Effects of renewable diesel exhaust on lung function and self-rated 1 symptoms for healthy volunteers in a human chamber exposure study 2

Background: Diesel engine exhaust causes adverse health effects. Meanwhile, the impact of renewable diesel exhaust on human health is less known. In this study, nasal patency, pulmonary function, and self-rated symptoms were assessed in 19 healthy volunteers after two separate 3- hour exposures to renewable diesel (hydrotreated vegetable oil [HVO]) exhaust, and exposure to 30 filtered air (FA) for comparison. The HVO exposures were generated with two modern non-road 31 vehicles (2019) having either: 1) no aftertreatment system (HVO PM+NOx ), or 2) an aftertreatment 32 system containing a diesel oxidation catalyst and a diesel particulate filter (HVO NOx ). The exposure 33 concentrations complied with current EU occupational exposure limits (OELs) of NO, NO 2 , formaldehyde, polycyclic aromatic hydrocarbons (PAHs), and future OELs of elemental carbon 35 (EC) from 2023.

Introduction compared to petroleum diesel (22,28). The solid PM fraction of the exhaust from diesel and HVO 86 is dominated by soot, which can be measured thermo-optically as elemental carbon (EC). A new 87 EU occupational exposure limit (OEL) for diesel engine exhaust, measured as EC, of 50 µg m -3 (29) 88 will be implemented in 2023 to reduce the exposure of the 3.6 million workers within the member 89 states (30). Because estimates of the life-time mortality risk from occupational exposure to diesel 90 for cancer alone (not including myocardial infarction or COPD) indicate that exposure levels need 91 to be kept extremely low (Vermeulen et al. 2014), it is of key importance to identify potential 92 adverse effects of the substitutes. As the substitution of petroleum diesel by renewable diesel is 93 increasing, a considerable number of people will be exposed to its exhaust. It is hence of interest to 94 understand the potential health effects of exhaust exposure from engines running on HVO, while 95 complying with the future OELs. organic compounds such as PAHs. An aftertreatment system can, for example, contain a diesel 100 oxidation catalyst (DOC) that oxidizes CO and organic compounds (31,32), and a diesel particle 101 filter (DPF) that oxidizes soot particles which removes significant amounts of PM (33). Hence use 102 of aftertreatment systems should reduce exposure to such emissions and their associated health 103 effects. However, a recent review assessing the effects of DPFs' use on health impacts in 104 occupational settings did not present conclusive results (34). It is thus of interest to investigate the 105 health impact from vehicles with different degrees of emission reduction technology. 106 107 Due to improved engine operation, modern diesel engines with or without aftertreatment systems 108 generally emit lower concentrations of pollutants (35) and reduce exhaust particle mass and size 109 (36). Particle characteristics such as size and morphology are of key importance in considering 110 possible health effects, as these characteristics determine where in the lungs the particles will 111 deposit. The deposition pattern in the lung depends on multiple aerosol characteristics and not 112 solely on the respirable PM mass concentration (37). 113 114 In the present study we investigate the human health effects from exposure to exhaust from two 115 modern non-road vehicles (wheel loaders). Due to their engine capacity, they fall under different 116 emission standards and were equipped with 1) no external aftertreatment device, or 2) a DOC in 117 combination with a DPF. The different emissions aftertreatment allowed for a comparison of the 118 exposure to NOx from diesel engines with and without a particulate fraction and other gaseous 119 pollutants. We aimed to evaluate self-rated symptoms, nasal patency, and pulmonary function 120 after exposure to HVO exhaust from modern non-road vehicles that complied with the EU OELs. 121 Additionally, we investigated the lung deposition of HVO exhaust particles with the deposition of 122 petroleum diesel from an older light-duty vehicle presented by Wierzbicka

126
Main findings 127 We compared the effects of exposure to exhaust from hydrotreated vegetable oil (HVO) (a 128 renewable diesel fuel) with filtered air (FA). The vehicle without an external aftertreatment system 129 (HVOPM+NOx) generated emissions of PM, NOx and organic compounds. The vehicle with an 130 external aftertreatment consisting of a DOC and DPF (HVONOx) emitted NOx with only negligible 131 concentrations of particles and measured organic components (hereafter referred to only NOx). 132 The two exposures to HVO caused mild self-rated irritations symptoms. 44% reported eye 133 irritation symptoms during the HVOPM+NOx exposure. The number of volunteers who reported 134 throat irritation symptoms was a factor 4.5 and 4 higher for HVOPM+NOx and HVONOx respectively, 135 compared to FA. In comparison to FA exposure, nasal obstruction (lower PNIF) occurred for the 136 HVONOx exposure. In the first of the following two sections we describe the exposure aerosol 137 characteristics and the HVO exhaust particle deposition in the airways in relation to diesel exhaust 138 particles. In the second, we describe the health effects in terms of self-rated symptoms and airway 139 function.

141
Exposure and lung deposition 142 143 Exposure aerosol characteristics 144 A summary of the average aerosol exposure concentrations and characteristics are presented in 145  Table 1. The average aerosol exposure characteristics during the three exposure scenarios are 146 presented in Fig. 1. The average PM1 concentration during the HVOPM+NOx exposures was 93±13 µg 147 m -3 and the average particle number (PN) concentration was 3.0 · 10 5 ± 0.3 · 10 5 cm -3 (Table 1). In 148 contrast, during HVONOx and FA exposures, the average exposure concentrations of PM1 were ~1 149 µg m -3 and PN <100 cm -3 (Table 1). For HVOPM+NOx, the elemental carbon (EC) fraction of total 150 carbon was 66±3% corresponding to an average EC concentration of 54±6 µg m -3 (Table 1). The 151 two vehicles were operated in a similar load/idle sequence, which is seen as increasing (during 152 load) and decreasing (during idle) PM1 mass (Fig. 1a) and NO concentrations during HVOPM+NOx 153 exposure, and increasing/decreasing NO2 levels for the HVONOx exposures (Fig. 1b).

155
The NO and NO2 concentrations were comparable for the two HVO exposures (no vehicle had 156 external NOx removal devices, such as selective catalytic reduction [SCR]) and on average below 157 the 8-hour OELs. During the HVONOx exposure, the average NO2 was slightly higher and the 158 average NO lower compared to HVOPM+NOx (Table 1). The variations in NO and NO2 from the 159 load/idle operation as described above can be seen in Fig. 1b. In addition, NO increased rapidly 160 after the cold start at the start of the HVONOx exposure and decreased shortly thereafter when the 161 diesel oxidation catalyst (DOC) of the vehicle has reached the operating temperature and started to 162 convert NO to NO2 more efficiently.

170
The particle number and mass size distribution of HVOPM+NOx, together with the effective density 171 of the soot agglomerates, are shown in Fig. 2a. The effective density decreased with increasing 172 mobility size, as the soot agglomerates became more open in their structure, and the diesel soot 173 power law function described by Park et al. (38) could be fitted to the experimental data (Fig. 2a).

174
The mass mobility exponent (Dfm) was on average 2.3 (where 3 corresponds to perfect spheres). An 175 example of the soot agglomerates generated during the HVOPM+NOx exposure is imaged by 176 transmission electron microscopy (TEM) in Fig. 2b. The average primary particle diameter of the 177 soot agglomerates of HVOPM+NOx was 24.5 ± 7.3 nm and an example is marked in Fig. 2b 3.0 · 10 5 9.0 · 10 1 7. All values are the average of all exposures (n=5-6) of a given type with ± 1 std. dev. Full compound analyses of PAHs and BTEX are found in Additional file B and C, respectively. a There were large uncertainties in the gravimetric mass analysis at low/no mass concentrations. The mass concentrations were in the range of the blank filters (-1±3 µg). b 33 native and alkylated, 10 oxy-and 17 nitro-PAHs were included in the analysis.
The calculated average inhaled deposited particle mass from nasal breathing during the 184 HVOPM+NOx exposure in the tracheobronchial and alveolar regions, from airway generation 1 to 24, 185 is presented in Fig. 3a. Deposited doses are given as deposited mass (µg) and as deposited mass 186 per lung tissue area (ng cm -2 ). The mass dose is largest in the distal airways (airway generations 187 >10), but when expressed as deposited mass per lung tissue area the largest dose is found in the 188 upper tracheobronchial region (airway generation <10). No deposition was analyzed for HVONOx 189 due to low PM and PN concentrations (Table 1).

191
A comparison is presented in Fig. 3b  deposited mass of the diesel aerosol was higher in comparison to HVOPM+NOx, but lower when using 197 the same exposure mass concentrations. The deposited fractions of the HVOPM+NOx aerosol mass 198 (i.e., the fraction of the deposited mass compared to the total inhaled mass concentration) were 199 around 40-50% higher compared to petroleum diesel in the tracheobronchial and alveolar regions.

200
The average accumulated deposited dose in the respiratory tract was 82 µg for HVOPM+NOx, 201 corresponding to an hourly average of 27 µg h -1 .

203
The inhaled deposited dose depends on multiple lung parameters (FRC, tidal volume, breathing  Table 2 presents the deposited HVOPM+NOx 206 particle dose expressed as mass, number and surface area compared to the petroleum diesel 207 exposure (using the original exposure concentrations). Please note the three times higher particle 208 mass concentration in the case of diesel in comparison to HVO when looking at values of deposited 209 dose by mass. The mass (and surface area) deposition fraction was a factor 1.5 higher for 210 HVOPM+NOx than diesel. Additionally, the deposited dose in terms of particle number was a factor 211 1.2 higher for HVOPM+NOx. As the particle number concentration was higher for HVOPM+NOx, despite 212 the lower mass concentration, this led to a higher number of particles deposited in the lungs for 213 HVO compared to diesel. The deposited mass fractions differ slightly from the multiple-path 214 particle dosimetry (MPPD) model due to model characteristics, see Additional File D for a 215 comparison of the deposition fraction depending on particle size for the two models. 216 217 Particle number concentration (µg m -3 ) 3.0 · 10 5 ± 0.3 · 10 5 3.9 · 10 5 ± 0.5 · 10 5 * Surface area concentration (µg m -3 ) 9.5 · 10 -5 ± 1.4 · 10 -5 3.5 · 10 -4 ± 0.7 · 10 -4 The proportion of volunteers who reported throat irritation was a factor 4.5 and 4 higher for 227

Mass
HVOPM+NOx and HVONOx, respectively, compared to FA. The difference was statistically significant 228 for HVOPM+NOx (p=0.011) and with borderline significance for HVONOx (p=0.062). The proportion 229 of reported eye irritation symptoms was around a factor 2.5 higher for HVOPM+NOx compared to FA 230 with a borderline significance (p=0.07). No volunteers reported chest tightness during the FA 231 exposure, while a few individuals did so during the HVOPM+NOx and HVONOx, respectively. 232 However, it should be noted that the reported symptom scores were generally low (mostly below 233 10 in a 0-100 VAS) for all categories. 234 235

Airway function 238
Peak nasal inspiratory flow (PNIF) and peak expiratory flow (PEF) 239 The changes in ΔPNIF and ΔPEF at each time point during the exposure are shown in Fig. 4, and 240 absolute values are presented in Additional file E. For both PNIF and PEF, there was an increasing 241 trend throughout the FA exposure while no such increase was seen for neither of the two HVO 242 exposures. The differences between average changes in PNIF and PEF measurements (ΔPNIF and 243 ΔPEF) during the two HVO exposure scenarios compared to FA are presented in Table 4. Although 244 no decrease in absolute PNIF values was found for the HVO exposures, we observed a statistically 245 significant decrement in ΔPNIF during HVONOx exposure compared to the FA exposure (-18.1 L 246 min -1 , p≤0.001), and a borderline significant decrement during HVOPM+NOx exposure (-7.4 L min -1 , 247 p=0.08). No difference in ΔPEF was found between the HVO exposures and FA. 248 249 0.60 Estimated average changes in ΔPNIF and ΔPEF (L min -1 ) during each exposure scenario (estimated mean) and differences between the two HVO exposures and FA exposure (beta). The beta values (L min -1 ) and significance (p-value) are based on the linear mixed model with exposure order correction. Values within brackets are the 95% CI.

252
Spirometry 253 The result of the FVC (forced vital capacity), FEV1 (forced expiratory volume in one second) and 254 FEV1/FVC (in L, z-score and as % of predicted) is presented in Additional file F. FEV1 and FVC 255 showed minimal and statistically insignificant differences after all exposure scenarios (Additional 256 file F1). A minimal (from 0.81 to 0.82) but statistically significant (p<0.05) increase in mean 257 FEV1/FVC was found after the HVONOx exposure. Compared to FA, no significant changes were 258 found after the HVO exposures (Additional file F2).

260 261
Oscillometry parameters The results of the oscillometry parameters reflecting reactance (X5, AX, FRES) and resistance (R5, 263 R19) are presented in Table 5. There was no statistically significant difference before and after any 264 exposure for any parameter (median). Weak statistical evidence (p=0.084) was found for a 265 decrease in reactance (X5) after HVONOx; however, similar trends were not seen for the related 266 parameters of AX and FRES which downplays the probability of a physiological effect on the lung. 267 Some volunteers had baseline oscillometry values deviating from the normal range (40,41) but 268 with normal spirometry measures; thus, the volunteers were further categorized into a "typical" 269 and "atypical" group based on their oscillometric measures (Additional file A). The atypical group 270 showed a higher proportion of having a history of symptoms and atopy (80% vs. 54%, p<0.05).

271
They were hypothesized to be more sensitive and have a different lung reaction to the exposures 272 than the typical group. However, no significant interactions between the typical/atypical groups 273 and PNIF or PEF were found. Neither were any statistically significant changes found for any 274 oscillometry parameters of the atypical/typical groups after the HVO exposures in comparison to 275 FA. 276 277 and eye irritation as well as bronchoconstriction in healthy volunteers exposed to petroleum diesel 309 at levels similar to the ones used in this study.

311
Respiratory function 312 Different patterns in PNIF values were found for the HVO and FA exposures (Fig. 4). In contrast to 313 FA exposure, PNIF did not increase during exposure to the HVO exhaust, hence indicating a nasal 314 obstruction during the two HVO exposures. The lack of an increase in PNIF was seen already 55 315 minutes into the exposure (Fig. 4). The decrements of ΔPNIF were larger during the HVONOx 316 exposure than during HVOPM+NOx in comparison to FA (Table 1). As the HVONOx exposure did not 317 contain any PM fraction (PN<100 cm -3 , PM~1µg m -3 ), the effects are attributed to the NO and NO2 318 exposure. In addition, the larger impact on the nasal patency may be related to the NO2 rather 319 than the NO concentration since NO was lower during the HVONOx exposure than during 320 HVOPM+NOx. The 3-hour average NO2 concentration was similar for the two HVO exposures but 321 fluctuated more for HVONOx and this caused short periods with higher concentrations (Fig. 1b). 322 However, the impact of NO2 on nasal patency is not known, and the uptake of NO2 generally 323 occurs deeper down in the lungs and causes effect on the small airways (44) and asthma-related 324 respiratory effects (reviewed in [45]). In this study no overall changes in lower airway function 325 (assessed with PEF, spirometry and FOT) were seen, while, for example, reduction in PEF has 326 been reported after petroleum diesel exhaust exposures of both higher (6) and lower (9) PM and 327 NOx exposure concentrations. However, in studies with NO2 exposures alone at similar 328 concentrations as in this study (1-3h, 0.1-4 ppm), NO2 has not caused any significant effects on 329 lung function (assessed by spirometry) in healthy subjects (46)(47)(48)(49).

331
Another cause for the increased nasal obstruction after the two HVO exposures could potentially 332 be due to local or pulmonary vasodilation induced by the NO exposure. NO is a known pulmonary  333 and systemic vasodilator, and when clinically administered it causes preferential pulmonary 334 vasodilation, which is used, for example, to treat hypoxemia and acute respiratory distress 335 syndrome (5-80 ppm) (50,51). However, as the decrement in PNIF (compared to FA) was lower for 336 HVOPM+NOx, which contained higher NO than HVONOx, we cannot attribute the effect solely to NO. 337 In addition, we cannot exclude that there is an interaction effect of the PM and gases, as the 338 changes in nasal patency were less pronounced during HVOPM+NOx than HVONOx.

340
The temporary changes in measured airway functions were small and unlikely to have clinical 341 importance in healthy persons from short-term exposure. Nevertheless, we cannot exclude a risk 342 from either short-term or long-term exposure on more sensitive persons, for example older people 343 or those with pre-existing lung or cardiovascular disease. The long-term effect of renewable diesel 344 exhaust is unknown, but the short-term responses in this study indicate that even exposure below 345 the future OELs is not completely without risk for negative health effects. Reduced PNIF is not a 346 measure of lung function, but clinically a long-term nasal obstruction caused by occupational 347 exposure could be considered indicative for the development of irritation asthma (52).

349
In previous studies, oscillometry measurements found early manifestations of lung disease before 350 these were measurable with spirometry (53,54). The subjects with baseline oscillometric values 351 just outside the normal range ("atypical group", Additional file A) were hence hypothesized to have 352 a different lung reaction than persons within the normal range. Differences between the 353 typical/atypical groups were investigated for oscillometric parameters, PNIF and PEF but due to 354 the small sample size, no clear conclusions can be drawn. The subjects with atypical oscillometry 355 measures showed a higher proportion of having a history of symptoms and atopy (80% vs. 54%, 356 p<0.05) from the initial medical assessment, which indicates that they may be more sensitive to 357 pollutants and allergens. However, studies with larger numbers of subjects need to be carried out 358 in order to draw any conclusions. It should also be noted that this group did not show any 359 indications of anomalies in the spirometry and for future controlled exposure studies, the 360 oscillometry measurement may increase the possibility to investigate small differences in lung 361 function. In addition, oscillometry may potentially be valuable in the assessment of lung function 362 effects related to occupational exposures for early detection and disease prevention.

364
Aerosol characteristics, deposited dose and occupational exposure limits

365
The total deposited mass dose of HVOPM+NOx during the 3-hours exposure (82 ± 32 µg, Fig. 3b) was 366 comparable to the hourly mass dose for people working outdoor in relatively polluted cities during 367 a similar time interval (50 µg PM2.5 h -1 , (55)). Compared to a previous exposure study on 368 petroleum diesel (15), the HVO particles generated by the modern diesel engine in this study had a 369 smaller mobility size which caused higher deposition fractions in terms of mass, surface area and 370 number. The difference in deposition is due to the higher deposition fraction (Additional file D) of 371 smaller particles which dominated HVOPM+NOx emissions (MMD 108nm) in comparison to 372 petroleum diesel (MMD 195 nm). It means that in the case of HVOPM+NOx, two times more particles 373 will deposit due to their smaller size in comparison to the compared petroleum diesel particles.

375
Modern diesel engines utilize improved combustion parameters of, for example, increased fuel 376 injection pressure and nozzle design, which reduce the size (but not necessarily the number 377 concentration) of the soot particles, which in turn reduces the soot mass emissions (36). As the 378 upcoming OEL for the EC from diesel engines is only expressed as mass (50 µg EC m -3 , from 379 2023), it may be more efficient in mitigating the inhaled and deposited dose of older diesel engine 380 emissions, but not necessarily as efficient in reducing the deposited dose from renewable fuels and 381 modern diesel engines without DPFs. For example, even though the PM1 mass concentration (93 382 µg m -3 ) was 3 times lower for HVO in this study, compared to the previous exposure study to 383 petroleum diesel (15) with PM1 276 µg m -3 , the average deposited mass (Table 2) was only a factor 384 0.8 lower (i.e., only 20% lower). Despite the lower PM1 mass, the number concentration was 385 similar and the surface area even higher due to the reduced particle size. OELs in terms of particle 386 number concentration can hence be more efficient in reducing the exposure to particle emissions 387 from renewable fuels and modern diesel engines similar to the ones used in this study. The particle 388 size distributions and number concentrations need to be assessed in exposure studies and not only 389 the mass in order to understand the deposition and dose dependent effects.

391
Even though the EU emission standards are continuously becoming more stringent with lower 392 allowed exhaust emissions, modern non-road vehicles lacking full emission aftertreatment 393 systems, like the two vehicles used in this study, will still be in use and pose a risk for hazardous 394 exposure. No vehicle in this study had an external NOx reduction unit, and the implementation of 395 one could potentially have reduced the NOx emissions and related health effects. From the results 396 in this short-exposure study, we cannot exclude the potential risk of short-and long-term effects 397 from exposure to renewable diesel exhaust from modern non-road vehicles that comply with the 398 latest emission standards and the future OELs. 399 400 Conclusion 401 We investigated the effects on airway function after exposure to exhaust from renewable diesel fuel 402 HVO from modern vehicles in comparison to filtered air. The vehicles were manufactured in 2019 403 and complied with the current non-road engine EU emission standards. The exposure levels were 404 kept below the EU OELs. Mild irritations symptoms (self-rated) were reported during the two 405 HVO exposures, with a slightly higher incidence number during the exposure from the vehicle 406 without an aftertreatment system (HVOPM+NOx). The data also suggested that some individuals 407 might be affected by exposure to HVO exhaust from modern work vehicles below the future EU 408 OELs. Compared to older diesel exhaust exposures, the deposited fraction in the respiratory tract 409 of HVO exhaust PM was higher, in terms of mass, number and surface area. The increase in 410 deposition was due to the smaller soot particle size.

412
In this study, the focus was on nasal patency and pulmonary function assessments. However, to In total 19 volunteers (9 f /10 m, age 20-55 years) were exposed to the two types of engine 426 emissions and particle free air during three separate 3-hour long sessions at least one week apart. 427 The exposure scenarios discussed in this publication were: 1) emissions from a wheel loader 428 without exhaust aftertreatment operated with HVO (HVOPM+NOx), 2) emissions from a wheel 429 loader with an aftertreatment system operated with HVO (HVONOx), and 3) filtered air (FA). We 430 would like to point out that the whole study also included an additional exposure scenario, namely, 431 exposure to aerosolized dry NaCl, the analysis and comparisons of which will be presented 432 separately. The exposures took place in a 22 m 3 stainless steel chamber with controlled relative 433 humidity, temperature and ventilation. The study was double-blind and a maximum of four 434 participants were exposed at the same time. All exposures took place on Tuesdays, Wednesdays  435 and Thursdays, between 9-12 a.m., with a minimum of one week between each exposure. Before 436 and immediately after each exposure, the participants went through medical examinations that 437 included spirometry and used the forced oscillation technique (FOT). Self-rated symptoms, PEF 438 (peak expiratory flow) and PNIF (peak nasal inspiratory flow) were registered four times: one time 439 before and three times during the exposure (

Aerosol generation 459
The exposures took place in a 22 m 3 stainless steel chamber with an air exchange rate of 4 460 exchanges/hour. The supply air used for dilution was filtered from particles with a HEPA (high-461 efficiency particulate absorbing) filter and from gases with an active carbon filter. The temperature 462 was kept at 26±1°C and the relative humidity at 33±4%.

464
Renewable diesel exposure 465 The renewable diesel exhaust exposure scenarios were generated with two types of modern off- Particle number size distribution and number concentration 502 The particle number size distributions in the range 9.8-430 nm (HVO exposures) or 19-914 nm 503 (FA exposures) were measured with a scanning mobility particle sizer (SMPS), including an 504 electrostatic classifier (TSI model 3082) and condensation particle counter (CPC, model 3775, 505 TSI). The aerodynamic size distribution of 0.5-20 µm was monitored with an aerodynamic particle 506 sizer (APS, model 3321, TSI) during the exposures to ensure that the particle number size 507 distributions maxima were captured with the SMPS. 508 509 Effective density (DMA-APM) The particle effective density was assessed using an aerosol particle mass analyzer (APM 3600, 511 Kanomax) in combination with a differential mobility analyzer (DMA, TSI Inc., U.S.A.) and a 512 condensation particle counter (CPC, model 3075, TSI Inc., U.S.A.) (56). The effective density was 513 measured at five DMA-selected particle mobility diameters: 50, 70, 100, 150 and 300 nm. Mobility 514 size (dp) selection was performed with the DMA. The APM measured the mass distribution of the (1) 520 Where ρPSL is the density of the PSL reference particles, VAPM is the measured arithmetic mean 521 voltage of the sampled particles for a given mobility diameter and RPM, and VAPM,PSL is the 522 theoretically calculated arithmetic mean voltage of the PSL reference particles for a given mobility 523 diameter and RPM. The DMA-APM system was calibrated with spherical PSL particles with a 524 known density of 1.05 g cm -3 .

526
The effective density of the HVOPM+NOx (soot particles) was fitted assuming a power law function,
(2) (57), where ′′ is a constant and Dfm the mass-mobility exponent. 528 = ′′ −3 (2) 529 The mass-mobility relationship was used to extrapolate a power law function for the mobility 530 equivalent particle diameters below 50 nm (up to the inherent material density of soot of 1.8 g cm -531 3 ) and above 300 nm. 532 533 Particle mass and surface area size distribution Mass size distributions were calculated by following Eq. (3), utilizing the particle number size 535 distribution (from the SMPS) and the experimentally determined effective density (ρeff, from the 536 APM) as a function of electrical mobility size (dp). 537 dM/dlogd p = πd p 3 6 * ρ eff (d p ) * dN/dlogd p (3) 538 Lognormal distributions were fitted to the mass size distribution up to 1 µm (PM1). 539 540 The surface area (SA) distributions were calculated using the model described by Rissler et al. (39),

541
which is based on DMA-APM measurements. From the DMA-APM, the mass of individual 542 agglomerates as a function of mobility particle size can be extracted if the effective density follows 543 the soot power law function (Eq. 2). The surface area of individual agglomerates is then calculated 544 by division of the mass of the agglomerate by the primary particle mass and surface area 545 (SApp=6/(ρpp*dpp))(39). The primary particle size (dpp) is obtained from TEM images, and the 546 inherent material density of soot (1.8 g cm -3 ) used for the primary particle density (ρpp). From the 547 surface area of individual agglomerates as a function of mobility particle size, the particle number 548 distribution (from the SMPS) can be converted to a particle surface area distribution. This method 549 accounts for the agglomerated soot structure and is described in more detail by Rissler et al. and NMAM 5040 diesel exhaust protocol. The limit of detection for EC (LOD) was 0.06 µg C cm -2 . Two 567 samples were collected in parallel, where one filter collected particle-free air after a Teflon filter 568 (Zefluor, pore size 1.0 µm) which was used to account for gas adsorption artifacts of the filter. Both 569 samples were collected after a PM1 cyclone at a flow rate of 5 L min -1 during the entire exposure 570 duration (180 min) and stored refrigerated (+6°C) until analysis. 571 To analyze the soot particle aggregate structure (morphology and primary particle size), samples 572 were collected with electrostatic precipitation using a nanometer aerosol sampler (model 3089, TSI) 573 on lacey carbon coated Cu-grids and analyzed with a transmission electron microscope (TEM, JEOL 574 3000F). The TEM was operated at 300kV and equipped with a Schottky FEG and 2x2k CCD. An 575 overview of the samples was first imaged at 10,000X magnification in order to ensure that the 576 sample was reasonably homogenous. The TEM images of HVOPM+NOx were analyzed for primary 577 particle size determination with the ImageJ software (58). The diameters of the clear primary 578 particles without overlap at the edges of the soot agglomerates were measured in TEM images with 579 a magnification minimum of 25,000X. The diameters of 81 primary particles were measured from 580 10 agglomerates. 581 582 PAH analysis 583 Samples for particulate PAH analysis were collected at a flow rate of 2 L min -1 during the entire 584 exposure duration (180 minutes) on Teflon filters (diameter 37 mm, pore size 2 µm (Teflo, Pall 585 Corporation, Port Washington, N.Y., U.S.A.). These filters were followed by XAD-2 tubes (SKC 586 Inc.) for sampling of gaseous PAHs. The samples were stored at -18°C prior to analysis. The 587 samples were analyzed for 33 native and alkylated PAHs (including the 16 U.S. EPA priority 588 PAHs), 17 nitrated, 10 oxygenated PAHs (nitro-PAHs and oxy-PAHs), and 6 dibenzothiophenes 589 (DBTs), as described by Gren et al. 2020 (22). In short, prior to extraction two labelled internal 590 standard mixtures containing 16 deuterated U.S. EPA priority PAHs were spiked to the filters and 591 XAD-2 adsorbent, respectively. Samples were extracted with 3 mL dichloromethane, cleaned using acetate, n-octane and n-nonane were also analyzed but not included in the total BTEX 610 concentration. The samples were collected on thermal desorption tubes (TENAX TA) and analyzed 611 by a thermal desorption GC-MS method. The sorbent tubes were heated to 250°C under a helium 612 flow for 5 minutes. The emitted compounds were refocused with a cold trap (-30°C) and then 613 quickly heated to 300°C in the thermal desorption instrument (Unity2 and Ultra, Markes) and 614 injected into the GC-MS (ThermoFisher Scientific). Target compounds were separated on a non-615 polar capillary column (TraceGold, TG-1MS, ThermoFisher Scientific) coupled to a mass 616 spectrometer (ISQ LT, ThermoFisher Scientific). 617 618 Model for particle deposition in the respiratory tract 619 Regional respiratory tract particle deposition fractions from nasal breathing were calculated for 620 the inhaled aerosols with the multiple-path particle dosimetry model (MPPD model version 3.04, 621 (59)). The input parameters are summarized in Table 9. 622 623 measures, thus the volunteers were further categorized into a "typical" and "atypical" group based 660 on their oscillometric measurements. The criteria were as follows: R5-19 ≥ 0.8 cmH2O s L -1 , X5 ≤ -661 1.8 cmH2O s L -1 , and Ax ≥ 14 cmH2O L -1 (Additional file A). These criteria were based on 662 characteristics of healthy and asthmatic subjects as published by Eddy et al. (41). The atypical 663 group showed a higher proportion of having a history of symptoms and atopy (80% vs. 54%, 664 p<0.05). They were hypothesized to be more sensitive and to have a different lung reaction than 665 the typical group, which was explored further in the analyses. 666 667 Self-rated symptoms 668 Similar to a previous study (6), self-rated symptoms of eye irritation, nose irritation (including 669 runny nose and nasal congestion), throat irritation and chest tightness/breathlessness were rated 670 by the volunteers themselves on a visual analog scale (VAS) (range 0 to 100 millimeter) before 671 exposure, and at 35, 95 and 155 min into the exposure during each exposure session (Table 7).

673
Statistical analysis For each self-rated symptom (eye, nose, throat, and chest), when a volunteer gave a higher score 675 than before the exposure at any time during the exposure, this person was recoded as "reported 676 symptoms", otherwise, "no reported symptoms". A person was then recoded as "reported any 677 symptom" if he/she reported any of the four symptoms during the exposure. The calculation was 678 performed for each exposure scenario separately. Descriptive analysis was used to count the 679 number of persons and corresponding proportion of persons with reported symptoms during 680 exposure for each exposure scenario. An 2 -test was used to investigate the difference in 681 proportion between the given exposure scenarios (each of the two exposure scenarios) in 682 comparison to FA exposure when applicable.

684
For PNIF and PEF measurements that were performed three times during each exposure, absolute 685 changes from before exposure were calculated on the individual level for each exposure scenario at 686 each time point. Linear mixed models were used to analyze the average changes in the selected 687 outcomes at given exposure scenarios versus changes at FA exposure. Subject ID, the exposure 688 scenarios and time points (1-4) were used to identify repeated measurements, and all models 689 included a random slope. The models further included exposure order (e.g., first or second time in 690 the chamber) as an adjustment since the order was imbalanced and a learning effect on the 691 measurement performance might have occurred. 692 For spirometry and FOT measurements, which were only performed before and after exposure, the 693 Wilcoxon signed-rank test was used to compare the differences between before and after exposure 694 at each exposure scenario.

696
Additionally, interaction terms between exposure scenarios and typical/atypical groups were 697 tested for PNIF, PEF and with the FOT in the linear mixed models described above, to see if the 698 atypical group with FOT measurements outside the normal range (Additional file A) showed 699 different exposure-related changes in nasal patency and pulmonary function. If the interaction 700