Nutrition and autophagy deficiency in critical illness

Purpose of review

Critical illness imposes a severe insult on the body, with various stressors triggering pronounced cell damage. This compromises cellular function, leading to a high risk of multiple organ failure. Autophagy can remove damaged molecules and organelles but appears insufficiently activated during critical illness. This review discusses insight into the role of autophagy in critical illness and the involvement of artificial feeding in insufficient autophagy activation in critical illness.

Recent findings

Animal studies manipulating autophagy have shown its protective effects against kidney, lung, liver, and intestinal injury after several critical insults. Autophagy activation also protected peripheral, respiratory, and cardiac muscle function, despite aggravated muscle atrophy. Its role in acute brain injury is more equivocal. Animal and patient studies showed that artificial feeding suppressed autophagy activation in critical illness, particularly with high protein/amino acid doses. Feeding-suppressed autophagy may explain short and long-term harm by early enhanced calorie/protein feeding in large randomized controlled trials.

Summary

Insufficient autophagy during critical illness is at least partly explained by feeding-induced suppression. This may explain why early enhanced nutrition failed to benefit critically ill patients or even induced harm. Safe, specific activation of autophagy avoiding prolonged starvation opens perspectives for improving outcomes of critical illness.

INTRODUCTION

Critical illness imposes a severe insult on the body, which goes hand-in-hand with pronounced cell damage that compromises cellular function and leads to a high risk of multiple organ failure [1]. Many factors contribute to the damage, including hypoxia, overwhelming inflammation, and metabolic insults [1,2].

The damage must be cleared to allow organ recovery, whereas failing to heal injured cells and organs may trigger additional damage. Autophagy is a mechanism with an important intracellular quality-control function that can remove damaged molecules and organelles [3,4^▪]. However, autophagy appeared insufficiently activated during critical illness to cope with the massive damage [5]. Autophagy is activated by fasting/starvation and inhibited by feeding [6]. This review discusses insight into the role of autophagy in critical illness and how artificial feeding may be involved in insufficient autophagy activation in critical illness. 

Box 1.

no caption available

THE PROCESS AND FUNCTIONS OF AUTOPHAGY

Autophagy is a process by which cells degrade their cytosol and organelles in lysosomes. Three types can be distinguished based on how cytoplasmic material is delivered to the lysosomes [7]. In microautophagy, lysosomes take up cytosolic components directly via lysosomal membrane invagination [8]. In chaperone-mediated autophagy, chaperones recognize an amino acid motif on the target proteins and deliver those to lysosomes via binding to a specific lysosome-associated membrane protein [9]. This review will focus on the third and most studied type, macroautophagy (Fig. 1) [3,10]. This process, further referred to as autophagy, starts with the formation of an isolation membrane or phagophore that engulfs portions of cytoplasm, which can also contain organelles. The isolation membrane elongates and finally forms a double-membrane vesicle called an autophagosome. The outer autophagosomal membrane then fuses with a lysosomal membrane to form an autolysosome, which contains all the enzymes needed to degrade the sequestered organelles and molecules and the inner autophagosomal membrane.

FIGURE 1.

Schematic presentation of the different phases of autophagy. Autophagy starts with the formation of a phagophore or isolation membrane (initiation phase), which is elongated to surround portions of cytoplasm and/or organelles leading to the formation of a double-membrane vesicle, called autophagosome (elongation phase). The mature autophagosome then fuses with a lysosome to form an autolysosome (maturation phase), which contains all the enzymes that are needed for degradation of the sequestered content as well as the inner autophagosomal membrane. This figure was created with BioRender.

Autophagy serves crucial functions. During fasting/starvation, autophagy's nonselective bulk degradation of cytoplasmic content provides an endogenous nutrient supply to maintain nutrient homeostasis [7,11^▪]. Fasting/starvation is the most potent physiological autophagy activator, whereas feeding inhibits the process [6,12]. Autophagy can also be selective [4^▪]. Selective autophagy serves an important intracellular quality-control function by degrading long-lived proteins, potentially toxic protein aggregates and damaged organelles, is involved in clearing invading pathogens, and has anti-inflammatory effects [3,4^▪,13,14]. Hence, it is not surprising that its disruption has been implicated in several diseases [3,7,11^▪,12].

AUTOPHAGY IN CRITICAL ILLNESS

The first studies on autophagy in a critical illness-related context reported an increase in autophagic vacuoles, morphological evidence of mitochondrial autophagy and/or increased expression of autophagy-related genes in the liver or heart of rodents or patients with sepsis or the diaphragm of mechanically ventilated brain-dead organ donors [15–19]. Whether this was an adaptive or harmful response remained unclear. A study on liver and muscle biopsies of prolonged critically ill patients subsequently suggested that autophagy is insufficiently activated [5], revealing a phenotype like mice with liver or muscle-selective autophagy inactivation [20,21]. Indeed, morphological and biochemical analyses showed that proteins involved in the initiation and elongation steps of autophagy were induced by critical illness, but mature autophagic vacuole formation was impaired, and proteins normally degraded by autophagy accumulated hugely. These findings were confirmed in the liver and kidney of prolonged critically ill rabbits, where the phenotype of insufficient autophagy appeared more pronounced in nonsurviving than in surviving animals and correlated with impaired mitochondrial function and worse organ damage [22]. Thus, these data suggested that incomplete clearance of cellular damage due to insufficient autophagy could explain the lack of recovery from organ failure in prolonged critical illness and opened perspectives for therapies that activate autophagy during critical illness.

Meanwhile, literature on autophagy in critical illnesses has exploded. Patient studies are scarce and primarily focus on autophagy markers as biomarkers, for example, as prognostic markers of outcome from acute kidney injury (AKI), sepsis with acute respiratory distress syndrome (ARDS), or acute pancreatitis [23–25]. Ample studies have been performed in animal models. Most are descriptive, but many interfered with the process and studied the impact on critical illness induced organ damage/dysfunction of pharmacological autophagy activation and/or of pharmacological inhibition or genetic inactivation of autophagy. The literature is not unequivocal, but overall, the balance mostly tilts toward studies emphasizing the importance of sufficient autophagy activation as an organ-protective strategy rather than autophagy activation as a detrimental process. Examples are given in the following paragraphs, focusing on different organs and conditions (Fig. 2).

FIGURE 2.

Organ-protective effects of autophagy during critical illness. Most studies that manipulated autophagy during critical illness emphasized the importance of sufficient autophagy activation as organ-protective strategy. This figure shows examples of conditions leading to critical illness, in which artificial stimulation of autophagy attenuated and/or artificial inhibition of autophagy aggravated damage to or dysfunction of the depicted organs. Data are more heterogeneous in the context of acute brain injury. Particularly in traumatic brain injury and cerebral ischemic injury both neuroprotective as well as harmful effects of autophagy have been documented in several studies, these conditions are therefore indicated in grey. Please note that this scheme is simplified, as including all literature was not possible and beyond the scope of this review. ARDS, acute respiratory distress syndrome; TBI, traumatic brain injury.

Autophagy has been shown to protect against the development of sepsis-associated AKI [26]. Pharmacological autophagy inhibition in mice subjected to cecal-ligation-and-puncture (CLP)-induced sepsis aggravated the increase in creatinine and urea and the pathological renal injury score [27], whereas pharmacological autophagy activation attenuated histologic tubular epithelial injury and improved renal function [28]. Also, activation of mitochondrial autophagy (mitophagy) reduced mitochondrial damage, improved ATP production, and inhibited reactive oxygen species (ROS) generation in AKI induced by ischemia-reperfusion, sepsis, cisplatin, or contrast agents [29^▪]. Autophagy modulation further proved promising for reducing acute lung injury in models of sepsis, ARDS, or lung fibrosis [30]. Autophagy activation has been implicated in lung protection by heme-oxygenase-1 or hydrogen sulfide in CLP-induced sepsis, where autophagy activation attenuated acute lung injury and autophagy inhibition showed the opposite effect [31^▪,32^▪]. Likewise, protection by aspirin against lung fibrosis in mice appeared mediated via autophagy activation, as it was abrogated by autophagy inhibition [33]. Several studies showed that autophagy also protects against sepsis-induced liver injury [34]. Impairing acute hepatic autophagy activation via local genetic inactivation worsened hepatic mitochondrial damage/dysfunction and aggravated liver damage after CLP in mice [35]. In other CLP mouse studies, autophagy enhancement alleviated liver injury [36,37]. Autophagy regulation protects against liver injury evoked by liver surgery associated ischemia-reperfusion [38]. In septic mice, pharmacological autophagy activation also reduced intestinal epithelial apoptosis and restored intestinal barrier function [39^▪]. Similarly, activation of intestinal autophagy alleviated severe burn-induced gut damage, with opposite effects by autophagy inhibition [40]. In severe acute pancreatitis, not only pancreatic tissue showed signs of impaired autophagy, but also spleen where mitophagy impairment appeared related to spleen injury [41,42].

Autophagy occurs following multiple types of muscle injury and is crucial for timely muscle recovery [43]. The importance of muscle autophagy in critical illness has also been underscored. Although autophagy may contribute to critical illness induced muscle atrophy [44–46], its activation overall protects muscle function. In mice with hind-limb ischemia, autophagy inhibition indeed had a dual effect, diminishing muscle peak force while decreasing fibrosis and myofiber size [46]. In rats subjected to CLP, autophagy activation improved, whereas autophagy inhibition worsened neuromuscular function [47]. Autophagy activation also largely prevented diaphragmatic force loss associated with mechanical ventilation [48]. Likewise, increased autophagy improved sepsis-induced cardiac dysfunction [49^▪]. CLP in mice increased formation but impaired degradation of autophagosomes in the left ventricle, indicating incomplete autophagy [50]. However, pharmacologically induced completion of autophagy restored cardiac performance, increased ATP levels, and decreased inflammation. Autophagy promotion appeared to mediate the alleviating effect of several compounds on myocardial apoptosis and cardiac dysfunction after CLP in rats [51^▪,52,53]. Resveratrol's protection against myocardial ischemia/reperfusion injury in rats may also be partly mediated via autophagic flux repair [54].

In the context of acute brain injury, the role of autophagy is more controversial, particularly in traumatic brain injury (TBI) and cerebral ischemic injury [55–57]. Several animal studies showed that autophagy activation is neuroprotective with reduced infarct volume, attenuated neuronal apoptosis and better functional recovery. However, other studies found benefits of autophagy inhibition, such as attenuated blood-brain barrier disruption, reduced neuronal damage and apoptosis, and better cognitive recovery. Extensive attention has been given to neuronal mitochondrial autophagy (mitophagy) in TBI, of which extent and spatiotemporal distribution could be related to injury type and severity, and of which stimulation might be beneficial or detrimental depending on cellular bioenergetics context [58]. Several studies on subarachnoid hemorrhage have shown that enhancing autophagy reduced, whereas inhibiting autophagy aggravated neuronal apoptosis [59]. In rodents with sepsis-associated encephalopathy after CLP, induction of autophagy in hippocampal neurons or mitophagy in cerebral microvascular endothelial cells reduced inflammation and pathological injury and ameliorated cognitive dysfunction [60,61].

The autophagy pathway has received ample attention in the context of the COVID-19 pandemic. Autophagy normally acts as an innate defensive system in virus-infected cells by delivering virions or viral components to lysosomes for degradation and eliciting antiviral immune responses to promote virus clearance [13,14]. However, the SARS-CoV-2 virus has evolved several mechanisms to escape autophagic degradation and hijack the autophagic machinery for its benefit [62,63]. Via autophagic removal of host factors, it can escape immune surveillance. Specific autophagic machinery components are involved in the endomembrane remodeling required for SARS-CoV-2 infection or in virus replication. At late stages, abundantly present viral proteins block the formation of degradative autolysosomes, promoting exocytosis-mediated viral egress. Thus, autophagy manipulation emerged as an interesting target to reduce viral load and combat SARS-CoV-2 infection [64,65]. Autophagy activation has also been put forward as a therapeutic strategy to inhibit NLRP3 inflammasome-mediated lung inflammation in COVID-19 [66]. However, the clinical benefit of autophagy-related therapy or drugs remains to be demonstrated [64,66,67].

NUTRITION AND AUTOPHAGY DEFICIENCY IN CRITICAL ILLNESS

Autophagy activation appeared insufficient during (prolonged) critical illness, despite increases in several stimulating factors, including hypoxia, oxidative stress, and endoplasmic reticulum stress [5]. As fasting activates autophagy and nutrient supply reverses this response [6], the typical continuous artificial feeding of critically ill patients emerged as a plausible cause of insufficient autophagy.

Artificial nutrition during critical illness

Critically ill patients cannot eat normally and often show enteral feeding intolerance, resulting in a rapidly accumulating macronutrient deficit, which in observational studies has been associated with adverse short-term outcomes [68–75]. Therefore, enhancement of macronutrient provision from early on has been advocated. However, several randomized controlled trials (RCTs) have shown that early full feeding, for example, via more energy-dense enteral nutrition formulae or supplemental parenteral nutrition completing insufficient enteral nutrition, did not benefit or even harmed critically ill patients [76^▪,77–86]. The EPaNIC and PEPaNIC-RCTs in adult and pediatric patients showed that early parenteral nutrition increased the risk of complications and delayed recovery compared with omitting parenteral nutrition in the first week (late parenteral nutrition) [77,78]. Criticisms that harm or lack of benefit would be explained by use of the parenteral feeding route, calculated energy targets, or too low amino acid doses were countered by large RCTs [76^▪,87–92,93^▪]. Harm by early parenteral nutrition in the EPaNIC and PEPaNIC-RCTs was statistically explained by the higher protein/amino acid doses rather than the glucose or lipid doses [94,95]. RCTs on enhanced protein intake did not find a benefit of higher doses [91,92,93^▪,96], and caution is warranted for adverse effects [93^▪,97]. The EFFORT-RCT indicated that higher protein doses might particularly harm patients with AKI and high organ failure scores [93^▪]. Also, higher amino acid doses with early parenteral nutrition did not promote anabolism and instead increased nitrogen wasting, which may burden the liver and kidney [95,98]. The most recent NUTRIREA-3-RCT demonstrated faster recovery and fewer complications with early calorie and protein restriction, further supporting short-term harm by early full feeding and high protein doses [99^▪]. Also in the long term, early enhanced feeding did not improve outcomes [81,100–102] or showed harm. The EPaNIC-RCT identified prolonged ICU stay and ICU-acquired neuromuscular complications, occurring more frequently with early parenteral nutrition than with late parenteral nutrition, as risk factors for worse long-term outcomes [77,103–107]. The PEPaNIC-RCT showed long-term developmental harm by early parenteral nutrition versus late parenteral nutrition [108–110].

Nutrition and autophagy deficiency in critical illness

About a decade ago, we showed in an animal model that early parenteral nutrition versus fasting during critical illness evoked an autophagy deficiency phenotype in the liver and muscle [111]. Autophagy substrates accumulated in fed versus fasted animals, accompanied by increased liver damage and worse myofiber vacuolization. The EPaNIC-RCT provided translation to patients, showing less efficient activation of autophagic quality-control of myofibers with early parenteral nutrition versus late parenteral nutrition, independently associated with more ICU-acquired weakness [103]. Autophagy suppression has also been implicated in parenteral nutrition associated lung injury [112]. Vice versa, autophagy promotion with (mimicking) short-term caloric restriction appeared associated with less cognitive dysfunction after TBI and better functional recovery after spinal cord injury in mice [113,114^▪].

Interestingly, the degree of autophagy in patients’ muscles correlated inversely with the amount of amino acid calories infused until biopsy [5]. An animal study also pinpointed amino acids as culprits since especially amino acid enriched nutrition suppressed autophagy compared with isocaloric glucose or lipid-enriched nutrition [111]. The strong autophagy-suppressing effect of amino acids [111,115] may explain why harm by early parenteral nutrition was statistically explained by higher doses of amino acids and not glucose or lipids [94,95]. Amino acids also antagonized autophagy activation in muscles of lipopolysaccharide-challenged neonatal pigs [116]. The impact of fatty acids on autophagy varies strongly. Palmitate was shown to impair autophagy [117]. However, docosahexaenoic acid and other (omega-3) polyunsaturated fatty acids (PUFAs) may induce neuroprotection against palmitate-induced lipotoxicity via enhanced autophagy [118]. Also, docosahexaenoic acid alleviated brain damage and behavioral dysfunction in mice with ischemic stroke, with neuroprotection abrogated by mitophagy inhibition [119]. Autophagy induction has further been implicated in neuroprotection by omega-3 PUFA supplementation after TBI in rats [120] and protection from lipopolysaccharide-induced cardiac dysfunction or ischemia-reperfusion renal injury in transgenic mice with high endogenous omega-3 PUFAs [121,122]. Octanoic acid upregulated autophagy and prevented lipopolysaccharide-induced acute liver injury [123].

Early fasting with better autophagy activation in critical illness has shown clear benefits but cannot be sustained. Maintaining fasting-induced benefits while avoiding prolonged starvation may be possible with intermittent fasting/fasting-mimicking diets [124,125]. As a first step towards such a strategy for ICU patients, we tested whether 12-h fasting can induce a metabolic fasting response in prolonged critically ill patients [126]. This fasting period increased serum bilirubin and ketone-bodies while decreasing insulin requirements and serum IGF-I. However, autophagy markers in blood were largely unaffected by fasting in patients and healthy individuals, but these measurements may not reflect autophagy at the tissue level. Intermittent fasting remains promising, as it improved behavioral performance and neuronal survival at the injured segment after spinal cord injury, which was prevented with autophagy inhibition, and exerted neuroprotective effects after TBI [127,128^▪]. Ketone bodies, which rise upon fasting, stimulate autophagy [126,129]. However, protection against sepsis-induced muscle weakness with ketone-bodies infusion appeared not mediated via autophagy stimulation [130].

The long-term impact of nutritional management in ICU on autophagy in relation to the long-term adverse legacy after critical illness [104–110,131^▪,132] remains unclear. Epigenetic abnormalities induced by critical illness or its nutritional management provide a plausible molecular basis for their adverse effects on long-term outcomes [133–136,137^▪]. Autophagy is prone to epigenetic regulation, and abnormal DNA methylation of genes involved in autophagy was observed in the muscles of critically ill patients versus matched controls [137^▪]. Altered autophagy regulation at the histone-modification level or noncoding RNAs may also be relevant in critical illness [138–140]. A small study invoked impaired muscle regeneration rather than autophagy as a mechanism contributing to long-term muscle weakness in critically ill patients but did not address any potential impact of nutritional management [141].

CONCLUSION

Autophagy is insufficiently activated in critically ill patients and animals receiving full feeding, which allows the accumulation of cellular damage, thus hampering organ function. Feeding-induced autophagy suppression may at least partially explain why early full nutrition failed to benefit critically ill patients and even induced harm. The search for specific autophagy activators or altered feeding strategies to activate autophagy in (prolonged) critical illness while avoiding prolonged starvation is ongoing.

Acknowledgements

None.

Financial support and sponsorship

This work was supported by the Research Foundation-Flanders (FWO, Project grant 0592.12 to I. Vanhorebeek and Fundamental Clinical Research Fellowship to M. Casaer), by the Methusalem program of the Flemish government (through the University of Leuven to I. Vanhorebeek, grant METH14/06), by the KU Leuven (C2 project funding to M. Casaer and J. Gunst), by the Clinical Research and Education Council of the University Hospitals Leuven (postdoctoral research fellowship to J. Gunst) and by the European Society of Intensive Care Medicine (Fundamental Research Award grant to J. Gunst).

Conflicts of interest

None.

REFERENCES AND RECOMMENDED READING

Papers of particular interest, published within the annual period of review, have been highlighted as:

• ▪ of special interest

• ▪▪ of outstanding interest