metabolic milieus have contributed to this important but small success. Furthermore, the capacity of calcium-independent phosphate binders in reducing serum phosphate and limit arterial calcification while maintaining PTH levels in both preclinical and clinical CKD-MBD settings^{132, 165} will very likely extend to diabetic patients with less severe¹⁶⁶ or no¹⁶⁷ renal insufficiency; this should be explicitly examined with consideration of the contextual modifiers discussed above. We must increasingly seek to exploit every opportunity to validate and vet our understanding of human vascular mineralized matrix metabolism – opportunities afforded by drug repurposing^{10, 57, 103}, phase IV or similar studies¹⁴², and by coordinating human molecular genetics^{164, 168}, metabolomics, and physiological phenotyping with preclinical model development^{137, 168}. Not every approach will be successful. Therefore, given finite research resources and the disease burden and its consequences^4–6, we must insist upon experimental design that embraces specific and relevant patient-oriented clinical settings to prioritize our pre-clinical research – with the goal of improving cardiovascular health and independence in our diabetic patients afflicted with arteriosclerotic disease.

Highlights.

Arteriosclerotic calcification is actively regulated in the settings of type 1 and Type 2 diabetes.

Preclinical disease models have revealed important roles for innate immune signaling and osteogenic morphogens in vascular mineralization.

Successful translation of an improved understanding of arteriosclerosis disease biology will require experimental design that symmetrically develops detailed arterial phenotyping and function in both preclinical and clinical research studies.