Chemosensory event-related potentials in 3M syndrome infants: an early biomarker based on EEG signal processing

3M syndrome is a rare disorder that involves the gene CUL7. CUL7 modulates odour detection, conditions the olfactory response (OR) and plays a role in olfactory system development. Despite this involvement, there are no direct studies on olfactory functional effects in 3M syndrome. The purpose of the present work was to analyse the cortical OR, through chemosensory event-related potentials (CSERP) and power spectra calculated by electroencephalogram (EEG) signals recorded in 3M infants: two twins (3M-N) and an additional subject (3M-O). The results suggest that olfactory processing is diversified. Comparison of N1 and LPC components indicated substantial differences in 3M syndrome that may be a consequence of a modified olfactory processing pattern. Moreover, the presence of delta rhythms in 3M-O and 3M-N clearly indicates their involvement with OR, since the delta rhythm is closely connected to chemosensory perception, in particular to olfactory perception.


Introduction
3M syndrome is a 'rare autosomal recessive dwarf syndrome' [1]. The distinctive features of this littleknown syndrome are limited prenatal growth, facial dysmorphism, the absence of microcephaly and cognitive impairment [2]. Since 3M syndrome is autosomal recessive, both inherited copies of the gene have mutations. Mutually exclusive genetic mutations in cullin-7 (CUL7), obscurin-like 1 (OBSL1) and coiled-coil domain-containing protein 8 (CCDC8), cause the pathology, as confirmed by a study conducted by Dan Hanson and collaborators [3]. They noted that, in terms of the clinical and biochemical 3M syndrome phenotype, children with CUL7 mutations are significantly shorter than those with OBSL1 or CCDC8 mutations. However, the aetiological mechanisms that lead to the observed growth disability in 3M syndrome remain unclear, but they are probably related to abnormalities in basic cell growth and changes in cellular responses to growth factor stimulation.
Although 3M syndrome is considered a relatively rare disease, it is probably an under-recognised condition; its main characteristics, including impaired pre-and post-natal growth, are shared with all gestational-age children with growth failure. This population includes many children who do not yet have a clear mechanism of growth impairment [4]. It is likely that 3M syndrome is often misdiagnosed or unrecognised due to normal mental development, mild dysmorphic facial features and good patient health. There are also numerous syndromes that may occur in a very similar manner to 3M syndrome, including: Seckel's syndrome, Meier-Gorlin syndrome and low osteo-dysplastic stature type I and II (MOPD I and II) [5], common normotepine syndromes (NPSS) and Silver-Russell syndrome (SRS) [6], whose genetic causes were successfully identified. 3M syndrome is markedly similar to Yakut short stature syndrome, which was identified in an isolated Serbian population [7]. In addition to presenting most of the physical characteristics typical of 3M syndrome, people affected by this disorder are often born with respiratory problems that can be life-threatening in childhood.
Residual clinical features (triangular face, pointed chin, mouth and prominent lips, fleshy nose with anteverted nostrils, short stature, large skull and prominent forehead) and clinical history (low birth weight) are typical of 3M syndrome [1]. 3M syndrome shows abnormal dermatoglyphics, specific skin ruts and ridges that take different forms and relationships by drawing figures of various types on digital surfaces, palms and feet. In a study conducted by Marik and co-workers [1], where two sisters affected by 3M syndrome were studied to probe their typical characteristics, other evident characteristics emerged after radiographic examination: slender bones and tubular ribs, back lumbar vertebral bodies, high vertebral bodies, occipital meningocele and short cortical neck. Kyphoscoliosis is also found in this syndrome.
Hypermobility and dislocations were observed in some patients, and a description of a fatal case of cerebral aneurysm suggests that 3M syndrome could be a generalised connective tissue disorder. The type of measurable low collagen in a patient's fibroblasts supports this possibility. The importance of skeletal relief allows clinicians to exclude with certainty other causes of short stature, such as bone dysplasias and mucopolysaccharidosis. Individuals with this condition may be at greater risk of developing swelling in the walls of blood vessels (aneurysms) in the brain [2]. Epidemiological data about 3M syndrome are not known. Today, approximately 200 cases have been reported worldwide [8].

CUL7 mutation and potential olfactory involvement
As mentioned above, genetically confirmed 3M syndrome patients carry mutations in CCDC8 (5%), OBSL1 (25%) or CUL7 (70%) [9]. CUL7 interacts with other cellular proteins and contributes to the formation of an E3 ubiquitin ligase complex that ubiquitinates specific targets. CUL7 mutations may disrupt insulin-like growth factor 1 (IGF-1) and growth hormone (GH) signalling pathways and contribute to growth alteration [8]. Insulin receptor substrate 1 (IRS1) is a target of the CUL7-SCF ubiquitin ligase. IRS1 is a signalling molecule that is a member of a family of adaptor molecules downstream of GH, IGF-1 and insulin receptors [9,10]. Insulin receptors are expressed in olfactory receptor neurones of rat olfactory mucosa, a fact that suggest insulin plays a role in odour detection modulation at the olfactory mucosa level [11,12].
CUL7-FBXW8 is a component of an E3 ubiquitin ligase that localises to the Golgi apparatus in neurones and is required for dendrite growth and organisation. Inhibition of this ligase in neurones alters Golgi morphology, impairs vesicle trafficking and disrupts dendrite morphogenesis and arborisation [13]. The ubiquitin ligase activity is linked to axon guidance during pathfinding in the development of olfactory system. MYCBP2-E3 ubiquitin ligase and its homologues play an important role in axon guidance and synaptogenesis in the olfactory system. In mice, MYCBP2 knockout alters a subpopulation of olfactory sensory axons project to the dorsal surface of the olfactory bulb, data that suggest MYCBP2-E3 ubiquitin ligase activity influences olfactory sensory neurone projections from the nasal cavity to the olfactory bulb [14].

Olfactory perception and chemosensory event-related potentials (CSERP) in infants
Olfactory perception is highly developed in newborns and infants. Recent research indicates that olfactory system activity is already present in 1-day-old newborns [15]. Macfarlane, in another study, showed that newborns can discriminate different odours at 6 day of age [16]. A longitudinal study demonstrated how olfactory memories persists from birth to early childhood and how it can influence the behavioural choices of the subjects [17]. On the other hand, Kuhn's research focuses on the olfactory environment of premature babies, who are particularly sensitive to olfactory and chemosensory stimulation [18]. Pleasant smells significantly and positively impact the behavioural and physiological development of the first stages of the subjects' lives. Smells can modulate nociception [19], for example, by inducing greater stability during painful procedures and lower severity of the central apnoea. Moreover, unpleasant or irritating odours promote disadvantageous evolutionary responses, such as decreased respiratory rate (up to apnoea) [20]. The sense of smell is also a compromised in children with cerebral malformations, genetic diseases (e.g., trisomy 13 or 18, Kallmann syndrome or Riley-Day syndrome), endocrinal disorders such as hypothyroidism and gonadal dysgeneses and in infants borne to diabetic mothers [21].
A recent work showed that it is possible to record olfactory event-related potentials (OERP) in infants [22]. OERPs and CSERPs are electrophysiological components that allow researchers to evaluate chemo-sensory and chemo-perceptual responses to olfactory stimuli. These evaluations, unlike other diagnostic and imaging tools, allow non-invasive, low-cost and excellent temporalstimulation-related tests [23]. The main difference between OERP and CSERP is that the former is elicited by a purely olfactory stimulation, while the latter is elicited by a chemical stimulation which may also include trigeminal activation [24].
Schriever and colleagues research, however, highlights the difficulty in observing OERPs in infants.
This phenomenon is likely because there are more recording artefacts. OERP components in infants are the same as in adults: early components N1 and P2 [22] and late positive components (LPC) [25]. N1 and P2 are the early sensorial components and are modulated by stimulus concentration and typology. LPC includes P3a and P3b and is modulated by the cognitive aspects of the stimulus (e.g., presentation frequency or stimulus salience) [25]. Moreover, time-frequency analysis highlights increased low frequencies (4-7 Hz) in a temporal range that corresponds to LPC [22].
Even though there are no previous ERP (and specifically CSERP) studies in 3M infants, one could hypothesise that the olfactory system could be dysfunctions in infants with CUL7 mutation [26].
Based on the integrated CSERP approach, the aim of this study was to investigate whether there are implications at the level of the olfactory perception both in OERP components and with regard to the main rhythms associated with rhinoencephalon [27] and entorhinal cortex [28] activity in 3M syndrome. Since no study has evaluated the use of CSERP or OERPs to investigate olfactory functional responses in 3M infants, olfactory function in this rare syndrome is poorly characterised.
Moreover, to the best of our knowledge, no study was conducted using electroencephalogram (EEG) signals from 3M subjects using signal processing and analysis strategies.
There are multiple potential benefits from this study. If the 3M syndrome subjects differ from the controls with respect to the olfactory response, early OERP screening, which represents an economic and non-invasive tool compared to genetic screening, could then lead to a possible subsequent genetic investigation (if it is positive). Furthermore, this research could allow us to deduce a functional CUL7 involvement in the human chemosensory/olfactory response, a prospect that has not yet studied.

Materials and methods
The research was conducted at the Neurology Unit of the Vito Fazzi Hospital in Lecce with subjects recruited at the Neonatal Intensive Care Unit (UTIN). Data collection was performed in compliance with the Code of Ethics of the World Medical Association (Helsinki Declaration) and authorized by the ASL-Lecce Ethics Committee. Written informed consent was obtained from the parents.

Subjects
Three subjects (males) with a 3M syndrome diagnosis were recruited for the study. The subjects were siblings: two 5-month-old twins (3M-N) and their 18-month-old brother (3M-O). CUL7 genetic analysis (exons 14-23/24) highlighted the presence of pathogenic variants c2781delC (p.Ser928Leufs*5) and c.4391 A>C (p.His1464Pro) in the state of compound heterozygosity. The diagnostic conclusion for all three siblings was 3M syndrome due to familiar mutations. The laboratory data is compatible with the segregation of the family pathology in the foetus. The 3M syndrome group presented the following medical history: prematurity, low birth weight (LBW), small size for gestational age, syndromic facies, triangular face, prominent frontal drafts, bulbous nose, flat angiomas of the median line, short neck and thorax, hypospadias and suspected bow curvature, fleshy and prominent heels, prenatal 3M diagnosis based on amniocentesis karyotype, glandular hypospadias, transient hypocalcaemia and transient oliguria. Moreover, the 3M subjects showed a larger cranial circumference (75-90°). Our sample size represents about 1,5% of the subjects described so far, because, since the syndrome was described, only approximately 200 cases have been described worldwide [8].
The control group was recruited with the criterion of having the same gestational and postconceptional ages as the 3M subjects, but no apparent clinical abnormalities from anamnestic data.
The controls consisted of two healthy 12-month-old male twins (HS-O) and two 4-month-old male twins (HS-N). For the 3M syndrome subjects, We chose not to merge the two twins with the older infant, but to consider them separately and perform independent comparisons, due to different characteristics from neonatal EEG and developmental brain behaviour [29]. Additionally, all infants were examined by the hospital paediatrician to rule out nasal congestion or other temporary respiratory diseases. Both healthy and 3M subjects were subjected to the neonatal auditory screening and the auditory brainstem response (ABR) and did not show any significant clinical abnormalities. Any other behavioural olfactory or chemosensory assessment was performed.
The OERP presentation paradigm consisted of sequences of olfactory stimulations; each stimulation lasted 340 ms, with an inter-stimulus interval (ISI) of 20 s. In total, the subjects were exposed to 20 stimulations (a sufficient number of stimulations, because the minimum number to elicit OERP is 8 [22]). In accordance with recommendations based on previous research, the ISI was greater than 10 s to avoid habituation [26]. The device used to record the presentation of odorous stimuli allowed us to measure, in a controlled and automated way, the CSERP evoked by olfactory stimuli synchronized to the acquisition of the EEG signal. The administration of the odorant, which took place through the olfactometer, was presented through a plexiglass tube that was positioned in the center of the two nostrils. The odorant was delivered as binarinal stimuli in front of the nose. During the electroencephalographic recording the children were seated in the arms of the mother, who in turn was sitting on a comfortable armchair placed inside the EEG recording room. The children were in a relaxed condition, they had a post-prandial condition (they eaten about an hour before the EEG recording) [31] and were in a waking state [32]. The choice of the eucalyptol odorant, which has a mixed component both olfactory and trigeminal, allowed us to keep the children in arousal during the CSERP recordings. [33][34][35][36]

EEG recording
The EEG signals were recorded using a Analyzer (Brain Products GmbH). The impedance was maintained below 8 kΩ, and the sampling rate was 256 Hz.

OERP preprocessing
The electrodes were online referenced to FCz, and offline they were postponed with a common offline reference [37]. The signal was filtered offline (0.01-50 Hz, 24 dB), and the artefact rejection threshold was set to > |125| [38]. Ocular rejection was performed by independent component analysis (ICA).
ERP epochs included a 100-ms pre-stimulus reference period and a 500-ms post-stimulus segment.
The peaks were automatically detected for all channels. The OERP components were labeled as N1 and LPC according to Pause et al. [25]. The latency windows were set to 100-400 ms for N1 and 350-600 ms for LPC [39,40]. The linear derivation process allows one to synthetize new channels from linear combinations of recording existing electrodes/channels [42].

EEG signal pre-processing
We further analysed the original EEG signals with signal processing strategies, since the sample was necessarily small and the study could be reduced exclusively to a single case. Thus, we investigated EEG rhythms on pieces (trials) of signal collected after each olfactory stimulation by searching for the presence of recurrent common trends in the 3M subjects with respect to the controls.
Signals were filtered with an off-line modality by considering only 1-s trials, starting with the onset of the olfactory stimulation, since the brain response exhibits a decaying signal that has a zero value 1 s after olfactory stimulation. First, to correct different amplification effects, each trial was normalised with respect to its baseline level, obtained by calculating the mean value of the power spectrum in a frequency band (65-75 Hz), which is usually only occupied by noise. Thus, all the resulting trials showed the same amplification. Then, the signal was subjected to a band-pass filter projected on a 2D scalp map, the topoplot [43]. In this way, we obtained a localisation of the component on the head. The shape of the resulting topoplots allows one to relate each component as an artefact [43,44]. The residual components were projected back to the signal space to obtain a filtered version of the signal that composed each trial.

OERP data analysis
Due to the sample size, we performed initial explorative and descriptive analyses to investigate N1 and LPC OERP components (see Tables 1 and 2). The explorative analysis showed an OERP trend for greater amplitudes in 3M syndrome subjects, and this trend was apparently more uniform and structured in 3M-O (Fig. 1) compared to 3M-N (Fig. 2).   There was no identifiable typified LPC [22] for 3M-N and HS-N in frontal right and for 3M-N in Cz.
All 3M-N ROIs exhibited shorter latencies compared to HS-N.

EEG spectral analysis
After pre-processing, each trial was analysed 0-1000 ms after onset (since the brain response signal is zero 1 s after olfactory stimulation). The resulting signals, each sampled for 1 s at 256 points, were However, the examined 3M subjects had low frequencies (≤ 4Hz) elicited by olfactory stimulation, while higher frequencies (> 4Hz) were mostly activated for healthy subjects. Fig. 3 highlights these aspects by reporting, for all subjects and for each channel, the percentage of trials for which 60% of the corresponding shared controls are reported in b). Vertical bars indicate the average threshold; differences are apparent between patients and controls.
This effect was most noticeable between 3M-O (row #1, column #1) and HS-O (row #1, columns #2 and #3) with respect to 3M-N (rows #2 and #3) and HS-N (row #2, columns #2 and #3). Moreover, some ROIs were more involved than others in this process, as shown in Fig. 4, which reports the percentage difference of the green area in Fig. 3 between 3M patients and corresponding controls (and separated by ROIs). Table 3 confirms that for 3M-O, the power spectrum was concentrated in the (0.01, 4] Hz interval (positive values, highlighted in "orange") for all ROIs with respect to both controls. However, for 3M-N patients, this behaviour was confirmed just for some specific ROIs (regions with discordant signs were not considered).   Table 3. Percentage difference, by ROIs, between the green regions (area of the power spectrum ≤ 4Hz) in Fig. 3   HS-N1 HS-N2

Discussion and conclusion
3M syndrome is extremely rare and difficult to diagnose. Its peculiarity lies in bone alterations and genetic variations, which, among various aspects that these changes modulate, also affects the olfactory response. Indeed, CUL7, a gene involved in the 3M syndrome, can modulate odour detection and condition the OR and plays a role in olfactory system development [3,[12][13][14]. Despite this involvement, there are no direct studies on the functional effects of this syndrome. This paucity of data is due to the fact that the syndrome is one of the rarest genetic disorders and evaluation of cortical responses to olfactory stimuli in infants and newborns is one of the less-frequent investigations within psychophysiology and cognitive neuroscience [22,46].
The purpose of the present work was to analyse the cortical olfactory response, recorded through OERP, in 3M syndrome infants. We first evaluated the OERP responses with a direct descriptive comparison (since this study is comparable to a single case study due to the small sample size) on the trends of the olfactory stimulus sensory and perceptive components. In particular, we analysed the sensory components N1 and LPC elicited by the stimulation paradigm [47]. The CSERP results demonstrated that the 3M-O infant exhibited increased N1 amplitudes and faster latencies.
Furthermore, we found a faster latency in Cz, which is positioned on the precentral gyrus [48], and greater amplitude in Pz, located in the middle parietal lobe. The precentral gyrus and parietal cortex are considered sites for olfactory working memory [49]. We interpret these findings as an indication of greater allocation of attentional resources, enhanced olfactory working memory and olfactory perception, visible in N1 component, that is involved in the sensorial detection of the olfactory stimulation [25]. We suppose that this enhancement could be related to the CUL7 alteration. LPC data followed the same results as N1, except for central right, where we observed a decreased amplitude.
The wider LPC is a consequence of the processing of olfactory information visible through the N1 component. The 3M-N twins also showed increased amplitude in precentral gyrus and faster N1 and LPC latencies, although the results in younger infants were apparently less defined and exhibited less LPC typing than for 3M-O and control groups. This minor typing is evident with the difficulty of identifying the LPC component both in frontal right, central right and Cz ROIs [22].
In 3M syndrome, olfactory processing appears to be clearly diversified. Specifically, comparison of the N1 and LPC components indicates substantial differences in 3M syndrome that may be a consequence of a modified olfactory processing pattern. Moreover, the 3M subjects showed different arousal localisations from olfactory stimulation, data that implicate much larger areas that range from the left hemisphere to the midline sites (i.e., Fz, Cz and Pz). These differences were more distributed and evident in the infant rather than younger twins, but in general they seemed to be constant with respect to the CSERP trend. As a further signal control, we performed a new analysis based on the assumption that the slow and high CSERP frequencies are related [50]. Indeed, we considered the rhythms within the signal and considered the greater cortical response, which in our case coincided, at a temporal level, with the CSERP-elicited response. These results demonstrated that the frequencies generated by olfactory stimulations were mostly present in the (0.01,8] Hz interval in 3M syndrome and healthy subjects. However, the behaviour observed in the examined subjects was that low frequencies, in particular δ (≤ 4Hz) were elicited by olfactory stimulation in 3M syndrome, while higher frequencies (> 4Hz) were mostly activated for healthy subjects. Moreover, we argue that some ROIs are more involved than others in this process. In particular, 3M-O showed involvement in all ROIs, although parietal, central-left, central, frontal and frontal left exhibited greater activation; 3M-N showed elevated activation in temporo-parietal (left and right), parietal, frontal left and frontal right. Overall, similar EEG patterns were present for the same clinical categorisation (e.g., 3M-O and 3M-N). The δ EEG rhythm appears to be more structured in 3M-O, although the general effect in 3M-N was the same, but less strong. The different age-related responses in 3M infants were more pronounced than those between controls, where similar patterns were maintained in OERP and EEG spectral analysis. Moreover, the presence of δ rhythms in 3M syndrome patients clearly implicates an olfactory response involvement, since this rhythm is closely connected to olfactory perception [51,52].
Although we were unable to perform robust statistical analysis due to the limited number of subjects, our results are relevant for basic research and clinicians. For basic research, these results highlight, for the first time in human infants, a functional aspect of the cortical olfactory response linked to the CUL7 gene. From the clinical point of view, these results suggest that a diagnostic evaluation of the cortical olfactory response at an early age may provide indications for subsequent genetic screening, which is more complex and expensive than an CSERP assessment. The limitations of this study concern the chemical nature of stimulation and the sample size. In fact, regarding the first limit, we did not use a purely olfactory stimulus (e.g., Phenethyl Alcohol) to prevent the child from relaxing and falling asleep during the EEG recording [36]. The administration of eucalyptus, in fact, on the one hand has allowed to keep the children in a state of mood increased-vigilance state , but on the other hand has ensured that the elicited component is of a mixed type (both olfactory and trigeminal) [24,36,53].
The sample size, even if is a limitation, also partly represent a strength. The small number of subjects actually represents a larger percentage of 3M syndrome subjects than the percentage that would usually be represented by afflicted individuals in clinical studies.