Acute arterial occlusion usually leads to tissue underperfusion, tissue ischaemia and may end with tissue necrosis necessitating limb amputation. However, gradually developing stenoses, leading to subacute occlusions, stimulate the growth of a bypass circulation leading to enlargement of pre-existing arteriolar connections that tap the high-pressure arterial network proximal and supply the low-pressure region distal from the occlusion with blood. The function of the occluded artery is thereby partially, sometimes completely, restored. Anecdotal evidence exists, that a ‘collateral’ circulation is able to compensate for the complete occlusion of the human thoracic aorta. Collateral arterial blood vessels had been extensively studied in the hearts of human patients and in experimental animals, also in the brain, kidney and peripheral circulation. The study of arterial growth following the occlusion of an artery is now termed ‘arteriogenesis’ (Figure 1).
Tissue ischaemia also leads to the proliferation of capillaries, termed ‘angiogenesis’. These vessels lie within the ischaemic region, and capillary connections to those from normally perfused neighbouring capillaries are scant. Furthermore, adding new small vessels of high resistance increases the total resistance of the vascular bed of the afflicted artery. Angiogenesis is therefore of little value for the functional replacement of occluded arteries.
The stimulus for growth of collateral vessels is the increase in fluid shear stress that is a consequence of the pressure difference created by the occlusion, which increases flow. This process is self-limiting because fluid shear stress falls with the increase in collateral vessel diameter. This may be the reason why the functional restoration of the occluded artery by collaterals is usually incomplete. The molecular mechanisms of shear stress-induced growth were studied in experimental animals with arteriovenous fistulas, known to produce vascular growth [1, 2]. Gene expression studies showed that several pathways are involved, that is, shear stress sensitive ion channels, such as TRPV4 [3, 4], the Rho-pathway utilizing actin-binding proteins  and the nitric oxide synthase (NOS) pathway . Blockage of the latter completely inhibits arteriogenesis.
Molecular mechanisms of arteriogenesis (collateral artery growth)
The physical stimulus, that is, fluid shear stress, acting upon the endothelial cells, caused by the viscous drag of the accelerated blood flow velocity, must be translated into a molecular signal that has to reach the vascular smooth muscle of the tunica media of the pre-existent arteriole. Because there exists no direct cellular contact between the endothelial and the smooth muscle cells, separated from each other by the internal elastic lamina and the basement membranes, the fluid shear stress transmitter can only be a diffusible molecule. The most likely candidate is nitric oxide (NO). It is produced by endothelial NOS (eNOS) upon activation by calcium (probably produced by the stress-activated Ca++ channel TRPV4) and phosphorylation and reaches the smooth muscle cells where it combines with soluble guanylate cyclase to produce cyclic guanosine monophosphate, which relaxes the smooth muscle and leads to vasodilatation. However, arteriogenesis is mainly characterized by cell proliferation, which leads to a marked increase in vessel diameter (up to fivefold) and length (tortuosity). This requires a mitogen and probably one that does not belong to the class of heparin-binding growth factors such as the fibroblast growth factors because targeted deletion of their genes does not impair collateral growth and neither do the class of vascular endothelial growth factors. The most probable candidate is the monocyte chemoattractant protein (MCP-1), which is produced in the smooth muscle layer after shear stress activation of the endothelium. It is at present unknown how the MCP-1 gene is indirectly activated by fluid shear stress. MCP-1 is secreted by the smooth muscle cells toward the lumen of the collateral arteriole as well as into the adventitia where it attracts monocytes: luminally excreted MCP-1 attracts monocytes from the flowing blood, and the adventitial MCP-1 attracts monocytes that originate from the nearest venules. Luminal attraction is not very efficient because NO dispels most monocytes. The adventitial accumulation of monocytes leads also to an activation of resident macrophages and to a transformation of monocytes into macrophages (Figure 2). These and the smooth muscle cells produce transmitters of inflammation such as tumour necrosis factor alpha and inducible NOS (iNOS). Targeted deletion of the eNOS gene leads to a loss of vasodilation but not of arteriogenesis. Targeted deletion of iNOS leads to impairment of arteriogenesis but not to complete inhibition. Only the deletion of both eNOS and iNOS leads to complete loss of collateral vessel remodelling upon arterial occlusion . We explain this as follows: with isolated eNOS knockout, the inflammatory component of arteriogenesis, represented by iNOS, is still intact and provides for only mildly impaired collateral growth (thinning of the arterial wall but no encroachment of diameter). The combined knockout eliminates both the vasodilatation as well as the growth stimulus.
Diameter enlargement by growth of pre-existent arterioles following occlusion of a major artery ends when the fluid shear stress had normalized. This occurs relatively early because any increment in radius is magnified by the fact that the radius appears in the third power and in the denominator of the stress equation. This is briefly counteracted in the beginning of the remodelling by the increase in flow, which is related to the fourth power of the radius according to Poissieulle's law. Although this ‘law’ is only partially valid for arteries, that is also the case for the stress equation.
Selection for high flow and fewer vessels
Murray's law states that vascular adaptation to altered physical stimuli are always tending to the most economic and efficient pathway. This is particularly the case in arteriogenesis because, initially, many small arterioles of the interconnecting network participate in the growth adaptation. However, many small vessels are not efficient to compensate for the loss of a major artery because of the comparatively high resistance that small vessels constitute. In the course of the establishment of an efficient collateral circulation, a few large vessels mature and most of the other small vessels regress. Collateral blood flow is thus diverted away from the small vessels and is increasingly carried by the few large vessels. The disappearance of numerous small vessels is caused by an inward growth of endothelium and smooth muscle that finally obstructs the lumen. It is tempting to speculate that the narrowing of the lumen of these vessels leads to ever increasing shear stress, the primary growth stimulus that had initiated the entire process but that now leads to the demise of the small ones. The critical decision point that makes the larger ones larger and the smaller ones smaller is the differential regulation of the matrix metalloproteinases that allow the outward remodelling of the larger collaterals but are down regulated in the small ones that become finally completely obstructed.
It is again tempting to speculate that reopening of an obstructed artery may not lead to occlusion of collaterals because the sudden decrease in shear stress may prevent occlusive growth. On the other hand, it is well known that low flow conditions, as in the arterial segment between two ligatures, lead to luminal obstruction by cellular proliferation.